Value of Local Data to Modelling Coastal Sediment Delivery in a Remote Unmodified Data-poor Tropical Island Environment

Globally accessible data are increasingly prevalent from remote sensing and broadly generalised studies. These datasets can be critical in remote and data-poor environments to predict water quality impacts from catchment land clearing driven by economic pressures for natural resource extraction. Efforts to collect field observations and calibrate detailed process-based models are both costly and time consuming representing a major challenge in remote data-poor regions. This study utilises the widely applied process-based SWAT model in a pristine, remote tropical catchment in the Solomon Islands comparing globally available input data with locally measured input data. Sediment delivery estimates to the receiving coastal lagoon from an uncalibrated model were first compared with calibrated model estimates. This calibrated model incorporated locally measured field data from the study area including metrological conditions, stream discharge rates, turbidity levels and total suspended solids concentration over a range of catchment flow events. The uncalibrated approach significantly underestimated coastal sediment delivery by at least two orders of magnitude compared to the calibrated model. The implications of the different sediment delivery estimates on the coastal receiving lagoon’s water quality was investigated using a three-dimensional, cohesive sediment transport model. This study highlights the need to undertake field campaigns to support model development in these remote, data-poor regions in order to increase confidence in the predicted impacts of catchment development.