Page 126 - 《高原气象》2026年第4期
P. 126
高 原 气 象 45 卷
1056
The predictability of extreme precipitation in Yunnan Province
with multimodel ensemble forecasts
Li Wenmao , Yan Hongming 2
1, 2
(1. Department of Atmospheric Sciences, Yunnan University, Kunming 650500, Yunnan, China;
2. Yunnan climate center, Kunming 650034, Yunnan, China)
Abstrac:: Extreme precipitation often leads to severe casualties and economic losses, making advancements in
its prediction critically important. Given the central role of numerical models in modern forecasting operations, a
deep understanding of their performance is fundamental to improving forecast accuracy. This study employs the
root mean square error (RMSE), spatial threat score (TS), and spatial correlation coefficient (COR) to evaluate
the predictability of 21 extreme precipitation events in Yunnan from 2014 to 2024, using ensemble forecasts from
the China Meteorological Administration (CMA), the National Centers for Environmental Prediction (NCEP),
and the European Centre for Medium-Range Weather Forecasts (ECMWF) within the TIGGE (THORPEX Inter‐
active Grand Global Ensemble) dataset. Overall predictability ranked as NCEP > ECMWF > CMA. The 90th
and 75th percentile ensemble forecasts (ES90 and ES75) showed high reference value across models. ES75 dem‐
onstrated better skill for precipitation intensity, while ES90 performed better in spatial precipitation distribution,
with ECMWF > NCEP > CMA in both cases. ES75 also exhibited superior spatial consistency with observations.
For precipitation area forecasting, the most skillful combinations were: ECMWF-ES75 for moderate rain,
ECMWF-ES90 for heavy rain, NCEP-ES90 for torrential rain, and CMA-ES90 for severe torrential rain. Sea‐
sonal variation was evident: ECMWF showed the highest predictability in winter, and NCEP in other seasons.
Predictability also varied by influencing weather system-ECMWF performed best for pre-trough precipitation,
while NCEP was superior for other system types. Model predictability for extreme precipitation primarily de‐
pended on the forecast skill of influencing systems. System spatial extent and location determined rainband pre‐
dictability, while system intensity and moisture transport governed precipitation magnitude predictability. Mod‐
els performed better for precipitation influenced by large-scale systems than by smaller-scale systems.
Key words: extreme precipitation events; ensemble forecast; model assessment; TIGGE

