Testing the ability of GATORM to predict the weather and air pollution during the SARMAP field study

The prognostic air quality / meteorological model, GATORM, was applied to study weather and air pollution during the August 3-6, 1990 SARMAP field campaign in Northern and Central California. The model was previously applied to the SCAQS field campaign in Southern California. The purpose of the current application was to test the prognostic ability of the model in the absence of nesting, nudging, and spinning up and to compare results without nesting with those with nesting. Model predictions of gas and meteorological parameters were compared with observations at 241 surface sites. Meteorological parameters were compared at 45 upper-air sites. In the absence of nesting, the gross error in near-surface temperatures, normalized over all 8600 day and night observations in the model domain, was less than 1% of observed kelvin temperatures. The normalized bias was effectively 0%. The gross error in near-surface relative humidity, normalized over all 5600 day and night observations in the domain was less than 30% of observed relative humidity values, and the bias was -4%. Most errors arose at a few stations. The gross errors in near-surface ozone mixing ratios, normalized over all 2900 measurements above 50 ppbv during the 4-day period, was near 30% with a bias of near -30%. This statistic indicates a general underprediction of ozone although time-series and peak ozone predictions were accurate at most sites. Time series plots of temperature and relative humidity especially, were strikingly accurate at a majority of sites. Prediction accuracy was obtained by careful initialization of temperatures, pressure gradients, and humidities from observations, special attention to soil moisture, and intensive radiative calculations. Comparisons of results from the baseline simulation with those from the nesting case will be shown.