Abstract:

The soil moisture product retrieved by single satellite has the disadvantage of discontinuous spatiotemporal coverage. In order to obtain spatiotemporal continuous satellite remote sensing soil moisture data in the Pearl river basin, with the volumetric soil moisture (VSM) product of soil moisture active/passive satellite (SMAP) as a reference, the cumulative distribution function (CDF) method was used to perform the matching bias correction for the VSM products from advanced microwave scanning radiometer 2 (AMSR2), soil moisture and ocean salinity (SMOS) and microwave radiation imager (MWRI), and the optimum interpolation method was used to fuse the data of these four VSM products to generate a spatiotemporal continuous daily fusion VSM product with a resolution of 10km in the Pearl river basin. The station observation data and reanalysis data were adopted to evaluate the fused VSM products. The results indicated that, (1) there were significant differences in the measurement range of soil moisture products retrieved from different satellites. The measurement ranges from high to low were SMOS, AMSR2, SMAP and MWRI, with maximum measurement values of 1.00, 0.99, 0.70 and 0.50m³·m⁻³, respectively. They were not suitable for simultaneous use in soil moisture monitoring. (2) There were deviations between multi-source satellite VSM products. SMOS VSM product had a negative bias compared to SMAP VSM product, with the smallest unbiased root mean square error and the highest correlation coefficient. AMSR2 VSM product had a positive bias compared to SMAP VSM product, and the correlation between these two satellite VSM products was relatively low. MWRI VSM product had a negative bias and the smallest correlation compared to SMAP VSM product. (3) The accuracy and stability of SMAP VSM product were better than those of AMSR2, SMOS and MWRI VSM products. The time series correlation between SMAP VSM product and in-situ data and reanalysis data was obviously better than the latter three satellite VSM products. (4) After CDF matching bias correction, the consistency between AMSR2, SMOS and MWRI VSM products and SMAP VSM product had been enhanced. Multi-source data fusion can correct the error of single satellite product, improve the correlations with in-situ data and reanalysis data, and compensate the spatiotemporal coverage continuity of remote sensing data.

Key words: Pearl river basin, Microwave remote sensing, Soil moisture, Bias correction, Data fusion