Global Land Suitability Maps for Agricultural Expansion at 300-Meter Resolution

10 arc-second resolution raster maps project global suitability for converting land to three agricultural categories. The data was created by Mirza Čengić of Radboud University Nijmegen using Artificial Neural Network models trained on conversion data from 2003 to 2013. Cross-validation indicated good model performance with a mean Area Under the Curve value of 0.95.

Use Cases

• Projecting future agricultural expansion worldwide based on high-resolution suitability scores.
• Assessing local environmental impacts of land conversion using fine-grained spatial patterns.
• Integrating suitability layers into global land change models to improve spatial resolution.
• Analyzing the influence of topography, climate, soil, and accessibility on agricultural land conversion.

Strengths

• High spatial resolution of 10 arc-seconds (~300 meters at the equator).
• Models demonstrated good performance with a mean cross-validation AUC of 0.95.
• Hindcasting from 1992 to 2002 revealed similar robust performance (AUC = 0.94).
• Data covers three distinct agricultural categories derived from ESA CCI land cover.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• The training data covers conversions from 2003 to 2013, which may not reflect recent trends.

Provenance

Source

Radboud University Nijmegen

Collection Method

Artificial Neural Network models relating historical land conversion locations to explanatory variables.

Time Range

Model training based on conversions from 2003 to 2013.

Geography

Global