Upscaling vegetation cover for runoff and erosion modeling

Upscaling vegetation cover for runoff and erosion modeling

J.P. Lesschen (1), L.H. Cammeraat (1), A. Meerkerk (2) and B. van Wesemael (2)

(1) Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, The
Netherlands (lesschen@science.uva.nl), (2) Geography Department, Université Catholique de
Louvain, Belgium

In semi-arid ecosystems the lack of available water and inter-plant competition leads
to sparse vegetation, often resulting in spotted or banded vegetation patterns. As a
result of the positive feedback between vegetation and water infiltration soil properties
become spatially heterogeneous as well with more organic matter and improved
soil physical properties under vegetated patches Consequently, vegetation is the key
factor controlling overland flow generation and the distribution of vegetation patches
also determines whether runoff becomes connected at the hillslope scale. Runoff and
erosion models should therefore include the influence of vegetation, and consequently
vegetation patterns have to be scaled up to run these models at catchment scales.
At plot scale we described vegetation patterns and erosion pathways and took detailed
aerial photographs of vegetation patterns. These images were classified and spatial
metrics of these patterns were calculated with FRAGSTATS. We found a linear relationship
between fractional vegetation cover and most spatial metrics, which offered
us a proxy to upscale vegetation patterns. From a QuickBird image we derived a fractional
vegetation cover map using a linear regression that had a R2 of 0.90. At hillslope
scale we mapped connectivity patterns after a large rain event in November 2006.
These connectivity patterns were compared with the fractional vegetation cover map.
The results showed that vegetation largely controls runoff and erosion patterns, not
only at the plot scale, but also at the hillslope scale, where 77% of all observed rills
were on areas with less than 30% vegetation cover.
We used the LAPSUS model to simulate runoff and erosion at the catchment scale
and included the fractional vegetation cover map as extra layer to take account for the
effect of vegetation patters by adapting the infiltration characteristics. The simulations
show more realistic runoff and erosion patterns, because runoff was less connected
due to the sink function of the vegetation patches. This demonstrates that erosion
models should incorporate the relevant sinks of the hydrological system to prevent
overestimation of runoff and erosion at broader scales.