                                August 6, 2021
                                       
MEMORANDUM

SUBJECT:     San Joaquin Valley, CA 1997 24-hour PM2.5 Nonattainment Area, Manteca Monitoring 
      Station: 2018 98[th] Percentile and 2018-2020 Design Value Calculations 

FROM: 	Dena Vallano, Region 9 Air Quality Analysis Office
		Brett Gantt, Office of Air Quality Planning and Standards

TO:                Docket EPA-R09-OAR-2021-0261
                                       
This memorandum addresses a data handling discrepancy in the EPA's Air Quality System (AQS) database and explains the methodology used by the EPA to determine the Manteca monitoring station (AQS ID: 06-077-2010) 2018 98[th] percentile concentration and 2018-2020 design value for the 1997 24-hour PM2.5 National Ambient Air Quality Standards (NAAQS). For this analysis, the EPA obtained data from data queries performed on July 23[rd], 2021 (AQS Combined Site Sample Values Report AMP355, Raw Data Report AMP350, and Design Value Report AMP480).

The PM2.5 98[th] percentile concentration and design value calculations listed in Table 1 compare the values currently in AQS with EPA calculations that follow the data handling procedures described in 40 CFR part 50, Appendix N Interpretation of the National Ambient Air Quality Standards for PM2.5. Specifically, the 24-hour PM2.5 averages measured at the Manteca monitoring station in calendar year 2018 were analyzed to calculate an annual 98[th] percentile concentration. This 2018 98[th] percentile concentration was then added to the 2019 and 2020 98[th] percentile concentrations from the Manteca monitoring station and averaged in accordance with 40 CFR part 50, Appendix N data handling procedures. These resulting 2018-2020 design values were then compared to the 1997 24-hour PM2.5 NAAQS of 65 ug/m[3]. Appendix N describes the data handling conventions and computations necessary for determining when the NAAQS for PM2.5 are met, specifically the primary and secondary annual and 24-hour PM2.5 NAAQS specified in §50.7, 50.13, and 50.18. Appendix N, §3.0(c) specifies how 24-hour average concentrations from submitted hourly PM2.5 concentration data will be computed in AQS. The validity of each 24-hour average concentration is determined by examining whether sufficient hourly values for the 24-hour period are available; if not, the 24-hour average concentration may still be considered valid if a substitution test is passed. Appendix N, §3.0(d) and (e) describe how site-level data are incorporated. Appendix N, §4.2 describes how the calculated number of creditable 24-hour samples for a given data year then determine whether the year of data meets data completeness requirements. Appendix N, §4.5 specifies the procedure for determining which value in the sorted series of daily values represents the 98[th] percentile concentration for that year.




Table 1. 2018 98[th] percentile concentrations and 2018-2020 design values for the 1997 24-hour PM2.5 NAAQS at the Manteca monitoring station (AQS ID: 06-077-2010) in San Joaquin Valley, CA
            AQS values not excluding exceptional events (ug/m[3])
       EPA-calculated values not excluding exceptional events (ug/m[3])
         EPA-calculated values excluding exceptional events (ug/m[3])
                            2018 98[th] percentile
                            2018-2020 design value
                            2018 98[th] percentile
                            2018-2020 design value
                            2018 98[th] percentile
                            2018-2020 design value
                                     96.9
                                     72[a]
                                     84.6
                                     68[b]
                                     84.6
                                  59[c][,][d]
[a] The 2019 98[th] percentile is 26.8 ug/m[3] and the 2020 98[th] percentile is 91.6 ug/m[3].
[b] The 2019 98[th] percentile is 26.8 ug/m[3] and the 2020 98[th] percentile is 91.6 ug/m[3].
[c]  Based on the design value calculation methodologies described in 40 CFR part 50, Appendix N, §4.2(b), the Manteca (AQS ID: 06-077-2010) 2018-2020 design value is considered invalid due to incomplete data in the 1[st], 2[nd], and 3[rd] quarters of 2019.
[d] The 2019 98[th] percentile is 26.8 ug/m[3] and the 2020 98[th] percentile is 66.9 ug/m[3].

For the Manteca monitoring station, the dates of November 12, 2018 and November 20, 2018 have 24-hour averages, 96.9 ug/m[3] and 97.0 ug/m[3] respectively, that are incorrectly considered valid creditable samples in AQS. AQS currently does not include the 1997 24-hour PM2.5 NAAQS as a Pollutant Standard. AQS reports for 24-hour PM2.5 data are only available for the 2006 24-hour PM2.5 NAAQS as a Pollutant Standard, and the rules applied within AQS to determine completeness, validity, and statistical calculations are those associated with the 2006 24-hour PM2.5 NAAQS. The EPA's interpretation of Appendix N is that the 1997 24-hour PM2.5 NAAQS of 65 ug/m[3] rather than the 2006 24-hour PM2.5 NAAQS of 35 ug/m[3] should be used when performing calculations to determine compliance with the 1997 24-hour PM2.5 NAAQS. Appendix N, §3.0(c) describes how to determine whether a 24-hour average value is creditable. If a 24-hour period is missing seven or more hours, it may be considered valid if, "after substituting zero for all missing hourly concentrations, the resulting 24-hour daily value is greater than the level of the 24-hour PM2.5 NAAQS". Both November 12, 2018 and November 20, 2018 missed seven or more hours of data due to instrument operational issues. Specifically, the missing data are flagged in AQS as "AH - Sample Flow Rate or CV out of Limits" and "AS  -  Poor Quality Assurance Results" on November 12, 2018 and November 20, 2018, respectively.  When zeros are substituted for the missing data, the resulting 24-hour value for both days is less than the level of the applicable NAAQS, i.e., less than 65 ug/m[3]. Based on the data handling protocols in Appendix N, §3.0(c), these two 24-hour averages should therefore not be considered creditable samples, not considered valid, and not be used to calculate the 98[th] percentile for calendar year 2018. The inclusion of these two 24-hour average values in the data record based on incorrectly using the 2006 24-hour PM2.5 NAAQS of 35 ug/m[3] results in a 2018 98[th] percentile concentration of 96.9 ug/m[3] and 2018-2020 24-hour design value of 72 ug/m[3], prior to exclusion of data impacted by potential exceptional events. The resulting corrected 2018 98[th] percentile concentration is 84.6 ug/m[3] and the 2018-2020 design value is 68 ug/m[3], prior to exclusion of data impacted by potential exceptional events. 

Furthermore, the exclusion of data at the Manteca monitoring station impacted by the August 20-24, 2020 exceptional event results in a final 2018-2020 design value of 59 ug/m[3]. The rightmost columns in Table 1 reflect the EPA's concurrence on a PM2.5 wildfire exceptional event that includes the event days of August 20, 2020, August 21, 2020, August 23, 2020, and August 24, 2020 for the Manteca monitoring station. This exceptional event is described in greater detail in the EPA's July 2021 PM2.5 exceptional event concurrence documents for San Joaquin Valley, CA addressing the August 20-24, 2020 exceptional event, as included in the docket (EPA-R09-OAR-2021-0261). For these four Manteca monitor-days, the 24-hour average PM2.5 concentrations were not counted as exceedances but were included in the data completeness calculations as the exceedances were valid scheduled sample days, consistent with data handling procedures in 40 CFR part 50 Appendix N, §3.0. The resulting, final 2018-2020 design value of 59 ug/m[3] is considered invalid due to incomplete data in the 1[st], 2[nd], and 3[rd] quarters of 2019 based on the design value calculation methodologies described in 40 CFR part 50, Appendix N, §4.2(b).


