新型遥感卫星土壤水分产品在不同地表覆盖下的精度验证与分析
Validation and Analysis of Remote Sensing Satellite Soil Moisture Data under Different Kinds of Vegetation
-
摘要: 近年来,土壤水分遥感卫星数据被广泛应用在气候、水文、农业等领域的研究中,微波土壤水分数据产品的适用性分析显得越来越重要。本研究分别以美国国家航空航天局的FLUXNET2015数据集和欧洲航天局发起并进行集中数据托管的ISMN网络为基准,运用统计学的方法(偏差、均方根误差和时间序列相关系数),在2012年7月至2014年12月,以IGBP(国际地圈生物圈计划)为标准的地表覆盖类型进行分类,在全球范围内的FLUXNET站点对SMOS-L3-SM和AMSR2-LPRM-SM两种遥感土壤水分产品与实测数据进行了对比验证,并在2016年1月至2017年12月,以30 m全球地表覆盖产品(GlobeLand30)为标准的地表覆盖类型进行分类,在全球范围内的ISMN站点对SMAP-L3-SM和AMSR2-LPRM-SM两种遥感土壤水分产品与实测数据进行了对比验证。结果表明:三种遥感土壤水分产品在草地等低矮植被类型低矮稀疏植被覆盖区域的土壤水分产品反演精度均要高于高大茂密植被覆盖的区域,且SMOS和AMSR2产品在地表覆盖为针叶林下反演精度较低,SMAP和AMSR2产品在地表覆盖为湿地下反演精度较低,SMAP产品的反演精度要高于AMSR2产品。Abstract: In recent years,soil moisture remote sensing satellite data are widely used in the study of climate,hydrology,agriculture and other fields. The applicability analysis of microwave soil moisture data products is becoming more and more important. Based on the fluxnet 2015 data set of NASA and the ISMN network initiated and managed by European Space Agency,this study uses statistical methods from July2012 to December 2014,based on IGBP(International geosphere biosphere program)as the standard of land cover type classification,two remote sensing soil moisture products SMOS-L3-SM and AMSR2-LPRM-SM are compared and verified with the measured data at FLUXNET sites around the world,and from January 2016 to December 2017,the soil moisture products of 30 meters.The global land cover product(globel and 30)is used as the standard of land cover classification,and the SMAP-L3-SM and AMSR2-LPRM-SM are compared with the measured data at ISMN sites around the world. The results show that the retrieval accuracy of the three remote sensing soil moisture products in the low and sparse vegetation coverage area is higher than that in the dense vegetation coverage area,and the retrieval accuracy of SMOS and AMSR2 products is lower when the surface coverage is coniferous forest,while the retrieval accuracy of SMAP and AMSR2 products is lower when the surface coverage is wet underground higher than AMSR2.