Analysis of sediment transport in the Mekong using the lumped BQART model

Abstract

Decreasing sediment concentrations in rivers have led to a decrease in deposition in delta regions worldwide. This has led to an increased flood risk in coastal zones with large potential damages. This decrease in sediment load has been mainly attributed to the construction of dams and reservoirs. Predicting the amount of sediment in a river is vital in order to propose measures to resolve this problem, but discrete models are expensive, have a large computation time and require large amounts of data. Lumped models do not have this problem, but may be less accurate. The BQART model, for example, is a semi-empirical model which estimates the sediment load of a river using several factors and is easy to use. However, it was intended to be used for whole basins at a decadal timescale. The aim of this study was to test whether the BQART model is accurate at a monthly timescale and whether it can be used for smaller catchment areas. A second aim was to evaluate whether adding a bank erosion term would improve the model’s predictions. In order to test this, the model was implemented for the Mekong, which is a river that has seen a large decrease in sediment concentration due to the construction of many dams in China, Laos and Vietnam. The river was divided into subcatchments, to which the BQART equation was applied at along with a new bank erosion term. This bank erosion term assumed a constant rate of bank retreat along the length of the major river branches. The calculated sediment loads for these catchments were added together to obtain a timeseries of the monthly sediment load in the Mekong. The model was then calibrated and validated using existing data. The model showed promising results and managed to predict the decreasing pattern in sediment load rather well. The average sediment load values were within the ranges found by previous studies. However, sediment load peaks were underestimated. The model also still has several limitations and problems that still need to be resolved, such as an overestimation of sediment contribution from the delta. The bank erosion term proved less useful due to the many assumptions that were made in order to implement it with the current lack of data. If these issues can be resolved, the sediment load can be more accurately predicted and the model can be used to get more insight into smaller scale sediment dynamics.