An Examination of the MOS Objective Temperature Prediction Model

Abstract

In this study, the performance of the Model Output Statistics (MOS) objective temperature forecasting for Albany, NY, during the period 1975?81 is examined by using various statistical technique. Both paired and unpaired statistical analysis procedures are used to evaluate the performance of the MOS model. Extreme value analysis is also utilized to examine the performance of MOS in predicting the higher observed temperatures. A relatively new statistical technique, called the ?bootstrap? method, is used to evaluate the model performance in simulating the extreme values. The results suggest that the MOS model has a warm bias in simulating the minimum temperatures and a cold bias in simulating the maxiimum temperatures. Finally, casts comprising the best and worst forecasts made by MOS are compared with surface weather maps to develop a picture of the synoptic situations that may be conducive to the production of errors in the MOS model predictions.