Modeling and projecting land-use and land-cover changes with Cellular Automaton in considering landscape trajectories: An improvement for simulation of plausible future states

International audienceThe modeling and projecting of land use change is essential to the assessment of consequent environmental impacts. In agricultural landscapes, land use patterns nearly always exhibit spatial autocorrelation, that is due in large part, to the clustered distribution of landscape features as hedgerows and wetlands, and also to the spatial interactions between land uses types itself. The importance of such structural spatial dependencies has to be taken into account when conducting land use projections, more especially as landscape features influence the precision of land use and land cover classifications of remote sensing imagery. The objective of this work is to improve land-use projections in considering landscape features in the modeling process. Cellular automata (CA), that provide a powerful tool for the dynamic modeling of land use changes, are a common method to take spatial interactions into account. They have been implemented in land use models that are able to simulate multiple land use types. This research adopts the spatial evolution concept embedded in CA and applies it to land-use and land-cover change study in one watershed inserted in an intensive agricultural area in Central Brittany, where water quality problems are often prevalent. A time-series of multi-scale and multi-temporal (including historical) satellite imagery and aerial photographs were used to determine both landscape features and the spatial characteristics and the temporal dynamics of land-use and land-cover over the period 1950 to 2003. Socio-economic and biophysical driving forces of observed changes have been established through a network of collaborating partners and agencies willing to share resources and eager to utilize developed techniques and model results. All these input data were complied, analysed and assessed in using spatial statistical techniques to quantify spatial dependencies. Cellular automaton modeling procedures were then applied to develop a spatially -explicit model- based simulations of future land use and cover change in considering that the evolving landscape frame slows down or accelerates changes according to the cases. Summary of neighbourhood conditions of each target cell reveals the dynamic processes of land use change constrained with the landscape frame and thus enhance our understanding on transition rules, the heart of a CA, in different types of landscapes. Model performance was evaluated in removing landscape features and in using shorter series of past observations. The model including landscape features as hedgerows network or wetlands distribution simulated the land-cover and land use at a higher accuracy than the model excluding landscape information for the three studied watersheds. In summary, our results showed that introducing landscape features improves simulations of land-use and land-cover future states, which will contribute to build more plausible scenarios of future changes