Landscape Evolution in a Savanna Ecosystem:

Soils and Drainage Development at Mpala Ranch, Kenya

Patricia A. Jellison and Andrew K. Johnston.


Introduction

This ongoing investigation involves soil erosion and drainage development in a savanna ecosystem. This work was the subject of a poster presentation at the 1994 meeting of the Geological Society of America. The field work described below took place at the Mpala Research Center , an ecological and environmental research station on the bushed grasslands of the Laikipia Plateau in north-central Kenya.

The southern third of Mpala is covered by a volcanic flow. Inspection of satellite image data shows that the flow is composed of two distinct layers. The upper layer lacks stream valleys, with the exception of a single headcut that developed during the interval 1975-1987. Instead, the remotely sensed images reveal broad patterns of overland flow. In contrast, the lower flow layer has active stream valley development.

Methods

Remotely-sensed images were combined with vector information digitized from published maps. Stream lengths were mapped from a 1987 SPOT satellite image and Survey of Kenya maps, last updated in 1975. A map was produced indicating those stream channels that had experienced growth. Field work took place in 1993 and 1994. Soil and vegetation reflectances were measured with a Barringer Hand-held ratioing radiometer, and positions were determined using a Magellan satellite navigation station. Following field work, ground-controlled training sites were transferred to image data to produce a supervised classification map of actively-eroding soils. Eroding and non-eroding areas were statistically compared with slope, soil type and proximity to active stream headcuts. ARC-INFO software was used for digitizing and georeferencing. Classification and image processing was performed with PCI and ERDAS.

Preliminary Results

Based on available topographic information, eroded terrain is not strongly correlated with regional slope. Higher topographic resolution is needed to assess small-scale slope dependence. Erosion is mainly in the form of gully development adjacent to intermittent streams. The spectral discriminibility of eroded areas is not due to bare soil, but is due to the combined response of soils with vigorous vegetation colonization of the gullies (predominantly Acacia sp.). In regions of impeded drainage (i.e. on basalts/black cotton soils), vegetation can colonize and thus define the headward extensions of waterways before actual erosion is visible in satellite images.

These suggest that in semi-arid bushed savanna, gully development and soil erosion may be discriminated based on the spectral signature of the vegetation that colonizes such areas. Rapidly-eroding areas are not characterized by bare soil, as in many other arid terrains. In addition, vegetation may in some cases be useful to discriminate the potential track and extent of headward erosion before the actual headcut is developed (e.g. in areas of impeded drainage). Future investigations will concentrate on land use and erosion, and the implications for land management and conservation.



CEPS Home
CEPS Research

Revised: April 2, 2001 (jh)
1994-2001 National Air and Space Museum