SPARROW regional regression for sediment yields in New Zealand rivers.

Abstract

This paper reports on the application of the SPARROW regression model for predicting mean annual sediment yield in streams throughout New Zealand. SPARROW generates sediment (or nutrient) sources as a function of subcatchment parameters such as areas of different erosion terrains, modifies the loads according to generation and land-to-water delivery factors, and routes the sediment down a dendritic stream network, allowing for stream and lake erosion or deposition. The parameters for the model are determined by nonlinear calibration to measured yields at stream monitoring stations (over 220 sites in New Zealand). Areas of different erosion terrains were the key source in the model, while the mean annual rainfall and slope were key source modifiers. Land cover also emerged as a significant source modifier. A model with these terms gave a regression R of 0.925 to the measured yields, in log space. Stream erosion and deposition terms were not statistically significant, so they would have to be imposed rather than calibrated to stream yield data.