R. E. Imhoff

TVA Atmospheric Sciences Department

November 22, 1996

Phone: (205)386-3801

Fax: (205)386-2499




This preliminary proposal is submitted by TVA for review and comments by the otag modeling groups and other interested parties. It should be viewed as a "straw man", it's purpose is to generate discussion of the purpose and methods of the proposed analysis But, as the schedule makes clear, time is extremely short for accomplishing this task. Comments will have to be received by November 29 to be considered in the revised proposal.

Back to Contents


The purpose of this proposed project is to evaluate the performance of the UAM-V model as applied in OTAG as compared to the same model with a finer grid resolution. Specifically to answer the following questions:

Is the performance of the model significantly different for a 4 km grid size. Does the model compare better to measured data for the same period? Are the results significantly different for the two grid sizes?

  1. When sensitivity control strategies are modeled with the two grid sizes is there a significant difference in the benefits and disbenefits outside the fine grid region?
  2. When a sensitivity control strategies are modeled with the two grid sizes is there a significant difference in the benefits and disbenefits within the fine grid region?
  3. Is there a bias in the relative benefits of NOx and VOC control strategies between the two grid sizes.

Back to Contents


OTAG's mission is to evaluate the effects of interregional transport of ozone and its precursors. An unknown factor is the effect of using a 12 km grid. Can we believe the OTAG modeling results for the 12 km grid in urban, suburban, and rural areas? Does any inaccuracy in modeling urban areas affect transport at downwind locations?

We know that there are factors which should make finer grid resolution simulations more accurate in some instances. With larger grid sizes:

  1. Emissions from dissimilar areas are mixed together yielding mixtures that do not actually occur. This could result, for instance, in too great a dilution of NOx with VOC's.
  2. There is an inability to simulate meteorological phenomena whose scale depends upon finer-than-12 km resolution. Examples are terrain-induced flows and lake/sea breezes.
  3. Artificial dispersion of small, high emitting source areas is caused by initial dilution into the 12 km grid cell.

The basic premise of this proposal is that grid modeling using a finer grid resolution will more accurately reproduce the chemical and physical parameters actually present.

Back to Contents


Previous modeling has shown that grid resolution can have significant effect on calculated concentrations.

[Journal references go here]

In addition, SAI has already conducted modeling for Southern Company Services and Cinergy Corporation to address the issue of the effect of using a 12 km grid resolution for three SIP domains (Atlanta, Cincinnati, and New England) which are part of the OTAG. Some of the conclusions from the report are (excerpted from pg. 5-1 through 5-4):

Base Case Comparison

Response to Reductions in Anthropogenic Emissions


This effort revealed much about what can be expected within an area covered by a finer grid. The inclusion of three regions allowed for comparison of effects in different locales. However, this modeling did leave unanswered the question about what is the effect of the fine grid on regional concentrations.

Back to Contents


At the October, 1996 OTAG Modeling Centers meeting, it was proposed that a regional fine grid model be run. Since TVA was already planning to conduct a similar modeling exercise on a regional domain around Nashville Tennessee based on the OTAG grid, it was proposed to see if the two efforts could be merged.

The TVA modeling is intended to furnish a test data set to evaluate the performance of various models and techniques against the Nashville SOS field study data. The OTAG July 1995 episode falls entirely within the Nashville SOS study period. The study has data from a network of surface stations and six aircraft making measurements specifically for diagnostic evaluations of models. Five meteorological profiles were operated throughout the study, and work was done to characterize the emissions during the study period. A high priority for the model inputs is to make them realistic so that model performance will be a function not of the input data, but as much as possible of the accuracy of the model. The modeling is be performed on a nested 4 km grid within the OTAG fine grid. Originally, this effort was to have followed the completion of the OTAG modeling.

The OTAG ultra-fine (UF) grid modeling should be as similar as possible to the rest of the OTAG modeling protocol. To be as directly comparable as possible to the OTAG modeling, meteorology and emissions input preparation should be done according to the OTAG protocol. However, we have learned improved techniques for processing the meteorology and emissions since the OTAG protocols were first set down. Also, the Nashville UF domain has hourly profile data to about 4 km height above the surface from five profilers stationed throughout the domain. Significant discrepancies between continuous emissions monitoring (CEM) data in the stacks of large point sources and the OTAG 1995 emissions inventory have been pointed out. Should the OTAG UF grid modeling use these improvements?

TVA believes that the model inputs should be made as accurate as possible - that all practical improvements in the input data be incorporated into the fine grid modeling. The expected increase in the accuracy of the model results due to improved inputs does not compromise the comparison with the main OTAG modeling. The improved inputs will also be used with the OTAG fine grid, and the UF grid results will be compared to these.

The utmost priority, however, must be given to completing the work - including the report - by the end of the OTAG process. Thus, a balance must be struck between accuracy of inputs and the schedule. This draft proposal should function as a discussion document for those engaged in the UF grid model work and those interested in the results of the work.

The remainder of this proposal will give the methods, responsibilities, and schedule for the Nashville UF grid modeling.

Back to Contents



A nested ultra-fine grid (UFG), approximate 4 km, will be placed on the OTAG 12-km grid (figure 1). The grid will consist of about 126 1/18 degree longitude cells (East-West) and 95 1/27 degree latitude cells (North-South) The Southwest corner of the UFG domain will be at 89W 34N, and the Northeast corner will be at 82W 37.5 N. Figure 2 shows the domain boundary relative to the FAA air traffic control areas for airports near large cities in the region. The domain will include the urban areas Nashville, Chattanooga, Knoxville, and several large point sources. Large urban areas near the domain, but outside include Memphis, Atlanta, and Charlotte.

The vertical layers for the UFG will be the same as for the OTAG fine grid. The met modeling will be done with more layers, and modeling subsequent to the OTAG grid resolution modeling can include as many layers as are needed. Here, since we are concerned mainly about the effect of horizontal grid resolution, the vertical layers will not be changed.


The UAM-V model will be run for the 1995 OTAG episode, July 7-18, 1995. The model will be run with a 2-way nest for the OTAG fine grid and the UFG and a 1-way nest for the OTAG coarse and fine grids. This means that changes in concentration caused by the UFG will be "communicated" to the OTAG fine grid. The fine grid concentrations will be different than for a normal OTAG simulation. The OTAG coarse grid, however will not be "aware" of the changes in concentration caused on the fine grid. The coarse grid is used to give boundary conditions to the fine grid. This means that the effect of the UFG can be evaluated on the fine grid, but not the coarse.

The model will be run with the 4 km resolution area source emissions and meteorology on the UFG. The chemistry and model parameters will be the same as for a normal OTAG run, with the possible exception of the PiG. The preferred option is to run without the PiG to eliminate the effect of the interaction of the PiG with the grid size, as was seen in SAI's earlier grid-size modeling. This interaction occurs since the release of material from the PiG system is governed by the relationship between the size of a puff and the grid cell that contains the puff. Smaller grid cells mean that the material in the puffs is released more quickly, and at different locations than in a larger grid run. This displacement of emission releases creates "noise" when comparing the grid cell concentrations that is not directly related to the physics and chemistry occurring in the grid system itself. The changes in concentration caused by the PiG system in OTAG has already been evaluated by TVA and others.

The model will be run by TVA on their DEC ALPHA workstation. The runs to be made are given in the Table 1. One option would be to make separate runs for low-level and elevated NOx releases. However since the emphasis is on the effects on a regional basis, the additional runs are probably not needed.

RUN IDSmallest Grid Size EmissionsEmissions Description
G04base495basD3 revised 95 base emissions
G04s5e495basD3-sens5e 30% low-level NOx and 60% elevated NOx reduction
G04s5d495basD3-sens5d 30% anthropogenic VOC reduction
G12base1295basD3 revised 95 base emissions
G12s5e1295basD3-sens5e 30% low-level NOx and 60% elevated NOx reduction
G12s5d1295basD3-sens5d 30% anthropogenic VOC reduction


The RAMS meteorological model will be run for the entire OTAG domain with a nested 4 km grid matching that for the UAM-V model runs. The hourly-average data from the five profilers around Nashville will be used to weakly nudge the simulation. Too strong a FDDA could result in distorted meteorology far from the profilers. The modeling will be performed by the Wisconsin Department of Natural Resources, possibly on their workstation. The preliminary estimate of the run time for the simulation on their workstation is 30 days; if we need the results faster, then a faster workstation will have to be found.


The emissions modeling for the 4 km grid will be performed by USEPA and their contractors. The 1995 emissions will be derived. Known problems in the data will be corrected as possible in the time available. The treatment of the sources and their emissions will be done as closely as possible to the OTAG protocol.


Analysis of the results and reporting will be done by a TVA contractor. The analyses to be performed will determine the effect of the UFG on fluxes and concentrations at distances from the UFG. In some ways the analysis could be viewed as the inverse of the analysis for the shrinking area and rollout geographic control that OTAG is performing. In those runs the emphasis is determining the effect of a change of emissions in an area which is a superset of the target area. The target area being that area for which the changes in concentration are evaluated. Some of the same techniques can be used to examine the effect of the UFG on regional concentrations. Suggested metrics are given in Table 2.

Average daily flux along UFG "downwind" boundary with both the UFG and 12 km modeling. Other flux planes may be placed further downwind on a case by case basis.
Maximum difference in concentration as a function of distance from the UFG.
The list of metrics for the rollout and shrinking area runs applied within the UFG Peak 1 hour
# grid cells > 80, 100,120,140,160
Max dif <-12, -12 to -6, -6 to -4, -4 to -1, -1 to +2, 2 to 4, 4 to 6, 6 to 12, >12
# grid cells > 12 ppb decrease and Base 1 >80, > 100, and > 120
# grid cells > 4 ppb decrease and Base 1 >80, > 100, and > 120
# grid cells > 4 ppb increase and Strat >80, > 100, and > 120
Weighted average ozone decrease
Peak 8-hour average
No. grid cells >80, >100, >120, >140, > 160 ppb 8-hour average
Max dif <-12, -12 to -6, -6 to -4, -4 to -1, -1 to +2, 2 to 4, 4 to 6, 6 to 12, >12 ppb 8-hour average


The schedule for the work is proposed as follows:

TaskNov DecemberJanuary February
252 9162330 61320 27310 1723
Met ModelingXX XXX X
Emissions ModelingXX XXX X
UAM-V Modeling XXX
Analysis of Results XX XX
Draft Report X

Figure 1. Proposed OTAG ultra-fine grid

Figure 2. Proposed OTAG ultra-fine grid relative to major metropolitan areas

Back to Contents

Submit your comments, feedback, questions, and ideas pertaining this page. Your input will be automatically added to the existing annotations. In order to add a new comment, you must be registered with the OTAG/AQA Peoples Page.