The application of land change models has increased considerably over the last few decades: starting from synthetic small scale applications that showed something remotely related to urbanization they are now evolved into computer programs that allow for high resolution simulation of land changes as a result of a myriad of drivers. Moreover, such models are increasingly more used for foresight activities, such as scenario studies or policy assessments, or at least these activities are used as a justification for the development and application of such model (while many other good reasons exist to employ land use models). However, the evaluation of such land use models, regardless of their intended use, has received much less attention, even though their quality is of critical importance for many applications. This presentation will discuss some land use models, and aspects of the evaluation thereof, including the differences between pattern and location accuracy, and the difference between process and predictive accuracy. In addition it will briefly point at issues such as multifinality, equifinality, uncertainty, sensitivity, reference models, to finally come to arrive at the point where we can discuss what is a valid land use model, in the context of a given application.
减灾与应急管理研究院
人与环境系统可持续性研究中心
地表过程与资源生态国家重点实验室