Modification and validation of the InVEST erosion model for application at a national scale

Area of Science:

• Environmental Science
• Geoscience
• Ecology

Background:

• Soil erosion is a significant ecological consequence of human activities like agriculture.
• Accurate erosion modeling is vital for targeting conservation efforts, but faces challenges in spatial scale and site-specific variation.
• Existing models require adequate parameterization and evaluation, which can be demanding.

Purpose of the Study:

• To adapt and assess an erosion model using the InVEST platform for Great Britain (GB).
• To integrate GB-specific data, including crop types, sub-annual periodicity, a GB-specific erosivity layer, and NDVI-derived cover management metrics.
• To evaluate the model's predictive accuracy against riverine sediment concentration data.

Main Methods:

• Adaptation and parameterization of the InVEST erosion model (based on the Revised Universal Soil Loss Equation - RUSLE) for Great Britain.
• Integration of GB-specific input data, multiple crop types, sub-annual periodicity, a GB-specific erosivity layer, and Normalized Difference Vegetation Index (NDVI) for cover management.
• Validation of model predictions against sediment concentration measurements in rivers.

Main Results:

• The adapted InVEST model demonstrated good validation against riverine sediment data, achieving 78% predictive accuracy on normalized data and a rank correlation of 0.46.
• Absolute model values were overestimated by approximately 28-fold, particularly at high erosion levels, potentially due to data limitations (lack of peak flow measurements) and RUSLE methodological issues.
• Normalized Difference Vegetation Index (NDVI) derived cover management factors represent a significant methodological improvement for RUSLE modeling.

Conclusions:

• The study provides a valuable tool for relative soil erosion risk mapping in Great Britain, aiding in the targeting of conservation actions.
• Improvements in national sediment load monitoring, including measurements during peak flows and integration of within-stream sedimentation functions, are recommended for future model training.
• NDVI-based cover management factors offer a reproducible methodological enhancement for RUSLE modeling applicable globally, while sub-annual modeling may benefit seasonal climates.