杜军凯, 贾仰文, 李晓星, 牛存稳, 刘欢, 仇亚琴. 基于TRMM卫星降水的太行山区降水时空分布格局[J]. 水科学进展, 2019, 30(1): 1-13. DOI: 10.14042/j.cnki.32.1309.2019.01.001
引用本文: 杜军凯, 贾仰文, 李晓星, 牛存稳, 刘欢, 仇亚琴. 基于TRMM卫星降水的太行山区降水时空分布格局[J]. 水科学进展, 2019, 30(1): 1-13. DOI: 10.14042/j.cnki.32.1309.2019.01.001
DU Junkai, JIA Yangwen, LI Xiaoxing, NIU Cunwen, LIU Huan, QIU Yaqin. Study on the spatial-temporal distribution pattern of precipitation in the Taihang Mountain region using TRMM data[J]. Advances in Water Science, 2019, 30(1): 1-13. DOI: 10.14042/j.cnki.32.1309.2019.01.001
Citation: DU Junkai, JIA Yangwen, LI Xiaoxing, NIU Cunwen, LIU Huan, QIU Yaqin. Study on the spatial-temporal distribution pattern of precipitation in the Taihang Mountain region using TRMM data[J]. Advances in Water Science, 2019, 30(1): 1-13. DOI: 10.14042/j.cnki.32.1309.2019.01.001

基于TRMM卫星降水的太行山区降水时空分布格局

Study on the spatial-temporal distribution pattern of precipitation in the Taihang Mountain region using TRMM data

  • 摘要: 基于TRMM 3B42V7数据,综合采用多元线性回归、偏最小二乘回归和地理加权回归3种方法,建立了太行山区卫星降水产品的降尺度校正模型,将遥感降水信息从0.25°×0.25°降尺度到0.05°×0.05°。在结果评估和优选的基础上,分析了"像元-集水区-全区"年、月降水的多时空尺度干湿季节分布和垂向分布特征,并从机理方面论证了研究的合理性。结果表明:①地理加权回归校正效果最优,可明显降低校正降水与实测降水系列的均方根误差和平均相对偏差且提高决定系数;偏最小二乘回归可降低两项误差,但对决定系数无提升;多元线性回归最差,各项指标均无改善。②处于夏季风迎风侧的东坡和南坡降水量普遍高于500 mm,背风侧的西坡和北坡降水量较低,最大年降水量位于东南坡海拔1 300~1 500 m的地带。③研究区7-9月降水量占全年的58.7%,干湿季节降水量之比为1:18,各集水区的变化范围为1:13~1:25。④季风风向影响降水中心的移动路径,各月降水量沿高程变化梯度区间为-5.2~6.7 mm/hm,且迎风坡降水的垂向分布更复杂。

     

    Abstract: A comprehensive understanding of the spatial and temporal distribution of precipitation in mountain areas is of great significance for improving the simulation accuracy of regional water cycle processes,as well as the level of water management. In view of the fact that there is a severe lack of ground monitoring in high altitude areas,as well as the disadvantage of downscaling of satellite precipitation products by a single method,this study selects the Taihang Mountain as the research region and establishes a downscaling correction model. The model consists of the validation module and the downscaling module,which includes the multiple linear regression,the partial least squares regression,and the geographically weighted regression. Using the model,the original TRMM data is scaled from 0.25° to 0.05°. The distribution of the wet and dry seasons and the vertical distribution of the annual and monthly precipitation of the "pixel-catchment-region" are analyzed on the basis of the evaluation and optimization of the downscaling results. The research results are as follows: ① The geographically weighted regression is the most effective,reducing the root mean square error and average relative deviation of the corrected and gauged precipitation series,as well as improving the coefficient of determination. The partial least squares regression can reduce the two errors,but it cannot improve the determination coefficient. The multiple linear regression cannot improve any of the three indicators. ② The precipitation on the east and south slopes on the windward side of the summer monsoon is generally higher than 500 mm,while that on the west and north slopes on the leeward side is lower. The zones of the maximum annual precipitation are located in the southeast slope at an altitude of 1 300—1 500 m. ③ The precipitation from July to September accounts for 58.7% of the entire year;the ratio of precipitation in the dry and wet season is 1∶18;and the ratio in each catchment ranges from 1∶13 to 1∶25. ④ The variations of wind direction of the monsoon affect the moving track of the precipitation center,and the vertical zonality is more complicated in the windward than in the leeward.

     

/

返回文章
返回