A comparison of nonlinear regression and neural network models for ground-level ozone forecasting

A hybrid nonlinear regression (NLR) model and a neural network (NN) model, each designed to forecast next-day maximum I-hr average ground-level O-3 co ncentrations in Louisville, KY, were compared for two O-3 seasons-1998 and 1999. The model predictions were compared for the forecast mode, using fore casted meteorological data as input, and for the hindcast I-node, using obs erved meteorological data as input. The two models performed nearly the sam e in the forecast mode. For the two seasons combined, the mean absolute for ecast error was 12.5 ppb for the NLR model and 12.3 ppb for the NN model. T he detection rate of 120 ppb threshold exceedances was 42% for each model i n the forecast mode. In the hindcast mode, the NLR model performed marginal ly better than the NN model. The mean absolute hindcast error was 11.1 ppb for the NLR model and 12.9 ppb for the NN model. The hindcast detection rat e was 92% for the NLR model and 75% for the NN model.