Scaling up of a mechanistic dynamic model in a GIS environment to model temperate grassland production at the regional scale

Regional‐scale ecological processes are mediated by processes occurring at local scales. Spatially explicit models are needed to understand the broad‐scale consequences of a large number of local processes, driven by factors which are heterogeneous at a broad scale. A Geographic Information System (GIS)‐based mechanistic model, which can be used as a flexible tool to investigate the regional‐scale effects of changes in environmental factors on herbage production from a Lolium perenne sward, was developed and used to investigate the consequences of aggregating driving variables at different resolutions. The model allows rigorous scaling up of previously existing field‐based modelling approaches within a GIS context. The model's mechanistic approach allows flexibility in the simulation of the separate effects of environmental driving variables and of cutting regimes. The driving variables (temperature, solar radiation, available soil moisture and soil nitrogen status) are scaled up using geostatistical techniques. The model is used to evaluate weekly changes in herbage production under environmental conditions, variable both in space and time, under different cutting regimes. It is shown that, in this model, aggregation at 1 km² is a good compromise between accuracy and practical feasibility, and that, while ignoring heterogeneity over many square kilometres can induce large errors, their magnitude and direction also depend on the model response curve to an input variable. The results obtained were consistent with the known trends in influential environmental factors. The programming within a GIS makes the model flexible in its application and, therefore, makes it easy to apply at a variety of scales.