Systemic change increases forecast uncertainty of land use change models

Cellular Automaton (CA) models of land use change are based on the assumption that the relationship between land use change and its explanatory processes is stationary. This means that model structure and parameterization are usually kept constant over time, ignoring potential systemic changes in this relationship resulting from societal changes, thereby overlooking a source of uncertainty. Evaluation of the stationarity of the relationship between land use and a set of spatial attributes has been done by others (e.g., Bakker and Veldkamp, 2012). These studies, however, use logistic regression, separate from the land use change model. Therefore, they do not gain information on how to implement the spatial attributes into the model. In addition, they often compare observations for only two points in time and do not check whether the change is statistically significant. To overcome these restrictions, we assimilate a time series of observations of real land use into a land use change CA (Verstegen et al., 2012), using a Bayesian data assimilation technique, the particle filter. The particle filter was used to update the prior knowledge about the parameterization and model structure, i.e. the selection and relative importance of the drivers of location of land use change. In a case study of sugar cane expansion in Brazil, optimal model structure and parameterization were determined for each point in time for which observations were available (all years from 2004 to 2012). A systemic change, i.e. a statistically significant deviation in model structure, was detected for the period 2006 to 2008. In this period the influence on the location of sugar cane expansion of the driver sugar cane in the neighborhood doubled, while the influence of slope and potential yield decreased by 75% and 25% respectively. Allowing these systemic changes to occur in our CA in the future (up to 2022) resulted in an increase in model forecast uncertainty by a factor two compared to the assumption of a stationary system. This means that the assumption of a constant model structure is not adequate and largely underestimates uncertainty in the forecast. Non-stationarity in land use change projections is challenging to model, because it is difficult to determine when the system will change and how. We believe that, in sight of these findings, land use change modelers should be more aware, and communicate more clearly, that what they try to project is at the limits, and perhaps beyond the limits, of what is still projectable. References Bakker, M., Veldkamp, A., 2012. Changing relationships between land use and environmental characteristics and their consequences for spatially explicit land-use change prediction. Journal of Land Use Science 7, 407-424. Verstegen, J.A., Karssenberg, D., van der Hilst, F., Faaij, A.P.C., 2012. Spatio-Temporal Uncertainty in Spatial Decision Support Systems: a Case Study of Changing Land Availability for Bioenergy Crops in Mozambique. Computers , Environment and Urban Systems 36, 30-42.