Early-warning signals of desertification

Landscape systems my undergo abrupt transitions as a result of a gradual change in system drivers. Such regime shifts, or critical transitions, are often considered undesirable because they cause large changes in the landscape that are often irreversible. A well known regime shift in land surface systems is desertification, i.e. the shift from a vegetated landscape to a largely unvegetated landscape, often with degraded soils and increased erosion. The process of desertification is often abrupt, while it may be driven by a rather gradual increase in grazing intensity. At a certain threshold grazing intensity, biomass starts to decrease, which results in increased throughfall and runoff, causing increased runoff erosion, reducing soil thickness, which again has a negative effect on biomass growth. This positive feedback loop results in a relatively abrupt degradation at the grazing intensity threshold.

It is notably hard to detect this upcoming regime shift, because mean values of the system state variables (e.g. soil thickness, discharge, vegetation biomass) show little change before a transition occurs. This problem has sparked research focused on finding alternative properties of the system that show a more marked change before a transition is coming. It has been shown that such so-called early-warning signals exist, more specifically higher-order statistics of state variables (e.g. instead of the mean value of discharge, the variance; instead of the mean vegetation biomass the spatial variation in biomass). Thus far, however, the existence of such early warning signals is mainly shown for virtual realities, i.e. modelled hillslopes. The aim of this study is to investigate the existence of early-warning signals in the real-world. You will address the questions of 1) What are the statistical properties of vegetation cover, soil moisture and/or discharge of various catchments at different stages of soil degradation or soil recovery? 2) Can differences in statistical properties be explained by the occurence of (or upcoming) system shifts?

You will answer these questions by a statistical analysis of timeseries of high-resolution remote sensing data, including soil moisture, leaf area index and vegetation cover and possibly hydrographs for the same area. We have access to a data set in an area close to Zaragoza, Spain, and cooperation with the research group in Zaragoza is an option if you choose for this topic. Results of this analysis will be combined with information on the occurence of soil degradation or recovery in the same area, possibly by making a field visit. If time allows (or if the study is done by two students), the study can be extended by a modelling study investigating the occurence of early-warning signals in similar, modelled, systems. Most of the data is already available.

Supervision: Dr Derek Karssenberg (Utrecht University)

In cooperation with: researchers in Spain (to be determined)

Location: Utrecht University, possibility to visit Spain to collect additional data

Period: to be determined

Number of students: 1-2

Program/track: Earth Surface Hydrology or Natural Hazards and Earth Observation

Prerequisites: courses in spatio-temporal modelling, geostatistics, remote sensing, hydrology, geomorphology, and/or natural hazards (content of project can be adjusted to your background)

Contact/info: Derek Karssenberg (d.karssenberg@uu.nl)