Spatial Validation of Land Use Change Models Using Multiple Assessment Techniques: A Case Study of Transition Potential Models

International audienceUsing land use and cover change (LUCC) modelsfor the urban growth planning, environmental assessment, anddecision-making needs the establishment of an appropriatelevel of confidence in their performance. The objective of thisresearch is to explore the importance of using multiple assess-ment techniques in order to fairly evaluate the performance ofland use models. An application is conducted by using theLand Change Modeler for Ecological Sustainability (LCM)which is an empirical and transition potential model. LCM isapplied to model the agricultural to developed areas transitionin Rennes metropolitan area (France). The land demand isestimated using the Markov Chain model; whereas, the tran-sition potential map is implemented using the Multi-LayerPerceptron Neural Network (MLP) method based on historicalchanges and driving variables. The model performance isassessed based on a variety of the most commonly used val-idation techniques. At the study area level, the correctness anddisagreement analysis shows that LCM performs better atpredicting the amount than the allocation of developed areas.Additionally, landscape metrics reveal that LCM tends to pre-dict a fragmented urban form, which seems evident because ofthe large number of the individual urban patches. At the mu-nicipality level, the error budget analysis shows that the modelperformance, which varies highly between different subareas,needs to be improved. Moreover, the cross-tabulation betweenthe transition potential map and both the observed and thepredicted agricultural to developed areas transitions revealsthat the order of the transition potential values does not per-fectly fit the observed change; whereas, the predicted changeis not solely limited to areas with high potential