基于雷达和遥感卫星的新疆区域降水反演

o obtain more accurate precipitation distribution in remote areas, this paper combines the high-resolution advantages of radar and the wide-coverage detection of satellites. By integrating radar and satellite-derived precipitation, we generate high-precision quantitative precipitation estimation (QPE) products. Taking the strong convective events in Xinjiang on August 12 and 13, 2023, as an example, we use radar reflectivity for precipitation inversion based on cloud classification and Z-R relationships. We input Himawari 9 satellite brightness temperature and IMERG precipitation into a BP neural network model to establish the relationship between average brightness temperature and average rainfall intensity. Subsequently, we use the instantaneous brightness temperature of the Himawari 9 satellite to invert momentary precipitation through the BP neural network model. We also propose two precipitation data fusion schemes: Scheme I uses a uniform correction value to integrate radar and satellite precipitation, while Scheme II further considers precipitation intensity levels for comparison. Finally, we obtain high-precision precipitation inversion products for Xinjiang. The results show that: (1) Cloud classification based on brightness temperature can finely estimate precipitation within the radar range, and brightness temperature differences can reduce the impact of non-precipitating clouds to some extent. (2) The root mean square errors (RMSE) of satellite precipitation inversion is 1.793 mmh1, with coefficient of determination (R2) of 0.572, indicating reasonable model accuracy. The binary classification score results show that the model can accurately invert precipitation in over 70% of the areas. (3) The fusion of precipitation by the two schemes has limited improvement in the accuracy of short-duration light rain. Scheme II outperforms Scheme I for short-duration moderate rain but shows a slight decline for short-duration heavy rain compared to Scheme I, indicating that the asynchrony between satellite observation and near-surface precipitation has some impact. (4) Under a 95% confidence interval, the P values for the RMSE and R2 differences between the two schemes and satellite inversion are all less than 0.005, while the P value for Scheme II compared to Scheme I is greater than 0.05. Both fusion schemes significantly improve the accuracy of satellite precipitation, but the improvement of Scheme II, which considers precipitation intensity levels, over Scheme I is minimal.