Global 1-km Gridded Soil Respiration Maps Derived from Machine Learning

Six global gridded products provide predicted annual soil respiration and heterotrophic respiration at 1-kilometer resolution. A quantile regression forest algorithm trained on the Soil Respiration Database version 3, with observations from 1961 to 2011, generated these maps and their associated uncertainty estimates. The dataset was produced by NASA to support carbon cycle research.

Use Cases

• Modeling spatial variability in soil carbon dioxide emissions using predicted annual soil respiration (Rs) maps.
• Assessing prediction confidence for carbon budget calculations using the associated uncertainty layers.
• Estimating heterotrophic respiration contributions to the carbon cycle using the two derived Rh maps.
• Analyzing long-term trends in soil respiration by leveraging the underlying training data spanning 1961-2011.
• Conducting local-scale carbon flux studies with the high 1-km spatial resolution of the gridded products.

Strengths

• High spatial resolution of 1 kilometer for detailed global analysis.
• Includes uncertainty estimates and prediction distribution quartiles, supporting robust error analysis.
• Based on a machine learning model trained on the global Soil Respiration Database (SRDB V3).
• Covers a significant temporal range in its source training data, from 1961 to 2011.

Limitations

• Specific column names and the total row/pixel count are not provided by any source.
• There is a conflict in the reported last updated date: Data.gov lists 2026-03 13, while NASA Earthdata lists 2011-12-31.
• The exact license terms are unclear, only specified as 'other-license-specified'.

Provenance

Source

National Aeronautics and Space Administration (NASA)

Collection Method

Quantile Regression Forest (QRF) machine learning algorithm trained on the global Soil Respiration Database (SRDB) version 3.

Time Range

1961 to 2011 (for source training observations)

Freshness

2026-03-13 01:53:58.278254

Geography

Global