Large-scale Passive Microwave Remote Sensing Soil Moisture Downscaling Based On Feature Space Method

Soil moisture is an important environmental factor controlling the surface water cycle system and atmospheric energy budget.Its change will have a potential impact on climate change and surface hydrothermal cycle.Therefore,accurately obtaining the spatial and temporal distribution data of large-scale soil moisture is an important basis for coping with climate change and maintaining the balance of the earth’s ecosystem.At present,there are many passive microwave soil moisture products with wide spatial coverage,but their spatial resolution is low,which is difficult to meet the actual needs.Downscaling of passive microwave remote sensing soil moisture products is an effective method to obtain high spatial resolution soil moisture data.In this context,this study takes China(without considering Hainan,Hong Kong,Macao,Taiwan and South China Sea Islands)as the study area,and uses MODIS(Moderate-Resolution Imaging Spectroradiometer)and SMAP(Soil Moisture Active Passive)remote sensing data as the input.Then,the spatial scale in traditional research is extended to the pixel scale based on the surface temperature-vegetation index(_sT-VI)feature spatial theoretical framework.Finally,the spatial scale conversion of soil moisture data from 36km resolution to 1km resolution is realized in China.The main work contents and results are as follows:(1)Designs a large-scale regional soil moisture downscaling parameterization scheme based on microwave data.Using the cosine function relationship between surface temperature and solar zenith,the key parameters are obtained pixel by pixel,and then the empirical boundary of _sT-VI feature space is obtained;on this basis,the theoretical boundary of _sT-VI feature space is determined driven by microwave data,and the 1 km resolution Modified Temperature Vegetation Dryness Index(MTVDI)is obtained.The parameter optimization results show that the theoretical boundary obtained by this scheme has the two-dimensional attribute of space-time,and the curve fitting effect of dry edge is better than that of wet edge.(2)The downscaling model of soil moisture in _sT-VI feature space based on pixel scale is constructed.The nonlinear relationship between MTVDI and soil moisture is established,and the soil moisture downscaling model based on pixel scale feature space method is developed.Then,based on the site-measured data,the applicability assessment of the model is carried out.The verification analysis shows that the correlation between the estimation results and the measured data is high,which has reached or even better than the accuracy level of the original soil moisture product.This shows that the model has good applicability in large-scale area,completely gets rid of the dependence on the measured data,and the continuous monitoring of soil moisture can be realized only by remote sensing technology.In addition,based on the comparison between the SMAP product and the high-resolution data,the downscaling method enriches the change information of soil moisture in the range of 36km×36km,and has a certain spatial correspondence with the original product.(3)The spatial-temporal distribution pattern and influencing factors of downscaling soil moisture are studied.From the perspective of temporal changes,winter and summer soil moisture have obvious contributions to the trend change of soil moisture on an interannual scale.From the perspective of spatial distribution,soil moisture has a significant geographical spatial differentiation pattern,showing a distribution pattern of high in the southeast and low in the northwest,and high in the coastal areas and low in the interior.In addition,the positive correlation between soil moisture and precipitation is the strongest,followed by vegetation coverage,and the weakest correlation with altitude.