Analysis of characteristics and evaluation of forecast accuracy for Super Typhoon Doksuri (2023)

Abstract

Super Typhoon Doksuri is a significant meteorological challenge for China this year due to its strong intensity and wide influence range, as well as significant and prolonged hazards. In this work, we studied Doksuri's main characteristics and assessed its forecast accuracy meticulously based on official forecasts, global models and regional models with lead times varying from 1 to 5 days. The results indicate that Typhoon Doksuri underwent rapid intensification and made landfall at 09:55 BJT on July 28 with a powerful intensity of 50 m s−1 confirmed by the real-time operational warnings issued by China Meteorological Administration (CMA). The typhoon also caused significant wind and rainfall impacts, with precipitation at several stations reaching historical extremes, ranking eighth in terms of total rainfall impact during the event. The evaluation of forecast accuracy for Doksuri suggests that Shanghai Multi-model Ensemble Method (SSTC) and Fengwu Model are the most effective for short-term track forecasts. Meanwhile, the forecasts from the European Centre for Medium-Range Weather Forecasts (ECMWF) and United Kingdom Meteorological Office (UKMO) are optimal for long-term predictions. It is worth noting that objective forecasts systematically underestimate the typhoon maximum intensity. The objective forecast is terribly poor when there is a sudden change in intensity. CMA-National Digital Forecast System (CMA-NDFS) provides a better reference value for typhoon accumulated rainfall forecasts, and regional models perform well in forecasting extreme rainfall. The analyses above assist forecasters in pinpointing challenges within typhoon predictions and gaining a comprehensive insight into the performance of each model. This improves the effective application of model products.

Keywords

• Typhoon doksuri
• Forecast errors
• Typhoon track forecasts
• Typhoon intensity forecasts
• TC-Reduced rainfall forecasts