| Vegetation and soil are very important elements in physical geography environment,which have great ecological significance and application value.Timely and accurate monitoring of soil moisture and vegetation moisture content is of great significance in fire detection,drought monitoring,observing land changes,and improving agricultural irrigation efficiency.The Normalized Microwave Reflection Index(NMRI)and Soil Moisture(SM)obtained based on the signal-to-noise ratio of Global Navigation Satellite System Interferometry and Reflection(GNSS-IR)technology can reflect the changes in vegetation moisture and soil moisture.The problem of GNSS-IR being limited to single station monitoring can be effectively solved through the fusion of GNSS-IR and multi-source data.However,the inversion based on GNSS-IR and multi-source data fusion is the output of a single product,and the data utilization rate is not high.Currently,research on synchronous inversion of vegetation moisture content and soil moisture has not been involved.In order to simultaneously obtain spatially continuous NMRI and SM products,this paper fully utilizes the advantages of combining GNSS-IR and MODIS data to construct a method for synchronously inverting vegetation moisture content and soil moisture through the fusion of GNSS-IR and multi-source data.The main work and innovations of this article are as follows:(1)Synthesize vegetation indices with higher temporal resolution using a combination of spectral reflectance calculations.This study addresses the issue of limited vegetation index when integrating GNSS-IR with multi-source data for unified temporal resolution.By utilizing band reflectance data to synthesize vegetation index products with higher temporal resolution,the temporal resolution of spatially continuous NMRI and SM products can be improved.(2)A method of selecting GNSS modeling stations according to the type of land cover is proposed.According to the correlation between NMRI and SM of each GNSS station,as well as the correlation between NMRI and different characteristic elements,the GNSS modeling station is selected in combination with the land cover type of the GNSS station.Finally,based on comprehensive feature selection,a modeling dataset is constructed by reasonably selecting feature elements for inverting vegetation moisture content and soil moisture.(3)Generated spatially continuous NMRI and SM products.Through cross validation for base model selection,and then using the Stacking algorithm for model fusion,a synchronous inversion model for vegetation moisture content and soil moisture was established.A total of12 consecutive NMRI and SM products with a spatial and temporal resolution of 8 days/500 m were synchronously generated.This article uses NMRI data from the PBO observation network and NASA-USDA global soil moisture data as reference values for vegetation moisture and soil moisture.184 modeling stations and 10 non modeling stations were used to quantitatively analyze the retrieved NMRI and SM products.The results showed that 60% and51% of the 184 modeling stations had a correlation R greater than 0.60,respectively.Among the 10 stations that did not participate in modeling,there were 6 and 4 stations with a correlation R greater than 0.60,respectively.Comparing NMRI and SM products with NDVI and LAI,as well as NASA-USDA global soil moisture images,it can be seen that NMRI and SM products have overall consistency in time and space.The time resolution of the spatially continuous NMRI and SM products generated in this article has been improved to 8 days based on existing research,which can more intuitively display the overall trend of vegetation moisture content and soil moisture. |
PDF Full Text Request