Table 1.
Overview of the four satellite soil moisture datasets being compared.
Table 2.
Indicator values for comparison of remote sensing data with GLDAS data.
Fig 1.
Correlation coefficients and data quantity distributions over China for passive microwave soil moisture products; grids with less than 5 data points are blank.
Fig 2.
Boxplots of mean DEM elevation for r and data quantity.
Fig 3.
Boxplots of DEM elevation ranges for Biasr, RMSE and ubRMSE.
Fig 4.
Boxplots of coefficients of correlation between SM data accuracy and land cover type for each product; abbreviations of landcover types are: PF paddy field, DL dry field, DF dense forest, DSS dwarf scrub and shrub, SF sparse forest, OF other forests, HCG high coverage grassland, MCG medium coverage grassland, LCG low coverage grassland, UL urban land, SA sand, GO Gobi, SAL saline alkali land, SL swampland, BL barren land, RG rock and gravel, OUL other unused land.
Fig 5.
Histograms of Biasr for different landcover types; the abbreviations are the same as for Fig 4.
Fig 6.
Boxplots of land cover information entropy (H) for r, Biasr and ubRMSE.
Fig 7.
Boxplots of temperature condition for r and ubRMSE.
Fig 8.
Boxplots of precipitation condition for r and ubRMSE.
Fig 9.
Distribution of parameters across China for SMOS, AMSR2 and FY-3C soil moisture products given by triple collocation analysis; results are only shown for grids which all three datasets were positively correlated.