Comments on Preliminary Proposal: Evaluation of the Effect of the use of a 12 Kilometer Grid on Otag Modeling Results

Comments on Preliminary Proposal: Evaluation of the Effect of the use of a 12

One of the primary purposes of model sensitivity runs, such as the one proposed,
is to aid in the evaluation of the model performance. This process can
help in identifying limitations of the model and insure that the basecase
runs match the measured ozone concentrations for the right reasons. Also,
sensitivity runs can help identify the causes for any biases in the basecase
simulations. Identifying their cause is critical to properly use the model results for policy
decisions. Evaluation of the model results have found several large
scale biases, which include, overestimation of high ozone in the Southern
part of the OTAG domain, underestimation of high ozone in the northern
part of the OTAG domain, overestimation of CO, and not properly simulating
the diurnal cycle of ozone outside of urban regions. Sensitivity analyses
should be conducted to address these large scale biases as well as those
issues listed in the proposal. Grid resolution is only one possible factor that influences
the model performance, others include vertical resolution, errors in transport,
turbulent mixing processes, dry and wet deposition, etc. Investigating
the influence of these other factors through sensitivity runs may provide
more information to OTAG then testing the grid resolution. Also, the impact
of grid resolution can already be investigated to some degree by comparing
the 12km and 36km results, but the influence of processes such as dry deposition
rates have not be analyzed. Therefore, before proceeding with the evaluation
of the effect of the grid resolution, I suggest that the possibility of
running other sensistivity runs examined.


Submitted by Bret Schichtel on 12/3/96 RecID: BretSchichtel

Re: Comments on Preliminary Proposal: Evaluation of the Effect of the use of a 12 Kilometer Grid on Otag Modeling Results

The Southeast modeling center has expended much effort in trying to determine the cause of the overestimation of peak ozone concentrations in the Southeast. You may recall that the SMC funded the development of the modification of the CB-4 chemistry model to incorporate the better isoprene reactions. They have also run the model with both RAMS and SAIMM meteorology, alternative RAMS modeling to increase mid-afternoon mixing heights in the southeast, and 11-layer fine-grid modeling. The revised CB-4 chemistry had the most beneficial effect on the model’s performance, the others did not clearly improve model results. Most of the work done by the SMC has been of a diagnostic nature rather than a sensitivity runs such as you describe. In sensitivity modeling a range of values, bracketing the expected range, for model parameters is input and the effect on the model’s output assessed. The main questions are “How sensitive is the model to these parameters? Do I have to expend a lot of energy getting this right?” In diagnostic model runs, as inputs are improved in response to problems noted in the model’s performance, detailed analysis of the effect of these changes on the model is made. The question here is, “What things is the model doing well or not well and what are the likely explanations for the problem areas? The sensitivity runs that you suggest are certainly very worthy of investigation. The deposition rates in particular have not received much scrutiny in OTAG, and yet as some of your work has made clear is very important to the inter-regional transport of ozone. We have been investigating the lifetime of ozone in the UAM-V model, but have not done any sensitivity analyses for deposition. These types of sensitivity runs should be made by members of the OTAG modeling community. Unfortunately, I think that the majority of the OTAG modeling resources have been expended in “cranking” out the large number of emissions sensitivity and control strategy runs needed rather than investigating model behavior. The proposed regional fine-grid model runs will be diagnostic rather than sensitivity runs. The fine-grid should improve both the atmospheric chemistry and meteorology in the model. The response of the model to these improvements will be assessed against the detailed measured chemistry data available from the 1995 Nashville / SOS field study. Getting information on the effect of 4 km grid resolution by comparing the existing 12 and 36 km grid results, unfortunately has a problem. We know that both the 36 and the 12-km resolutions are unsatisfactory for producing acceptable results in urban areas. The effect of grid resolution on chemistry is non-linear and we cannot extrapolate the results to a finer scale. There are also meteorological phenomena that are not well modeled at either the 12 or 36 km resolution.


Submitted by Robert Imhoff on 12/3/96 RecID: RobertImhoff

---

---