December 26, 1991 MEMORANDUM SUBJECT: Information Copy of El Paso-Juarez PM-10 Modeling Scoping Study FROM: William G. Laxton, Director Technical Support Division (MD-14) John Calcagni, Director Air Quality Management Division (MD-15) TO: A. Stanley Meiburg, Director Air, Pesticides & Toxics Division, Region VI (6T) This is in response to your memorandum of November 27, 1991. The Model Clearinghouse has completed its review of the subject PM-10 Modeling Scoping Study and has the following comments. We agree with the contractor's (Systems Applications International, SAI) recommendations on selection of models for use in PM-10 SIP development in this airshed. The rationale provided supports the selection of the Urban Airshed Model (UAM) for episode modeling, the Diagnostic Windfield Model (DWM) for preparing windfield data for use in UAM, and the Chemical Mass Balance (CMB) model for receptor modeling. The report is generally well written and professional. However, the success of the SAI proposal is dependent on a number of assumptions/conclusions not all of which can be evaluated from the information available at this time. For example, one of the critical references, which is cited several times in the SAI report, is a personal communication (Enfield and Church, 1991); alternative references to written documentation should be provided for these citations. An assessment of the need for "hot spot" modeling should be included, especially since hot spot modeling is not possible with UAM. A better description of the intended receptor model/dispersion model reconciliation process, based on EPA guidance, is needed and should be included in the modeling protocol. Finally, a performance evaluation of the reconciled dispersion model using an independent data set is appropriate and should also be included in the modeling protocol. More detailed comments on these and other issues are being sent directly to Jim Yarbrough. If you have questions or need additional information, please contact Desmond Bailey at FTS 629-5248 or me. Attachments cc: G. Blais T. Coulter D. Wilson J. Yarbrough Model Clearinghouse Comments on Modeling Program for PM-10 State Implementation Plan Development for the El Paso/Ciudad Juarez Airshed Issue Conclusion or Assumption Comment Secondary particulate formation The conclusion is based, in part, on is not significant and, thus, PM-10 analysis of denuder samples from one fractions of sulfate and nitrate site for the December 1990 episode. can be attributed to their sources Other than the lack of other sup- by a linear rollback calculation porting material there is no reason based on SO2 and NOx emissions. to doubt the validity of the (pp. 6, 33) conclusion. Vehicular, biomass combustion, This citation refers to a personal unpaved roads and smelters are communication which should be the primary emission sources documented. contributing to the PM-10 problem (Einfield and Church, 1991). (p. 10) High PM-10 episodes in wintertime If this is not the case then bias annual average PM-10 alternative strategies will be concentrations. Thus, control needed to assess attainment of the strategies to attain the short- annual standard. UAM is not suited term NAAQS will also ensure for long-term (annual average) attainment of the annual standard. analyses. (pp. 10, 33) Source apportionment based on As stated (p. 33), if CMB results analysis of short-term samples do not confirm this assumption, then will be representative of the it may be necessary to employ a long- higher PM-10 concentration events term dispersion model ... The that influence the annual average. resources required ... however, would (p. 33) be considerable. UAM will be modified to accept more This should not be a problem. than one aerosol species and thus allow determination of source contributions without multiple simulations. (pp. 19, 33) Point, area and mobile source This is probably the most uncertain emission inventories, particularly element in the proposal. Detailed for Juarez, will be developed and procedures should be prepared to will be of a quality demanded by ensure an accurate and complete highly detailed, gridded dispersion inventory. This is especially crit- models. (pp. 36-37) ical, given the problems encountered in the past (i.e., by Alliance Technologies). Maximum advantage can be taken of Early on coordination with the UAM the ongoing UAM applications to ozone application will be necessary the El Paso ozone problem. to ensure maximum use of information (pp. 32, 34) and intermediate products (e.g., grids) for both applications. The ongoing assessments of Quality meteorological data is meteorological data quality will critical. As stated (p. 38), if show that data are sufficient and data quality is poor ... then acceptable for use in episode and another intensive field study may long-term modeling, as necessary. be necessary. (pp. 30, 38) DWM performance for the El Paso Experience gained from the ozone region will be improved through application should be useful in this full use or better selection of task. input parameters (Douglas, 1991). (p. 37) Receptor Modeling/Sampling Issues 1. Source profile data acquisition On p. 23, last paragraph, it is stated that "each source class is tested to determine a representative chemical species profile ..." On p. 36, 3rd paragraph, it states that "it is recommended that region-specific source profiles be developed for El Paso ..." On p. 37, 1st paragraph, it is stated that "EPA and SEDUE are considering joint stack testing of major sources in the airshed to supply additional CMB ... information." However, on p. 24, 2nd paragraph, there is the implication that existing source profile libraries will be used in CMB analysis. The importance of site specific source fingerprints cannot be overemphasized. Library profiles, available through EPA's Speciation Data System (SPECIATE) or otherwise, are usually dated, non-site specific, and may be of poor or questionable quality. To the extent possible, all significant sources or PM-10, especially fugitive and area sources, should be carefully speciated in order for CMB to apportion properly. The intention to do this should be detailed in Task 8 (p. 43). 2. Ambient sampling On the top of p. 33, reference is made to "both the fine and coarse PM10 fractions." The cutpoints for these fractions, however, is not specified in the report. Presumably, the cutpoints will be at 2.5 and 10æm. Extreme care should be taken to assure that accurate and consistent flow rates are maintained in the dichotomous samplers to attain the desired cutpoint. This dichotomous sampling should be useful in assessing the importance of dry deposition in the El Paso airshed; particles in the 2.5 - 10æm range settle 20-40 times as fast as those in the 0 - 2.5æm range. On p. 34, 1st full paragraph, it is suggested to modify UAM to output 12- hour averages. Obviously, this would only be appropriate if ambient samples are 12-hour. In most SLAMS/NAMS systems, PM-10 is sampled for 24-hour periods. This apparent discrepancy should be clarified. If secondary particle formation is considered to be important, great care must be taken in sampling; 80% of NO3-1 is typically volatilized in conventional sampling processes. 3. Dispersion model/receptor model reconciliation As discussed on p. 35, reconciliation should be part of the overall performance evaluation process. Indeed, this is stipulated in EPA's PM-10 SIP Development Guideline (June 1987; EPA-450/2-86-001). A reconciliation protocol should be established and detailed as a task following #9: CMB Application (p. 43). Guidance for such a reconciliation effort may be found in EPA's Protocol for Reconciling Differences among Receptor and Dispersion Models (March 1987; EPA-450/4-87-008). 4. Quality control/quality assurance To add emphasis to the discussion on p. 39-40, great care should be taken to assure consistent quality control in the sampling and analyses performed for CMB calculations. In particular, the same analytical methods used for assaying elements in the ambient samples should be used for those in the source profiles. 5. Miscellaneous On p. 36, 4th paragraph, it is unclear how CMB is to be used to "demonstrate attainment of the ... standard." Such application connotes a predictive capability of the model. CMB is not a predictive model in the sense of dispersion models; it is classically used to attribute emissions to particular sources (or source categories) and as such is useful for refining the emission inventory used in a dispersion modeling analysis.