Scholarlink

<p>The present study is undertaken to evaluate the skill of the NCMRWF regional ensemble prediction system (EPS) in simulating intense precipitation events over India, which is run at a convective-scale resolution of ~ 4&#xa0;km. Five heavy rainfall cases that occurred during the monsoon and post-monsoon seasons have been considered. 24-hrs accumulated rainfall indicates that the performance of the model in predicting extreme precipitation is better during the monsoon period compared to the post-monsoon period. The skill of the ensemble mean in forecasting extreme precipitation is dictated by the amount of ensemble spread. A larger spread among the members leaves the ensemble mean with lesser predicting skill at higher thresholds of rainfall. While the probabilistic forecasts give a measure of the uncertainty in the precipitation forecast, the individual forecasts of the members can be vital in assessing the intensity of the rainfall. Verification of station rainfall clearly shows that the EPS forecasts are superior to deterministic forecasts in predicting intense precipitation at all lead times. CNTL exhibits higher RMSE than EPS at all lead times and the EPS forecasts are more reliable during the monsoon than the post-monsoon scenario. The deterministic skill scores like POD, FAR and ETS suggest that the forecast skill of the ensemble mean diminishes with increasing threshold and lead time compared to the deterministic forecast. Further, the over-prediction of precipitation by the explicit convection schemes in the deterministic forecast is considerably reduced in the EPS owing to fewer false alarms in the EPS. Despite the mixed results in the forecast skill of the EPS during monsoon and post-monsoon, the lesser error (CRPS &lt; MAE) in the EPS forecasts and better discrimination skill of events and non-events by the EPS does necessitates the use of a convective-scale ensemble prediction system for prediction of intense rainfall events over the Indian region and thereby substantiates the use of huge computing resources involved in maintaining an EPS.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Assessing the performance of a high-resolution regional ensemble prediction system in predicting intense precipitation events over India

  • 期刊论文

  • 发表日期：

    2025年12月9日

查看全文
  • S. Kiran Prasad,
  • Ashish Routray,
  • Gauri Shanker,
  • V. S. Prasad

摘要

The present study is undertaken to evaluate the skill of the NCMRWF regional ensemble prediction system (EPS) in simulating intense precipitation events over India, which is run at a convective-scale resolution of ~ 4 km. Five heavy rainfall cases that occurred during the monsoon and post-monsoon seasons have been considered. 24-hrs accumulated rainfall indicates that the performance of the model in predicting extreme precipitation is better during the monsoon period compared to the post-monsoon period. The skill of the ensemble mean in forecasting extreme precipitation is dictated by the amount of ensemble spread. A larger spread among the members leaves the ensemble mean with lesser predicting skill at higher thresholds of rainfall. While the probabilistic forecasts give a measure of the uncertainty in the precipitation forecast, the individual forecasts of the members can be vital in assessing the intensity of the rainfall. Verification of station rainfall clearly shows that the EPS forecasts are superior to deterministic forecasts in predicting intense precipitation at all lead times. CNTL exhibits higher RMSE than EPS at all lead times and the EPS forecasts are more reliable during the monsoon than the post-monsoon scenario. The deterministic skill scores like POD, FAR and ETS suggest that the forecast skill of the ensemble mean diminishes with increasing threshold and lead time compared to the deterministic forecast. Further, the over-prediction of precipitation by the explicit convection schemes in the deterministic forecast is considerably reduced in the EPS owing to fewer false alarms in the EPS. Despite the mixed results in the forecast skill of the EPS during monsoon and post-monsoon, the lesser error (CRPS < MAE) in the EPS forecasts and better discrimination skill of events and non-events by the EPS does necessitates the use of a convective-scale ensemble prediction system for prediction of intense rainfall events over the Indian region and thereby substantiates the use of huge computing resources involved in maintaining an EPS.