基于指数滤波模型的青藏高原根系层土壤水分估算研究  

Estimation of Rootzone Soil Moisture over the Qinghai-Xizang Plateau using the Exponential Filter Model

作　　者：宋雨曦 苏建宾 文军 郑东海 SONG Yuxi;SU Jianbin;WEN Jun;ZHENG Donghai(Key Laboratory of Plateau Atmosphere and Environment,Sichuan Province,College of Atmospheric Sciences,Chengdu University of Information Technology,Chengdu 610225,Sichuan,China;National Tibetan Plateau Data Center,State Key Laboratory of Tibetan Plateau Earth System,Resource and Environment,Institute of Tibetan Plateau Research,Chinese Academy of Sciences,Beijing 100101,China)

机构地区：[1]成都信息工程大学大气科学学院,高原大气与环境四川省重点实验室,四川成都610225 [2]中国科学院青藏高原研究所,青藏高原地球系统与资源环境全国重点实验室,国家青藏高原科学数据中心,北京100101

出　　处：《高原气象》2025年第1期134-149,共16页Plateau Meteorology

基　　金：国家重点研发计划项目(2023YFB3905801)。

摘　　要：高精度、长时序的土壤水分数据对青藏高原陆-气相互作用研究十分重要。然而,当前青藏高原土壤水分研究大多聚焦表层土壤水分的卫星反演,根系层土壤水分的估算研究相对匮乏。基于青藏高原五个土壤水分观测网络(狮泉河、帕里、那曲、黑河上游、玛曲)数据,本文系统评估了指数滤波模型在青藏高原不同气候与下垫面区域估算根系层(10 cm/20 cm/40 cm)土壤水分的适用性,探讨了不同环境因素(土壤、气候、植被)对模型特征时长参数T值的影响,最后评估了利用三种方法(各观测网络或青藏高原站点最优T值(Topt)中位数、随机森林模型)获取的区域T值估算青藏高原根系层土壤水分的可靠性。研究发现:(1)随着土壤深度的增加,根系层与表层土壤水分的相关性降低,而其空间异质性增大,导致指数滤波模型的应用精度降低,而模型T值及其空间异质性增大;(2)空间上随着降水量和土壤含水量的增加,根系层与表层土壤水分的相关性增大,而其空间异质性降低,使得指数滤波模型的应用精度上升,而其在区域内不同站点应用精度的差异呈现降低趋势;(3)土壤性质尤其是砂土含量是控制青藏高原区域T值空间分布的主要因素;(4)T值的获取对指数滤波模型在青藏高原的应用精度影响较小,当前常用的区域Topt中位数和随机森林模型均能获取相对合理的区域T值和取得一致且相对准确的模拟结果。本研究的发现有望推动利用指数滤波模型基于卫星表层土壤水分资料准确获取青藏高原根系层土壤水分。High-precision and long time-series soil moisture(SM)data are crucial for quantifying the land-atmosphere interactions on the Qinghai-Xizang Plateau(QXP).However,most of current studies on the Tibetan Plateau mainly focus on retrieving surface SM based on the satellite data,with relative lack of studies on estimating rootzone SM(RZSM).Based on data collected from five SM observation networks on the QXP(i.e.,Shiquanhe,Pali,Naqu,upper reaches of Heihe River,and Maqu),this paper systematically evaluates the applicability of the exponential filter model for estimating RZSM(i.e.,10,20,40 cm)in different climatic and land areas of the QXP.In addition,we explore the impacts of different environmental factors(e.g.,soil properties,climate,and vegetation)on the estimated key model parameter,i.e.,characteristic time length T.Moreover,the reliability of regional-scale T-value obtained using three methods(i.e.,using the median value of optimal Tvalue obtained for each observation network or the whole Qinghai-Xizang Plateau,or the random forest model)for estimating RZSM on the QXP was assessed.The results showed that:(1)With the increase of soil depth,the correlation between rootzone and surface SM decreased while its spatial heterogeneity increased.This results in a decrease in the accuracy of the exponential filter model,but the obtained T-value and its spatial heterogeneity increased.(2)Spatially,with the increase of precipitation and SM content,the correlation between rootzone and surface SM increased while its spatial heterogeneity decreased.This leads to an increase in the accuracy of the exponential filter model,while the difference in accuracy of the model applied to different sites shows a decreasing trend.(3)Soil properties,especially the sand content,are the main factors controlling the spatial distribution of T-value on the QXP.(4)The different methods for obtaining regional-scale T-value have little influence on the accuracy of the exponential filter model in estimating the RZSM on the QXP.Specifically,both the co

关 键 词：青藏高原 指数滤波模型 根系层土壤水分 不同气候与下垫面区域 

分 类 号：P461[天文地球—大气科学及气象学]