Kaho'olawe, the eighth largest island in the Hawaiian chain, has experience
d major surface erosion during the last century due to overgrazing and mili
tary activity. In this paper we report the results from a series of physica
lly-based numerical simulations of hydrologic response to large rainfall ev
ents, that were performed for Kaho'olawe. An event-based rainfall-runoff an
d erosion model (i.e., KINEROS) was applied, in a geographic information sy
stem (GIS) framework, to make quantitative and distributed estimates of inf
iltration, Horton overland now generation, and erosion. A digital elevation
model (DEM) was employed to delineate individual catchments across the isl
and, and define areas of uniform slope within each catchment. A Landsat MSS
scene of Kaho'olawe was used to classify the island into three distinct la
ndcover categories via the normalized difference vegetation index (NDVI), T
he landcover categories were then used to further approximate the spatial p
attern of near-surface soil hydraulic properties across the island. Distrib
uted estimates of saturated hydraulic conductivity and sorptivity were obta
ined by ordinary kriging of 135 field measurements from the island. Sensiti
vity analysis was performed to characterize what impart the uncertainty in
soil-hydraulic property and rainfall data had on simulated runoff. Horton o
verland now and the related transport limited erosion were shown to be high
ly variable. The introduction of vegetation to denuded areas on Kaho'olawe
was found to significantly reduce erosion.