There is an ongoing desire to have spatially explicit uncertainty information for outputs of landscape models. However, the increasing complexity and sophistication of models makes this a difficult task for a number of reasons. Principal among these is the reality that many complex landscape models are integrative structures populated by third-party components that address individual aspects of landscape process. Whereas the individual components are generally well validated, the validation of models that address interactions of multiple landscape processes is often nearly impossible.

Nonetheless, it is important for model producers to be able to provide spatially explicit information to model users about the quality of model outputs. The goal of this paper is to describe one method that was used for a landscape model based on hydrological processes that also includes a variety of landscape dynamics – e.g., soil erosion, streamflow, biomass production.

The approach adopted was to statistically validate the hydrological model in a non-spatially explicit manner using ground-based data. In addition, uncertainty maps of the areas of interest were produced. These maps indicated five levels of uncertainty based on the spatial extent of measured data, generalised scientific literature on landscape and hydrological systems, and the availability of local expert knowledge.

This information was presented for only variables related to hydrological processes. Nonetheless, because hydrological processes are the key driver of other landscape processes, it was concluded that this information was extremely valuable for informing non-technical model consumers about the spatial reliability of a number of model outputs.