Ozone modeling using neural networks

Ozone models for the city of Tulsa were developed using neural network mode ling techniques. The neural models were developed using meteorological data from the Oklahoma Mesonet and ozone, nitric oxide, and nitrogen dioxide (N O2) data from Environmental Protection Agency monitoring sites in the Tulsa area. An initial model trained with only eight surface meteorological inpu t variables and NO2 was able to simulate ozone concentrations with a col-re lation coefficient of 0.77. The trained model was then used to evaluate the sensitivity to the primary variables that affect ozone concentrations. The most important variables (NO2, temperature, solar radiation, and relative humidity) showed response curves with strong nonlinear codependencies. Inco rporation of ozone concentrations from the previous 3 days into the model i ncreased the correlation coefficient to 0.82. As expected, the ozone concen trations correlated best with the most recent (1-day previous) values. The model's correlation coefficient was increased to 0.88 by the incorporation of upper-air data from the National Weather Service's Nested Grid Model. Se nsitivity analysis for the upper-air variables indicated unusual positive c orrelations between ozone and the relative humidity from 500 hPa to the tro popause in addition to the other expected correlations with upper-air tempe ratures, vertical wind velocity, and 1000-500-hPa layer thickness. The neur al model results are encouraging for the further use of these systems to ev aluate complex parameter cosensitivities, and for the use of these systems in automated ozone forecast systems.