GSSM1km: Global Daily 1km Surface Soil Moisture, 2000-2020

GSSM1km provides surface soil moisture (0-5 cm) at a 1 km spatial and daily temporal resolution globally from 2000 to 2020. The dataset was generated by Han using physics-informed machine learning, integrating data from the International Soil Moisture Network, remote sensing, and meteorological sources. Its performance shows a root mean square error of 0.05 cm³/cm³ and a correlation coefficient of 0.9 on the testing set.

Use Cases

• Analyzing long-term soil moisture trends based on the 21-year daily time series.
• Investigating climate extremes like droughts based on high-resolution (1 km) global coverage.
• Validating other soil moisture products based on the inter-comparisons mentioned in the description.
• Studying the influence of key predictors like the Antecedent Precipitation Evaporation Index (APEI) on soil moisture dynamics.

Strengths

• Covers 21 years (2000-2020) of daily data.
• High spatial resolution of 1 km for global coverage.
• Demonstrated performance with a testing set RMSE of 0.05 cm³/cm³ and correlation of 0.9.
• Integrates 18 predictors, with APEI identified as the most important.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The full global dataset is 779GB, which may present storage and processing challenges.
• Last update date is unknown; freshness unverified.

Provenance

Source

Generated by Han using International Soil Moisture Network (ISMN), remote sensing, and meteorological data.

Collection Method

Physics-informed machine learning guided by knowledge of physical processes impacting soil moisture dynamics.

Time Range

2000-2020

Geography

Global