MEMORANDUM TO:OTAG Regional and Urban Scale Modeling Workgroup Members FROM:Michael Koerber SUBJECT:Summary of OTAG Modeling Conference Calls and Meetings for the Weeks of May 20, May 27, and June 3 DATE:June 11, 1996 During the weeks of May 20, May 27, and June 3, several OTAG modeling conference calls and meetings were scheduled. A brief summary of these calls and meetings is provided below. Modeling Centers Calls May 23 Conference Call The purpose of this call was to review the progress of the basecase, diagnostic, and sensitivity modeling. MW: Basecase B run is underway. Several diagnostic runs have been completed (e.g., dirty initial concentrations, rainfall, clouds, deposition, and vertical diffusivities). The proxy and the Group B, Group D, and Group E sensitivity runs will be modeled. U.S. EPA: Basecase B run will start in the next day or two. A few diagnostic runs will be performed (e.g., dirty initial concentrations and rainfall). The proxy (through July 11) and, possibly, the Group B sensitivity runs will be modeled. SE: Basecase B run will start in the next day or two. Several diagnostic runs will be performed (e.g., dirty initial concentrations and vertical diffusivities). The proxy may also be modeled. NEMAC: Basecase B run will start soon. No diagnostic runs are planned. The proxy and Sens 5c will be modeled. Two issues related to the air quality data were noted. First, data should reflect a single time zone (EST). Because AIRS data are generally reported as LST, data for the central time zone sites need to be adjusted. Dyntel will provide a program to each Center to make this adjustment. Second, a subset of key sites will be used for generating time series plots for sensitivity/strategy modeling. Comments on the subset of 33 sites (10 in the NE, SE, and MW regions, and 3 in the SW region) were requested. The subjective/objective measures for sensitivity/strategy modeling were reviewed. A "core" set of measures will be prepared for each sensitivity/strategy run. The full list of measures will be prepared for only certain key runs. A draft list of core measures will be distributed to the Centers for comment. It was suggested that the modeling episodes be shortened to save time and to focus the analysis on the key (highest) days. For 1988, U.S. EPA proposed dropping July 12 to 15. For the other three episodes, each Center thought that no more than the last day or two could be dropped. This issue will be discussed further with the Workgroup. May 30 Conference Call The primary purpose of this call was to review the progress of the modeling. Based on the diagnostic analyses completed to date, the MW and SE Modeling Centers were unable to identify any specific model or model input improvements. The next round of basecase modeling will involve primarily the CB-IV update. During the month of June, several model runs were suggested: Basecase C, other diagnostic analyses, and 1990 scenario. June 5/6 Conference Call The purpose of this call was to follow up on the discussions at the June 3 Centers meeting and the June 4 Workgroup meeting. In particular, the action items for reaching closure on the basecase were reviewed. These items are as summarized below. Emissions: U.S. EPA to document BEIS2 calculations 6/14 Meteorology: Wet Deposition (SAI to deliver new code) 6/11 Rainfall (SAI to prepare new 1991, 1995 files) ??? Kv's (ALAPCO and NC to provide aloft profiles) 6/10 (NC to compare SAIMM and RAMS) 6/10 Winds (SE to decide on MET model) 6/10 AQ Data: ESE Issues (ALAPCO to resolve with R. Husar) 6/12 Time Zone (Dyntel to fix) 6/14 Site Classifications (Dyntel to add classifications 6/14 to database; SAI to modify UVPS) Chemistry: CB-IV Update (SAI to deliver new code) 6/12 In addition, several modifications to the post-processing software were identified (e.g., flux calculations, number cell calculations, and generate table of objective measures). These modifications are needed for producing sensitivity/strategy modeling results. ALAPCO, U.S. EPA, and SAI will pursue these modifications. The need to shorten the episodes was discussed again. It was agreed that there would be no change for 1991, 1993, and 1995. For 1988, all days will be modeled for the basecase and final strategy modeling, but only July 1 to 11 will be modeled for most of the sensitivity/strategy modeling. Also, it was agreed that at least the first two days of each episode would not be presented (i.e., ramp-up days should not be used to assess basecase or sensitivity/strategy scenarios). June 3 Modeling Centers Meeting The purposes of this meeting were to review the Basecase B and diagnostic modeling and to decide if the current basecase is acceptable. If it is not, then the Modeling Workgroup needs to determine what needs to be done to reach closure. 1988: Mean observed and predicted (Basecase A) concentrations show good agreement in the MW for all days and in the NE for the first several days, and overprediction in the SE for all days and in the NE for the last half of the episode. These results suggest no overall bias trend. The problem in the SE, in particular, may be due to meteorological effects (e.g., inability of SAIMM to reproduce, or UAM-V to handle, certain features). The dirty initial concentration run shows little effect after the first few days. The performance statistics improved slightly during the first few days. The rainfall run showed small concentration differences and similar model performance statistics. Basecase B reflects changes in emissions and rainfall inputs. The performance statistics are only slightly different. The prediction problem in the SE, especially later in the episode, is still apparent. 1991: Basecase B reflects changes in emissions and various meteorological inputs (kv and winds). The performance statistics are only slightly different. Although the peaks are lower, the overprediction is a little greater. Spatial plots reflect a broad area of elevated ozone in the Ohio River Valley. Scatterplots indicate good agreement between predicted and observed daily peaks. The dirty initial concentration run shows little effect by Day 4. Given that the first few days are used only for ramp-up purposes, a change in the initial concentrations is not considered necessary. The vertical diffusivity run indicates that most of the difference between Basecase A and Basecase B is due to the new kv's (i.e., on the order of 5 ppb to 15 ppb throughout the region). The cloud run shows lower ozone due to higher attenuation factors. The deposition run also shows lower ozone. Although these two changes lead to "better" model performance, no modification to the current cloud attenuation factors and deposition velocities appears to be warranted. The rainfall run shows a small change in ozone over a limited spatial area. 1993: Basecase B reflects changes in emissions and various meteorological inputs (kv and winds). The performance statistics are slightly different. The peaks are higher (in Atlanta), and the overprediction is a little greater. Spatial plots reflect high ozone throughout the SE, especially in Atlanta. Scatterplots indicate good agreement between predicted and observed daily peaks in the MW and NE and overprediction in the SE and SW. Key findings include major overprediction in urban areas and better model performance in rural areas than urban areas, and RAMS and SAIMM meteorological inputs produce similar results. The SE Modeling Center still wants to perform a more detailed evaluation (time series plots, other species. . .) and focus on individual subdomains. The vertical diffusivity run indicates that most of the difference between Basecase A and Basecase B is due to the new kv's. The rainfall run shows as much as an 8 ppb change on July 23 and 15 ppb on July 28 (rain versus no rain). A concern was raised about whether the wet deposition algorithm was working in Grid A (fine grid). SAI will look into this issue. Several additional diagnostic analyses are planned (SAIMM winds, dirty initial concentrations, modified SAIMM kv's, minimum wind speed = 1.5 m/sec, CB-IV update, Atlanta subdomain modeling). 1995: Basecase B reflects changes in emissions and various meteorological inputs (kv and winds). As with the 1991 and 1993 modeling, the new kv's were responsible for most of the difference between Basecase A and Basecase B predictions. Performance statistics are only slightly different. Although the peaks are lower, the overprediction is a little greater. Scatterplots indicate good agreement between predicted and observed daily peaks in the NE and SE and overprediction in the MW and SW. Time series plots were presented for CO, isoprene, and ozone. The CO plots for three sites in North Carolina show fairly good agreement, especially during the second half of the episode. The isoprene plots indicate underprediction. The ozone plots for three NE sites show good agreement. Ozone predictions in layer 3 were found to be fairly high, which is important for transport assessments. Each of the key model inputs was reviewed: Emissions: The Centers agreed that the Basecase B inventory can be considered the final basecase inventory. Although a few minor problems exist, another round of basecase emissions modeling is unnecessary. The problems will, however, be corrected in the 1990 inventory to be used as the basis for any strategy modeling. Meteorology: SAI will examine the wet deposition algorithm to make sure that the model is working in the fine grid. ALAPCO will discuss with SAI the development of new rainfall files for 1991 and 1993. Vertical diffusivities will be examined further. The SE Modeling Center will determine the appropriate meteorological inputs for the July 1993 episode. No change in the current cloud attenuation factors will be made. Chemistry: The CB-IV update will be tested for all four episodes. A decision to use the updated code will be made by mid-June. Air Quality: The SE Modeling Center has raised several concerns with the air quality file prepared by R. Husar (e.g., site classifications and duplicate sites). ALAPCO will work with R. Husar to resolve these concerns. Dyntel will help correct the time zone problem for the central time zone sites. Finalization of the basecase by July 1 was determined to be a reasonable goal. Another round of basecase modeling will be performed with the following changes: CB-IV update (if delivered in time, and if testing is positive); meteorological revisions (rainfall file, wet deposition algorithm, and, if necessary, kv); and use of final PiG list. May 29 Modeling Workgroup Conference Call The primary purpose of this call was to get status reports on the basecase, sensitivity, and strategy modeling. Upcoming Calls and Meetings The next Workgroup call will be June 19, 1996. The next Workgroup/Subgroup meeting will be June 3 and 4 in Washington, D.C. (The Policy Group will meet on June 5 in Washington, D.C.) Status Reports From Modeling Centers MW: Basecase B run is underway. Several diagnostic runs have been completed (e.g., dirty initial concentrations, rainfall, clouds, deposition, and vertical diffusivities). The proxy and the Group B, Group D, and Group E sensitivity runs will be modeled. Three problems with the sensitivity modeling were noted (i.e., no min. kv is used; top 250 stacks are modeled as PiG, not full list of 300 stacks; and proxy inventory reflects excessive utility growth and excessive SE utility weekend emissions). Due to these problems, the usefulness of the existing sensitivity runs is questionable. U.S. EPA: Basecase B run will start in the next day or two. A few diagnostic runs will be performed (e.g., dirty initial concentrations and rainfall). The proxy (through July 11) and, possibly, the Group B sensitivity runs will be modeled. SE: Basecase B run will start in the next day or two. Several diagnostic runs will be performed (e.g., rainfall, minimum wind speed, SAIMM winds, and Atlanta analyses). The proxy may also be modeled. The status of the CB-IV update was reviewed. An initial Atlanta test run should be completed shortly. The full SUPROXA domain will then be modeled for the July 1993 episode. Documentation is being prepared for the peer review group. The new source code will be provided to the Modeling Centers following completion of the testing by SAI. NEMAC: Basecase B run will start soon. No diagnostic runs are planned. The proxy and Sens 5c will be modeled. At the June meeting, the results of the new basecase modeling, the CB-IV chemistry update, and the diagnostic analyses will be discussed to determine the next step in the modeling. If basecase modeling can be wrapped-up (and model performance is acceptable), then the Modeling Workgroup will be able to move forward with sensitivity and strategy modeling. Additional Status Reports Stakeholder Modeling: TVA had nothing new to report. SCS is continuing to support the SE Modeling Center. MOG is continuing with its SAQM run. No results are available yet. Data Clearinghouse: North Carolina has received the necessary funding from U.S. EPA. A letter of intent was given to MCNC to allow it to purchase equipment. The clearinghouse should be operational by the end of June. Animation: The Photochemical Modelers have decided to use PAVE for animating model results. MCNC is planning to deliver PAVE to the Modeling Centers in June. Availability of Modeling Results: The Workgroup Co-Chairs will each have a full set of the standard model documentation. This material is too voluminous to distribute to the entire workgroup. Select portions will be handed-out at the Workgroup meetings. Several options were discussed for making the full set of modeling results available (e.g., MCNC, Walcoff, U.S. EPA Regional Offices). A list of modeling reports will be prepared and posted on the TTN. Documentation: Emissions reports are being prepared by EHPechan (National Inventory) and ALAPCO (EMS-95 Modeling). Modeling reports will be prepared by each Modeling Center (diagnostic and sensitivity modeling) and U.S. EPA's contractor (basecase modeling). June 4 Modeling Workgroup Meeting Mike Koerber prefaced the presentation of the Basecase B modeling results with three slides concerning (1) the expectations of the OTAG model performance evaluation; (2) the key factors in assessing regional model performance; and (3) the acceptability of OTAG model performance. A goal of the June 4 meeting is to decide if the Workgroup is comfortable with model performance. Questions to be answered include the following: Is another round of basecase modeling necessary? Are additional diagnostic analyses needed? What must be done to bring closure to the basecase modeling? Each Modeling Center then presented the following information for its episode: OVERVIEW OF EPISODE Emissions (bar charts with VOC and NOx totals) STATISTICS (Base B2) Tabular summary of ave accuracy of peak, normalized bias, and normalized gross error for SUPROXA and regions (all days) SCATTERPLOTS (Base B2) Full Data Set: Plot for SUPROXA and regions (1 to 2 key days) Key Sites: Plot for SUPROXA and regions with urban and rural sites identified (1 to 2 key days) SPATIAL HATCHPLOTS (Base B2) Plot for SUPROXA (1 to 2 key days) TIME SERIES PLOTS (Base B2) (Optional) SPATIAL ISOPLETH DIFFERENCE PLOTS (Base B2 - Base A2) Plot for SUPROXA (1 to 2 key days) The material is discussed above in the summary of the June 3 Modeling Centers meeting. CB-IV Update: Gary Whitten reviewed SAI's work to update the CB-IV chemical mechanism. He noted that the new chemistry appears to be "better" because it has lower reactivity. Initial testing shows little difference with one-day smog chamber data, but should have more effect for a multi-day simulation. Other Modeling: Cyril Durrenberger (TNRCC) reviewed Texas' regional scale modeling with UAM-V in the Gulf area. Ralph Morris (Environ) reviewed Ohio's plans for local urban scale modeling and possibly regional scale modeling. UAM-X, with its source apportionment algorithm, would be used for the regional scale modeling. Tom Tesche (Alpine Geophysics) reviewed the status of MOG's modeling work. He noted that RAMS and MM5 results for the July 1991 episode are comparable. SAQM runs are under way and should be available shortly. In conclusion, the following summary of the basecase modeling was provided: o The Modeling Workgroup is committed to the modeling process, which calls for first getting a good basecase. o The Modeling Workgroup also recognizes that OTAG is not a research program; the Workgroup is here to provide answers. o Initial basecase modeling results are encouraging, but some troubling features are noted. o Diagnostic analyses indicate the model is responding as expected, and do not appreciably change model performance. o What we know now suggests that model performance is reasonable, but a few issues still need to be resolved (emissions, air quality data, chemistry, and meteorology). o The Modeling Workgroup plans to resolve these issues in the next few weeks and produce a final basecase by July 1. A joint session with the Implementation and Strategies Issues (ISI) Workgroup was held to review the results of the initial sensitivity modeling. The results were presented primarily for informational purpose. Due to problems with the proxy inventory and the lack of a final model basecase, it is premature to reach any final conclusions. It is worth noting, however, that these results are directionally consistent with previous modeling studies. The Modeling Workgroup would prefer to wait for BASE1 and the final basecase before proceeding with additional sensitivity runs. Other options exist if the ISI Workgroup needs answers sooner (e.g., "patch" the proxy inventory). After some discussion, this issue was not resolved. It will be discussed further at the June 5 Policy Group meeting.