Measurement and prediction of natural and anthropogenic sediment sources, St. John, U.S. Virgin Islands

A quantitative understanding of both natural and anthropogenic sediment sources is needed to accurately assess and predict the potentially adverse effects of land development on aquatic ecosystems. The main objective of this study was to quantify sediment production and delivery rates in a dry tropical environment on the island of St. John in the eastern Caribbean. One to three years of measurements were used to determine values and empirical functions for estimating sediment production from streambanks, treethrow, undisturbed hillslopes, zero-order subcatchments, unpaved road surfaces, and road cutslopes. Sediment production also was measured from both undisturbed and roaded first-order subcatchments. Among natural sources of sediment, streambanks had the highest mean erosion rate at 100 Mg ha yr. The uprooting of trees along stream margins is estimated to generate approximately of 0.2 Mg of sediment per kilometer of stream per year, or about 0.1 Mg ha yr for a stream corridor that consists of a 9-m wide channel and a 3-m wide buffer zone. Undisturbed 40 m hillslope plots generated 0.01 to 0.27 Mg ha yr. Mean sediment yields from undisturbed zeroand first-order catchments were only 0.01 and 0.08 Mg ha yr, respectively. Unpaved roads that were graded at least every other year had sediment production rates ranging from 57 Mg ha yr for a road with a 2% slope to 580 Mg ha yr for a road with a 21% slope. Sediment production rates from ungraded roads were about 40% lower than those from recently graded roads, while production rates from steep abandoned roads were only 12 Mg ha yr. Cutslope sediment production rates ranged from 20 to 170 Mg ha yr, but their contribution to sediment yields at the road segment scale was relatively small. Since unpaved roads increase hillslope-scale sediment production rates by several orders of magnitude, the first-order catchments with unpaved roads had sediment yields that were at least five times higher than undisturbed catchments. The relative importance of each sediment source varies from catchment to catchment as a result of the abundance and spatial distribution of landscape types. The values and predictive functions developed in this study have been incorporated into a GIS-based model to predict catchmentscale sediment yields. Application of this model to three basins in St. John suggest that unpaved roads are currently the dominant sediment source, and that they are responsible for increasing watershed-scale sediment yields by 3–9 times relative to undisturbed conditions. Both the data from the present study and the GIS model can help estimate sediment production and catchment-scale sediment yields in similar environments. © 2007 Elsevier B.V. All rights reserved.