Evaluation of Hydrologic Models to Predict Sediment Export With Changing Land Use in Leeward Hawaiian Watersheds

Land-based sediments are a key threat to shallow coral reef ecosystems in Hawaii. Estimating sediment export is a critical step to being able to connect future land use changes with changes in sediment released to the coastal zone. However, empirically- and process-based hydrological models have proven difficult to adapt to Hawaii's geography, adding significant uncertainty to using available decision support tools. Four soil loss and sediment yield models, InVEST, N-SPECT, SWAT and GSSHA, were compared. Data including precipitation, flow discharge, and suspended sediment concentration were compiled from four leeward watersheds in the Hawaiian Islands. These were combined with the most recently available GIS data on soils, rainfall, land use and 10-m elevation. Results show that annual sediment export is typically underpredicted by an order of magnitude in the models. Moreover, soil loss predictions are spatially incongruent with field observations. Model results overestimate soil loss in the steep forested zones, where field observations show source material to be limited, and are not able to adequately capture human- and animal-disturbed material that connects hydrologically with the stream network. We suggest that the differences stem from a mismatch of processes that source sediments, including stream channel erosion and storage and shallow landslides, which are not included in all the models that are typically used for decision support. Moreover, different modeling platforms use different transport equations, which have not been validated for steep, mountainous watersheds. Changes in land use, such as new developments or cover crops, are obscured by models which consider steeply-sloped areas to be the primary source of sediment. The comparison suggests that decision support tools for Hawaii need a different approach for predicting sediment export with changing land use.