Fang et al. (2025) A global 400-m high-resolution soil moisture dataset derived from multi-sensor remote sensing observations

Identification

• Journal: Scientific Data
• Year: 2025
• Date: 2025-12-15
• Authors: Bin Fang, Venkataraman Lakshmi, Christopher Hain, Vikalp Mishra
• DOI: 10.1038/s41597-025-06356-z

Research Groups

• Department of Civil and Environmental Engineering, University of Virginia, Charlottesville, VA, USA
• National Aeronautics and Space Administration, Marshall Space Flight Center, Huntsville, AL, USA

Short Summary

This study developed and validated a global 400-meter resolution soil moisture (SM) dataset by downscaling the 9-kilometer SMAP product using VIIRS land surface temperature and leaf area index. The resulting 400-meter product demonstrated improved accuracy and better captured spatial variability compared to the 1-kilometer and original 9-kilometer SMAP products when validated against in situ observations.

Objective

• To develop and implement a global 400-meter high-resolution soil moisture dataset by downscaling the 9-kilometer SMAP enhanced Level-2 radiometer product using a visible/infrared (VIS/IR) algorithm that incorporates VIIRS land surface temperature (LST) and leaf area index (LAI) products.

Study Configuration

• Spatial Scale: Global, covering 180° W to 180° E and 86° N to 86° S, with a grid spacing of 400.36 meters. Downscaling from 9 kilometers to 400 meters.
• Temporal Scale: Daily data from April 1, 2015, to December 31, 2024.

Methodology and Data

• Models used:
  • VIS/IR downscaling algorithm based on apparent thermal inertia (SM-temperature difference, θ – ΔT).
  • Linear regression model for parameterizing the θ – ΔT relationship using GLDAS Noah model outputs.
  • Hybrid bias-correction approach combining additive and ratio-based corrections.
  • SMAP Single Channel Algorithm (SCA) and Backus-Gilbert optimal interpolation method for 9-kilometer SMAP product.
• Data sources:
  • Satellite:
    • Soil Moisture Active Passive (SMAP) enhanced Level-2 radiometer half-orbit 9-kilometer SM product (L-band passive microwave).
    • Visible Infrared Imaging Radiometer Suite (VIIRS) daily LST (375 meters) and 8-day LAI (500 meters).
    • Long-Term Data Record (LTDR) AVHRR NDVI Version 5 (0.05° resolution).
    • Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG) Version 6 daily dataset (0.1° resolution).
  • Observation:
    • International Soil Moisture Network (ISMN) in situ 0-5 centimeter SM measurements from 31 global networks.
  • Reanalysis/Model outputs:
    • Global Land Data Assimilation System (GLDAS) V2.0 (1981-1999) and V2.1 (2000-2020) Noah land surface model outputs (surface skin temperature and 0-10 centimeter SM content).

Main Results

• The 400-meter downscaled SM product demonstrated improved accuracy with an overall unbiased Root Mean Square Error (ubRMSE) of 0.072 m³/m³ and Mean Absolute Error (MAE) of 0.066 m³/m³, outperforming the 1-kilometer (ubRMSE 0.073 m³/m³, MAE 0.067 m³/m³) and 9-kilometer (ubRMSE 0.075 m³/m³, MAE 0.069 m³/m³) SMAP products.
• The 400-meter product better preserved spatial variability, with an average spatial standard deviation (SSD) of 0.041 m³/m³, which was closer to the in situ benchmark (0.1 m³/m³) compared to the 1-kilometer (0.039 m³/m³) and 9-kilometer (0.033 m³/m³) products.
• A slight degradation in the coefficient of determination (R²) was observed for the downscaled products (0.406 for 400 meters, 0.409 for 1 kilometer) compared to the 9-kilometer product (0.426), attributed to the bias-variance trade-off inherent in downscaling.
• The downscaled products effectively captured finer-scale spatial patterns of SM, supporting local and regional hydrological applications.
• Data availability for the global downscaled SMAP SM product is approximately between 65° N and 45° S, with missing data in regions affected by persistent cloud cover, dense vegetation, or radio-frequency interference.

Contributions

• Generated the first global remotely sensed soil moisture product at 400-meter resolution, derived from VIIRS-based downscaling, significantly improving upon previously published 1-kilometer downscaled SMAP SM products.
• Introduced a novel hybrid bias-correction approach that combines additive and ratio-based corrections, enhancing the robustness of the 400-meter downscaled dataset and reducing blocky artifacts compared to earlier additive-only methods.
• Provided a comprehensive assessment of the VIS/IR downscaling algorithm's performance across multiple spatial scales using global in situ soil moisture data from the International Soil Moisture Network (ISMN), highlighting its implications for fine-scale hydrological applications.

Funding

• NASA Terrestrial Hydrology - NASA Award Number 80NSSC19K0993.

Citation

@article{Fang2025global,
  author = {Fang, Bin and Lakshmi, Venkataraman and Hain, Christopher and Mishra, Vikalp},
  title = {A global 400-m high-resolution soil moisture dataset derived from multi-sensor remote sensing observations},
  journal = {Scientific Data},
  year = {2025},
  doi = {10.1038/s41597-025-06356-z},
  url = {https://doi.org/10.1038/s41597-025-06356-z}
}