Evaluation of the Surface Prognoses of Cyclones and Anticyclones of the JMA and FNOC Models over East Asia and the Western Pacific during the 1983 Mei-Yu Season

Abstract

This study compares the systematic errors of 36-h surface cyclone and anticyclone forecasts for two operational numerical weather prediction models over East Asia and the western North Pacific Ocean: the U.S. Navy's Operational Global Atmospheric Prediction System (NOGAPS), and Japan Meteorological Agency's Fine-mesh Limited Area Model (JFLM). The study is carried out for the 1983 Mei-Yu season (May?July), which is the wettest season over East Asia based on nontyphoon-produced rainfall. All available 0000 and 1200 GMT forecast runs are evaluated against an independent dataset of subjective analysis produced operationally by the Central Weather Bureau, Taipei. The mean position errors, mean central pressure errors and forecast skill indices for both cyclones and anticyclones in the NOGAPS and JFLM models are examined. Both NOGAPS and JFLM models are more likely to underforecast than to overforecast the existence and/or genesis of both cyclones and anticyclones. However, over the Tibetan Plateau and its vicinity, both models tend to overforecast the existence and/or genesis of cyclones. They also forecast both cyclones and anticyclones too slow and too far to the north. Diurnal variations in central pressure errors suggest that the error source is the lack of radiation processes in the JFLM and too strong a diurnal cycle of radiation processes in NOGAPS. Also, the failure to treat adequately the bulk effects of cumulus convection seems to be primarily responsible for the poor forecasts of oceanic cyclone development.