In-situ and triple-collocation based assessments of CYGNSS-R soil moisture compared with satellite and merged estimates quasi-globally

Abstract

Soil moisture is essential for a wide range of agricultural production and climate change. Compared with soil moisture retrieved through satellite remote sensing or merged products, soil moisture products retrieved using the Cyclone Global Navigation Satellite System observations interferometric Reflectometry (CYGNSS-R) have received extensive attention. The soil moisture products retrieved from the CYGNSS-R observation have high temporal and spatial revisit characteristics. In this study, we provide a comprehensive evaluation of CYGNSS-R-based surface soil moisture (SSM), Soil Moisture Active Passive (SMAP) SSM, and the European Space Agency’s Climate Change Initiative (ESA CCI) SSM at a quasi-global scale. An in-situ validation based on stations from International Soil Moisture Network and an evaluation strategy based on triple collocation is conducted to comprehensively assess the accuracy of heterologous SSM products. The new evaluation triplet includes the CYGNSS-R-based, merged, and in-situ SSM products TC method (CMID-TC). By analyzing the temporal and spatial distribution of the errors and the distribution in different land cover types, the accuracy of CYGNSS soil moisture product is assessed and compared. Results show that CYGNSS-R-based SSM achieve overall higher absolute accuracy than SMAP satellite-based SSM. CYGNSS-R-based SSM achieved worse accuracy indicators than SMAP SSM in grassland, savanna, shrubland, cropland, and forest land cover and had slightly better accuracy than SMAP in barren pixels. CYGNSS has demonstrated the potential to improve the accuracy of fusion products through its high accuracy and ability to capture temporal correlations in TC analyses. The complementarity in the accuracy and spatial coverage between CYGNSS-R SSM and ESA CCI SSM is shown considering diverse land cover. In general, CYGNSS-R SSM shows the potential to capture higher accuracy and temporal trends in lightly vegetated areas. The TC-based assessments proposed in this study can provide a reference for evaluating the accuracy of soil moisture products and comparing the accuracy of soil moisture from different sources. Furthermore, the error verification results can provide a reference for improving the integration of soil moisture products from different sources by comparing the accuracy in different locations and land cover.