基于Delta分位数映射法的青藏高原中东部IMERG卫星降水误差订正

Bias Correction of IMERG Satellite Precipitation in the Central and Eastern Qinghai-Xizang Plateau based on Quantile Delta Mapping Method

可靠的降水资料对理解青藏高原水量平衡和水循环过程尤为重要。IMERG(Integrated MultisatellitE Retrievals for Global Precipitation Measurement)是新一代卫星降水产品,具有更广的覆盖范围与更高的时空分辨率,但在高原复杂地形条件下仍然存在较大的不确定性。鉴于此,本研究应用Delta分位数映射法(Quantile Delta Mapping,QDM),对IMERG日降水数据进行偏差订正,使用2001-2010年的中国区域地面气象要素数据集(China Meteorological Forcing Dataset,CMFD)降水数据和IMERG日降水产品分季节建立传递函数,对2011-2014年的IMERG逐日降水进行订正。研究结果表明:(1)Delta分位数映射法能够有效订正IMERG的降水频率、数值和空间分布,对极端降水和负偏差较大区域的订正效果更为明显。订正后的IMERG降水概率分布更加接近观测概率分布,降水偏差也更符合正态分布,改进了对全年和季节降水空间分布的刻画,提高了月降水的精度。(2)订正后日降水量均方根误差由1.49 mm·d^(-1)降低到1.26 mm·d^(-1),精度提高了15.44%;订正后的日降水在不同降水量级的临界成功指数CSI、命中率POD、误报率FAR、准确率Precision和F评分Fscore均有提高,降低了微量和暴量降水的空报率。(3)对极端降水的订正效果显著,降水强度SDII以及极强降水量R95p和R99p的均值更接近观测值;有效提高了对极端降水空间分布的表征,极端降水偏差从30%以上降低到5%以内;SDII、R95p和R99p的均方根误差从1.59 mm·d^(-1)、6.54 mm·d^(-1)、14.89 mm·d^(-1)降为0.65 mm·d^(-1)、3.01 mm·d^(-1)、8.99 mm·d^(-1),精度分别提高了59.12%、53.98%和39.62%。本研究验证了Delta分位数映射法在青藏高原的适用性,有利于为青藏高原气象和水文研究获取更精确的降水数据。

Reliable precipitation information is particularly important for understanding the water balance and water cycle processes in the Qinghai-Xizang Plateau.As a new generation of satellite precipitation data,IMERG(Integrated Multi-satellitE Retrievals for Global Precipitation Measurement)represents an advanced iteration in satellite precipitation data,characterized by expanded coverage and heightened spatiotemporal resolution.However,due to the complex terrain in the Qinghai-Xizang Plateau,IMERG still has great uncertainty in the plateau region.In view of this,this study performs bias correction of IMERG daily precipitation data based on the Quantile Delta Mapping(QDM)method.The transfer function is established seasonally using CMFD(China Meteorological Forcing Dataset)precipitation and IMERG daily precipitation data during 2001-2010 to correct IMERG daily precipitation during 2011-2014.The results show that:(1)The QDM method can effectively correct the frequency,value,and spatial distribution of IMERG precipitation products,and the corrections are more effective for extreme precipitation and large negative deviation regions.The probability distribution of corrected IMERG daily precipitation is closer to that of the observations,and the precipitation deviation is more in line with the normal distribution.Monthly precipitation and the spatial distribution characteristics of annual and seasonal precipitation are improved.(2)The root mean square error of the corrected daily precipitation is reduced from 1.49 mm·d^(-1) to 1.26 mm·d^(-1) with an improvement by 15.44%.The critical success index CSI,probability of detection POD,false alarm rate FAR,precision rate,and Fscore of the corrected daily precipitation in different precipitation intensities are improved,and the Bias score of tiny and torrential precipitation is enhanced.(3)After correction,the extreme precipitation is significantly improved,and the mean values of the simple daily intensity index(SDII)and percentile-based threshold indices(R95p and R99p)are closer to the observed values.The spatial distribution of extreme precipitation is effectively displayed,and the bias of extreme precipitation is reduced from over 30%to within 5%.The root mean square error of SDII,R95p,and R99p is reduced from 1.59,6.54,and 14.89 mm·d^(-1) to 0.65,3.01,and 8.99 mm·d^(-1) with the accuracy improved by 59.12%,53.98%,and 39.62%,respectively.This study verifies the applicability of the Quantile Delta Mapping method in the Tibetan Plateau,which is beneficial for obtaining more accurate precipitation data for meteorological and hydrological studies in the region.