Comparison of empirical and process-based modelling to quantify soil-supported ecosystem services on the Saclay plateau (France)

Following the rapid development of models to assess and map ecosystem services (ES) in the last decades, there is an increasing need for comparative studies testing their efficiency and accuracy against field data. The representation of soils in these models is often oversimplified and remains a major source of uncertainty in ES assessment and mapping. In this context, a first objective of the present article is to develop two approaches of increasing complexity (empirical versus process-based) integrating a realistic representation of soils, to map two provisioning (biomass and water) and two regulating (water quality and global climate) soil-supported ES on the Saclay plateau (France). A second objective is to compare them according to their relative ability to replicate direct measurements (absolute accuracy) in a cultivated soil and to translate gradients in soil properties into differences in levels of the soil-supported ES (relative accuracy). On the basis of currently available models, the soil-supported ES can be assessed only under a homogeneous simplified crop rotation. For the fixed land management used in this research, the soil type significantly influences the levels of the considered services. The process-based modelling is effective only in deep, homogeneous, and cultivated soils. By contrast, the empirical modelling is effective over a larger range of soils, but mostly for provisioning services. The ability to integrate a realistic representation of soils in ES assessment and mapping with common tools and models is currently limited in terms of land-use, soil diversity, or types of ES. Increasing the accuracy of ES assessment and mapping requires pursuing the effort towards developing and validating models that explicitly take into account soil processes and properties.

Automated summary

With the rapid development of models assessing and mapping ecosystem services, there is a growing need for comparative studies testing their efficiency and accuracy against field data. The representation of soils in these models is often oversimplified and remains a major source of uncertainty. The study found that process-based modeling is more effective in deep, homogeneous, and cultivated soils, while empirical modeling works better for a wider range of soils.