Mapping high-resolution soil properties with geoadditive models.
Basque Center for Applied Mathematics (BCAM)
Agriculture and Fishing
Neiker was interested in elaborating high-resolution maps of carbon stocks and soil texture properties in different land use at 0-30cm depth in the Basque Country.
Challenges and goals
The goals of the project were to compute the soil erosivity factor (R-factor) in MJ mm ha-¹ h-¹ yr-¹ , to identify areas of susceptible erosion, to relate with climate and environmental variables, to predict organic carbon stock (Mg C/ha) and to predict soil texture properties (%sand, %clay and %silt).
Mathematical and computational methods
- Geoadditive models are used as an unified framework for spatial prediction with smooth effects of climate variables such as average temperature, min/max temperature and precipitation.
- Carbon stocks are predicted using pedotransfer functions.
- Soil texture data are estimated using a compositional data approach using an additive log-ratio transformation and a multivariate Gaussian distribution.
Results and Benefits
An unified mathematical framework for spatial prediction of rain erosivity factor and soil properties at high-resolution is developed.
These maps contribute to agricultural planning of crops, forest management and environmental protection.
Stock carbon and soil texture maps are publicly available at GeoEuskadi.