Lessons learned for spatial modelling of ecosystem services in support of ecosystem accounting

Assessment of ecosystem services through spatial modelling plays a key role in ecosystem accounting. Spatial models for ecosystem services try to capture spatial heterogeneity with high accuracy. This endeavour, however, faces several practical constraints. In this article we analyse the trade-offs between accurately representing spatial heterogeneity of ecosystem services and the practical constraints of modelling ecosystem services. By doing so we aim to explore the boundary conditions for best practice of spatial ecosystem accounting. We distinguished seven types of spatial ES modelling methods, including four types of look-up tables, causal relationships, spatial interpolation, and environmental regression. We classified 29 spatial ES models according to a judgement of accuracy and modelling feasibility. Best practice of spatial ES models varies depending on the reliability requirements of different policy applications and decision contexts. We propose that in best practice for ecosystem accounting an approach should be adopted that provides sufficient accuracy at acceptable costs given heterogeneity of the respective service. Furthermore, we suggest that different policy applications require different accuracy and different spatial modelling approaches. Societal investment in higher data availability of ecosystem services make models of a specific accuracy more feasible or would enable achievement of higher accuracy with comparable feasibility.