Spatial analysis of factors controlling the presence of closed depressions and gullies under forest: Application of rare event logistic regression

Although numerous studies have reported on past geomorphic change, it is a common difficulty to identify the relative role of driving human or climatic forces. To that end, the possibilities of rare events logistic regression are explored in a 1300-hectare large, ancient forest in central Belgium. Here, 43 permanent gullies and 71 closed depressions were observed that were probably caused by an anthropogenically induced land use change in the past. This human origin was confirmed for a limited number of case-study areas by recent geomorphic research, with dating and soil profile analysis. The objective of this study is to analyze the factors controlling the position of all gullies and closed depressions in the study area and investigate their mutual relation. This will indicate whether or not the results obtained for the small case-study areas can be extrapolated to the rest of the geomorphic features found in the whole study area.The obtained logistic regression models show that both anthropogenic and physical factors are important in the spatial distribution of both geomorphic features. For the prediction of the occurrence of closed depressions, anthropogenic factors are most important however. The probability of finding a depression is larger close to reported prehistoric and Roman archaeological sites, on well-developed, undisturbed Luvisols and on northeast- to southeast-facing slopes. The local slope gradient of the soil surface is the most important factor in the multivariate model for predicting the presence of the permanent gullies. The occurrence probability of permanent gullies is also larger on northeast- to southeast-facing slopes. The anthropogenic factor expressed by the distance to the closed depressions is significant as a univariate model. This suggests that a critical slope gradient is a prerequisite for gully incision, but gully formation still needs to be triggered by a human-induced land use change. It can therefore be concluded that both the gullies and the closed depressions in the Meerdaal Forest are related to past human activity, either directly, through mining of loess or indirectly, through runoff production on cropland. This study also showed the potential of rare events logistic regression for analyzing the spatial pattern of geomorphological features in relation to environmental factors.