An evaluating indicator for urban sprawl simulation and prediction

The research on urban development simulation and prediction plays an important role in urban planning and effective use of land. Many models have been built on land expansion, using different simulation and prediction techniques. To determine the feasibility of a method, criteria are needed to evaluate the result. Most currently-used criteria consider only quantitative factors, neglecting the spatial factors. However, generating equal results in area of lands does not guarantee the similar accuracy of two predictions. In this research, spatial factors are considered and a new evaluating indicator which compares not only the areas of the prediction and the actual land, but also the diversity of the spatial distribution is proposed. The result of this evaluating indicator named modified-Kappa coefficient provides a better criterion for model evaluation.