On an Influence of Forecast-Analysis Cycle on the Forecast Performance

Abstract

 The operational hemispheric spectral forecast model was changed in several aspects at Japan Meteorological Agency (JMA). The revised forecast model was used to obtain the initial guess for the analysis in the forecast-analysis cycle as well as the new forecast. The performance of the new model was examined by comparing the forecast results with those by the old model. The new analysis obtained through the new forecast-analysis cycle was used as the initial condition for the new forecast model, while the old routine analysis was used for the old forecast model. Improvements on the performance of the new forecast system were clearly found in the objective skill scores. Subjective examinations of the forecast maps also indicated an improvement of the new model forecasts.
 An experiment was carried out in order to identify which component of the model change contributes to the improvement of the forecast skill. All components of the change in the forecast model have small positive impacts on the improvement of the forecast on average. It was found that the improvement of the analysis through the forecast-analysis cycle most contributes to the improvement of the forecast skill. The reason for the improvement of the analysis is also discussed.