LADAMER Land Degradation Assessment in Mediterranean Europe

The two main activities of RIKS in the Ladamer project are related to land use modelling and calibration/validation methods. For the land use modelling there has been progress both on the macroand the micro level land use modelling. For the calibration and validation task the focus has been on the development of goodness-of-fit measures for cellular models. Macro level land use modelling The macro level models operate at the NUTS 2 level they set the socio-economic environment for land use change at the cellular level. The main function of the macro models is to give a good prediction of the socio-economic status for each region in the model at different moments in the future. These predictions are based on scenarios that give the nationwide trends in time and historical data from the EUROSTAT REGIO database. Different variations of four types of models that are found in the literature are considered: Constant share model Shift share model (14 variations) Single constrained gravity model (2 main variations) Double constrained gravity model (2 main variations) Furthermore, a new densification model has been developed. The densification module intends to predict land use claims (acreages of land for different land use types) this module functions as an intermediate between the macro model and the micro model. The macro model produces results in the form of socio-economic figures (employment and population), whereas the micro model takes land use acreages as input. The new model is based upon a continuous equilibrium of land use pressure. Micro level land use modelling At the micro level of Ladamer a model is applied that has been used before in other EU projects (Moland, Murbandy, Medaction) it is the Constrained Cellular Automata (CCA). A characteristic of the CCA is that it uses many parameters, especially for the calculation of the neighbourhood effect. The large amount of parameters may be an obstacle in the calibration process. In order to reduce the complicatedness and to ease calibration the number of parameters in the Ladamer model will be reduced. This is possible by restricting the possible curves of the Neighbourhood Rules to composite exponential curves. First experiments with the reduce set of parameters yield possible results and two conclusions: 1. The reduced parameter set can substitute the original full parameter set without compromising the model quality. 2. With the reduced parameter set the calibration of the model can be faster and more intuitive. For the cellular model it is necessary to have a land use / land cover map at a 1 km resolution. For this purpose the Corine dataset has been aggregated from the 250m grid to a 1000m grid.