An Analysis of a Commercial GNSS-R Soil Moisture Dataset

• Published in: IEEE journal of selected topics in applied earth observations and remote sensing Vol. 17; pp. 15480 - 15493
• Main Authors: Al-Khaldi, Mohammad M., Johnson, Joel T., Horton, Dustin, McKague, Darren S., Twigg, Dorina, Russel, Anthony, Policelli, Frederick S., Ouellette, Jeffrey D., Bindlish, Rajat, Park, Jeonghwan
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Active satellites; Bistatic radar systems; CubeSats; Datasets; Flood forecasting; Global navigation satellite system; global navigation satellite systems reflectometry (GNSS-R); Inland waters; Moisture content; Navigation; Navigational satellites; Observatories; Pixels; Receivers; Reflectivity; rough surface scattering; Rough surfaces; Satellite constellations; Satellite observation; Satellites; Scattering; SmallSats; Soil; Soil analysis; Soil investigations; Soil moisture; Surface roughness; Surface treatment; Water bodies; Water pollution; Weather forecasting
• ISSN: 1939-1404; 2151-1535
• DOI: 10.1109/JSTARS.2024.3449773

An analysis of a Level-2 (L2) soil moisture record extending from 1 May 2021 to 1 January 2024 derived from Spire, Inc.'s Global Navigation Satellite System Reflectometry (GNSS-R) observatories is presented. The product's sensitivity to large scale soil moisture variability is demonstrated using an example of a 2022 flood in Pakistan. Product consistency among the constellation's multiple satellites is also investigated; no clear evidence of intersatellite biases is observed. Further comparisons are performed with soil moisture datasets from the Soil Moisture Active Passive (SMAP) and Cyclone Global Navigation Satellite System (CYGNSS) missions, from the European Center for Medium-Range Weather Forecasts Reanalysis v5 (ERA5), and from in situ International Soil Moisture Network (ISMN) sites. Although an overall product correlation with SMAP soil moisture of approximately 85<inline-formula><tex-math notation="LaTeX">\%</tex-math></inline-formula> is determined, per-pixel correlations vary significantly and per-pixel root-mean-square errors (RMSE) can range from 0.02 to 0.09 (cm<inline-formula><tex-math notation="LaTeX">^{3}</tex-math></inline-formula>/cm<inline-formula><tex-math notation="LaTeX">^{3}</tex-math></inline-formula>) depending on land class. The importance of applying the product's quality flags is also demonstrated. The influence of other calibration effects and inland water body contamination on these results is also discussed.