The Importance of Scale in Land Use Models: Experiments in Data Conversion, Data Resampling, Resolution and Neighborhood Extent

Abstract

The investigation and modeling of land use dynamics can be conducted at different scales based on the objective of the study. However, few studies have looked at comparing various scale aspects, such as spatial resolution and the related neighborhood effect, for practical case study applications. In this chapter, we contribute to this under-explored area with a detailed study of how changes in the data preparation procedures and the scale decisions made in setting up a land use model can affect its performance. For these purposes we used a Cellular Automata (CA) based land use model, which we applied to the Madrid region in Spain. In order to discover the most appropriate method for preparing input data, different vector-to-raster conversion and resampling strategies were tested with reference to 4 statistics. For vector-to-raster conversion, the cell center method was found to give the best results across all of the statistics. Furthermore, direct conversion from the original vector map to raster format at the desired cell size was found to give better results than resampling to the desired cell size from a different cell size. We also tested the effect of changing spatial resolution and cell neighborhood distance on a model’s goodness-of-fit to real data using a range of location and pattern metrics. Although differences were noted in the simulations, all the applications fitted the data satisfactorily. Nevertheless, the 50 × 50 m cell resolution applications were visually much more realistic, perhaps because this resolution was used in the initial calibration of the model. The results indicate that data conversion issues have a major effect on the quality of the input data. Additionally, models of this type appear to be much less sensitive to scale changes, either through cell resolution changes, neighborhood changes, or both, than is usually suggested by the literature.