                         FEDERAL IMPLEMENTATION PLAN 
                          FOR REGIONAL HAZE FOR THE 
                         UNITED STATES VIRGIN ISLANDS
                                       

Table of Contents
1.0	Background and Overview	1-1
1.1	General Background of the Clean Air Act and the Federal Regional Haze Rule	1-1
1.2	Virgin Islands Class I Area	1-2
2.0	Visibility Monitoring in the Virgin Islands	2-3
2.1	IMPROVE Program Objectives	2-4
2.2	Monitoring Information and Strategy for the Virgin Islands National Park Class I Areas	2-5
3.0	Assesment of Baseline and Natural Conditions	3-7
3.1	Requirement, Data, and Methods Used	3-7
3.2	Virgin Island Baseline Visibility	3-9
3.3	Natural Visibility	3-12
3.4	Contributions to Visibility Impairment in the Virgin Islands	3-13
3.4.1	Area of Influence for the Virgin Islands National Park	3-13
3.4.2	Pollutants Contributing to Visibility Impairment (2000-2004 Baseline Data)	3-18
3.5	Class I Areas Affected by Emissions from the Virgin Islands	3-20
4.0	Emissions Inventory and Modeling Analysis	4-21
4.1	Sources of Emissions	4-22
4.2	Modeled Baseline Emissions	4-23
4.2.1	HOVENSA St. Croix	4-23
4.2.2	St. John Point Sources - Generators	4-24
4.2.3	St. John Marine Sources	4-24
4.2.4	St. John Marine Sources with Emissions Reductions	4-25
4.2.5	St. John Paved and Unpaved Road Fugitive Sources	4-25
4.2.6	Aggregate St. John Burning	4-25
4.2.7	St. John Residential Boilers	4-25
4.2.8	St. John Miscellaneous Industrial Processes -- Construction	4-26
4.2.9	St. John Motor Vehicle Emission Sources	4-26
4.2.10	Sample Puerto Rico Source	4-26
4.2.11	St. Thomas Sources	4-26
4.2.12	Other St. Croix Sources	4-27
4.2.13	Aggregate British Virgin Islands Sources	4-27
4.2.14	Simulated BVI Sources	4-27
4.2.15	St. John Off-road Sources	4-28
4.3	Emissions Inventory Characteristics	4-28
4.3.1	Sulfur Dioxide (SO2)	4-28
4.3.2	Nitrogen Oxides (NOx)	4-29
4.3.3	Particulate Matter (PM)	4-30
4.4	Modeling Results	4-32
4.5	Future Year Emission Control Inventories	4-38
5.0	Evaluation of Reasonable Control Measures	5-39
5.1.1	HOVENSA	5-39
5.1.2	St. John Point Sources -Generators	5-39
5.1.3	St. John Marine Sources	5-40
5.1.4	St. John Paved and Unpaved Road Fugitive Sources	5-40
5.1.5	Aggregate St. John Burning	5-40
5.1.6	St. John Residential Boilers	5-41
5.1.7	St. John Miscellaneous Industrial Processes -- Construction	5-41
5.1.8	St. John Motor Vehicle Emission Sources	5-41
5.1.9	Sample Puerto Rico Source	5-41
5.1.10	St. Thomas Sources	5-42
5.1.11	Other St. Croix Sources	5-42
5.1.12	Aggregate British Virgin Islands Combustion Sources	5-42
5.1.13	Simulated BVI Sources	5-42
5.1.14	St. John Off-Road Sources	5-43
6.0	Best Available Retrofit Technology	6-43
6.1	The BART Rule	6-43
6.2	Potential BART Sources in the Virgin Islands	6-44
6.2.1	Cap-Outs and Shutdowns	6-46
6.3	Sources Subject to BART	6-46
6.4	Anticipated Visibility Improvement as a result of BART	6-47
6.4.1	Reasonably Attributable Visibility Impairment	6-48
6.5	Determination of BART Requirements for Identified BART-Eligible Sources and Analysis of Best System for Each Source	6-48
6.5.1	Five Factor Analysis for Each BART Source	6-48
6.5.2	BART Determinations	6-62
6.5.3	BART Determinations including Other Types of Standards	6-63
6.6	Description of BART Alternative for Any Source	6-63
6.7	Schedule for BART Implementation	6-64
7.0	Long-Term Strategy	7-64
7.1	Overview of the Long-Term Strategy Development Process	7-65
7.2	Emission Reductions Due to Ongoing Air Pollution Programs	7-65
7.2.1	Other Point Source Controls Expected by 2018 Due to Ongoing Air Pollution Control Programs	7-66
7.2.2	Controls on Non-road Sources Expected by 2018 due to Ongoing Air Pollution Control Programs	7-67
7.2.3	Mobile Source Controls Expected by 2018 due to Ongoing Air Pollution Control Programs	7-67
7.2.4	Source Retirement and Replacement Schedules	7-68
7.3	Additional Reasonable Strategies	7-69
7.3.1	Analysis of the Four Statutory Factors	7-69
7.3.2	Best Available Retrofit Technology	7-70
7.3.3	Changes to Emissions by 2018	7-70
7.4	Additional Measures Considered	7-72
7.4.1	Caribbean Emissions Control Area	7-72
7.4.2	Measures to Mitigate the Impacts of Construction Activities	7-73
7.4.3	Agricultural and Forestry Smoke Management	7-73
7.5	Enforceability of Emission Limitations and Control Measures	7-73
7.6	Consultation on the Long-Term Strategy	7-74
8.0	Reasonable Progress Goals	8-74
8.1	Calculation of Uniform Rate of Progress	8-75
8.2	Reasonable Progress Goals for the Class I Area in the State	8-76
8.2.1	Consideration of Other Air Quality Requirements	8-77
8.2.2	Additional Reasonable Controls within Virgin Islands	8-78
8.2.3	Visibility Impacts of Additional Reasonable Controls	8-78
8.2.4	Impacts of BART Controls in the Virgin Islands	8-78
8.2.5	Impacts of other Federal Controls in the Virgin Islands	8-78
8.2.6	Results of Best and Final Modeling	8-79
8.3	Demonstration that the Goals are based on Reasonable Controls	8-80
9.0	Consultation and Coordination	9-81
9.1	Regional Planning	9-81
9.2	State/Tribe and Federal Land Manager Coordination	9-82
10.0	Comprehensive Periodic Implementation Plan Revisions	10-83
11.0	Commitment to Determine the Adequacy of the Existing Plan	11-84
12.0	Summary Of The Federal Plan For Visibility In The National Park On St. John In The United States Virgin Islands	12-85


Tables
Table 3-1: IMPROVE Information for Virgin Islands and Selected Class I Areas	3-9
Table 3-2: Baseline Visibility for the Twenty Percent Worst Days and Twenty Percent Best Days for 5 Years (from 2000-2004) in Selected Class I Areas	3-11
Table 3-3: Summary of Baseline Visibility and Natural Conditions for the Twenty Percent Worst and Twenty Percent Best Visibility Days	3-13
Table 3-4: Summary of Trajectory Analyses for Top Four Days with the Highest Visibility Impairment for Each Species 2000 - 2004	3-15
Table 4-1: Summary of Emissions St. John, Virgin Islands	4-22
Table 4-2: Results of Modeling the Virgin Islands Emission Inventory's Impact at the IMPROVE site on St. John  -  All Emission Sources Combined.	4-35
Table 6-1: BART-eligible Sources in the Virgin Islands:	6-36
Table 6-2:  Individual BART-Eligible Source Visibility Impacts on Virgin Islands Class I Area	6-36
Table 6-3: Sulfur Dioxide Emissions Evaluations for Potential BART Sources	6-36
Table 6-4: Nitrogen Oxides Emissions Evaluation for Potential BART Sources	6-36
Table 6-5: PM10 Emissions Evaluation for Potential BART Sources	6-36
Table 7-1: Emissions from Point, Area and Mobile Sources on St. John, Virgin Islands (SO2 tpy)	7-36
Table 7-2: Emissions from Point, Area and Mobile Sources on St. John, Virgin Islands (NOx tpy)	7-36
Table 7-3: Emissions from Point, Area and Mobile Sources on St. John, Virgin Islands	7-36
Table 7-4: Emissions from HOVENSA (tons per year)	7-36
Table 8-1: Uniform Rate of Progress Calculation	8-36
Table 12-1: Virgin Islands National Park Visibility Improvement Due to the FIP	12-36

Figures
Figure 2-1: The IMPROVE Monitoring Site at the Virgin Islands National Park	2-6
Figure 2-2: Virgin Islands National Park IMPROVE Monitor Location	2-6
Figure 3-1: Wind Rose for St Thomas, Virgin Islands, based on 2007 - 2010 National Weather Service Data	3-14
Figure 3-2: Back Trajectories of Day with Highest Impact on Visibility from Fine Soils  -  notice the orgin for two layers of wind in the Saharan Desert.	3-17
Figure 3-3: Above: Percent Contributions to Total Light Extinction from all particles, Baseline Conditions (2004 data used as an example) Below: List of particles contributing to reduced visibility, color-coded, with possible sources listed.	3-18
Figure 3-4: 2004 Visibility for St. John IMPROVE Data  -  higher values are poorer visibility	3-19
Figure 3-5: Contribution of particulate to reductions in visibility on the days with the best visibility. (2004 data.)	3-20
Figure 4-1: Modeled SO2 Emissions on St. John Compared to Surrounding Areas.	4-29
Figure 4-2: Modeled NOx Emissions on St. John Compared to Surrounding Areas	4-30
Figure 4-3: Modeled PM2.5 Emissions at St. John and Surrounding Areas	4-32
Figure 4-4: Yellow Diamonds are Locations of Receptors for Modeling	4-34
Figure 4-5: Modeled Impacts at IMPROVE Monitoring Site on St. John by Emissions Grouped Type of Emission	4-36
Figure 8-1: Projected Progress Goals and Modeled Progress at Virgin Islands National Park	8-36

Appendices

Appendix A  -  Reserved for FLM Comments
Appendix B  -  Natural Background and Reasonable Progress Calculations
Appendix C1  -  Preliminary Contractor Report on Air Modeling for the Virgin Islands Regional Haze Analysis
Appendix C2  -  Draft Final Contractor Report on Air Modeling for the Virgin Islands Regional Haze Analysis
Appendix C3 - Final Contractor Report on Air Modeling for the Virgin Islands Regional Haze Analysis (Reserved)
Appendix D  -  BART Analysis
Appendix E  -  Contractor Report on St. John Emissions Inventory
Appendix F  -  Trajectory Analyses
Appendix G - December 20, 2006 Section 110 letter, Virgin Islands' lack of contribution to transported pollutants




Background and Overview
	This document is a Federal plan that identifies human-caused interference with visibility in the United States National Park on St. John in the Virgin Islands.  It describes the progress toward the national goal of improving visibility on days with poor visibility and preventing good visibility days from becoming worse. The United States Environmental Protection Agency (EPA) is working with the government of the Territory of the United States Virgin Islands to prepare this Federal Implementation Plan (FIP). This Plan will need to be revised and extended no later than 2018, and at any point this Federal Plan can become a State Implementation Plan of the Virgin Islands' Government.  
General Background of the Clean Air Act and the Federal Regional Haze Rule
In amendments to the Clean Air Act (CAA) in 1977, Congress added Section 169 (42 U.S.C. 7491) setting forth the following national visibility goal:
      Congress hereby declares as a national goal the prevention of any future, and the remedying of any existing, impairment of visibility in mandatory Class I Federal areas which impairment results from man-made air pollution.
Mandatory Class I areas, as defined in the CAA, included the Virgin Islands National Park, which is the only Class I area in the Caribbean Islands. 
From 1977 to present, modest steps were taken to address the visibility problems in Class I areas. In 1980, the Environmental Protection Agency implemented rules affecting individual sources of pollution (45 FR 80084). If a plume from a source affected the visibility in a Class I area, emissions from that source would be reduced to improve the view. New sources had to assess their impact on visibility and visibility reduction was one of many factors that determined the level of control for new sources of air pollution.
Congress amended the CAA in 1990, adding Section 169B (42 U.S.C. 7492), authorizing further research and regular assessments of the improvements in visibility. In 1993, the National Academy of Sciences concluded that "current scientific knowledge is adequate and control technologies are available for taking regulatory action to improve and protect visibility."
The CAA's requirements for improving visibility by reducing human-caused emissions that can impair visibility of scenic views were the basis for the EPA's Regional Haze Rule (RHR), which was adopted July 1, 1999, and went into effect on August 30, 1999. This rule seeks to address the combined visibility effects of various pollution sources over wide geographic regions. While better known for its establishing regional visibility organizations, this Rule also applies to regional sources of haze that affect the Class I area on St. John in the Virgin Islands. 
After several legal challenges to the RHR, the EPA finalized the Regional Haze Regulations.
The EPA is working with the government of the Virgin Islands to prepare this regional haze FIP. In various other regions of the continental United States, states have joined to work together on developing the information needed to support their haze plans. Since the Virgin Islands are separated from the continental United States by over a thousand miles, this plan was developed without the assistance of the multi-state planning organizations that the continental states organized. The EPA supported the work of these multi-state planning organizations; because of the impact these states had on air quality in neighboring states, much work needed to be done with extensive inventories and modeling. The negative aspect of having little interaction with the continental states, from an air pollution point of view, was that the Virgin Islands was not able to leverage its resources by combining them with those from other states to help lay the groundwork for a State Implementation Plan (SIP) for haze in the Virgin Islands. Thus, the EPA is taking the lead, working with the Virgin Islands' government, to prepare a FIP to address anthropogenic visibility impairment in the National Park on St. John.  
Virgin Islands Class I Area
The U.S. Virgin Islands contains one Class I area, the Virgin Islands National Park. The Friends of the Virgin Islands National Park website describes the beauty of the park, which covers more than half of the Island of St. John, plus Hassel Island (near St. Thomas). The park was established by law in 1956. In 1962, the enabling legislation was amended to add 5,650 acres of submerged land "...in order to preserve for the benefit of the public significant coral gardens, marine life, and seascapes in the vicinity there of..." In 1978, the legislation establishing Virgin Islands National Park was again amended to add Hassel Island, located in Charlotte Amalie harbor adjoining St. Thomas, to the Park.
Within its borders lie protected bays of crystal blue-green waters teeming with coral reef life, white sandy beaches shaded by seagrape trees, coconut palms, and tropical forests providing habitat for over 800 species of plants. The Park protects a rich cultural history, with relics from the Pre-Colombian Amerindian Civilization, remains of the Danish Colonial Sugar Plantations, and reminders of African Slavery and the Subsistence Culture that followed during the 100 years after Emancipation.
Scenic views in the Park showcase the green hillside, sparkling blue water and tropical blue skies. When visibility is reduced, the sky is a milky white and colors of landmarks are muted.  Most times, reduced visibility is caused by dust raised in the Sahara Desert and transported across the Atlantic Ocean via a deep layer of trade winds moving from east to west. This plan cannot reverse reduced visibility caused by Saharan Dust or volcanic ash from Montserrat.  The goal of the FIP is to evaluate and remedy the causes of reduced visibility due to human sources. These sources include sulfates and nitrates from combustion in refineries, power plants, boilers and ships. 
Many regional haze plans across the United States assess the impact of a state on Class I areas where their emissions cause or contribute to visibility reduction. However, emissions in other parts of the United States do not cause or contribute to visibility reductions in the Virgin Islands. In addition, there are no other Class I areas in the Caribbean.  The Virgin Islands are sufficiently distant from other Class I locations in the United States that the sources in the Virgin Islands do not affect other Class I areas, the closest of which is the Everglades National Park, 1,200 miles away in Florida. [See correspondence regarding Sec. 110a2d, December 20, 2006, from VIDPNR, in Appendix.]
 
Visibility Monitoring in the Virgin Islands
The Interagency Monitoring of Protected Visual Environments (IMPROVE) program was initiated in 1985 to establish current visibility conditions, track changes in visibility, and help determine the causes of visibility impairment in Class I Areas. The IMPROVE network collects particulate matter data at air monitoring sites in the Virgin Islands National Park and other Class I areas, which are used in calculations that quantify the obscuration of visibility from the captured particles. These monitors collect information on the quantity and chemical composition of the particles, because both of these qualities affect light scattering and visibility. 
Visibility conditions representative of those within the Virgin Islands National Park are monitored by the federally operated IMPROVE program. The IMPROVE monitoring program was established to measure visibility impairment in mandatory Class I areas throughout the United States. This monitoring is designed to aid the creation of Federal and State implementation plans for the protection of the visibility in Class I areas stipulated in the 1977 amendments to the CAA. Data from the IMPROVE monitoring program have been collected since 1990 at the Virgin Islands National Park.
The IMPROVE monitoring sites are operated and maintained through a formal cooperative relationship between the U.S. Environmental Protection Agency, the U.S. National Park Service, the U.S. Fish and Wildlife Service, the Bureau of Land Management, and the U.S. Forest Service. In 1991, several additional organizations joined the effort. These organizations include the National Association of Clean Air Agencies (formerly State and Territorial Air Pollution Program Administrators and the Association of Local Air Pollution Control Officials), Western States Air Resources Council, Mid-Atlantic Regional Air Management Association, and the Northeast States for Coordinated Air Use Management.
IMPROVE Program Objectives
Data collected at these sites are used by land managers, industry planners, scientists, public interest groups, and air quality regulators to understand and protect the visual air quality resource in Class I areas. Most importantly, the IMPROVE program scientifically documents for American citizens the visual air quality of their wilderness areas and national parks. Program objectives include:
         * Establish current visibility and aerosol conditions in mandatory Class I areas.
         * Identify chemical species and emission sources responsible for existing anthropogenic visibility impairment.
         * Document long-term trends for assessing progress towards the national visibility goals.
         * Provide regional haze monitoring representing all visibility-protected federal Class I areas where practical, as required by EPA's Regional Haze Rule.
Monitoring Information and Strategy for the Virgin Islands National Park Class I Areas
Section 51.308(d)(4) of EPA's Regional Haze Rule requires a monitoring strategy for measuring, characterizing, and reporting regional haze visibility impairment that is representative of all mandatory Class I areas within the Virgin Islands. The monitoring strategy for the Virgin Islands relies upon the continued availability of the IMPROVE network.
As part of the IMPROVE network, visual air quality in the Virgin Islands National Park has been monitored using an aerosol sampler (1990  -  present) and a nephelometer (1998  -  2005). The IMPROVE monitor for the Virgin Islands National park (indicated as VIIS1 in the IMPROVE monitoring network database) is located at the Biosphere Reserve Complex at Lind Point on the island of St. John. The monitor is at an elevation of 46 meters, a latitude of 18.3363°, and a longitude of -64.7962°. Figure 2-1 is a northward facing photograph of the monitoring site and Figures 2-2 and 2-3 show the location of the monitoring site on the island of St. John.
In addition to the regional haze monitor, a National Atmospheric Deposition/National Trends Network (NADP/NTN) wet deposition monitor and a Clean Air Status and Trends (CASTNet) dry deposition monitor have been operating at the Virgin Islands National Park since 1998. Ozone was continuously monitored from 1998 through 2003. Concentrations of ozone were well below past and present ozone air quality health standards.

                                       
Figure 2-1: The IMPROVE Monitoring Site at the Virgin Islands National Park
                                       
Figure 2-2: Virgin Islands National Park IMPROVE Monitor Location

Figure 2-3. Location of IMPROVE monitor in the Virgin Islands National Park on St. John. Blue markers are the grid from the air quality modeling.

Assesment of Baseline and Natural Conditions
Requirement, Data, and Methods Used
Under the CAA, the Regional Haze Plans must contain measures to make reasonable progress toward the goal of achieving natural visibility, which EPA's Regional Haze Rule notes is to be accomplished by 2064. Each State or Territory containing a Class I area must determine baseline and natural visibility conditions for their Class I area(s) in consultation with the following: 1) the FLMs, and 2) States identified as containing sources whose emissions contribute to visibility impairment in Class I areas. Comparing baseline conditions to natural visibility conditions determines the uniform rate of progress that must be considered as haze plans set reasonable progress goals for each Class I area. 
The Clean Air Act also stipulates that haze plans should achieve these visibility goals by improving the visibility on the worst days for visibility, while ensuring that visibility does not degrade on the days with the best visibility. To address this requirement, the average visibility impairment over the worst one-fifth (top twenty percent impaired days) and the best one-fifth (best twenty percent) days are calculated. 
Data from the Virgin Islands National Park IMPROVE monitor (VIIS1) are representative of the Class I Area in the Virgin Islands, as described in the Monitoring Section of this FIP (Section 2).  This haze plan shall use the IMPROVE data to calculate the baseline and natural conditions for the Virgin Islands National Park in accordance with 40 CFR 51.308(d)(2)(i).
In addition to data from the Virgin Islands National Park monitor, data from additional IMPROVE monitors (Table 3-1) are presented throughout this FIP to provide context for the Virgin Island data. Data from Brigantine Wilderness Area and Acadia National Park are presented as sites with humid climates that are considered clean areas, but affected by emissions across the eastern and midwestern United States.  Florida's Everglades National Park data show a park that has a tropical climate similar to the Virgin Islands; however, the Everglades are affected by more sources of pollution than the Virgin Islands. Arizona's Grand Canyon National Park is included because it is representative of a remote Class I area with a dry climate that is distanced from major urban areas and sources of pollution. While the Virgin Islands National Park is remote from major urban areas and has less industry nearby, like the Grand Canyon National Park, the Virgin Islands' more humid climate allows particulates, especially sulfates, to have a greater effect on reducing visibility than at the Grand Canyon. 
In September 2003, EPA issued guidance for the calculation of natural background and baseline visibility conditions. The guidance provided a default method and describes certain refinements that States could evaluate to tailor these estimates to a specific Class I area, if it was poorly represented by the default method. In 2006, the IMPROVE Steering Committee adopted an alternative reconstructed extinction equation to revise certain aspects of the default method. The revised aspects of the extinction equation were scientifically well understood, and the Committee determined that revisions improved the performance of the equation at reproducing observed visibility at Class I sites. 
The baseline and natural conditions reported herein were calculated using the alternative method approved by the IMPROVE Steering Committee in 2006 because these equations represent the most updated understanding of the science.
Table 3-1: IMPROVE Information for Virgin Islands and Selected Class I Areas
Class I Area
IMPROVE Site
Location 
(latitude and longitude)
State
Virgin Islands National Park
VIIS1
18.34N, 64.80W
US Virgin Islands
Acadia National Park 
ACAD1
44.38N, 68.26W
Maine 
Brigantine Wilderness Area
BRIG1
39.47N, 74.45W
New Jersey 
Everglades National Park
EVER1
25.39N, 80.68W
Florida
Hance Camp at Grand Canyon National Park
GRCA2
35.97N, 111.98W
Arizona
Source: VIEWS (http://vista.cira.colostate.edu/improve/), prepared on 10/14/11

Virgin Island Baseline Visibility
                          How visibility is measured 
                         or "What's a deciview?"
EPA uses a unit called a `deciview' to quantify visibility for the haze plans. The deciview scale varies in a similar way to how the human eye sees how visibility changes. For example, a change in visibility of one deciview unit is noticeable to a typical human observer. Higher deciviews values mean worst visibility and lower values mean clearer air.  

To calculate visibility in deciviews, we can use measurements of how far an observer can see (visual range) or calculate deciviews from extinction in megameters[-1]. Extinction is the amount of obscuration occurring because of particles that fill the air that make it harder to see distant objects. A formula coverts visual range or extinction to deciviews to account for how the eye sees changes in visibility.  















The IMPROVE program has calculated the baseline (2000-2004) and natural visibility conditions for the twenty percent best and worst visibility days at each IMPROVE monitoring site, including the Virgin Island National Park site. These values are posted on the Visibility Information Exchange Web System (VIEWS). The VIEWS site lists the worst visibility days as a 5-year average for 2000-2004 using the alternative IMPROVE algorithm approved in 2006 by the IMPROVE Steering Committee. The VIEWS site also provides the natural background visibility for the twenty percent worst and the twenty percent best visibility days. 
Table 3-2 presents the baseline visibility for the twenty percent best and worst visibility days for a select subset of Class I areas. 
The baseline visibility conditions in the Virgin Islands National Park are better than conditions in the eastern United States, as seen in the data in Table 3-2. However, even the best days in the Virgin Islands have more obscuration than most days in dry western locations, like the Grand Canyon National Park. As mentioned in Section 3.1, the difference in natural background visibility between the Virgin Islands National Park and the Grand Canyon National Park is largely due to differences in humidity. As humidity increases, the light scattering properties of the sulfate component of particulate matter is enhanced, which results in decreased visibility. Hence, sites in dry regions, like the western United States, have better visibility than sites in humid regions, like the southeastern United States and the Virgin Islands. The western United States also have the advantage of fewer sources of pollutants that affect visibility than the eastern United States, and lack of impact from transported Saharan Dust.

Table 3-2: Baseline Visibility for the Twenty Percent Worst Days and Twenty Percent Best Days for 5 Years (from 2000-2004) in Selected Class I Areas
Class I Area 
(IMPROVE Monitor)
Year
20 % Worst Days
Deciviews (dv)
20 % Best Days
Deciviews (dv)
Virgin Island National Park (VIIS1)
2000
19.57*
9.93*

2001
16.76
9.08

2002
16.39
8.31

2003
17.38
8.14

2004
17.56
8.61
5-Year Average
 
17.53
8.81
 
 
 
 
Acadia National Park (ACAD1)
2000
21.64
8.89

2001
23.28
8.87

2002
23.91
8.77

2003
23.65
8.77

2004
21.98
8.56
5-Year Average
 
22.89
8.77
 
 
 
 
Brigantine Wilderness Area (BRIG1)
2000
28.95
14.26

2001
28.38
13.82

2002
29.31
14.83

2003
29.79
14.39

2004
28.59
14.36
5-Year Average
 
29.01
14.33
 
 
 
 
Everglades National Park (EVER1)
2000
* *
* *

2001
20.93
11.45

2002
20.64
10.97

2003
24.44
12.22

2004
23.21
12.1
5-Year Average
 
22.31
11.69
 
 
 
 
Hance Camp at Grand Canyon National Park (GRCA2)
2000
11.12
2.89

2001
* *
* *

2002
11.62
1.98

2003
12.74
1.82

2004
11.18
1.98
5-Year Average
 
11.67
2.17
 
 
 
 
Values, in deciviews, are the average for each 20% group.
*For the Virgin Islands site, the year 2000 data have too many missing samples to be part of the valid multi-year average.  While 2000 is included in this table's average value, data from 2000 is not included in calculating the values used to determine the starting point and reasonable progress rate used later in this FIP.
*  * Data do not exist for this monitoring year. However, according to the EPA document entitled, Guidance for Tracking Progress Under the Regional Haze Rule, states that, "...a minimum of 3 years of data meeting these completeness requirements is sufficient to calculate the 5-year averages within each 5-year period...the 3-year completeness criterion allows for the calculation of baseline conditions at sites with less than 5 years of data." 
Source: VIEWS (http://vista.cira.colostate.edu/improve/Data/IMPROVE/summary_data.htm), prepared on 10/14/11.


Natural Visibility
Natural visibility is the visibility for each Class I area representative of the conditions before human activities affected air quality in the area. Certain natural phenomena can reduce visibility. A Clean Air Act goal is to remedy visibility impairment resulting from human activity. 
A 5-year average (2000 to 2004) visibility in deciviews was calculated for each selected Class I Area for the twenty percent best and twenty percent worst days in accordance with 40 CFR 51.308(d)(2) and detailed in the EPA document Guidance for Tracking Progress Under the Regional Haze Rule. The deciview visibilities for these worst and best days are based on calculations and data included in Appendix B of this FIP. 
For the Virgin Islands site, the year 2000 is missing some data and is not used in this calculation. However, four years of valid data are acceptable, according to EPA's Regional Haze Rule.
This calculated natural visibility does not include the impact of locally important natural obscurers of visibility. In the Virgin Islands, for example, Saharan Dust can noticeably reduce visibility. Less obvious sources are airborne sea salt, and marine life - phytoplankton that produce sulfates (Tombach, 2005).
During the May 28, 2008 meeting with the FLM, EPA Region 2 staff discussed the calculation of natural background conditions in the Virgin Islands National Park. At this meeting, EPA Region 2 representatives noted that the default calculation of natural background conditions did not include the large impact of non-anthropogenic impacts from Saharan Dust and biogenic emissions. The FLM agreed that the information presented (and described in this FIP) showed the large influence of Sahara Dust on visibility in the Virgin Islands National Park. However, the FLM noted that more work would need to be done to more precisely calculate the impact on natural sources on visibility in the Virgin Islands National Park. The FLM suggested that the current implementation plan focus on implementing reasonable controls on sources that affect visibility in the Virgin Islands National Park. The FLM further suggested that the issues relating to the impact of natural sources on visibility in the Virgin Islands National Park should be researched for possible revised natural background conditions in the 2018 comprehensive implementation plan revision. 
Table 3-3 displays the baseline visibility for the twenty percent worst and the twenty percent best visibility days based on the 5-year average for 2000-2004, natural visibility for the twenty percent worst and the twenty percent best visibility days, and the difference between baseline and natural visibility conditions for each selected Class I area.
Table 3-3: Summary of Baseline Visibility and Natural Conditions for the Twenty Percent Worst and Twenty Percent Best Visibility Days
Class I Area
2000-2004 
Baseline (dv)
Natural 
Conditions (dv)
Difference (dv)

Worst 
20 %
Best 
20%
Worst 
20 %
Best 
20%
Worst 
20 %
Best 
20%
Virgin Island National Park*
17.02
8.81
10.68
4.41
6.34
4.40
Acadia National Park
22.89
8.77
12.43
4.66
10.46
4.11
Brigantine Wilderness Area
29.01
14.33
12.24
5.51
16.77
8.82
Everglades National Park
22.31
11.69
12.15
5.22
10.16
6.47
Hance Camp at Grand Canyon National Park
11.67
2.17
7.04
0.31
4.63
1.86
* Using valid data from 2001- 2004.
Source: VIEWS (http://vista.cira.colostate.edu/improve/Data/IMPROVE/summary_data.htm), prepared on 10/14/11.

EPA Region 2 staff informed the FLM about progress on the haze plan for the Virgin Islands through the series of national conference calls sponsored by EPA's Office of Air Quality Planning and Standards.  These calls were held as often as twice a month.  EPA Region 2 set up a formal consultation meeting via conference call with representatives of the FLM on May 28, 2008 to discuss the calculation of baseline and natural background conditions in the Virgin Islands National Park. The FLM agreed with the calculation of the baseline conditions and the estimate of progress needed to reach natural conditions. EPA Region 2 met with the FLMs on May 2, 2012 via conference call to discuss the haze plan the EPA was likely to propose and to listen to their comments on the plan.

Contributions to Visibility Impairment in the Virgin Islands
Area of Influence for the Virgin Islands National Park
Many of the National Parks and other areas covered by haze plans are affected by sources of human pollution nearby and from hundreds of miles away from the parks. The Virgin Islands National Park is not affected to a large extent from anthropogenic sources outside the Virgin Islands. The Virgin Islands are upwind of Puerto Rico (see the wind rose for St. Thomas in Figure 3-1). Even the large refinery on the island of St. Croix, 40 miles south of St. John, is located crosswind from St. John. Trajectory analyses for the days with the worst visibility in the park are available in the docket that supports this Plan.  Results are summarized in Table 3-4. 

Figure 3-1: Wind Rose for St Thomas, Virgin Islands, based on 2007 - 2010 National Weather Service Data 


Table 3-4: Summary of Trajectory Analyses for Top Four Days with the Highest Visibility Impairment for Each Species 2000 - 2004
Species
Total Light Extinction (bext)
Source Region (Height of Trajectory in Meters Above Ground Level)


Local
Regional
Long-Distance
All Aerosols Combined
79.5
St. John (100, 500)
Anguilla (100)
NW Africa (500, 1000)

77.4
St. John (100, 500, 1000)
None
NW Africa (1000)

71.8
St. John (100, 500, 1000)
Anguila (500)
Central Africa (1000)

71.1
St. John (100, 500)
Anguila (100) 
St. Martin (500)
African Coast (1000)
Fine Soil
19.9
None
None
Central Africa (500, 1000)

17.9
St. John (100, 500, 1000)
None
NW Africa (1000)

17.3
None
Guadeloupe (100)
Central Africa (1000)

16.9
St. John (1000)
Guadeloupe (100, 500)
Central Africa (1000)
Sea Salt
24.2
St. John (100, 500, 1000)
None
None

21.8
St. John (100)
Anguila (100)
None

18.9
St. John (100, 500)
Anguila (100) 
St. Martin (500)
African Coast (1000)

17.9
St. John (100, 500, 1000)
None
None
Sulfate
42.0
St. John (100)
Puerto Rico (1000)
None

34.9
St. Thomas (100)
Puerto Rico (100, 500) Dominican Rep. (1000)
Florida (500)

26.8
None
Guadeloupe (100)
Central Africa (1000)

26.6
None
Montserrat (500)
Western U.S. (1000)
Nitrate
5.2
St. John (100, 500, 1000)
Anguila (500)
Barbuda (1000)
None

4.8
None
St. Martin (100)
Barbuda (500)
Canada (100, 500, 1000)

4.52
St. John (100, 500)
Anguila (100) 
St. Martin (500)
African Coast (1000)

4.52
Puerto Rico (500, 1000)
Dominica (100)
Guadeloupe (500)
None
Elemental Carbon
15.5
St. John (500, 1000)
BVI (100)
None
Central Africa (1000)
African Coast (500)

14.0
None
St. Martin (100)
Mideast U.S. (100)
Mexico, Florida (500, 1000)

13.3
None
St. Martin (100, 500)
None

10.3
None
BVI (100, 500)
Canada, N.E. U.S. (100)
Organic Carbon
18.4
None
St. Martin (100)
Mideast U.S. (100)
Mexico, Florida (500, 1000)

17.2
St. John (500, 1000)
BVI (100)
None
Central Africa (1000)
African Coast (500)

7.09
St. John (100, 500, 1000)
None
Central Africa (1000)

7.05
None
St. Martin (100, 500)
None
Coarse PM
34.4
None
None
Central Africa (500, 1000)

33.4
St. John (100, 500)
Anguilla (100)
NW Africa (500, 1000)

30.5
St. John (100, 500, 1000)
BVI (100, 500, 1000)
W Africa (100, 500, 1000)

28.8
St. John (500, 1000)
BVI (100, 500)
None
We use the trajectory analysis to find likely source areas for the visibility problems included the top four days with the greatest impact on visibility from each kind of particulate that contributes to reduced visibility. On the worst days, we would expect to find that the air came from areas that contributed to the reduced visibility on those days. We used the HYSPLIT model, which uses weather data from every three hours, converted to a grid for each hour of the day. The user specifies the starting point (in our case, the monitoring site on St. John), how long to run the model (2 weeks) and the height above ground at the monitoring site.  We used three heights above the monitoring site within the well-mixed tropical surface layer, since tropical air is well mixed over land at these heights and different heights can have different trajectories when wind direction changes with height. 
  Since the trade winds mostly blow from east to west across the Virgin Islands, it is not surprising that when the winds are traced backwards from the sampling site on the western end of St. John, the path that the air takes almost always passes over the island of St. John. Many trajectories pass over neighboring islands like Antigua and St. Martin. Some trajectories pass over the islands of the British Virgin Islands, located to the northeast of St. John. However, there is not any particular pattern where days with the highest amounts of light-scattering particles have trajectories that commonly pass through one or more of these likely sources of human air pollution. For example, only one of the days with high levels of haze that may be from combustion has trajectories that pass over St. Croix, home to a major industrial center. These trajectories show that even on days when visibility is reduced due to sulfates, nitrates, and carbon (likely suspects from sources of combustion), the air affecting the park usually does not pass over St. Croix, St. Thomas, or Puerto Rico. 
	More often, days with high concentrations of fine soil or coarse particulates in the atmosphere occur when air travels from Africa's Sahara Desert. For example, Figure 3-2 shows a set of backward trajectories originating at the Virgin Islands National park for 100, 500, and 1,000 meters above ground level for September 26, 2002, which is typical of days with high visibility impairment due to fine soils. This illustrates how a significant causes of reduced visibility in the Virgin Islands National Park is dust transported from the Sahara Desert, located over 5,000 miles away in Africa. The Virgin Islands can also be affected by other various natural sources, such as sulfate produced by plankton in the ocean and ash from the Montserrat Volcano, located over 200 miles away to the southeast. The effects of these natural sources on visibility can be very noticeable on some days, especially because there are not as many anthropogenic sources of pollution in the Islands in comparison with the more urban areas of the continental United States.


Figure 3-2: Back Trajectories of Day with Highest Impact on Visibility from Fine Soils  -  notice the orgin for two layers of wind in the Saharan Desert.

      More examples of trajectories are in the Appendix and the docket for this action has all of the data used for the trajectories.
      
Pollutants Contributing to Visibility Impairment (2000-2004 Baseline Data) 
Section 51.308(d)(3)(iv) of 40 CFR part 51 requires each State to identify all anthropogenic sources of visibility impairment considered by the State in developing its long-term strategy. EPA's Guidance for Setting Reasonable Progress Goals Under the Regional Haze Program (June, 2007) notes that this process begins with the identification of key pollutants and source categories that contribute to visibility impairment at the Class I area(s) affected by emissions from the State.
Figure 3-3 shows how much each of the types of particles on the filters from the IMPROVE site contribute to reducing visibility. This graph is the average impact of each component on extinction, calculated from the 2004. The largest contributor to haze in the Virgin Islands is coarse mass (cm_bext  -  the all particles larger than 2.5 microns) which accounts for 36 percent of total interference with visibility on the twenty percent haziest days at the Virgin Islands National Park. The next largest contributor is sea salt (seasalt_bext) at 20 percent; then sulfate (amnso4f_bext) at 19 percent; soils (soilf_bext) were the fourth largest contributor at 13 percent. 
                                       
Figure 3-3: Above: Percent Contributions to Total Light Extinction from all particles, Baseline Conditions (2004 data used as an example) Below: List of particles contributing to reduced visibility, color-coded, with possible sources listed.



There is nothing to be done about portion of light extinction attributable to sea salt, as it is an entirely from sea spray generated by wave action and winds. 
As shown in the graph, below, the days with the highest contributions to reduced visibility have the highest amounts of coarse particulates and fine soil, which indicate the presence of Saharan Dust (Data from 2004, plotted from the VIEWS data generator web site.) 

Figure 3-4:2004 Visibility for St. John IMPROVE Data  -  higher values are poorer visibility
As noted by Tombach and Brewer (2005), African dust transported from the Saharan desert can represent more than 40 percent of the total soil emissions throughout the southeastern United States and an even larger component for the Virgin Islands. While impacts from Saharan dust are episodic, impacts can be seen 9 months out of the year in the Virgin Islands and can easily represent more than 3 ug/m[3] of the total fine particulate mass (Perry et al., 1997). Looking at the mass of particles in the fine particulate fraction, soils make up 5.90 ug/m[3] of the total mass of PM2.5 on the twenty percent worst visibility days during the baseline period, which is more than half of the total fine mass. Part of the soils component could be attributed to tilling or other disruptions of the ground that would allow dust to escape into the air and blown by the wind. Motor vehicles or construction are two other sources of disruption resulting in dust lofted into the winds. For this reason, EPA commissioned a search for emissions on the island of St. John, which will be discussed more in the section on emission inventories.
                                       
Figure 3-5: Contribution of particulate to reductions in visibility on the days with the best visibility. (2004 data.)
On the clearest days, with the best visibility, the small amount of interference with seeing the view is mostly from coarse particulates, followed closely by sulfates. The impact of fine soil, a sign of transported dust, is very small on these clean days.
 Since little can be done to control sea salt or Saharan dust emissions, sulfate emerges as the most controllable of the major constituents of haze-forming fine particle pollution in the Virgin Islands. As a result of this larger role of sulfate in the formation of regional haze in the Virgin Islands and sulfates' unique ability to scatter light more effectively in a humid atmosphere, control of sulfur emissions (if feasible and cost-effective) would have the largest effect on visibility in the Islands, per ton of emissions reduced. However, the effects of sulfur controls may be overwhelmed by the impact of natural sulfate and Saharan Dust.
Class I Areas Affected by Emissions from the Virgin Islands
There are no other Class I areas in the Caribbean islands, including Puerto Rico. The U.S. Virgin Islands are located more than 1,200 miles from the closest Class I area in the continental United States (Everglades National Park). As described in section 4, emissions from the Virgin Islands are much lower in comparison to most urban areas in the continental United States. If the east to west trade winds were to bring emissions from the Virgin Islands, any such emissions reaching the continental United States would be highly diluted after having traveled so far across the Atlantic Ocean. Therefore, it is reasonable to conclude that emissions from the Virgin Islands do not significantly affect other Class I areas in the United States. In fact, EPA and the government of the Territory of the Virgin Islands made that determination when deciding that the Islands did not need a long-range transport SIP under section 110 of the Clean Air Act.  
 

Emissions Inventory and Modeling Analysis
Section 51.308(d)(4)(v) of EPA's RHR requires a statewide emission inventory of pollutants that are reasonably anticipated to cause or contribute to visibility impairment in any mandatory Class I area. The pollutants that affect fine particle formation, and thus contribute to regional haze are sulfur oxides (SOx), nitrogen oxides (NOx), ammonia (NH3), volatile organic compounds (VOCs), and particles with an aerodynamic diameter less than or equal to 10 and 2.5 um (i.e., primary PM10 and PM2.5, respectively). Because emissions from the Virgin Islands in available databases (i.e., the National Emissions Inventory) are not as developed as those from the continental United States, an independent list of emission sources throughout the Virgin Islands and Puerto Rico was developed, as described below. The pollutants inventoried include sulfur dioxide (SO2), NOx, PM10 (modeled as PM2.5), PM2.5 and elemental carbon (EC).
EPA noted in its proposed rule on BART evaluations (69 FR 25184) that EPA determined in the NOx SIP Call notice (63 FR 57356) that VOCs do not contribute in a large amount to the formation of particulate matter in rural areas. EPA also noted in its final guidance on BART determinations (70 FR 39144) that EPA does not agree that rural VOC emissions can be completely ruled out as a significant contributor to particle formation and may play a significant role where NOx emissions are high. The complicated and rapidly evolving state of knowledge of particle formation by VOCs lead EPA to the conclusion that any determination of whether VOC emissions are precursors to PM2.5 formation should be completed on a case-by-case basis. In the case of the Virgin Islands, where development of an emission inventory for VOC is in its early stages, EPA will defer evaluation of the impact of VOC emissions' contribution to visibility reduction to the next round of visibility plans covering 2018 to 2028.
Sources of Emissions
Section 51.308(d)(3)(iv) of 40 CFR part 51 requires the Virgin Islands to identify all anthropogenic sources of visibility impairment in developing its FIP and long-term strategy.
Due to its proximity to the National Park, an emission inventory for the island of St. John, and of major sources nearby islands, is most important for this haze plan. For the reasons described above in Section 3.0, the emphasis in developing this FIP was placed on sources of sulfur emissions, SO2, which are precursors to sulfate. Emissions inventory analysis shows that point sources dominated the 2002 inventory of SO2 emissions. The largest source category of sulfur dioxide in the region is the Hovensa Refinery, followed by electric generating units (EGUs). These facilities are also the largest contributors to NOx emissions as well.
Additional SO2 source categories analyzed include marine sources and the mobile source sector. 

Table 4-1: Summary of Emissions St. John, Virgin Islands
by Pollutant in Tons per Year
Emissions on St. John
                                Sulfur Dioxide
                               Nitrogen Dioxide
                              Particulate Matter
                               Elemental Carbon
2002 Base Inventory
                                    156.73
                                    826.68
                                     83.88
                                     0.84
2018 Post-control inventory
                                     76.77
                                    572.09
                                     76.60
                                     0.77




The post-control inventory is the base inventory minus the emissions reductions anticipated for the ships that sail to and from St. John.

Typically, for most haze plans, emissions are summed for the entire state (in this case, all of the Virgin Islands).  But since a complete inventory is not available for St. Thomas and St. Croix, the FIP inventory for the base and post-control runs will include only the emissions on the island of St. John and ship emissions in the nearby waters. For modeling the impact of sources in and near the Virgin Islands, the largest emission sources are modeled from point sources on St. Thomas and St. Croix, as well as notable emissions from nearby islands not part of the United States and a sample major point source from Puerto Rico. The inventory is described in more detail in the upcoming sections.

Modeled Baseline Emissions
Section 51.308(d)(3)(iii) of 40 CFR part 51 requires the Virgin Islands haze plan to identify the baseline emissions information on which the long-term strategy is based.
The EPA provided lists of sources for the contractor to use in modeling regional haze for the Virgin Islands.  Sources of these lists include: the base 2002 EPA inventory developed using surrogate emissions for air toxics modeling, the emission inventory developed for EPA for sources on St. John (including nearby marine vessels), point source inventories from new source review applications, and EPA estimates of emissions from other nearby islands, as well as an example of a major source in Puerto Rico. In summary, these inventories include local emission sources on the island of St. John as well as other large sources in the islands with the potential to affect visibility at the Virgin Islands National Park. The modeled emissions are in the Appendix from the air modeling contractor and are summarized in Section 7.3.  
The sources were split into 15 groups, each of which is explained in more detail below. Details regarding the emission rate calculations for these sources can be found in the final report for the Virgin Islands National Park Regional Haze Modeling Analysis.
HOVENSA St. Croix
	Sources in this group were selected from Table 6-4 of the AEG Anguilla Project Report.  In the original modeling (prior to modeling BART-eligible sources), this group contained seven sources (six of the largest sources and one aggregate of the remaining sources), representing a total of approximately 75% of St. Croix's total emissions.  Since there were over 170 sources on St. Croix, the model was limited to modeling all the emissions combined into major emission points which were the electric utilities and the top 3 emitters of PM, SO2, or NOX.  St. Croix sources were split between Groups 1 and 6, with Group 1 representing only the HOVENSA refinery, and Group 6 representing Water and Power Authority (WAPA) St. Croix sources. St. Croix WAPA were modeled separately, as described in Section 4.2.12. The aggregated source consisted of the non-BART sources.  Emission rates for the aggregate source do not include any BART-eligible sources (i.e., no double counting of emissions).
	For this modeling effort, BART-eligible boilers, heaters, flares, and reciprocating gas compressors at HOVENSA were explicitly modeled.  The result was 65 BART-eligible point sources.  
	Emissions of the flares were provided as a sum for the entire facility, with potential to emit rates of 1200 tpy SO2, 1897 tpy NOX, and negligible PM emissions.  Ten flares were identified at HOVENSA in the AEG Anguilla inventory, with zero emissions for two of them.  Per email correspondence on May 11, 2012, only five flares are BART-eligible.  A proportion of the total emissions (5 BART flares / 8 total flares) was used to approximate the BART flare emissions.
St. John Point Sources - Generators
These sources represent actual point sources on St. John. This group contains six sources: Caneel Bay Resort, Gallows Point Resort, Grand Bay Resort, Myrah Keating Smith Community Health Center, Sirenusa, and Virgin Islands Water and Power Authority. The modeling included emissions from a back-up power generator at Caneel Bay Resort, two propane laundry dryers and a backup power generator at Gallows Bay Resort, a backup power Generator at the Myrah Keating Smith Clinic, and a backup power generator at the Virgin Islands Water and Power Authority. Fuel use data was collected from each facility and was used to estimate emissions.
St. John Marine Sources
This group represents the marine sources at St. John, including ferries, cruise ships, and cargo ships. Emissions are based on ship call data pulled from the Virgin Islands Water and Power Authority website. Emissions from all three source types were modeled as point sources located at the Cruz Bay docks. The stack parameters for the three different types of ships were selected based on typical cruise ship data specifications.
St. John Marine Sources with Emissions Reductions
This group represents an alternative set of emission rates for the marine sources at St. John. The source parameters are identical to Group 3, but emissions of particulate matter (PM) and SO2 have been reduced by 85 percent and 80 percent, respectively. These reductions are based on the agreement to implement the Caribbean ECA (Emissions Control Area) for ocean vessels. Beginning in 2015, fuel used by all vessels operating in these areas cannot exceed 0.1 percent fuel sulfur (1,000 ppm). This requirement is expected to reduce PM and SOx emissions by more than 85 percent. Beginning in 2016, new engines on vessels operating in these areas must use emission controls that achieve an 80 percent reduction in NOx emissions. 
 This group is not modeled simultaneously with Group 3, but is instead used to examine the benefits of reductions in emissions from marine sources.
St. John Paved and Unpaved Road Fugitive Sources
This group was modeled as a single area source representing fugitive dust emissions from roadways on St. John. The emissions were calculated based on total vehicle miles traveled (VMT) on St. John in 2002 based on data from the Federal Highway Administration and emission factors taken from Sections 13.2.1 and 13.2.2 of U.S. EPA's AP-42 emissions factor compilation. More recent studies indicate that emissions of fugitive dust kicked up from the roads by traffic are overestimated by a factor of five.  So the inventory used in modeling has reduced the emissions from this category by 80 percent from the AP-42 calculated values.
Aggregate St. John Burning
This group was modeled as a single area source and represents all sources of fires on St. John, including open burning, forest fires, and structure fires. Emissions were based on information acquired through a questionnaire sent to the Virgin Islands Fire Service in November 2009. The area is conservatively modeled as 8.6km by 3.4km rectangle (about 29.4 km[2]) that is contained mostly within the land mass of St. John.
St. John Residential Boilers
This group was modeled as a single area source and represents emissions from residential boiler use on St. John. Essentially, these emissions are from heating hot water for homes. Emissions were based on fuel usage data from Development of 2002 Regional Haze Area, Point, Non road Mobile and On road Mobile Source Inventories for the St. John, VI, Final Report June 2010.  The source was modeled as a 3 km x 3 km square region located at Cruz Bay settlement, St. John.
St. John Miscellaneous Industrial Processes -- Construction
This group was modeled as a single area source and represents emissions from other industrial sources on St. John. This group includes emissions from the Majestic Concrete facility and fugitive dust from residential construction. Emissions from Majestic Concrete were based on information acquired through a questionnaire sent to Majestic Concrete in February of 2010. Emissions from residential construction were based on the number of residential building permits issued on St. John in 2002. The area source was modeled as a 3 km x 3 km square region located at Cruz Bay settlement, St. John.
St. John Motor Vehicle Emission Sources
This group represents motor vehicle emissions on St. John. The emissions were based on highway mobile source activity data and emissions developed in support of the 2002 NEI. The area source was modeled as a 5 km x 10.4 km rectangular region, representing an area equal to that of the total land area of St. John.
Sample Puerto Rico Source
This group represents the AES Puerto Rico, L.P. 454 megawatt (MW) coal-fired power plant in Guayama, Puerto Rico. The stack parameters are based on a similarly sized coal-fired boiler. For modeling purposes, it was placed at the eastern end of Puerto Rico, on the shore facing the Virgin Islands.  This way, if the source had a negligible impact on St. John, it would confirm that sources on Puerto Rico have no readily discernible effect on reducing visibility in the Virgin Islands and controls in Puerto Rico would not be effective at improving visibility in the Class I area on St. John.
St. Thomas Sources
Emissions from this group are based on the Virgin Islands Water & Power Authority's 39 MW turbine that burns no. 2 distillate fuel oil. The stack parameters were based on an engineering estimate for similarly sized turbines.
Other St. Croix Sources
This group represents all St. Croix emissions except for HOVENSA. The sources are from Table 6-4 of the AEG Anguilla Power Renewable Energy Power Generation Project (AEG, 2010). The project details over 170 separate point sources. In order to reduce the computational demand on the model, seven actual sources were modeled and emissions for the remaining sources were modeled as a single point source having the same stack parameters and location as the largest emitter of PM10 on St. Croix -- the Hovensa wet scrubber. The group comprises seven sources that represent approximately 75 percent of St. Croix's total emissions. The St. Croix sources were split between groups 1 and 6, with group 1 representing the HOVENSA refinery.

Aggregate British Virgin Islands Sources
This group represents emissions from total oil usage on the British Virgin Islands (BVI). Oil usage data was taken from the Central Intelligence Agency (CIA) World Factbook. Since the Department of Energy (DOE) indicates that residual oil is not consumed in significant quantity in the BVIs, only a distillate emission factor was estimated. The distillate oil emission rates were calculated based on AP-42 emission factors assuming sulfur content of 0.5 percent. This group was modeled as a single area source representative of the island of Tortola, the closest of the BVIs to St. John. The source has dimensions of 2.5 km x 2.5 km and represents Road Town, the largest city on Tortola.
Simulated BVI Sources
Because of concerns of the residents of St. John that open burning on Tortula in the British Virgin Islands might be reducing visibility, we decided to model the effects of open burning on Tortula on visibility on St. John. This group represents general land clearing via open burning and was modeled as a single area source. The emission factors were taken from AP-42 Chapter 11.9, Table 11.9-1 for overburden. The modeled area was 100 m x 100 m and the location was selected based on sensible and possible locations of future land clearing.

St. John Off-road Sources
	These are emissions from the engines of off-road vehicles on St. John. The data are from a 2002 non-road mobile emission inventory developed by EPA Headquarters staff, using default factors. These emissions were modeled as an area source on St. John.

Emissions Inventory Characteristics
This section describes emission characteristics by pollutant and source type (e.g., point, area, and mobile) for the island of St. John in the U.S. Virgin Islands. The modeled emissions at St. John are also compared to the aggregate modeled emissions from surrounding areas within the modeling domain with the potential to affect regional haze at St. John (i.e., Puerto Rico, BVI, St. Thomas, and St. Croix). 
Note that St. John has a complete emission inventory, and the emissions from nearby areas are not complete, but it include the most significant sources.  For St. Thomas and St. Croix, the inventory includes point sources only. The emissions for Puerto Rico are from a typical power plant, so we could determine if a major point source on Puerto Rico had any effect on visibility on St. John.  The emissions from the British Virgin Islands - fuel combustion and open burning - are simulated as area sources.
Sulfur Dioxide (SO2)
Sulfur dioxide is the primary precursor pollutant for sulfate particles. Sulfate particles account for about 19 percent of the particle-related light extinction on the worst twenty percent days in the Virgin Islands. Hence, SO2 emissions are a reasonable choice for maximizing reductions in human-caused regional haze. 
Figure 4-1 shows that SO2 emissions are high on St. Croix, primarily due to emissions from the HOVENSA Refinery. Non-road sources, which consist primarily of marine vessel sources, dominate SO2 emissions on St. John. 

Figure 4-1: Modeled SO2 Emissions on St. John Compared to Surrounding Areas. 

Nitrogen Oxides (NOx)
NOx emissions contribute to visibility impairment by forming light scattering nitrate particles. Nitrate particles formed from combustion of sulfur in fuels account for about 14 percent of the visibility impairment on the worst twenty percent haze days in the Virgin Islands.
Figure 4-2 shows the percent contribution from different source categories to the modeled NOx emissions at St. John and the aggregate NOx emissions of the surrounding areas within the modeling domain. Similar to SO2, non-road sources dominate NOx emissions at St. John, mostly from marine vessel emissions. Mobile sources are the next largest contributor to the modeled NOx emissions from St. John, contributing approximately 7 percent of the total modeled emissions. The modeled emissions of NOx on St. Croix are higher than the other islands, due primarily to emissions from the Hovensa refinery.
                                       
Figure 4-2: Modeled NOx Emissions on St. John Compared to Surrounding Areas 
               Top graph: Total emissions modeled for each area.
Bottom graph: Percentage of four source categories  -  note that only St. John has a complete inventory of Point, Area, Non-road and On-road Mobile; the British Virgin Islands was modeled as an area source only and only point source emissions are in the inventory for Puerto Rico, St. Croix and St. Thomas.

Particulate Matter (PM)
Particles suspended in the ambient air are categorized in a variety of ways depending on their composition and genesis. Directly-emitted or "primary" particles are those that are emitted in non-gaseous form directly from the source. Secondary particles form in the atmosphere through chemical reactions involving precursor pollutants like SO2 and NOx.
A further distinction is made between particles with an aerodynamic diameter less than or equal to 10 micrometers and smaller particles with an aerodynamic diameter less than or equal to 2.5 micrometers (i.e., primary PM10 and PM2.5, respectively). For the purpose of the Virgin Islands regional haze modeling, all PM emissions were modeled as PM2.5. This may overestimate the impact of particles on haze, since PM2.5, that is, fine particulate less than 2.5 microns, reduces visibility more than particles larger than 2.5 microns.
Figure 4-3 shows that non-road sources dominate primary PM emissions at St. John, with area and mobile emissions both contributing between 7 and 8 percent of the total modeled PM emissions. Area sources dominate the PM emissions at BVI while only point sources were modeled from Puerto Rico, St. Croix, and St. Thomas. The largest source of PM emissions from St. John is reentrained dust from paved and unpaved roadways.
In addition to total PM2.5, elemental carbon particles were also modeled. For the purposes of the Virgin Islands regional haze modeling, emissions of elemental carbon were modeled as 10 percent of total PM emissions. EC is a PM species that is emitted directly into the air from various combustion activities such as fireplaces, cooking, forest fires, gasoline engines, and agricultural activities, but are especially prevalent in diesel exhaust and woodsmoke. EC particles typically reside in the smaller size modes and tend to reduce visibility by absorbing rather than scattering light.
                                       
                                       
Figure 4-3: Modeled PM2.5 Emissions at St. John and Surrounding Areas 
Top graph: Total emissions modeled for each area. Bottom graph: Percentage of four source categories  -  note that only St. John has a complete inventory of Point, Area, Non-road and On-road Mobile; the British Virgin Islands was modeled as an area source only and only point source emissions are in the inventory for Puerto Rico, St. Croix and St. Thomas.
Modeling Results
	An EPA contractor used these emission inventories and the CALPUFF model to predict the impact of all these sources together and each category of sources separately (see Appendix for the full report from the contractor). Meteorological data was used from 2007-2010 because those years had more weather stations, in order to provide better predictions.
	The BART Guidelines provide that we may use the CALPUFF modeling system or another appropriate model to predict the visibility impacts from a single source on a Class I area and to, therefore, determine whether an individual source is anticipated to cause or contribute to impairment of visibility in Class I areas, i.e., "is subject to BART." The Guidelines state that we find CALPUFF is the best regulatory modeling application currently available for predicting a single source's contribution to visibility impairment (70 FR 39162 (July 6, 2005)).  The BART Guidelines also recommend that a modeling protocol be developed for making individual source attributions, which in this case is the EPA-approved work plan developed by the contractor. 
	To determine whether each BART-eligible source has a significant impact on visibility, we propose to use the CALPUFF modeling to estimate daily visibility impacts above estimated natural conditions at the Class I area, which is the Virgin Islands National Park, covering much of St. John as well as Hassel Island near St. Thomas. There are no other Class I areas within 300 kilometers (km) of any BART-eligible facility in the Virgin Islands.  Emissions were modeled with four years worth of meteorological data, from 2007 through 2010. We used these years because more meteorological stations were available and the output provided from the modeling was closer to the actual monitored data than the period 2001 to 2004. The modeling evaluated the impact of three BART sources on the Class I area.  EPA believes that this modeling provides a reasonable estimate of daily visibility impacts above estimated natural conditions at the Class I area. Therefore, we propose to use the results of this CALPUFF modeling to determine whether each BART-eligible source has a significant impact on visibility. 
	The next figure is a map with St. John on the right side and St. Thomas on the left side showing receptor locations, that is, where the model predicted concentrations of pollutants.  Then the model converted these concentrations into visibility impairment estimates. These receptors are placed inside the National Park, which is the Class I area. Impact of pollutants on visibility varies across the Park. This FIP will describe the impacts of the sources on the receptor at the IMPROVE monitoring site in the National Park on St. John, as impacts on other locations in the Park do not change the conclusions of this Plan.


Figure 4-4 Yellow Diamonds are Locations of Receptors for Modeling 
Receptors for the CALPUFF modeling, covering the Class I area  -  Virgin Islands National Park  -  consisting of much of the island of St. John (on the right) and Hassel Island (in the harbor of the island of St. Thomas, on the left).  For this FIP, we used impacts at the IMPROVE monitor on St. John, and some analysis was done of the impacts on Hassel Island near St. Thomas.

Modeling impact results are in the following table. This is the impact of all sources together predicted for the location of the IMPROVE monitor on St. John.

Table 4-2: Results of Modeling the Virgin Islands Emission Inventory's Impact at the IMPROVE site on St. John  -  All Emission Sources Combined.
                      Modeling Results at IMPROVE Site 
                                 BEXT (Source)
                                  BEXT (BKG)
                                 BEXT (Total)
                                Percent CHANGE
                                   BEXT SO4
                                   BEXT NO3
                                    BEXT EC
                                   BEXT PMF
                                      dv 
                                    (total)
                                      dv 
                                     (bkg)
                                     ∆dv
                                   (source)
                                       
                                       
                                    Mm[-1]
                                    Mm[-1]
                                    Mm[-1]
                                       %
                                    Mm[-1]
                                    Mm[-1]
                                    Mm[-1]
                                    Mm[-1]
                                       
                                       
                                       
                                     2007
                                 Best 20% Days
                                     2.68
                                     15.64
                                     18.33
                                     17.16
                                     0.08
                                     0.00
                                     0.24
                                     2.36
                                     6.05
                                     4.47
                                     1.58
                                     2007 
                                Worst 20% Days
                                     17.54
                                     15.67
                                     33.23
                                    112.10
                                     3.09
                                     0.62
                                     1.26
                                     12.57
                                     11.86
                                     4.49
                                     7.37
                                     2008
                                 Best 20% Days
                                     2.80
                                     15.62
                                     18.42
                                     17.95
                                     0.05
                                     0.00
                                     0.25
                                     2.50
                                     6.11
                                     4.46
                                     1.65
                                     2008 
                                Worst 20% Days
                                     10.73
                                     15.71
                                     26.45
                                     68.36
                                     1.05
                                     0.21
                                     0.86
                                     8.61
                                     9.65
                                     4.52
                                     5.13
                                     2009
                                 Best 20% Days
                                     2.89
                                     15.62
                                     18.51
                                     18.49
                                     0.04
                                     0.00
                                     0.26
                                     2.59
                                     6.15
                                     4.46
                                     1.70
                                     2009 
                                Worst 20% Days
                                     12.53
                                     15.70
                                     28.24
                                     79.88
                                     1.77
                                     0.32
                                     0.95
                                     9.49
                                     10.19
                                     4.51
                                     5.68
                                     2010
                                 Best 20% Days
                                     2.95
                                     15.58
                                     18.53
                                     18.95
                                     0.01
                                     0.00
                                     0.27
                                     2.67
                                     6.17
                                     4.43
                                     1.73
                                     2010 
                                Worst 20% Days
                                     8.83
                                     15.70
                                     24.54
                                     56.32
                                     2.32
                                     0.51
                                     0.55
                                     5.46
                                     8.85
                                     4.51
                                     4.34
                             Average Best 20% Days
                                     2.83
                                       
                                     18.45
                                       
                                     0.05
                                     0.00
                                     0.26
                                     2.53
                                     6.12
                                       
                                     1.67
                            Average Worst  20% Days
                                     12.41
                                       
                                     28.12
                                       
                                     2.06
                                     0.42
                                     0.91
                                     9.03
                                     10.14
                                       
                                     5.63

Key to Modeling Results Table: BEXT is beta extinction; the amount of light removed along the light's path to an observer over a standard distance. The larger the number, the worse the visibility. In this case, light lost over per one million meters (giving a unit of Mm-1, usually called `inverse megameters')
dv is deciviews, a unit of visibility impairment where  a change of 1 dv is an amount of change people can see. 
See the box in section 3.2 for a comparison of the units used for visibility.
   * BEXT (Source)  -  Modeled beta extinction coefficient from the source's emissions
   * BEXT (BKG)  -  Background beta extinction coefficient (from a formula of annual background)
   * BEXT (Total)  -  beta extinction coefficient of modeled plus background visibility impairment
   * Percent Change  -  percent of source's impact compared to background (e.g., 100% means the impact is equal to the background visibility, that would double the visibility impairment over the estimated background
   * BEXT SO4  -  beta extinction coefficient of modeled impact of sulfate on visibility
   * BEXT NO3  -  beta extinction coefficient of modeling impact of nitrates on visibility
   * BEXT EC  -  beta extinction coefficient from IMPROVE monitor elemental carbon;
   * BEXT PMF  -  beta extinction coefficient from IMPROVE monitor fine particulate matter;
   * dv (total)  -  deciview haze index from modeled emissions + calculated background;
   * dv (bkg)  -  estimated background deciview haze index
   * ∆dv (source)  -  change in deciview haze index due to the impact of the individual source or source categories on visibility

   Modeling results are shown for the IMPROVE monitoring site on St. John. Data are also available for a receptor on Hassel Island, in the harbor on the southern end of St. Thomas. See the modeling report for details.

The following graph shows the relative impact of the modeled source categories on visibility at the IMPROVE monitor on St. John using meteorology from 2007 - 2010.

Figure 4-5: Modeled Impacts at IMPROVE Monitoring Site on St. John by Emissions Grouped Type of Emission
Most of the impact on visibility from human emissions is due to dirt lofted into the air by construction and from motor vehicles kicking up dust from paved and unpaved roads on St. John. Most of this dust is made up of coarse particles, which have an effect over relatively short distances, but being close to the monitoring site can have a large effect. The modeling does not include effects from natural sources, like Saharan Dust or sulfate from phytoplankton in the ocean.

Future Year Emission Control Inventories
Projection-year inventories typically consist of: control of or reduction of emissions from sources and, increases or decreases due to growth or contraction of the economy or of source emissions.
The future-year emission reductions for the U.S. Virgin Islands Haze Plan are from the reductions in HOVENSA due to the consent decree and reductions in emissions from marine vessels from implementation of the ECA. 
Emission reductions that we haven't quantified are likely to occur in the following areas:
         * replacement of old motor vehicles with newer, cleaner vehicles would decrease emissions; 
         * use of cleaner fuels in homes, businesses and vehicles would decrease emissions;
         * increases or decreases in population or tourism could increase or decrease emissions. 
These changes are more difficult to assess in the Virgin Islands than for most locations because of its remote location and lack of direct ties with other states. The Virgin Islands may experience a mix of vehicles further from the national averages and have less similar rates of vehicle turnover than other states. Federal regulations for new motor vehicles require new models of vehicles have reduced emissions, cleaner fuels and additional controls on off-road vehicles may be take longer to occur in the Virgin Islands than in the continental United States.  The fleet of vehicles in the Virgin Islands may not change as rapidly due to the isolated location and may change less predictably due to the small fleet of vehicles.
Thus, this plan does not attempt to quantify emissions changes due to growth or decreases due to default emission reductions that other plans use. For the 2018 future case, post control, emission inventory, this plan only adjusts emissions to account for control strategies specifically related to particular sources in the Virgin Islands.    
  
Evaluation of Reasonable Control Measures
HOVENSA
	The emissions from the HOVENSA refinery on St. Croix have a relativity small impact on visibility on St. John for a large refinery only 40 miles away. A facility that may have a noticeable effect on visibility with a predicted impact of 1.06 dv on visibility in the Class I area on St. John must have a review to determine if reasonable controls exist. For HOVENSA, at this time, EPA will accept that the reductions negotiated in the Consent Decree are `reasonable' controls for this facility. This facility is mostly shutdown at this time of the FIP, so developing controls beyond the consent decree is not reasonable at this time. If the facility is going to be reopened, before reactivating HOVENSA must prepare and submit to EPA for approval an analysis of reasonable emission reductions following EPA's guidelines for reasonable reductions for Regional Haze.
      .  
St. John Point Sources -Generators
These sources are power generators, most for backup electric power. The impact of these sources, combined, of 0.30 dv, is low enough not to need a full review for reasonable controls. Most of these sources do not operate all the time. The only full-time source in this category is a laundry drier that runs on propane, which has relatively low emissions, and practically no emissions of sulfur, so there is no sulfate produced and not much contribution to visibility obstruction.  

St. John Marine Sources
These sources include ferries, cruise ships, and cargo ships that dock at St. John. The Caribbean ECA (Emissions Control Area) emission reductions for ocean vessels are included in the future case modeling. Beginning in 2015, fuel used by all vessels operating in these areas cannot exceed 0.1 percent fuel sulfur (1,000 ppm). This requirement is expected to reduce PM and SOx emissions by more than 85 percent. Beginning in 2016, new engines on vessels operating in these areas must use emission controls that achieve an 80 percent reduction in NOx emissions. While these reductions are not enforceable as part of this FIP, EPA expects them to occur and they will be included in the reductions expected in the period through 2018.

St. John Paved and Unpaved Road Fugitive Sources
Fugitive dust emissions from roadways on St. John contribute to visibility reduction via the coarse particulate fraction. Paving dirt roads in the National Park is not reasonable, as it would detract from the rustic, primitive nature of the Park. Implementation plans often include street sweeping to reduce reentrained dust from roadways. On St. John, based on past experience, the road dust is mostly dirt washed onto the roads in heavy, tropical downpours.  These downpours erode the steep terrain of the island onto the roads and require more than street sweeping and pavement wetting to reduce reentraining this dirt. Thus, this source is not reasonable for control measures at this time. EPA will discuss with VIDPNR possibilities for reducing erosion from hillsides, which would also reduce silt washed into the ocean that may affect the growth of fragile coral reefs.  

Aggregate St. John Burning
This group covers all sources of fires on St. John, including open burning, forest fires, and structure fires. This group has a small effect on modeled visibility, less than half a deciview.  EPA will review with VIDPNR existing rules for containment of open burning on St. John, but we expect that reductions in emissions due to better control of open burning will not be significant enough to noticeably improve visibility.    

St. John Residential Boilers
Emissions in this group are from residential heating, mostly for hot water, on St. John. Their contribution to modeled haze of 0.01 dv is too small to consider for reasonable controls to reduce haze.  

St. John Miscellaneous Industrial Processes -- Construction
This group's emissions are particulate emissions on St. John from home construction and a concrete facility and are predicted by modeling to cause a noticeable reduction in visibility. The major controllable source, a concrete facility, has less than four percent of the emissions in this category and so controls on concrete production would not improve visibility substantially. EPA will review with the VIDPNR its rules for suppression of particulate emissions during construction and determine if the existing rules are sufficient or can be modified or if enforcement can be increased to reduce emissions from construction on St. John. 
St. John Motor Vehicle Emission Sources
These sources are motor vehicle emissions on St. John, with a 0.08 dv impact on visibility on St. John. While cleaner vehicles in the future will reduce mobile source emissions in all locations, St. John's fleet of vehicles is small, the makeup of the present fleet may be different from United States national defaults, the rate of vehicle turnover is affected by the remote location with no connecting roads to the outside world, and the difficulty of forecasting a future rate of increase or decrease in use of motor vehicles on St. John. This all adds up to making the uncertainties in projecting future emissions likely to be almost as large as the present estimated emissions themselves. Reductions are likely as new vehicles are brought to the island, but the small impact of motor vehicle emissions on visibility make any future change small as well.
 
Sample Puerto Rico Source
We modeled a major power plant on Puerto Rico as if it was on the east coast of Puerto Rico, nearer to the Virgin Islands, to see if any major source on Puerto Rico could have an impact on visibility on St. John.  Modeling predicts a lack of impact of this major source on St. John.  This is likely to be due to Puerto Rico being downwind of St. John almost all the time. Thus, emission reductions from Puerto Rico will not have a quantifiable effect on visibility on St. John.

St. Thomas Sources
The emissions from the Virgin Islands Water & Power Authority on St. Thomas are downwind of St. John, accounting for VIWAPA's small (0.04 dv) impact on visibility on St. John. Controls would result in small, perhaps too small to quantify, reductions.

Other St. Croix Sources
Emissions from other sources on St. Croix other than HOVENSA have less than a half a deciview impact on visibility on St. John. With 170 separate point sources, reductions would have to occur at many different locations to quantify even a small improvement in visibility on St. John, so controls on these sources are not reasonable.

Aggregate British Virgin Islands Combustion Sources
The impact of oil usage in the British Virgin Islands amounts to only 0.36 dv on St. John. Even if the impact of these emissions were larger, emission reductions would need to be negotiated with the British Government. International controls are beyond the scope of this FIP.  

Simulated BVI Sources
	Because of concerns of the residents of St. John that open burning on Tortula in the British Virgin Islands might be reducing visibility, we decided to simulate the effects of open burning of Tortula on visibility on St. John. This group represents a hypothetical source of emissions from general land clearing via open burning. The impact was too small to contribute to visibility impairment on St. John.

St. John Off-Road Sources
	These are direct emissions from off-road vehicles on the island of St. John. With a total predicted impact of 0.18 dv, these emissions are not contributing to noticeable impairment of visibility.

	In summary, only emissions from HOVENSA are large enough and have possibilities for reasonable control.  EPA has determined that the existing consent decree is, by its nature, a starting point for reasonable controls for this facility. If the facility reopens, EPA requires that HOVENSA perform a complete evaluation to determine what levels of control are reasonable, by evaluating all sources at HOVENSA. A useful methodology would be EPA's multi-factor analysis as described in EPA's Regional Haze Rule.

Best Available Retrofit Technology
As required by 40 CFR §51.308(e), the plan evaluates sources within the Virgin Islands to determine if they are BART-eligible and if they may reasonably be anticipated to cause or contribute to any impairment of visibility in any mandatory Class I Federal area. The BART evaluation determines if controls should be implemented on any sources subject to BART within in the Virgin Islands.  
BART is an element of EPA's LTS, as well as a requirement to evaluate controls for older sources that affect Class I areas, for the first implementation period.  The BART regional haze requirement consists of three steps:  (a) identification of all the BART-eligible sources; (b) an assessment of whether the BART-eligible sources are subject to BART; and (c) the determination of the BART controls. 

The BART Rule
The BART requirements pertain to large facilities in each of 26 source categories that meet certain criteria, including industrial boilers, paper and pulp plants, cement kilns, and other large stationary sources. The BART program applies to units installed and operated between 1962 and 1977 with the potential to emit more than 250 tons per year of a visibility impairing pollutant. Each potential BART unit must be modeled to determine its impact on the Class I area.  If a unit is determined to be BART-eligible, the unit must undergo a case-by-case analysis to determine if new emission limits are appropriate to limit its impact on Class I areas. The BART requirements are intended to reduce emissions specifically from large sources that, due to age, were exempted from new source performance standards (NSPS) requirements of the Clean Air Act. In June 2005, EPA adopted the final BART rule. 
To meet the BART rule, the EPA has developed an inventory of sources within the Territory of the Virgin Islands that could be eligible for controls. The rule also includes information on:
         Methods to determine if a source is "reasonably anticipated to cause or contribute to haze"
         Methodology for conducting BART control analysis
         Presumptive limits for electricity generating units (EGUs) larger than 750 Megawatts
Beyond the specific elements listed above, the final BART rule provides a great deal of flexibility in implementing the BART program.
The following sections provide information on how this plan meets the core requirements of the BART regulations.
Potential BART Sources in the Virgin Islands
The first component of a BART evaluation is to identify all the BART eligible sources within the United States Virgin Islands ("Virgin Islands" or "Territory").  While the Virgin Island's Department of Planning and Natural Resources (VIDPNR), the Territory's environmental agency, did not submit a SIP, EPA's evaluation process of identifying BART-eligible sources included a review of Title V permits, a review of Title V applications received from VIDPNR, and direct communications with HOVENSA, LLC, one of the BART-eligible sources.  To establish which facilities are BART-eligible, EPA evaluated eligibility criteria for combustion and other process units at the following eight sources throughout the Territory: 
   * HOVENSA, LLC (St. Croix)
   * Three of the Virgin Islands Water and Power Authority (VI WAPA) facilities  -  one on each of the islands (St. Croix, St. Thomas and St. John)
   * St. Croix Renaissance Group, LLLP (St. Croix)
   * Wyndham Sugar Bay Beach Club & Resort (St. Thomas)
   * Divi Carina Bay Hotel (St. Croix)
   * Buccaneer Hotel (St. Croix)
A BART-eligible source is one that meets all of the following criteria.  These criteria are in section 169A(b)(2)(A) of the Act, codified in 40 CFR Part 51, Appendix Y:
   * One or more emissions units at the facility are within one of the 26 categories listed in the BART Guidelines (70 FR 39104, 39158-39159 (July 6, 2005));
   * The emission unit(s) began operation after August 6, 1962 and were still in existence on August 7, 1977
      
   * Potential emissions of SO2, NOx, and PM10 from subject units are 250 tons or more per year. 
Table 6-1: BART-eligible Sources in the Virgin Islands:
Facilities 
Units
BART Source
Category 
Location 
Summary of Evaluation
VI WAPA 
2 boilers and 
2 combustion turbines
Fossil fuel-fired steam electric plant of  > 250 mm BTU/hr 
St. Thomas
Impact on visibility is very small: not subject to BART  
VI WAPA
2 boilers
Fossil fuel-fired steam electric plant of  > 250 mm BTU/hr
St. Croix
Impact on visibility is very small: not subject to BART  
HOVENSA
8 boilers
Petroleum 
Refinery
St. Croix
Sulfur emissions control not cost effective. Nitrogen emissions control cost effective, but small impact on visibility: subject to BART analysis

9 combustion turbines




64 process heaters




11 reciprocating gas compressors




1 tail gas treatment unit




3 tail gas incinerators




5 flares




water intake pumps and desalination water pump



      Summary of EPA's Evaluation
      
 EPA identified three of the eight sources, including multiple combustion or process units at each source, as BART-eligible.  The three BART-eligible sources identified by EPA as potentially impacting the Class I area, are summarized in Table 6-1.  None of the remaining five sources met these criteria and therefore were removed from consideration for BART review.  EPA evaluated the three BART-eligible sources to determine if BART controls should be required for each of the BART-eligible stacks from these sources.  Table 6-1 includes a summary of the evaluation, which is described in more detail in the following subsections.



Cap-Outs and Shutdowns
Some BART-eligible facilities with relatively small emission but with potential emissions that exceeded the statutory threshold of 250 tons per year or more may be exempted from BART requirements if the actual emissions of visibility impairing pollutants are well under 250 tons in any year. Such facilities may have accepted a permit limitation, restricting their emissions by law to less than 250 tons per year. There are no sources in the Virgin Islands that qualified for this option. 
In January 2012, HOVENSA, LLC announced it would be shutting down its refinery located on the Virgin Islands. Until the facility revises its Title V permits, the emission inventories for the base and future case, and for BART analyses, will still consider the emissions from a fully operating HOVENSA refinery, complying with its consent decree, and full emissions will be the basis for any BART analysis that may be required. 
Sources Subject to BART
According to Section III of the 2005 Regional Haze Rule, once the state has compiled its list of BART-eligible sources, it needs to determine whether to make BART determinations for all of the sources or to consider exempting some of them from BART because they may not reasonably be anticipated to cause or contribute to any visibility impairment in a Class I area.
The BART Guidelines recommend addressing SO2, NOx, and PM10 as visibility-impairment pollutants.  The Guidelines note that states can decide whether to evaluate VOC or ammonia emissions.  EPA is not developing additional strategies for VOC or ammonia emissions in its FIP.  EPA proposes to determine that the lack of tools available to estimate emissions and subsequently model VOC and ammonia effects on visibility inhibits EPA from addressing BART for these.  
Furthermore, EPA noted in its proposed rule on BART evaluations (69 FR 25184) that EPA determined in the NOx SIP Call notice for the eastern continental United States (63 FR 57356) that VOCs do not contribute in a large amount to the formation of particulate matter in rural areas, as cited by EPA in its proposed BART guidance.  While EPA noted in its final guidance on BART determinations ( 70 FR 39144) that EPA does not agree that rural VOC emissions can be ruled out as a significant contributor to particle formation, and may play a significant role where NOx emissions are high, the complicated and rapidly evolving state of knowledge of particle formation by VOCs leads EPA to the conclusion that determination of the approach to follow in considering whether VOC emissions are precursors to PM2.5 formation is a case-by-case approach.  In the case of the Virgin Islands, where development of an emission inventory for VOC is in its early stages, EPA will defer evaluation of the impact of VOC emissions' contribution to visibility reduction to the next round of visibility plans covering 2018 to 2028.
Anticipated Visibility Improvement as a result of BART
The EPA conducted a modeling analysis of BART-eligible sources using CALPUFF in order to assess the degree of visibility improvement which could result from installation of BART controls. The EPA considered the results of this analysis in its determination of BART for individual sources.
Because there are only three BART-eligible sources in the Virgin Islands, which impact only one Class I area, so EPA evaluated each source by itself.   As shown in Table 6-2, EPA will exempt two of the three BART-eligible sources in the Virgin Islands from further review under the BART requirements, since their impacts are very small.
This leaves HOVENSA as having the only BART-eligible stacks in the Virgin Islands. 
Table 6-2:  Individual BART-Eligible Source Visibility Impacts on Virgin Islands Class I Area
Facility and Location
Class I Area
And Locations of Modeling Receptor
Average 4-year
98[th] Percentile
Visibility Impact, (deciviews)
Subject to BART Analysis?

VI WAPA
St. Thomas
St. John
Hassel Island,
St Thomas
0.06
0.04
No
VI WAPA
St. Croix
St. John
Hassel Island,
St Thomas
0.09
0.10
No
HOVENSA
St. Croix
St. John
Hassel Island,
St. Thomas
1.88
2.25
Yes

Reasonably Attributable Visibility Impairment
Section 31.302 (c) provides for general plan requirement in cases where the affected FLM has notified the State that Reasonably Attributable Visibility Impairment (RAVI) exists in a Class I Area in the state. There are no RAVI sources in the Virgin Islands. The FIP incorporates the provisions of 40 CFR 52.26, 52.29, to address RAVI in the Virgin Islands.

Determination of BART Requirements for Identified BART-Eligible Sources and Analysis of Best System for Each Source
Five Factor Analysis for Each BART Source
The BART cost analysis for each BART-eligible source is included in the EPA contractor's report in an Appendix.  The BART analysis includes a modeling analysis to include consideration of the degree of improvement in visibility (included in an appendix), cost of controls, energy and non-air quality environmental impacts, existing controls at the source, and the remaining useful life of the source. The BART-eligible sources in the Virgin Islands are shown in Tables 6-2 through 6-4 for each visibility impairing pollutant.  These Tables include the contractor's evaluation of possible BART controls. BART is the emission limit for each pollutant based on the EPA's evaluation of degree of reduction achievable through the application of the best system of continuous emission reduction, taking into consideration the technology available, the costs of compliance, the energy and the non-air quality environmental impacts of compliance, any pollution control equipment in use or in existence at the source, the remaining useful life of the source, and the degree of improvement in visibility which may reasonably be anticipated to result from the use of such technology. 
EPA determined BART for each BART-eligible source using the methodology in the Guidelines for Best Available Control Retrofit Technology (BART) Determinations under the Regional Haze Rules, 40 CFR Part 51, Appendix Y. 
	There are no BART-eligible sources on the island of St. John, where the National Park is located.
Table 6-3: Sulfur Dioxide Emissions Evaluations for Potential BART Sources
                                Source and Unit
                                     ID #
                                   Location
                           Baseline Emissions (tpy)
                   Emissions at Maximum Utilization Capacity
                           BART Level of Control (%)
                        Emissions After Controls (tpy)
                          Emissions Reductions (tpy)
                                Type of Control
                               Hovensa Boiler 3
                                    B-1153
                                   St Croix
                                     330.1
                                     330.1
                                     55.0
                                     148.5
                                     181.5
                                  DIFF, SNCR
                               Hovensa Boiler 5
                                    B-1155
                                   St Croix
                                     484.9
                                     484.9
                                     55.0
                                     218.2
                                     266.7
                                  DIFF, SNCR
                               Hovensa Boiler 6
                                    B-3301
                                   St Croix
                                     330.8
                                     330.8
                                     55.0
                                     148.9
                                     182.0
                                  DIFF, SNCR
                               Hovensa Boiler 7
                                    B-3302
                                   St Croix
                                     330.8
                                     330.8
                                     55.0
                                     148.9
                                     182.0
                                  DIFF, SNCR
                              Hovensa Turbine GT8
                                    G-3408
                                   St Croix
                                     167.6
                                     167.6
                                     55.0
                                     75.4
                                     92.2
                                   DIFF, SCR
                               Hovensa Boiler 1
                                    B-1151
                                   St Croix
                                     330.1
                                     330.1
                                     55.0
                                     148.5
                                     181.5
                                  DIFF, SNCR
                               Hovensa Boiler 4
                                    B-1154
                                   St Croix
                                     322.5
                                     322.5
                                     55.0
                                     145.1
                                     177.4
                                  DIFF, SNCR
                               Hovensa Boiler 8
                                    B-3303
                                   St Croix
                                     640.1
                                     640.1
                                     55.0
                                     288.0
                                     352.1
                                  DIFF, SNCR
                               Hovensa Boiler 9
                                    B-3304
                                   St Croix
                                     640.1
                                     640.1
                                     55.0
                                     288.0
                                     352.1
                                  DIFF, SNCR
                              Hovensa Turbine GT1
                                    G-1101E
                                   St Croix
                                     135.5
                                     135.5
                                     55.0
                                     61.0
                                     74.5
                                   DIFF, SCR
                              Hovensa Turbine GT2
                                    G-1101F
                                   St Croix
                                     135.5
                                     135.5
                                     55.0
                                     61.0
                                     74.5
                                   DIFF, SCR
                              Hovensa Turbine GT3
                                    G-1101G
                                   St Croix
                                     135.5
                                     135.5
                                     55.0
                                     61.0
                                     74.5
                                   DIFF, SCR
                              Hovensa Turbine GT4
                                    G-3404
                                   St Croix
                                     161.0
                                     161.0
                                     55.0
                                     72.4
                                     88.5
                                   DIFF, SCR
                              Hovensa Turbine GT5
                                    G-3405
                                   St Croix
                                     161.0
                                     161.0
                                     55.0
                                     72.4
                                     88.5
                                   DIFF, SCR
                              Hovensa Turbine GT6
                                    G-3406
                                   St Croix
                                     161.0
                                     161.0
                                     55.0
                                     72.4
                                     88.5
                                   DIFF, SCR
                              Hovensa Turbine GT7
                                    G-3407
                                   St Croix
                                     161.0
                                     161.0
                                     55.0
                                     72.4
                                     88.5
                                   DIFF, SCR
                              Hovensa Turbine GT9
                                    G-3409
                                   St Croix
                                     52.2
                                     52.2
                                     55.0
                                     23.5
                                     28.7
                                   DIFF, SCR
                               VI WAPA Boiler 11
                                      11
                                   St Thomas
                                     646.0
                                     646.0
                                     55.0
                                     290.7
                                     355.3
                                  DIFF, SNCR
                               VI WAPA Boiler 13
                                      13
                                   St Thomas
                                    1,116.0
                                    1,116.0
                                     55.0
                                     502.2
                                     613.8
                                  DIFF, SNCR
                              VI WAPA Turbine 12
                                      12
                                   St Thomas
                                     381.0
                                     381.0
                                     55.0
                                     171.5
                                     209.6
                                   DIFF, SCR
                              VI WAPA Turbine 14
                                      14
                                   St Thomas
                                     392.0
                                     392.0
                                     55.0
                                     176.4
                                     215.6
                                   DIFF, SCR
                               VI WAPA Boiler 10
                                      10
                                   St Croix
                                     396.0
                                     396.0
                                     55.0
                                     178.2
                                     217.8
                                  DIFF, SNCR
                               VI WAPA Boiler 11
                                      11
                                   St Croix
                                     712.0
                                     712.0
                                     55.0
                                     320.4
                                     391.6
                                  DIFF, SNCR
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       
                                       




Table 6-4: Nitrogen Oxides Emissions Evaluation for Potential BART Sources
                                Source and Unit
                                     ID #
                                   Location
                           Baseline Emissions (tpy)
                   Emissions at Maximum Utilization Capacity
                           BART Level of Control (%)
                        Emissions After Controls (tpy)
                          Emissions Reductions (tpy)
                                Type of Control
                               Hovensa Boiler 3
                                    B-1153
                                   St Croix
                                     450.6
                                     450.6
                                     65.0
                                     157.7
                                     292.9
                                  DIFF, SNCR
                               Hovensa Boiler 5
                                    B-1155
                                   St Croix
                                     676.9
                                     676.9
                                     65.0
                                     236.9
                                     440.0
                                  DIFF, SNCR
                               Hovensa Boiler 6
                                    B-3301
                                   St Croix
                                     435.3
                                     435.3
                                     65.0
                                     152.3
                                     282.9
                                  DIFF, SNCR
                               Hovensa Boiler 7
                                    B-3302
                                   St Croix
                                     435.3
                                     435.3
                                     65.0
                                     152.3
                                     282.9
                                  DIFF, SNCR
                              Hovensa Turbine GT8
                                    G-3408
                                   St Croix
                                    1002.1
                                    1002.1
                                     80.0
                                     200.4
                                     801.7
                                   DIFF, SCR
                               Hovensa Boiler 1
                                    B-1151
                                   St Croix
                                     450.6
                                     450.6
                                     65.0
                                     157.7
                                     292.9
                                  DIFF, SNCR
                               Hovensa Boiler 4
                                    B-1154
                                   St Croix
                                     443.5
                                     443.5
                                     65.0
                                     155.2
                                     288.3
                                  DIFF, SNCR
                               Hovensa Boiler 8
                                    B-3303
                                   St Croix
                                     559.8
                                     559.8
                                     65.0
                                     195.9
                                     363.9
                                  DIFF, SNCR
                               Hovensa Boiler 9
                                    B-3304
                                   St Croix
                                     559.8
                                     559.8
                                     65.0
                                     195.9
                                     363.9
                                  DIFF, SNCR
                              Hovensa Turbine GT1
                                    G-1101E
                                   St Croix
                                     805.7
                                     805.7
                                     80.0
                                     161.1
                                     644.6
                                   DIFF, SCR
                              Hovensa Turbine GT2
                                    G-1101F
                                   St Croix
                                     805.7
                                     805.7
                                     80.0
                                     161.1
                                     644.6
                                   DIFF, SCR
                              Hovensa Turbine GT3
                                    G-1101G
                                   St Croix
                                     805.7
                                     805.7
                                     80.0
                                     161.1
                                     644.6
                                   DIFF, SCR
                              Hovensa Turbine GT4
                                    G-3404
                                   St Croix
                                     809.5
                                     809.5
                                     80.0
                                     161.9
                                     647.6
                                   DIFF, SCR
                              Hovensa Turbine GT5
                                    G-3405
                                   St Croix
                                     766.5
                                     766.5
                                     80.0
                                     153.3
                                     613.2
                                   DIFF, SCR
                              Hovensa Turbine GT6
                                    G-3406
                                   St Croix
                                     766.5
                                     766.5
                                     80.0
                                     153.3
                                     613.2
                                   DIFF, SCR
                              Hovensa Turbine GT7
                                    G-3407
                                   St Croix
                                     766.5
                                     766.5
                                     80.0
                                     153.3
                                     613.2
                                   DIFF, SCR
                              Hovensa Turbine GT9
                                    G-3409
                                   St Croix
                                     150.2
                                     150.2
                                     80.0
                                     30.0
                                     120.2
                                   DIFF, SCR
                               VI WAPA Boiler 11
                                      11
                                   St Thomas
                                     371.0
                                     371.0
                                     65.0
                                     129.9
                                     241.2
                                  DIFF, SNCR
                               VI WAPA Boiler 13
                                      13
                                   St Thomas
                                     670.0
                                     670.0
                                     65.0
                                     234.5
                                     435.5
                                  DIFF, SNCR
                              VI WAPA Turbine 12
                                      12
                                   St Thomas
                                    1288.0
                                    1288.0
                                     80.0
                                     257.6
                                    1030.4
                                   DIFF, SCR
                              VI WAPA Turbine 14
                                      14
                                   St Thomas
                                    1326.0
                                    1326.0
                                     80.0
                                     265.2
                                    1060.8
                                   DIFF, SCR
                               VI WAPA Boiler 10
                                      10
                                   St Croix
                                     359.0
                                     359.0
                                     65.0
                                     125.7
                                     233.4
                                  DIFF, SNCR
                               VI WAPA Boiler 11
                                      11
                                   St Croix
                                     646.0
                                     646.0
                                     65.0
                                     226.1
                                     419.9
                                  DIFF, SNCR

Table 6-5: PM10 Emissions Evaluation for Potential BART Sources
                                Source and Unit
                                     ID #
                                   Location
                           Baseline Emissions (tpy)
                   Emissions at Maximum Utilization Capacity
                           BART Level of Control (%)
                        Emissions After Controls (tpy)
                          Emissions Reductions (tpy)
                                Type of Control
                               Hovensa Boiler 3
                                    B-1153
                                   St Croix
                                     40.6
                                     40.6
                                     99.0
                                      0.4
                                     40.2
                                  DIFF, SNCR
                               Hovensa Boiler 5
                                    B-1155
                                   St Croix
                                     60.7
                                     60.7
                                     99.0
                                      0.6
                                     60.1
                                  DIFF, SNCR
                               Hovensa Boiler 6
                                    B-3301
                                   St Croix
                                     40.6
                                     40.6
                                     99.0
                                      0.4
                                     40.2
                                  DIFF, SNCR
                               Hovensa Boiler 7
                                    B-3302
                                   St Croix
                                     40.6
                                     40.6
                                     99.0
                                      0.4
                                     40.2
                                  DIFF, SNCR
                              Hovensa Turbine GT8
                                    G-3408
                                   St Croix
                                     15.1
                                     15.1
                                     99.0
                                      0.2
                                     14.9
                                   DIFF, SCR
                               Hovensa Boiler 1
                                    B-1151
                                   St Croix
                                     40.6
                                     40.6
                                     99.0
                                      0.4
                                     40.2
                                  DIFF, SNCR
                               Hovensa Boiler 4
                                    B-1154
                                   St Croix
                                     39.7
                                     39.7
                                     99.0
                                      0.4
                                     39.3
                                  DIFF, SNCR
                               Hovensa Boiler 8
                                    B-3303
                                   St Croix
                                     78.6
                                     78.6
                                     99.0
                                      0.8
                                     77.8
                                  DIFF, SNCR
                               Hovensa Boiler 9
                                    B-3304
                                   St Croix
                                     78.6
                                     78.6
                                     99.0
                                      0.8
                                     77.8
                                  DIFF, SNCR
                              Hovensa Turbine GT1
                                    G-1101E
                                   St Croix
                                     12.2
                                     12.2
                                     99.0
                                      0.1
                                     12.0
                                   DIFF, SCR
                              Hovensa Turbine GT2
                                    G-1101F
                                   St Croix
                                     12.2
                                     12.2
                                     99.0
                                      0.1
                                     12.0
                                   DIFF, SCR
                              Hovensa Turbine GT3
                                    G-1101G
                                   St Croix
                                     12.2
                                     12.2
                                     99.0
                                      0.1
                                     12.0
                                   DIFF, SCR
                              Hovensa Turbine GT4
                                    G-3404
                                   St Croix
                                     12.9
                                     12.9
                                     99.0
                                      0.1
                                     12.8
                                   DIFF, SCR
                              Hovensa Turbine GT5
                                    G-3405
                                   St Croix
                                     12.9
                                     12.9
                                     99.0
                                      0.1
                                     12.8
                                   DIFF, SCR
                              Hovensa Turbine GT6
                                    G-3406
                                   St Croix
                                     12.9
                                     12.9
                                     99.0
                                      0.1
                                     12.8
                                   DIFF, SCR
                              Hovensa Turbine GT7
                                    G-3407
                                   St Croix
                                     12.9
                                     12.9
                                     99.0
                                      0.1
                                     12.8
                                   DIFF, SCR
                              Hovensa Turbine GT9
                                    G-3409
                                   St Croix
                                     14.0
                                     14.0
                                     99.0
                                      0.1
                                     13.9
                                   DIFF, SCR
                               VI WAPA Boiler 11
                                      11
                                   St Thomas
                                     60.0
                                     60.0
                                     99.0
                                      0.6
                                     59.4
                                  DIFF, SNCR
                               VI WAPA Boiler 13
                                      13
                                   St Thomas
                                     28.0
                                     28.0
                                     99.0
                                      0.3
                                     27.7
                                  DIFF, SNCR
                              VI WAPA Turbine 12
                                      12
                                   St Thomas
                                     15.0
                                     15.0
                                     99.0
                                      0.2
                                     14.9
                                   DIFF, SCR
                              VI WAPA Turbine 14
                                      14
                                   St Thomas
                                     16.0
                                     16.0
                                     99.0
                                      0.2
                                     15.8
                                   DIFF, SCR
                               VI WAPA Boiler 10
                                      10
                                   St Croix
                                     45.0
                                     45.0
                                     99.0
                                      0.5
                                     44.6
                                  DIFF, SNCR
                               VI WAPA Boiler 11
                                      11
                                   St Croix
                                     81.0
                                     81.0
                                     99.0
                                      0.8
                                     80.2
                                  DIFF, SNCR

Steam Boilers
HOVENSA owns and operates nine steam boilers that are capable of combusting either refinery fuel gas (RFG) or No. 6 fuel oil and the heat input to the boilers is in the range of 205 to 405 mm BTU/hr.  
One of the boilers (Boiler 10) was constructed in 1999 and therefore is not BART-eligible.  EPA has determined that the following boilers are subject to BART: Boilers 1 (B-1151), 3 (B-1153), 4 (B-1154), 5 (B-1155), 6 (B-3301), 7 (B-3302), 8 (B-3303, and 9 (B-3304).  HOVENSA has indicated that Boiler 2 is no longer in service.  SO2 emissions are controlled by a permit limiting the sulfur content of No. 6 fuel oil to 0.50% or 1.0% depending upon wind conditions.  In addition, the June 2011 consent decree will lower SO2 emissions by requiring that the combustion of RFG, containing hydrogen sulfide (H2S), meet the requirements of the New Source Performance Standards (NSPS) at Part 60, subparts J and Ja: Boilers 1-5 subject to BART must meet subpart Ja requirements that limits the H2S content of RFG to 162 ppmv (3-hour rolling average) and 60 ppmv (daily on a 365 day rolling average); Boilers 6-9 subject to BART must meet the subpart J requirements that limits the H2S content of RFG to 0.10 grains per dry standard cubic feet (equivalent to about 162 ppmv).  The June 2011 consent decree will also lower the sulfur content of No. 6 fuel oil to 0.55% maximum , 0.50% annually, and to a low limit 0.30% should wind conditions warrant this (as defined in the CD).  There are no existing controls for NOx and PM emissions from the BART-eligible boilers.  
For control of SO2, NOx and PM emissions, based upon EPA's analysis, EPA is proposing that current operations represent BART for each of the boilers subject to BART.  For SO2 and PM control, EPA's contractor evaluated Duct Injection and Fabric Filters (DIFF) using lime as the alkaline reagent.  DIFF is a semi-wet flue gas desulfurization (FGD) process, with additional water added to humidify and cool the gas stream to the desired reaction temperature.  The fabric filter is the PM control device that would collect the sulfur reactant products and any PM formed directly from the combustion process.  EPA determined that the cost effectiveness for the eight boilers subject to BART varied in the range of about $19,100 to $39,600 per ton of SO2 and PM reduced.  EPA has determined that for the DIFF controls evaluated for the boilers subject to BART, Boilers 1 and 3-9, the cost-effective values are too high to be effective control.   In addition, it is EPA's opinion that if maximum controls had been evaluated, such as lime or limestone wet FGD, the cost effectiveness would be even higher than for the DIFF controls evaluated.  Therefore, EPA determines that for SO2 and PM controls, current operation is considered as BART.  
	For control of NOx emissions, EPA's contractor evaluated selective non-catalytic reduction (SNCR) using ammonia as the reagent.  EPA has determined that implementation of SNCR controls for boilers subject to BART are cost effective.  The actual cost effectiveness for the boilers is in the range of about $710 to $860 per ton of NOx removed.  As summarized in Table 8, the visibility impact (98[th] percentile, 4 year average, see modeling report in the Appendix) of all BART-eligible sources at HOVENSA on two receptors in the Class I area, St. John and Hassel Island near St. Thomas are 1.91 dv and 2.35 dv, respectively for all pollutants.  EPA further analyzed the contribution of various chemical species and components on the visibility impacts and has established that the contribution of NOx compounds from all potential BART sources at HOVENSA is about 5% which would be equivalent to about a 0.09 dv visibility impact at St. John and 0.91 for Hassel Island from all HOVENSA units subject to BART.  EPA has not analyzed the portion of the NOx contribution specifically from only the 8 boilers subject to BART since HOVENSA is not currently operating the refinery process units.  Since HOVENSA is not currently operating as a refinery, EPA proposes to determine that HOVENSA is not subject to implementing any new NOx controls at this time and therefore EPA proposes to determine that current operation of the boilers subject to BART is considered BART for controlling NOx emissions.  Regardless, as discussed in the RPG section, EPA is proposing to require HOVENSA to do a new reasonable measure analysis (if they resume operations) for those units that will be in operation.
Combustion Turbines
HOVENSA owns and operates eleven combustion turbines that are capable of combusting two or more of the following fuel combinations: refinery fuel gas (RFG), liquefied petroleum gas (LPG) and distillate oil.  Two of the turbines (GT 10 and GT 13) were constructed in 1993 and 2009 and are therefore not BART-eligible.  EPA has determined that the following nine turbines are subject to BART: GT1 (G-1101E), GT2 (G-1101F), GT3 (G-1101G), GT4 (G-3404), GT5 (G-3405), GT6 (G-3406), GT7 (G-3407), GT8 (G-3408), and GT9 (G-3409).  These turbines generate electricity for use at the facility, each in the range of 17.5  -  24 MW for a total nominal generating capacity of about 169 MW.  SO2 emissions are controlled by limiting the fuel sulfur content as follows: distillate oil has a permit sulfur limit of 0.20%; LPG does not contain any sulfur; RFG sulfur content will be limited by the CD that requires the combustion of RFG with limits on the H2S content in accordance with the NSPS requirements at subpart J or Ja; turbines GT Nos. 1, 2 and 3 must meet subpart Ja requirements limiting the H2S content of the RFG to 162 ppmv (3-hour rolling average) and 60 ppmv (daily on a 365 day rolling average); turbines GT Nos. 4 through 9 must meet subpart J requirements limiting the H2S content of the RFG to 0.10 grains per dry standard cubic feet (equivalent to 162 ppmv).  HOVENSA has indicated to EPA that turbines GT1 through GT3 typically combusts LPG; turbines GT4 through GT 8  typically combusts RFG that is already compliant with Subpart J limits for H2S; and for GT9, the permit limits this turbine's gaseous fuels to propane and butane. For NOx, only turbine GT9 has implemented control technology (steam injection).  For PM, none of the turbines subject to BART have any controls.  
For control of SO2, NOx and PM emissions, based upon EPA's analysis, EPA is proposing that current operations represent BART for each of the nine combustion turbines subject to BART.  For SO2 and PM control, as with the boilers discussed above, EPA's contractor evaluated Duct Injection and Fabric Filters (DIFF) using lime as the alkaline reagent.  Based upon this analysis, EPA has determined that the DIFF controls evaluated for the nine combustion turbines subject to BART are not cost effective because it exceeds a reasonable cost effectiveness for this kind of source.  EPA determined that the cost effectiveness for the nine combustion turbines subject to BART varied in the range of about $122,300 (8 turbines) to $359,186 (1 turbine) per ton of SO2 and PM reduced.  The cost effectiveness values for the combustion turbines are much higher than for the boilers because the SO2 emissions from the boilers are much higher (by a factor of 2 to 4 times) than from the turbines.  Therefore, EPA determines that for SO2 and PM controls, current operation is considered as BART.
For control of NOx emissions, as discussed above for the boilers, EPA's contractor evaluated selective non-catalytic reduction (SNCR) using ammonia as the reagent.  EPA has determined, except for turbine GT9, that implementation of SNCR controls for eight turbines subject to BART, GT1 through GT8, the cost in dollars per ton is low enough to be cost-effective. The actual cost effectiveness for the turbines GT1 through GT 8 is in the range of about $1750 to $1890 per ton of NOx removed.  The actual cost effectiveness for turbine GT9 is about $9,500/ton and is too high to be cost effective.  The cost effectiveness of turbine GT9 is much higher than the other eight turbines because the NOx emissions are much lower due to NOx controls installed on GT9.  
Even though eight of the nine turbines subject to BART have cost effective values within acceptable limits of EPA's cost benchmark, EPA has determined, for the same reasons discussed above for the boilers, that the visibility impact due to NOx emissions is only about 0.09 dv from all HOVENSA units subject to BART, and therefore the implementation of any new NOx controls (even SNCR or SCR) on the combustion turbines would not have enough of a improvement in visibility on the Class I area in the Virgin Islands.  Therefore, EPA is determining that current operations of the nine turbines subject to BART are considered BART for controlling NOx emissions.    
Process Heaters 
HOVENSA owns seventy process heaters of which twenty-one were shut down in early 2011.  Of the seventy heaters, EPA has determined that sixty-four are subject to BART whereas the remaining six heaters were constructed after 1977 and are therefore not BART-eligible.  Of the sixty-four process heaters subject to BART, fifteen are capable of combusting either RFG or No. 6 fuel oil whereas the remaining forty-nine heaters combust only RFG.  EPA has determined that the following list of sixty-four process heaters are subject to BART:  H-101, H-104, H-200, H-201, H-202, H-401 A-C, H-1401 A and B, H-1500, H 1501, H-160, H-600 through H-606, H-800A and B, H-801, H-2101, H-2102, H-2201A and B, H-2202, H-2400, H-2401, H-2501, H-4502 through H-4505, H-3101A and B, H-4101A and B, H-4401, H-4402, H-4451 through H-4455, H-4201, H-4202, H-5401, H-5402, H-5451 through H-5455, H-4601A and B, H-4602, H-4301A and B, H-4302, H-5301A and B, and H-5302.  
For the fifteen heaters capable of combusting No. 6 fuel oil, SO2 emissions are controlled by permit limiting the sulfur content of No. 6 fuel oil to 0.50% or 1.0% depending upon wind conditions.  The June 2011 CD may lower SO2 emissions by establishing the sulfur content of No. 6 fuel oil to 0.55% maximum, 0.50% annually, and to a low limit 0.30% should wind conditions warrant this (as defined in the CD).  In addition, the CD requires process heaters to limit the H2S content of RFG in accordance with the NSPS at Part 60, at either subpart J (about 162 ppmv H2S) or Ja (162 ppmv H2S on a 3 hour rolling average and 60 ppmv H2S daily on a 365 day rolling average).  Appendix A of the CD identifies which process heaters are subject to either subparts J or Ja.  None of the process heaters subject to BART have any controls for either NOx or PM.     
For control of SO2, NOx and PM emissions, based upon EPA's analysis, EPA is proposing that current operations represent BART for each of the sixty-four process heaters subject to BART.  Although EPA's contractor determined cost effectiveness for only the boilers and combustion turbines, EPA has concluded that, for control of SO2, NOx and PM, there is sufficient information to make a determination that current operation represents BART for each of the process heaters subject to BART.  For the SO2 and PM BART determination, EPA notes that the SO2 emissions, heat input and fuel type for the each of the six largest process heaters is similar to that of most of the boilers which EPA's contractor established cost effectiveness values that EPA believes are too high to be cost effective.  It is EPA's judgment from this size comparison between the boilers and the six largest heaters that the cost effectiveness for the process heaters would be less than for the boilers but still excessive.  The great majority of the remainder of the process heaters combust only RFG, have a smaller heat input (each by a factor of about 2.75 average) and have lower SO2 emissions (each by a factor of about 7.8 on average) than the six large heaters.  Based upon this comparison, EPA would expect that the remaining smaller process heaters will also have a cost effectiveness that are sufficiently excessive to make these potential controls not cost effective.  Therefore, for SO2 and PM control, EPA proposes to determine that the process heaters subject to BART are all not cost effective for controlling SO2 and PM emissions and that current operation is considered BART.   
As discussed above for the boilers and combustion turbines, EPA determined that implementation of controls on NOx emissions from all BART units at HOVENSA have an insignificant visibility impact on the Class I areas in the Virgin Islands and therefore EPA is proposing to determine this is also true for the process heaters.  Therefore EPA proposes that current operation of the process heaters subject to BART is considered as BART for controlling NOx emissions.    
Reciprocating gas compressors
HOVENSA owns eleven compressors of which five were shut down in early 2011.  EPA has determined that the following eleven compressors are subject to BART:  C-200A through C, C-1500 A through C, C-2400 A and B, and C-4601 A through C.  Five of the compressors (C-200 A through C and C-2400 A and B) are equipped with catalytic converters (similar to those used on cars) for controlling NOx emissions.  These eleven compressors combust only liquefied petroleum gas (LPG) and therefore there are no emissions of SO2 and PM emissions are small, as indicated by emissions data from HOVENSA.  
For control of SO2, NOx and PM emissions, based upon EPA's analysis EPA is proposing that current operations represent BART for each of the eleven reciprocating gas compressors subject to BART.  Since there are no SO2 emissions and PM emissions are small, EPA is proposing that current operation using LPG fuel represents BART for control of SO2 and PM emissions from the eleven reciprocating gas compressors subject to BART.  NOx reductions will not have a useful impact on improving visibility.
Tail Gas Treatment Units
HOVENSA owns and operates two Tail Gas Treatment (TGT) units and components of each unit combust RFG.  EPA has determined that one of the TGT units, No. 2 Beavon (which includes H-4761(Reduced Gas Generator) and T-4761 (Stack)), is subject to BART whereas the other unit, No. 1 Beavon (which includes H-1061 (Reduced Gas Generator) and T-1061 (Stack)) was constructed after 1977 and is therefore not BART-eligible.  The TGT units ("Beavons") are control devices for the sulfur recovery plants and the two TGT units process tail gas from the four sulfur recovery units (SRU) in order to lower emissions of sulfur compounds and to recover additional sulfur.  The TGT unit subject to BART (No. 2 Beavon including H-4761 and T-4761) processes tail gas from No. 3 SRU and No. 4 SRU and emissions of SO2, NOx and PM are low (potential to emit (PTE): 2 tpy SO2 and 4.0 tpy NOx) as compared to other units at this refinery facility (e.g., Boiler No. 1 PTE: 330 tpy SO2 and 450 tpy NOx).  Since the emissions are so low for the TGT unit subject to BART, it is EPA's judgment that any BART analysis for additional controls would have resulted in cost effectiveness values too to be cost effective for BART.  Furthermore, as discussed above for other units subject to BART, any reduction of NOx emissions would not have any significant impact on visibility and therefore current operations represents BART for control of NOx emissions.  Also, it should be noted that the CD requires that (1) by the end of 2014, HOVENSA install a second TGT unit to service the No. 3 and No. 4 SRUs; and (2) by April 1, 2015, the H-4761 section of the TGT unit must be in compliance with the NSPS Part 60, subpart Ja concerning the H2S concentration limit of the RFG.  
Therefore, based upon the above analysis, EPA is proposing that current operations without any new controls represent BART for the TGT unit subject to BART.  
Tail Gas Incinerators
HOVENSA owns and operates three Tail Gas Incinerators (TGI) and each combusts RFG.  EPA has determined that all three of the TGI units (H-4745, H-1032 and H-1042) are subject to BART.  The TGI units are part of the sulfur recovery plants that include the Tail Gas Treatment (TGT) units described above as well as for flare units as well as other units that are not needed for consideration for this BART determination.  It is EPA's understanding that: (1) two incinerators (H-1032 and H-1042) are associated with No.1 and 2 SRUs and these TGIs are used during startup, shutdown and malfunction of those SRUs or during a malfunction of the associated TGT unit (No. 1 Beavon which includes H-1061 and T-1061); and (2) one incinerator (H-4745) is associated with No. 3 and 4 SRUs and is used during periods of maintenance or malfunction of  the associated TGT unit (No. 2 Beavon which includes H-4761 and T-4761) as well as during startup,  shutdown or malfunction of these SRU units.  
TGI unit H-4745 has much higher annual emissions (PTE: 900 tpy SO2 and 28 tpy NOx; actual emissions: 228 tpy SO2 and 7.0 tpy NOx) than each of the other two incinerators.   However, as discussed above for the TGT unit, the CD requires that, by the end of 2014, HOVENSA is to install a second TGT unit to service the No. 3 and No. 4 SRUs.  Under this scenario there would be two TGT units (No. 2 Beavon which includes H-4761 (Reduced Gas Generator) and T-4761 (Stack) and a new TGT unit) servicing No. 3 and 4 SRUs, and because of this redundancy in TGT units, the use of incinerator H-4745 might only be used for startup, shutdown and malfunction of the associated TGT unit (either the No. 2 Beavon system or the new TGT unit).   Under this redundancy situation, HOVENSA expects that the PTE for SO2 for incinerator H-4745 would be reduced from 900 tpy to a small fraction of this current PTE and, in addition, SO2 emissions would meet the NSPS emission limits under Part 60 subparts J or Ja.
Since TGI units H-1032 and H-1042 have very low air pollutant emissions and TGI unit H-4745 would also have low air pollutant emissions once in compliance with the CD, it is EPA's judgment that any BART analysis for additional controls would have resulted in high cost effectiveness values.   Furthermore, as discussed above for other units (boilers, turbines etc) subject to BART, any reduction of NOx emissions would not have any significant impact on visibility and therefore EPA proposes that current operation represents BART for control of NOx emissions.  Also, it should be noted that the CD requires that each of the three TGI units subject to BART must be in compliance with the NSPS Part 60, subpart Ja such that the emissions of SO2 are limited to not more than 250 ppmv (at zero percent excess air)  -  TGI unit H-4745 must comply by April 1, 2015 and TGI units H-1032 and H-1042 must comply by December 31, 2011.  
Therefore, EPA is proposing that current operations without any new controls of SO2, NOx and PM emissions represent BART for the three TGI units subject to BART.  

Flares  
HOVENSA owns and operates eight flares throughout the refinery.  HOVENSA indicates that the flares are used primarily as safety devices for purposes of relieving accidental over-pressuring.   EPA has determined that five of the flares (No 2 (H-1105), No. 3 (H-1104), No. 5 (H-3351), No. 6 (H-3352), No. 7 (H-3301)), all located within the sulfur recovery plant, are subject to BART.  Flares 1 and 4 are no longer in service.  Each flare subject to BART includes a steam injection system at the flare tip for the purpose of improving combustion and thereby eliminating smoking conditions.  HOVENSA reports that the facility PTE for all eight flares is as follows: 1200 tpy SO2, 1897 tpy NOx and negligible emissions of PM.  For all the flares, including those subject to BART, HOVENSA's best estimate of the PTE for each flare is that emissions are divided approximately equally among each flare, as follows: 150 tpy SO2, 237 tpy NOx and negligible emissions of PM.  However, HOVENSA also reports that the PTE emissions are much higher than actual emissions which have the following average emissions for the five flares subject to BART: 1.46 tpy SO2, 6.7 tpy NOx and 1.0 typ PM.  
The June 2011 CD requires that each of the five BART flares subject to the NSPS Part 60 subparts J and Ja requirements implement "flare gas recovery" to comply with the CD.  Flare gas recovery is a system to reduce SO2 emissions by the installation of compressors on the flare line to recycle the flare gas back into process units for removal of H2S.    Since the CD will further reduce SO2 emissions and actual emissions from the flares are very low and EPA is not aware of any desulfurization technologies to treat flare gas emissions, EPA is proposing that current operations are considered as BART for both SO2 and PM.  Furthermore, as discussed above for other units (boilers, turbines etc) subject to BART, any reduction of NOx emissions would not have any significant impact on visibility and therefore, EPA is proposing that current operations are considered as BART for NOx.   
Therefore, EPA is proposing that current operations without any new controls of SO2, NOx and PM emissions represent BART for the five flare units subject to BART.  

Water Intake Pumps and Desalination Water Pump
HOVENSA owns and operates four Water Intake Pumps and one Desalination Water Pump and each combusts either diesel or No. 2 fuel oil.  EPA has determined that each of the Water Intake Pumps (PD-1062 through PD-1065) and the Desalination Water Pump (PD-1620) are subject to BART.  HOVENSA has confirmed that each of these pumps is an emergency standby pump that is used only when there is a complete power failure at the refinery.  HOVENSA reports that actual emissions are negligible and PTE emissions are small.  EPA determined that the average PTEs are as follows: 1.8 tpy SO2, 38 tpy NOx, and 2.7 tpy PM.  The pumps normally in use are electric driven pumps and therefore have no air pollutant emissions.   
Since these five pumps are used very infrequently and actual emissions are negligible it is EPA's judgment that any detailed cost analysis would conclude that implementation of any technologies for controlling emissions of SO2, NOx or PM are not cost effective.  Therefore, EPA is proposing that current operation represents BART for these five pumps subject to BART.  

BART Determinations
In summary, EPA's BART evaluation of the boilers, turbines, process heaters, and several other source categories that are subject to BART at HOVENSA has determined that no additional control for SO2, NOx and PM emissions is consistent with BART.  Most stacks are not amenable to cost effective controls. EPA is proposing that current operations represent BART for HOVENSA.  As such, EPA's Federal plan includes the establishment of emission limits for SO2, NOx and PM equivalent to the potential to emit (PTE) for each unit subject to BART, as derived from HOVENSA's permit limit conditions.  EPA's Federal plan includes these PTE limits in the spreadsheets found in the Attachments to the FIP. 
For the stacks that have low cost effectiveness for controls for NOx, EPA further analyzed the contribution of various chemical species and components on the visibility impacts. The contribution of NOx compounds from all potential BART sources at HOVENSA to visibility impairment on the average 98[th] percentile day is about a 0.09 dv visibility impact at St. John and 0.91 dvfor Hassel Island from all HOVENSA units subject to BART.  The implementation of any new NOx controls (even SNCR or SCR) would not result in sufficient improvements in visibility on St. John.  Therefore, EPA is determining that current operations of the BART-eligible sources at HOVENSA, including the nine turbines, are considered BART for controlling NOx emissions.    
BART Analysis for WAPA Facilities on St. Croix and St. Thomas
WAPA power plants on St. Croix and St. Thomas are potential BART sources, with emission sources that are large enough to qualify for BART and were built at during the BART-eligible period. However, the impact of the WAPA power plant on St. Croix and the WAPA power plant on St. Thomas have an impact of 0.09 and 0.04 dv, respectively, for the average 98[th] percentile metric, on visibility at the IMPROVE monitor on St. John. An analysis of potential controls is listed in the accompanying tables. However, the improvement in visibility from these reductions would be even less than the less than 0.1 dv total impact of these potential BART sources, so EPA has determined that they are not eligible for BART controls, due to how little they add to improving visibility at the National Park in St. John.
BART Determinations including Other Types of Standards
As allowed by 40 CFR §51.308(e)(1)(iii) of the Regional Haze Rule the EPA has determined in establishing BART that technological or economic limitations on the applicability of measurement methodology would make the imposition of an emission standard infeasible. A design, equipment, work practice, or other operational standard, or combination thereof can be prescribed to require the application of BART for each source. These standards, to the degree possible, set forth the emission reduction to be achieved by implementation of such design, equipment, work practice or operation, and provide for compliance by means that achieve equivalent results. 
For this FIP, existing controls were determined to be BART.  No other standards were needed to determine that BART controls were not feasible. 
Description of BART Alternative for Any Source
In section (e)(2) of the RHR provides that a state may opt to implement an emissions trading program or other alternative measure rather than to require sources subject to BART to install, operate, and maintain BART. To do so, the State must demonstrate that emissions trading program or other alternative measure will achieve greater reasonable progress than would be achieved through the installation and operation of BART. To make this demonstration, the State must submit an implementation plan containing the elements listed in Section (e)(2).
No alternative measures are available for the Virgin Islands FIP, since there are no sources in other states that affect the Virgin Islands' Class I area. 
Schedule for BART Implementation
As provided in 40 CFR §51.308(e)(1)(iv) BART must be in operation for each applicable source no later than five years after FIP approval. If these potential BART sources in the Virgin Islands were to be subject to BART, each would be required to install and operate BART as expeditiously as practicable but in no event later than five years after approval of the FIP or plan revision by EPA. 
As provided in 40 CFR §51.308(e)(1)(v), the Title V operating permits for BART sources must include a requirement that each source maintain the control equipment and establish procedures to ensure such equipment is properly operated and maintained. This requirement would have been included in the Title V operating permit for each source subject to BART. 

LONG-TERM STRATEGY
Section 51.308(d)(3) of 40 CFR part 51 requires the Virgin Islands to submit a long-term strategy that addresses regional haze visibility impairment for each mandatory Class I Federal area within and outside the Territory that may be affected by emissions from within the Territory. The long-term strategy must include enforceable emissions limitations, compliance schedules, and other measures necessary to achieve the reasonable progress goals established for the Class I. 
This section describes how the FIP for the Virgin Islands meets the long-term strategy requirements. This long-term strategy addresses visibility impairment for the Virgins Islands National Park, which is the only Class I Area that is affected by emissions from within the Virgin Islands.
The long-term strategy described below includes enforceable emissions limitations, compliance schedules, and other measures necessary to achieve the reasonable progress goals established for the Class I to the extent that it is reasonable for the Virgin Islands (or EPA on its behalf) to adopt them before the date this FIP is finalized. Additional measures may be reasonable to adopt at a later date after further consideration and review. In this FIP, EPA is using existing measures to address reasonable progress. 
Overview of the Long-Term Strategy Development Process
Section 51.308(d)(3)(iii) of 40 CFR part 51 requires us to document the technical basis for the State's/Tribe's apportionment of emission reductions necessary to meet reasonable progress goals in each Class I area affected by the Territory's emissions.
Modeling shows that Puerto Rico, the portion of the United States nearest to the Virgin Islands, does not affect visibility in the Class I area of the Virgin Islands National Park on St. John. We demonstrated this using dispersion modeling for a representative major source, which we located on the closest point on Puerto Rico to simulate the maximum possible impact on visibility in the Virgin Islands. The modeling predicts a less than 0.01 DV impact from a potential major source on Puerto Rico, so sources or groups of sources in the continental United States, which is much further away from the Virgin Islands than Puerto Rico, will not have an impact on visibility in the Virgin Islands.
 The trajectory and modeling results show that there may be some small effect from international sources (e.g., Antigua, St. Martin, British Virgin Islands); reductions from these sources would be a matter of international policy. Emissions from natural sources (e.g., Saharan Dust, Montserrat Volcano), cannot be controlled. Therefore, the ability of the Virgin Islands to coordinate emission reductions from sources on a regionwide scale is limited. 
Because of this limitation, this FIP for the Virgin Islands relies on reductions that will occur in the Virgin Islands to achieve reasonable progress toward improving visibility in the Virgin Islands National Park. The following sections discuss the pollutants, source regions, and types of sources considered in developing this long-term strategy to assure the attainment of reasonable progress goals. 
Emission Reductions Due to Ongoing Air Pollution Programs 
Section 51.308(d)(3)(v)(A) of 40 CFR part 51 requires State/Tribes to consider emission reductions from ongoing pollution control programs. In developing its Long Term Strategy, the Virgin Islands FIP considered emission control programs being implemented between the baseline period and 2018, as discussed below. However, the effect of these programs, except for the HOVENSA consent decree and emission reductions from ocean vessels, were not included in the modeling, since EPA and VIDPNR do not have sufficient information on these source categories and how they will change by 2018. In addition, the remoteness of the Virgin Islands, especially the island of St. John (which is mostly a National Park), makes it difficult to project the effects of vehicle turnover and improvements to diesel engine fleets. Thus, these measures are included as possible contributors to improved visibility for the National Park on St. John.
Other Point Source Controls Expected by 2018 Due to Ongoing Air Pollution Control Programs
Point source controls applicable to the Virgin Islands include the following:
EPA's Refinery Enforcement Initiative. Since March 2000, the EPA has reached settlements with 29 U.S. companies responsible for 90 percent of petroleum refining in the nation. Through multi-issue, multi-facility settlements or detailed investigations and aggressive enforcement, this national initiative addresses the most significant Clean Air Act compliance concerns affecting the petroleum refining industry. HOVENSA, LLC was one such company that entered into a settlement with the EPA for its refinery on St. Croix. EPA's settlements require significant reductions of SO2 and nitrogen oxide, as well as emission reductions of benzene, VOCs, and particulate matter. 
VOC 2-, 4-, 7-, and 10-year MACT Standards. Various point source MACTs and associated emission reductions were implemented in many parts of the United States, but were not considered in the development of projection emissions inventory for the Virgin Islands. 
Combustion Turbine MACT. The projection inventories do not include the NOx co-benefit effects of the MACT regulations for Gas Turbines or stationary Reciprocating Internal Combustion Engines, which the EPA estimates to be small compared to the overall inventory. 
Industrial Boiler/Process Heater/RICE MACTs. The EPA issued final rules to substantially reduce emissions of toxic air pollutants from industrial, commercial and institutional boilers, process heaters and from stationary reciprocating internal combustion engines (RICE). These rules reduce emissions of a number of toxic air pollutants, including hydrogen chloride, manganese, lead, arsenic and mercury by 2013. This rule also reduces emissions of NOx and VOCs in conjunction with the toxic air pollutant reductions (http://www.epa.gov/ttn/atw/rice/ricepg.html).
Controls on Non-road Sources Expected by 2018 due to Ongoing Air Pollution Control Programs
Non-road source controls that may be applicable to the Virgin Islands include the following:
Non-road Diesel Rule. This rule sets standards that will reduce emissions by more than 90 percent from non-road diesel equipment, and reduce sulfur levels by 99 percent from current levels in non-road diesel fuel starting in 2007. This step will apply to most non-road diesel fuel in 2010 and to fuel used in locomotives and marine vessels in 2012 (http://www.epa.gov/nonroad-diesel/). Reductions due to this program may reduce impairment to visibility by a small amount, but it was not modeled due to uncertainties in the emissions from this source category. 
Mobile Source Controls Expected by 2018 due to Ongoing Air Pollution Control Programs
Mobile source controls applicable to the Virgin Islands include the following:
Heavy Duty Diesel (2007) Engine Standard. EPA set a PM emissions standard for new heavy-duty engines of 0.01 grams per brake-horsepower-hour (g/bhp-hr), to take full effect for diesel engines in the 2007 model year. This rule also includes standards for NOx and non-methane hydrocarbons (NMHC) of 0.20 g/bhp-hr and 0.14 g/bhp-hr, respectively. These NOx and NMHC standards will be phased in together between 2007 and 2010 for diesel engines. Sulfur in diesel fuel must be lowered to enable modern pollution-control technology to be effective on these trucks and buses. EPA will require a 97 percent reduction in the sulfur content of highway diesel fuel from its current level of 500 parts per million (low sulfur diesel, or LSD) to 15 parts per million (ultra-low sulfur diesel, or ULSD) (http://www.epa.gov/otaq/highway-diesel/index.htm).
Tier 2 Motor Vehicle Standards. Tier 2 is a fleet averaging program, modeled after the California LEV II standards. Manufacturers can produce vehicles with emissions ranging from relatively high to zero, but the mix of vehicles a manufacturer sells each year must have average NOx emissions below a specified value. Tier 2 standards became effective in the 2005 model year (http://www.epa.gov/tier2/finalrule.htm).
Large Spark Ignition and Recreational Vehicle Rule. EPA has adopted new standards for emissions of NOx, hydrocarbons (HC), and carbon monoxide (CO) from several groups of previously unregulated non-road engines. Included in these are large industrial spark-ignition engines and recreational vehicles. Non-road spark-ignition engines are those powered by gasoline, liquid propane, or compressed natural gas rated over 19 kilowatts (kW) (25 horsepower). These engines are used in commercial and industrial applications, including forklifts, electric generators, airport baggage transport vehicles, and a variety of farm and construction applications. Non-road recreational vehicles include snowmobiles, off-highway motorcycles, and all-terrain vehicles. These rules were initially effective in 2004 and should be fully phased-in by 2012 (http://epa.gov/nonroad/2002/cleanrec-final.htm).
Reductions from these source categories are not estimated for modeling because of the uncertainties in the makeup of the emission inventories for these sources. 
Source Retirement and Replacement Schedules 
Section 51.308(d)(3)(v)(D) of 40 CFR part 51 requires us to consider source retirement and replacement schedules in developing reasonable progress goals. In January 2012, it was announced that the HOVENSA refinery would be ceasing refinery operations and converting to an oil storage terminal by mid-February of 2012. Modeling suggests the emissions from HOVENSA sources are responsible for 1.06 dv of visibility impairment in the Virgin Islands National Park at St. John. The closure of this facility would provide much of the reduction needed to meet reasonable progress target for 2018, although this FIP is not using these reductions because HOVENSA has not surrendered its emissions permits. So for the long-term plan, HOVENSA is modeled as if it continues to operate, but with the emission reductions agreed to in the consent decree. 
Additional source retirements or replacements are not expected by 2018.  Any retirement and replacement must comply with existing SIP requirements including those pertaining to PSD and New Source Review.
Additional Reasonable Strategies
Section 51.308(d)(3) of 40 CFR part 51 requires Plans to include enforceable emissions limitations, compliance schedules, and other measures as necessary to achieve the reasonable progress goals established by States/Tribes with Class I areas. The plan must include reasonable measures and identify the visibility improvement that will result from those measures.
Since existing measures and agreements are the reasonable measures for this FIP for the Virgin Islands, the Section 51.308(d)(3) compliance schedules are not need for this Plan.
As discussed in the next section, additional controls for this long-term strategy are limited, in part, due to the compressed schedule for this FIP. 
Analysis of the Four Statutory Factors
The CAA (Sec. 169A) requires that Haze Plans provide for reasonable progress toward attaining the goal of preventing any future, and remedying any existing impairment of visibility in mandatory Class I Federal areas. In determining reasonable progress, States shall consider the four factors listed below:
         * Costs of compliance 
         * Time necessary for compliance
         * Energy and non-air quality environmental impacts of compliance
         * Remaining useful life of any potentially affected sources.
Regulations implementing this requirement have been promulgated in 40 CFR Part 51, Subpart P  -  Protection of Visibility. Section 51.308 specifically addresses the regional haze program requirements. Guidance on implementing the reasonable progress requirements have been developed by the EPA, Guidance for Setting Reasonable Progress Goals under the Regional Haze Program, revised June 1, 2007.
In judging the adequacy of visibility improvement, the EPA has considered all current regulations and programs, and then determined the amount of additional improvement that may be necessary to represent reasonable progress. For the Virgin Islands, visibility improvements will not reach the reasonable progress goal, but existing controls are all that can reasonably be done to improve visibility in the first period ending in 2018.   
With respect to the four-factor analysis, the implementation of any additional control strategies is hampered by the compressed schedule that this FIP was developed under. Sources would have only a few years to comply with any additional control programs. Based on timeframes presented in regional haze SIPs for other States, this would be a minimum amount of time needed for the simplest of additional strategies. It is unreasonable to expect any additional control in this period. 
Best Available Retrofit Technology
BART controls at sources potentially subject to BART are identified in Section 6 of this FIP.  EPA has determined BART to be existing controls, as additional controls are neither cost effective nor to have much impact on improving visibility in the National Park.
Changes to Emissions by 2018
The emissions inventory for the Virgin Islands Haze Plan has been completed for the island of St. John. Due to the small impact from sources on St. Thomas and St. Croix, except for HOVENSA, and the emerging state of the emissions inventory for the Virgin Islands, the official emission inventory for regional haze plan will be tabulated for sources on the island of St. John and surrounding waters. Emissions change from the 2004 base year to the 2018 inventory for the island of St. John due to reduced emissions from marine vessels. While these reductions in marine vessel emissions are not part of the enforceable plan for the Virgin Islands Haze Plan, they are included as EPA anticipates that the marine vessel controls will be in place before 2018.  Other emission reductions are possible, such as introduction of newer, cleaner motor vehicles and reduced sulfur in fuel and controls on off-road vehicles. But these reductions in vehicle emissions are hard to quantify for a remote location like St. John, so emission reductions from these sources will not be quantified in the emission inventory for St. John for 2018.
Visibility on St. John will also be improved by the emission reductions from the consent decree at HOVENSA. This change in emissions is tabulated separately from the emission inventory for St. John, as they occur on the island of St. Croix, not on St. John. Other emissions on St. Croix and St. Thomas are not anticipated to change or the change will be too small to affect visibility on St. John, so this haze plan will not include them in the emissions summaries.  A summary of these changes is in emissions of SO2 is provided in Table 7-1, NOx in Table 7-2, and PM in Table 7.3 for the island of St. John. HOVENSA emission changes from the consent decree are in Table 7.4. 
Table 7-1: Emissions from Point, Area and Mobile Sources on St. John, Virgin Islands (SO2 tpy)
                                 Source Sector
                                 Baseline 2002
                         2018 (with measures for RPG)
                                     Point
                                     43.11
                                     43.11
                                     Area
                                     0.05
                                     0.05
                                Non-Road Mobile
                                     17.89
                                     17.89
                                On-Road Mobile
                                     1.61
                                     1.61
                                Marine Vessels
                                     94.06
                                     14.11
                                     Total
                                    156.72
                                     76.77

Table 7-2: Emissions from Point, Area and Mobile Sources on St. John, Virgin Islands (NOx tpy)
                                 Source Sector
                                 Baseline 2002
                         2018 (with measures for RPG)
                                     Point
                                    477.66
                                    477.66
                                     Area
                                     3.69
                                     3.69
                                Non-Road Mobile
                                     2.07
                                     2.07
                                On-Road Mobile
                                     25.03
                                     25.03
                                Marine Vessels
                                    318.23
                                     63.65
                                     Total
                                    826.68
                                     572.1
      
Table 7-3: Emissions from Point, Area and Mobile Sources on St. John, Virgin Islands 
(PM tpy)
                                 Source Sector
                                 Baseline 2002
                         2018 (with measures for RPG)
                                     Point
                                     34.33
                                     34.33
                                     Area
                                     38.32
                                     38.32
                                Non-Road Mobile
                                     1.93
                                     1.93
                                On-Road Mobile
                                     0.73
                                     0.73
                                Marine Vessels
                                     8.57
                                     1.28
                                     Total
                                     83.88
                                     76.59



Table 7-4: Emissions from HOVENSA (tons per year)

SO2
NOx
PM
EC*
HOVENSA Base 2002
12,778
26,362
2,207
22.1
HOVENSA Post Control 2018
9,318
21,331
2,192
21.9
* Elemental Carbon emissions were assumed to be 1/100 of total PM emissions.
These emissions are included in the modeling analysis to determine the improvement in visibility at the Class I area in the Virgin Islands.  Other emissions sources are described in the modeling analysis in the Appendix.

Additional Measures Considered
These sources include ferries, cruise ships, and cargo ships that dock at St. John. The Caribbean ECA (Emissions Control Area) emission reductions for ocean vessels are included in the future case modeling. Beginning in 2015, fuel used by all vessels operating in these areas cannot exceed 0.1 percent fuel sulfur (1,000 ppm). This requirement is expected to reduce PM and SOx emissions by more than 85 percent. Beginning in 2016, new engines on vessels operating in these areas must use emission controls that achieve an 80 percent reduction in NOx emissions. While these reductions are not enforceable as part of this FIP, EPA expects them to occur and they will be included in the reductions expected in the period through 2018.
Caribbean Emissions Control Area
The Caribbean ECA (Emissions Control Area) emission reductions for ocean vessels are included as an improvement in air quality for the future case modeling. Beginning in 2015, fuel used by all vessels operating in these areas cannot exceed 0.1 percent fuel sulfur (1,000 ppm). This requirement is expected to reduce PM and SOx emissions by more than 85 percent. Beginning in 2016, new engines on vessels operating in these areas must use emission controls that achieve an 80 percent reduction in NOx emissions. While these emission reductions were not developed specifically for this FIP, the use of the proper fuel and emission controls will be enforceable by EPA when this program goes into effect.  These emission reductions will result in 0.03 dv improvement in visibility on the twenty percent worst visibility days at the IMPROVE site on St. John.  
Measures to Mitigate the Impacts of Construction Activities
Section 51.308(d)(3)(v)(B) of 40 CFR part 51 requires the Virgin Islands to consider measures to mitigate the impacts of construction activities. 
As noted previously, the majority of the soils contribution to visibility impairment on the twenty percent worst days is attributable to transport of Saharan dust, and not necessarily local sources. In preparation for the next SIP revision in 2018, the EPA will work with VIDPNR to determine the effectiveness of better construction practices on St. John and quantify the contribution of construction activity to visibility impairment and possible mitigation strategies. 
Agricultural and Forestry Smoke Management
Section 51.308(d)(3)(v)(E) of 40 CFR part 51 requires Haze Plans to consider smoke management techniques for the purposes of agricultural and forestry management in developing reasonable progress goals. The Virgin Islands has an open burning regulation.
Smoke Management Programs are only required when smoke impacts from fires managed for resource benefits contribute significantly to regional haze. The results of the emissions inventory on St. John indicate that emissions from agricultural, managed, and prescribed burning are small sources of pollution. It is unlikely that fires for agricultural or forestry management cause large impacts on visibility in the Virgin Islands National Park. On rare occasions, when smoke from major fires degrades the air quality and visibility in the Virgin Islands. However, these fires are generally unwanted wildfires that are not subject to smoke management programs. 
Enforceability of Emission Limitations and Control Measures
Section 51.308(d)(3)(v)(F) of 40 CFR part 51 requires Haze Plans to ensure that emission limitations and control measures used to meet reasonable progress goals are enforceable. 
BART eligible sources have existing permits that are enforceable by EPA.
Federal measures, such as the Federal Motor Vehicle Control Program, are enforceable by EPA. 
The marine vessel emission reductions under ECA are enforceable by EPA.
EPA will continue to work with the Virgin Islands to evaluate the measures included in the long-term strategy, especially if HOVENSA reopens.  EPA and the Virgin Islands will determine whether controls, particularly at HOVENSA, are reasonable to adopt and implement by 2018 and will evaluate these and other potential measures in preparation for the next revision of the haze plan in 2018. 
EPA will require that if HOVENSA reopens its refinery, HOVENSA shall evaluate all of its emission points to determine if reasonable controls are available for those emission points.
Consultation on the Long-Term Strategy
Section 51.308(d)(3)(i) 40 CFR part 51 requires consultation with other States/Tribes to develop coordinated emission management strategies. This requirement applies both where emissions from the Virgin Islands are reasonably anticipated to contribute to visibility impairment in Class I areas outside the Virgin Islands and when emissions from other States/Tribes are reasonably anticipated to contribute to visibility impairment in Class I areas within the Virgin Islands.
As noted in Section 3, the Virgin Islands have little impact on visibility impairment in other States and other States have little to no impact on visibility impairment in Virgin Islands. Therefore, consultation was limited to interaction with EPA Region 2 and the FLMs. The coordination with FLMs on long-term strategy development is described in sections 3.3, 8.0 and 9.2 of this plan.

Reasonable Progress Goals
For each Class I area within the State/Tribe, 40 CFR 51.308 (d)(1) requires the state (in this case, EPA for the Virgin Islands) to establish reasonable progress goals (RPG) (expressed in deciviews) that provide for reasonable progress towards achieving natural visibility. The EPA released guidance on June 7, 2007 to use in setting reasonable progress goals. The goals must provide improvement in visibility for the most impaired days, and ensure no degradation in visibility for the least impaired days over the SIP period.  
Under 40 CFR 51.308 (d)(1)(iv), consultation is required in developing reasonable progress goals. As discussed in Section 3, because of the Virgin Islands' remote location, there is little to no impact from the Virgin Islands on visibility on other Class I areas or impacts of other states on Virgin Islands National Park. As such, the Virgin Islands did not need to be consulted during the establishment of reasonable progress goals for other Class I Areas and other States were not consulted in the development of the RPG for the Virgin Islands. In lieu of coordination with other States, the required discussions about the RPG that occurred between EPA Region 2 officials, EPA Headquarters and the appropriate FLMs will suffice for consultation.
In developing the RPG, the Class I State must also consider four factors (cost, time needed, energy and non-air quality environmental impacts, and remaining useful life). This plan discusses the uniform rate of improvement, the improvement in visibility needed to achieve the RPG for 2018 and how the effect of the planned measures compare to the RPG for the period covered by the implementation plan. 
Calculation of Uniform Rate of Progress 
As a benchmark to aid in developing reasonable progress goals, EPA Region 2 compared the baseline visibility conditions to natural visibility condition in the Class I area and determined the uniform rate of visibility improvement (in deciviews) that would need to be maintained during each implementation period in order to attain natural visibility condition by 2064. Table 7-1 presents the baseline visibility, natural visibility, and required uniform rate of progress for the Virgin Islands National Park. Additional Class I areas are listed in Table 7-1 for comparison. 
Table 8-1: Uniform Rate of Progress Calculation 
                                 Class I Area
                  (2000-2004) Baseline Visibility (deciviews)
                               (20% Worst Days)
                         Natural Visibility Conditions
                               (20% Worst Days)
                Deciview Improvement Needed by 2018 (from 2004)
                   Total Deciview Improvement Needed by 2064
                     Uniform Rate of Improvement Annually
                          Virgin Island National Park
                                     17.02
                                     10.68
                                     1.48
                                     6.34
                                     0.11
                             Acadia National Park
                                     22.9
                                     12.4
                                      2.4
                                     10.5
                                     0.17
                             Brigantine Wilderness
                                     29.0
                                     12.2
                                      3.9
                                     16.8
                                     0.28
                           Everglades National Park
                                     22.3
                                     12.2
                                      2.4
                                     10.2
                                     0.17
                   Hance Camp at Grand Canyon National Park
                                     11.7
                                      7.0
                                      1.1
                                      4.6
                                     0.08

Both natural conditions and baseline visibility for the 5-year period from 2000 through 2004 were calculated in conformance with the alternative, updated method recommended by the IMPROVE Steering Committee. Progress toward the 2018 target will be calculated based on 5-year averages calculated in a nationally consistent manner consistent with EPA's Guidance for Tracking Progress Under the Regional Haze Rule (EPA-454/B-03-004, September 2003) as updated by the alternative method for calculating regional haze recommended by the IMPROVE Steering Committee.
	Based on a base case visibility for the twenty percent worst days of 17.02 DV and a natural background calculated for worst twenty percent days of 10.68 DV, the difference is the Reasonable Progress Goal for 2064: 6.34 DV. According to EPA's Regional Haze Rule, a reduction of 6.34 DV by 2064 is to be accomplished by reducing visibility impairment by at least the rate of uniform reduction in each of the ten-year period covered by the haze plans.  The reduction needed by 2018 from 2004 (14 years for the first period) is calculated thus: 6.34 dv divided by 60 and multiplied by 14 = 1.48dv.

Reasonable Progress Goals for the Class I Area in the State 
In accordance with the requirements of 40 CFR Section 51.308(d)(1), this Regional Haze Plan establishes reasonable progress goals for the Class I area in the Virgin Islands for the period of the implementation plan. The RPG for the Virgin Islands for the twenty percent worst days is 1.48 dv, of which the modeling predicts that visibility will be improved by 0.16 dv for the first period ending in 2018.  
The Clean Air Act's plan for reducing regional haze is twofold:  1) Set the natural visibility target and insure that every 10 years the Virgin Islands has a plan to reduce emissions enough to remain on track to meet the target, and 2) Find all reasonable reductions in emissions that reduce visibility in the protected area. There is a tremendous amount of uncertainty in estimating how much of the Virgin Islands' visibility impairment is due to natural sources, and this plan has used default natural background concentrations, and not tried to adjust for the effects of Saharan Dust or other natural sources. It is likely that the effects of dust from the Sahara Desert overwhelm the effects of local sources on visibility in the Virgin Islands National Plan in a way not seen in any other part of the United States, since the Virgin Islands are nearer to Africa than any other location in the United States.  
So, this plan determines the amount of emission reductions for the first plan period through 2018 based on the application of reasonably available emission reductions from sources that affect visibility in the Virgin Islands National Park on St. John. 
In order to be on track to improve visibility to meet the CAA goal of reaching natural visibility levels, this plan estimates the amount of progress needed by 2018 based on the calculated natural level of visibility. In the case of the Virgin Islands, the emission reductions do not improve visibility to meet the reasonable progress goal line for 2018. However, the amount of progress estimate for 2018 may be an overestimate due to the underestimation of natural visibility conditions. This calculation does not include the full impact of Saharan Dust and other natural emissions that interfere with visibility. Therefore, this plan will focus on obtaining the maximum amount of reasonable reductions. For the next 10 years' plan, EPA and the Virgin Islands Department of Natural Resources will need to develop better estimates of the natural background visibility in the Virgin Islands and adjust the goal accordingly, while still reducing emissions as much as reasonably possible in the meantime. 
To determine the RPG in deciviews, EPA Region 2 prepared emissions data and conducted modeling before and after certain control measures. The control measures reflected in these reasonable progress goals include the Caribbean ECA and the HOVENSA Consent Decree.
Consideration of Other Air Quality Requirements 
40 CFR 51.308(d)(1)(vi) of EPA's Clean Air Visibility Rules requires that reasonable progress goals represent at least the visibility improvement expected from implementation of other CAA programs during the applicable planning period.
As documented in the emissions inventory and long-term strategy sections of this FIP, the modeling that formed the basis for reasonable progress goals in the Virgin Islands National Park could not estimate how other programs required by the CAA reduce emissions. Further information can be found in those sections of this FIP.
Additional Reasonable Controls within Virgin Islands 
A full list of all Federal control measures considered is presented in Section 7. No additional measures are determined to be reasonable for the Virgin Islands. Implementation of any measures that have not been reviewed would be hampered by the compressed schedule that this FIP was developed under. Sources would have little time to comply with any additional control programs recommended in this FIP. Based on timeframes presented by States in their regional haze SIPs, this is minimum amount of time needed for the simplest of additional strategies. It is unreasonable to plan for any additional control in this period.
Visibility Impacts of Additional Reasonable Controls
This section explains assumptions used to model the impact of potential control strategies and describes the combined potential visibility benefits of all the strategies based on CALPUFF modeling and emissions analysis. As with all modeling, emissions estimates and modeling results for 2018 entail uncertainty, and further evaluation may be conducted as part of the SIP report required in 5 years under 40 CFR 51.308(g).
Impacts of BART Controls in the Virgin Islands
As discussed in Section 5, the Virgin Islands have three facilities with units that potentially could be subject to BART. As explained in this FIP, no BART controls beyond existing controls are being proposed or adopted.  
Impacts of other Federal Controls in the Virgin Islands
The implementation of low sulfur fuels in the non-road sector (i.e., marine vessels) under the Caribbean ECA will provide an 85 percent reduction in SO2 and an additional 80 percent reduction in NOx from this source sector. This represents a 0.03 deciview improvement on the twenty percent worst days. 
The settlement reached with HOVENSA, LLC as a result of EPA's Refinery Enforcement Initiative will result in an estimated 15 percent reduction in SO2 and a 16 percent reduction in NOx from the facility. This represents a 0.13 deciview reduction on the twenty percent worst days.
Additional controls that will affect the Virgin Islands are noted in Section 6, but were not included in the modeling for calculating improvement to visibility in the Class I area because of the limited data available on emissions in the Virgin Islands. All of the measures laid out in Section 6 will contribute to reductions in anthropogenic emission that may further improve visibility conditions at the Virgin Islands National Park. 
Results of Best and Final Modeling
The starting point for indicating progress achieved by measures included in this FIP is the 2000-2004 baseline visibility at affected Class I areas. To calculate the baseline visibility for affected Class I areas, using 2000-2004 IMPROVE monitoring data, the deciview value for the twenty percent best days in each year with adequate data were averaged together, producing a single average deciview value for the best days. Similarly, the deciview values for the twenty percent worst days in each year with adequate data were averaged together, producing a single average deciview value for the worst days.
In order to estimate the visibility impacts of the measures discussed above, the EPA combined CALPUFF modeling results and emissions scaling analysis to assess the visibility improvement in the Virgin Islands. Documentation of the CALPUFF modeling is contained in Appendix C. Based on currently available information, this analysis provides an estimate of visibility improvement that could be achieved by 2018 through reasonable measures described above. 
Figure 7-1 illustrates the predicted visibility improvement by 2018 resulting from the implementation of the long-term strategy, where quantifiable. This improvement is compared to the Uniform Rate of Progress for affected Class I areas. The Virgin Islands are projected to fall short of the uniform rate of progress goal for 2018. However, emission reductions from several control methods noted in this section could not be quantified due to the limited emissions information for the Virgin Islands. EPA has included reasonable measures in this FIP. In addition, the emissions are not expected to increase impairment of best day visibility relative to the baseline.
In addition to improving visibility on the 20% worst visibility days, states are also required to protect visibility on the 20% best days at the Class I areas. EPA's calculations show that visibility on the 20% best days is projected to be protected in this first planning period at the Virgin Islands National Park as emission controls will decrease emissions though 2018.



Figure 8-1: Projected Progress Goals and Modeled Progress at Virgin Islands National Park 
Demonstration that the Goals are based on Reasonable Controls
Section 51.308 (d)(1)(i)(A) of 40 CFR part 51 requires that in establishing reasonable progress goals for each Class I area, the Plan must consider the costs of compliance, the time necessary for compliance, the energy and non-air quality environmental impacts of compliance, and the remaining useful life of any potentially affected sources, and the SIP must include a demonstration showing how these factors were taken into consideration in setting the goal. These factors are sometimes termed the "four statutory factors," since their consideration is required by the Clean Air Act.
Focus on SO2
The conceptual model developed for the Virgin Islands indicated that the dominant anthropogenic contributor to visibility impairment in the Virgin Islands National Park was particulate sulfate formed from emissions of SO2. While other pollutants will need to be addressed in order to achieve the national visibility goals, SO2 is responsible for 19 percent of light extinction on the twenty percent worst days. Therefore, it is reasonable that the additional measures considered in setting the reasonable progress goals described in Section 9.3 required reductions of SO2 emissions. Because of the relative lack of anthropogenic emission sources in the Islands, natural interference with visibility, mostly due to Saharan Dust, predominates, but can't be controlled. 
Contributing Sources
As discussed in Section 1, the Virgin Island's distance from major land areas with major sources of pollution limits the number of sources affecting their Class I area. The emission inventory in Section 4 shows that point sources on St. Thomas and St. Croix are significant sources of SO2 emissions.  Evaluation of these sources' emissions for reasonable controls is in section 5. The largest of these sources included potentially BART-eligible sources and the evaluation of BART controls is discussed in section 6. Potential BART controls were either not reasonable or had little impact on improving visibility in the Virgin Islands.

Consultation and Coordination
Regional Planning
In 1999, EPA and affected States/Tribes on the continental United States agreed to create five Regional Planning Organizations (RPOs) to facilitate interstate coordination on Regional Haze SIP/Tribal Implementation Plans. The RPOs and States/Tribes within each RPO are required to consult on the development of emission management strategies directed towards visibility improvement in affected Class 1 areas. The RPOs have also allowed States to pool resources and collectively develop information, such as emissions inventories and modeling, to support regional haze plans. 
Because the Virgin Islands are separated from the continental United States by over a thousand miles, they were not included in any RPO. As a result, this plan was developed without the assistance that the multi-state RPOs typically provide. Because the Virgin Islands government could not leverage its resources by working with other states to lay the groundwork for the their regional haze plan, EPA has opted to work alongside the Virgin Islands' government to prepare this regional haze plan as a Federal Implementation Plan 
State/Tribe and Federal Land Manager Coordination
Section 51.308(i) of 40 CFR requires coordination between States/Tribes and the Federal Land Managers (FLMs). EPA Region 2 has discussed the technical documents developed by EPA Region 2 and included in this FIP with the FLMs. EPA Region 2 has provided agency contacts to the FLMs as required. In the development of this plan, the FLMs were consulted in accordance with the provisions of 40 CFR 51.308(i)(2). EPA Region 2 has consulted via conference call meetings with the FLMs.  The National Park Service has expressed their understanding of the proposed plan and reserves the opportunity to comment during the comment period upon EPA's proposal of this FIP.  EPA, in consultation with the Virgin Islands Government, will respond to comments from the FLM in accordance with 40 CFR 51.308(i)(3). The FLM's comments and EPA's responses will be included in Appendix A of this plan.  
Section 51.308(i)(4) of 40 CFR requires procedures for continuing consultation between the State/Tribe and FLMs on the implementation of the visibility protection program. The Virgin Islands and EPA Region 2 will consult with the FLM(s) on the status of the following implementation items:  
   1. Implementation of emissions strategies identified in the FIP as contributing to achieving improvement in the worst-day visibility.
   2. Summary of major new source permits issued.
   3. Status of actions in the Virgin Islands to meet commitments for completing any future assessments or rulemakings on sources identified as likely contributors to visibility impairment, but not directly addressed in the most recent FIP revision. 
   4. Any changes to the monitoring strategy or monitoring stations status that may affect tracking of reasonable progress. 
   5. Work underway for preparing the 5-year review and / or 10-year revision.
   6. Items for FLMs to consider or provide support for in preparation for any visibility protection SIP revisions (based on a 5-year review or the 10-year revision schedule under EPA's RHR)  
   7. Summary of topics discussion (meetings, emails, other records) covered in ongoing communications between the Virgin Islands Government, EPA and FLMs regarding implementation of the visibility program.
The consultation will be coordinated with the designated visibility protection program coordinators for the National Park Service.
EPA Region 2 and the Virgin Islands will continue to coordinate and consult with the FLMs during the development of future progress reports and plan revisions, as well as during the implementation of programs having the potential to contribute to visibility impairment in the mandatory Class I areas.

Comprehensive Periodic Implementation Plan Revisions 
Pursuant to the requirements of 40 CFR 51.308(a) and (b), EPA Region 2 will implement this FIP to meet the requirements of EPA's Regional Haze rules that were adopted to comply with requirements set forth in the Clean Air Act. Elements of this Plan address the Core Requirements pursuant to 40 CFR 51.308(d) and the BART components of 40 CFR 50.308(e). In addition, this FIP addresses Regional Planning, State/Tribe and Federal Land Manager coordination, and contains a commitment to provide Plan revisions and adequacy determinations.
Section 51.308(f) of 40 CFR requires a State/Tribe to revise its regional haze implementation plan and submit a plan revision to EPA by July 31, 2018 and every 10 years thereafter. In accordance with the requirements listed in 40 CFR 51.308(f) of the Federal rule for regional haze, EPA expects the Virgin Islands government to revise and submit a regional haze implementation plan by July 31, 2018 and every 10 years thereafter.  
In addition, 40 CFR 51.308(g) requires periodic reports evaluating progress towards the reasonable progress goals established for each mandatory Class I area. In accordance with the requirements listed in 40 CFR 51.308(g) of the Federal rule for regional haze, the EPA expects the Virgin Islands government to submit a report on reasonable progress every 5 years following the initial submittal of this FIP. The report will be in the form of a SIP revision submitted 5 years after FIP submission. The reasonable progress report will evaluate the progress made towards the reasonable progress goal for the mandatory Class I area located within the Virgin Islands. All requirements listed in 40 CFR 51.308(g) shall be addressed in the FIP revision for reasonable progress.
Section 51.308(d)(4)(v) of 40 CFR requires periodic updates of the emission inventory. The EPA expects that the Virgin Islands government will update and expand upon the existing inventory.

Commitment to Determine the Adequacy of the Existing Plan
As required by 40 CFR 51.308(h), depending on the findings of the 5-year progress report, required under 40 CFR 51.308(g), the EPA expects that the Government of the Territory of the Virgin Islands will take one of the following actions at the same time the 5-year progress report is submitted: 
(1) If EPA and the Virgin Islands determine that the existing implementation plan requires no further substantive revision in order to achieve established goals for visibility improvement and emissions reductions, the EPA will make a negative declaration that further revision of the existing implementation plan is not needed.
(2) If EPA or the Virgin Islands determine that the implementation plan is or may be inadequate to ensure reasonable progress due to emissions from sources in another country, the EPA will work with the Virgin Islands to assess the amount of impairment from other counties and determine if consultation is needed with the other country or countries.  
(3) If EPA or the Virgin Islands determines that the implementation plan is or may be inadequate to ensure reasonable progress due to emissions from sources within the Virgin Islands, and if reasonable measures are available, the EPA expects that the Virgin Islands will submit a revision to the implementation plan to address the plan's deficiencies within 1 year.
The findings of the 5-year progress report will determine which actions are appropriate and necessary.

SUMMARY OF THE FEDERAL PLAN FOR VISIBILITY IN THE NATIONAL PARK ON ST. JOHN IN THE UNITED STATES VIRGIN ISLANDS
EPA is adopting a Regional Haze Federal Implementation Plan to begin the process for reducing man-made interference with visibility to meet the Federal Regional Haze Rule's goal of not man-made pollution interfering with visibility by 2064.  This FIP also addresses coordination with the Virgin Islands Government and the National Park Service as Federal Land Manager for the Virgin Islands National Park.  EPA expects the Virgin Islands Government to determine if the Plan is working and to update the plan within five years and when the next plan is due in 2018.  
	The EPA used models to calculate how much local air pollution has affected visibility in Virgin Islands National Park, and is implementing this plan to reduce regional haze in the National Park using controls already in place and on the way. 
	Most of the interference with visibility in the Park is due to the effects of dust from the Sahara Desert. However, EPA developed this haze plan by evaluating the reasonableness and effectiveness of reducing emissions based on the amount of air quality improvement that would occur. Emission controls were not rejected because of their impact compared to the effects of natural sources of visibility reduction affecting the Caribbean, but because they would be too expensive, not practicable, or not improve visibility. We cannot count the improvement in visibility from the shutdown of HOVENSA in this Plan, as HOVENSA must give up their air quality permits before we can take credit for up to 1.04 dv in visibility improvement. 
	 
Table 12-1: Virgin Islands National Park Visibility Improvement Due to the FIP
Virgin Islands National Park

       IMPROVE Measured Visibility Impairment in for the 20% worst days
                                  (Deciviews)
                                       
         Modeled Visibility Impairment for 20% worst days (Deciviews)
                             All sources, 4 years
                              Measured Baseline 
                                     17.02
                                     10.14
                         Calculated Natural Background
                                     10.68
                                       -
                            Progress Goal for 2064
                                     6.34
                                       -
                              2018 Goal / Modeled
                                     15.54
                                     9.98
                   Reduction 2004-2018:     Needed / Modeled
                                     1.48
                                     0.16

	In summary, EPA's calculated a progress goal of improving visibility by 1.48 dv.  The air quality improvements from the HOVENSA consent decree and the expected reductions in marine vessel emissions are modeled as a reduction of 0.16 dv. A number of measures, mostly related to motor vehicle emissions, are not quantifiable and no other measures were cost effective on a dollar per ton basis or they did not improve visibility very much.  
	EPA expects that the Virgin Islands' Government will determine if local action can reduce impacts of local sources on visibility in the Virgin Islands. In addition, before the HOVENSA refining facility on St. Croix reopens, it will first need to reevaluate its emissions to determine if reasonable controls should be implemented. The methods in EPA's Regional Haze Rule should be a useful guide to a successful evaluation of reasonable measures.

