Scaling simulation models for spatially heterogeneous ecosystems with diffusive transportation

The behavioral dependence of vegetation simulation models for spatially heterogeneous grasslands on simulation resolution was investigated. The dependence can be largely attributed to the non-linearity of the models. We showed that increasing scale or decreasing spatial resolution tended to overestimate the changing rate of an ecosystem using our landscape simulation model for alkaline grasslands in northeast China. A technique for scaling up simulation models with diffusive transportation was developed in this study by means of expanding the nonlinear driving functions in the model. The analysis showed that a simulation model for spatially heterogeneous landscapes might necessitate modification of both its mathematical structure and parameterization when applied to different scales. The scaling coefficients derived in this study were shown to be proportional to the variances or covariance of the spatially referenced variables, and can be estimated by running the model at a fine resolution for selected samples of the coarser grid cells. The technique was applied to a grassland landscape in northeast China and the results were compared with five-year observations on community succession. The comparison indicated that the proposed technique could effectively reduce overall scaling error of the model by as much as 80%, depending on the scaling difference between the fine and the coarse resolutions as well as the sampling scheme used.