Premium
Estimation of Surface Soil Moisture With Downscaled Land Surface Temperatures Using a Data Fusion Approach for Heterogeneous Agricultural Land
Author(s) -
Bai Liangliang,
Long Di,
Yan La
Publication year - 2019
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/2018wr024162
Subject(s) - environmental science , remote sensing , mean squared error , image resolution , water content , temporal resolution , geography , geology , computer science , statistics , physics , mathematics , geotechnical engineering , quantum mechanics , artificial intelligence
Field‐scale surface soil moisture (SSM, 0–10 cm), which is closely linked with land surface temperature (LST), is particularly important to agricultural water resource management. Active and passive microwave remote sensing‐based SSM retrievals on the order of kilometer squared resolutions are difficult to apply to heterogeneous agricultural land surfaces that may need SSM data at a resolution of 30 m. In this study, the High‐resolution Urban Thermal Sharpener and Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model were applied to downscale optical and thermal remote sensing data simultaneously by blending Landsat and MODIS red‐near infrared‐LST data, with the ultimate goal to generate field‐scale SSM values from the trapezoidal approach. To evaluate the performance of the downscaled LST E (based on the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model method) and SSM, an irrigation district (Area 1) in Inner Mongolia and an irrigation district in the North China Plain (Area 2) with varying spatial heterogeneity were selected as the testbeds. Results indicated that the downscaled LST E was highly consistent with synchronous Landsat LST H and in situ LST measurements in Area 1, with the root‐mean‐square error ranging from 0.73 to 2.75 K. Compared with the MODIS SSM, the average root‐mean‐square error of the downscaled SSM improved from 0.048 to 0.038 cm 3 /cm 3 for both areas. The downscaled LST E and SSM developed in this study enhance the spatiotemporal resolutions of the SSM estimates, maximizing the potential of remotely sensed information for agricultural water resource management.