多源降水融合产品在新疆暖季的适用性评估

Applicability Assessment of Multi-Source Precipitation Fusion Products in the Warm Season in Xinjiang

摘要:

基于新疆1 889个自动气象站降水观测资料，检验评估了由中国区域融合降水分析系统（CMPAS）发布的2020—2023年5—9月（暖季）多源降水融合产品在新疆的适用性，结果表明：2022—2023年多源降水融合产品与实测降水的各项检验指标较2020—2021年明显变优，近2年二者的相关系数和TS评分已分别从0.86和0.648上升至0.939和0.854，增幅为9.2%和31.8%，均方根误差从0.692 mm降低至0.484 mm，降幅为30%，空报率从0.325降至0.101，降幅高达68.9%，漏报率变化不大。暖季各月中，8月降水融合产品综合质量最好；在实测降水较大时段（15—21时），降水融合产品与实测降水的相关系数较大且相对稳定，但由于空报率、漏报率较高导致TS评分略低。多源降水融合产品与实测降水的检验指标随降水强度增大而变差，但差距呈逐年减小的趋势。降水融合产品质量较差区域主要集中在北疆西北部、天山山区中段南北两侧、南疆西部、昆仑山北坡、阿勒泰山区及东疆地区。2022—2023年与前2年相比，降水融合产品与实测降水的各项指标在不同时间尺度、不同降水等级及不同海拔高度的检验评估中质量均明显变优，说明2021年下半年降水融合产品算法的优化对产品质量提升效果显著。

Abstract:

Based on the precipitation observation data from 1 889 automatic meteorological stations in Xinjiang, this study assesses the applicability of the multi-source precipitation fusion product (with a spatial resolution of 1 km and a temporal resolution of 1 hour) released by the CMA Multi-source Precipitation Analysis System (CMPAS) for the warm season (May to September) from 2020 to 2023. The results show that the product has significantly improved in 2022-2023 compared to the 2020-2021 period across all evaluation indicators. The correlation coefficient (COR) and TS score in 2022-2023 increased from 0.86 and 0.648 to 0.939 and 0.854, respectively, representing an increase of 9.2% and 31.8%. The root mean square error decreased from 0.692 mm to 0.484 mm, a reduction of 30%. The false alarm rate dramatically decreased from 0.325 to 0.101, a reduction of 68.9%, while the missed rate showed no significant change. Among the months of the warm season, the comprehensive quality of the precipitation fusion product was the best in August. During periods of high actual precipitation (15:00-21:00), the correlation coefficient between the precipitation fusion product and the actual precipitation was relatively high and stable. However, the TS score was slightly lower due to the higher false alarm rate and missed rates. The performance of the multi-source precipitation fusion product deteriorated with increasing precipitation intensity, but the gap was trending downward annually. The areas with relatively poor precipitation fusion product quality are mainly located in the northwest of northern Xinjiang, the middle section of the Tianshan Mountains (both northern and southern slopes), the western part of southern Xinjiang, the northern slope of the Kunlun Mountains, the Altay Mountains, and the eastern Xinjiang region. Compared to the 2020-2021 period, the quality of the precipitation fusion product improved significantly in 2022-2023 across various time scales, precipitation intensities, and altitudinal ranges. This indicates that the optimization of the precipitation fusion product algorithm in the second half of 2021 has significantly enhanced the product accuracy and reliability.