An Object-Based Approach for Quantitative Verification of Quasi-Stationary Band-Shaped Precipitation Systems, “Senjo-Kousuitai,” Forecasts

Abstract

This study introduces an object-based approach for assessing the prediction accuracy of quasi-stationary band-shaped precipitation systems (QSBPSs), known as “Senjo-Kousuitai.” Unlike a conventional grid-based verification approach, the proposed method integrates the unique characteristics of QSBPSs, including their shape, size, and duration, to provide a comprehensive evaluation of forecast accuracy. Quantitative prediction verification was performed by analyzing the features of QSBPSs based on the error between the observation and numerical weather prediction using the operational forecast model. The proposed approach was applied to Japan and its surrounding areas from June to September 2024, during which 41 QSBPSs cases were identified based on observational data. A total of 157 QSBPSs cases, including multiple forecasts of the same observed events, were detected using the local forecast model operated by the Japan Meteorological Agency with a 2 km horizontal resolution and 18-hour forecasts eight times daily. The proposed approach was applied to these observed and predicted cases to evaluate prediction accuracy. In addition, a new index, the “SCS: Senjo-Kousuitai (QSBPSs) Composite Score,” was introduced to evaluate forecast performance across the entire period. The preliminary results obtained by applying this method indicate its effectiveness in quantitatively assessing the accuracy of numerical weather predictions for QSBPSs.