Interpolating gaps in river suspended sediment records using artificial neural networks
Sediment transport is important in the management of rivers, catchments and floodplains. Sediment is monitored by measuring the suspended sediment concentration in rivers, but these measurements can be interrupted by various sensor malfunctions, leading to gaps in the record. An artificial neural network is developed for predicting suspended sediment concentration in these gaps, and is trained using high sampling rate contiguous records of quickflow and sediment concentration. The approach is evaluated by application to records from the Motueka River in the South Island of New Zealand.