Page 285 - 《高原气象》2026年第3期
P. 285
3 期 徐 渊等:基于华南区域模式的江西省短时强降水预报偏差特征及成因分析 893
Bias Characteristics and Causes of Short-Time Heavy Rainfall Forecasts
in Jiangxi Province Based on South China Limited-Area Model
XU Yuan, CHEN Yunhui, HUANG Yujing, FU Yu, OUYANG Hongxiang
(Jiangxi Meteorological Observatory, Nanchang 330096, Jiangxi, China)
Abstract: The self-developed limited-area model for South China, CMA-GD(R3), provides technical support
for weather forecasting operations in the Pan-South China region, particularly for small-scale and short-duration
strong convective weather events. This study investigates whether and how CMA-GD(R3) can be utilized in
short-time heavy rainfall (STHR) forecasting in Jiangxi Province, aims to improve both its usage rate and fore‐
casting accuracy. The forecast bias characteristics of CMA-GD(R3) in STHR events in Jiangxi Province are ana‐
lyzed by categorizing the circulation background based on thermal and dynamic structures, and the causes of
forecast biases in different types of events are also elucidated through representative cases. The results show that
CMA-GD(R3) generally demonstrates good forecast capability for STHR in Jiangxi Province, with forecast fre‐
quency comparable to observation. However, it significantly underestimates the intensity of more extreme STHR
events, which exhibit lower predictability. The statistical and causal analysis of STHR biases in events of differ‐
ent types further indicates that CMA-GD(R3) performs better in forecasting barocline frontogenesis events com‐
pared to others, with sample points distributed relatively evenly on both sides of the diagonal. The forecast bias
in STHR is associated with the forecast errors in the location and intensity of the front. When there are spatial
forecast errors in the front, the STHR area should be adjusted in the opposite direction. However, there is a no‐
ticeable underestimation for high-level cold-advection-forced events, and the forecast initiated at 08:00 on the to‐
day performs better than the one initiated at 20:00 (Beijing Time) on the previous day. The model generally per‐
forms well in forecasting large-scale circulation background such as cold vortex or trough, with bias related to
forecast errors in local factors like the low-level humidity. Model development should focus on improving the
physical representation of this type of events. The forecast performance for low-level warm-advection-forced
events is also relatively good, but the forecast frequency is systematically overestimated and there is still a notice‐
able underestimation of extreme STHR. The accuracy of forecasting the double low-level jets is one of the key
sensitive factors affecting forecast performance. When the model underestimates (overestimates) the double low-
level jets, the forecast magnitude of STHR should be increased (decreased) in critical areas such as the exit and
coupling zones of the jets. In addition, the forecasts of quasi-barotropic events show relatively good performance
in predicting the magnitude of extreme STHR, but attention should be given to correcting the time and location.
The forecasts for elevated thunderstorms events are unstable, requiring adjustments during the nowcasting peri‐
od. The results of this study can deepen the understanding of CMA-GD(R3) forecasting behavior for STHR in
the Pan-South China region, and provide a reference basis for further targeted interpretation and application in
Jiangxi Province.
Key words: short-time heavy rainfall; CMA-GD(R3); forecast bias; circulation type; interpretation and appli‐
cation

