The 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.