|Authors||C.M.I. MacGillivray, S.K. Donovan|
|Title||A Relative Potential Erosion Detection (PED) model for the upper Buff Bay catchment, parish of Portland, Jamaica: A Geographical Information System application|
|Journal||Scripta Geologica. Special Issue|
|Keywords||Jamaica; Buff Bay; Geographical Information Systems; IDRISI; geomorphology; soil; erosion|
|Abstract||This research introduces a Geographical Information Systems (GIS) model that predicts the location and relative susceptibility of humid subtropical hillslopes to sheetwash erosion. The extent of the erosion was based on the conservation potential of the existing vegetation cover. This is an original deductive and deterministic model (Potential Erosion Detection, PED) incorporating regionally applicable physical and land use factors thought to be influential. These were climate (agroclimatic zones), topography (aspect and slope angle), soil (texture, drainage, depth, aggregation), vegetation cover and land use (tillage activity). |
The study looked at surface erosion as a perceived problem in a post-colonial economy. The processes, cause and effect of erosion were considered, and socio-economic factors discussed. Data collection and the design of the model recognised potential errors and uncertainties.
This research was initiated in the upper basin of the Buff Bay River (Portland) in the Blue Mountains of Jamaica. The results revealed a steep terrain, erodible soils and half of the watershed had little understorey and low litter levels. The dominance of coffee ensured that just under half of the research area had soils that were regularly disturbed. The model estimated that 30 % of the upper watershed had the potential for moderate to extreme erosion, contrasting with much higher previous estimates. The statistical dominance of each factor was analysed, showing that soil erodibility dominated the top erosion classes, followed by slope angle, then land use. Alternative soil and vegetative cover parameters for application of the model to other watersheds were also compared.
The reliability of the model was analysed using a number of local empirical relationships between erosion and influential factors. The erosion stake Cumulative Erosion Potential values were not significantly related to the PED model scores, but limited results differentiation weakened this approach. The research met two of the three objectives and provided an important preliminary conservation model for the local agencies involved in watershed management.
|Note||prepared for publication by S.K. Donovan|
|Download paper|| http://www.repository.naturalis.nl/document/45041 |