Comparison of regionalization approaches in parameterizing sediment rating curve in ungauged catchments for subsequent instantaneous sediment yield prediction

- We compare regionalization approaches in the context of sediment yield modeling.
- We assess the catchment similarity index computed by multidimensional scaling.
- The sparse network of basin monitoring favors the catchment similarity approach.
- The combined formulation of catchment similarity and regression is proposed.
- We establish a regional sediment yield model for use in the Lower Mekong Basin.

SummaryObservations of sediment loading are lacking for rivers in many parts of the world, particularly in developing countries. There is a need for better understanding of what models might be available to offer estimates of sediment yield in such poorly gauged or ungauged areas. Therefore, the aim of this research is to determine an ideal regionalization methodology for estimating sediment rating curves in ungauged catchments for instantaneous suspended sediment yield (SSY) prediction. A comparison of three regionalization approaches (catchment similarity, regression and spatial proximity) was carried out in 16 catchments, in the Lower Mekong Basin, after discarding one with the evidence that the data point is not homogeneous. The highest quality results are provided by the catchment similarity approach in which a single donor catchment was selected in accordance with an optimum catchment similarity index computed by the multidimensional scaling technique. The regression-based approach is intermediate. The sparse network of basin monitoring in the study area does not favor the spatial proximity approach, producing the worst regionalization solution. The overall predictive accuracy was further improved through a combined formulation of catchment similarity and regression. This reveals that different approaches have different advantages and therefore, using only one of them cannot fully encompass a wide range of catchment heterogeneity. Since the ideal regionalization methodology performs satisfactorily in all 16 sites, a regional model with a relatively simple geomorphic framework was established for estimating catchment-scale SSY in ungauged rivers of the basin.