Calibration and validation of the Land Use Scanner allocation algorithms

5 Calibration and validation of the Land Use Scanner allocation algorithms The Land Use Scanner is a spatial model that simulates future land use. Since its development in 1997, it has been applied in many policy-related land-use projects. In 2005, a completely revised version became available. In this version, land use can be modelled on a reduced scale, the smallest resolution now being 100 metres. Furthermore, the new version offers the possibility to model homogenous or discrete cells (containing only one type of land use) in addition to the heterogeneous or continuous cells (with more types of land use per cell), that were already available in the previous version. Each of these model versions uses its own algorithm to allocate land-use types to individual cells. In this report, both algorithms are described and their spatial allocation performance is calibrated using multinomial logistic regression, to specify the weights of the factors included in the local cell-based definition of suitability. Furthermore, the two model algorithms were validated by applying the calibrated suitability definition in a subsequent time period. This validation exercise indicates that both model algorithms provide sensible spatial patterns. In fact, the two different modelling approaches produce very comparable results, given equal starting points. In general, we conclude that the model is well-suited to simulate possible future spatial patterns in the scenario or policy-optimisation studies that are typically carried out by the Netherlands Environmental Assessment Agency. The current focus on the spatial allocation behaviour of the model implies that additional modelling aspects relating to the complete modelling chain in which the Land Use Scanner is applied, are not considered. These and other more conceptual modelling issues are covered in the model improvement project that the Netherlands Environmental Assessment Agency started in 2008.