Characterizing sediment load variability in the red river system using empirical orthogonal function analysis: Implications for water resources management in data poor regions

This paper introduces a novel data-based modelling approach, Empirical Orthogonal Function (EOF), to characterize the sediment load variability at the downstream section of the Red River system in Vietnam. We modified the original EOF and performed it on the demeaned datasets of monthly total sediment load. Rating curves between the discharge and sediment load were used to validate the performance of EOF analysis. The results showed significant similarity between simulated and observed patterns. A dramatic decrease in sediment load has been observed via both the EOF and rating curve methods across the observation period (-1958-2021), with two dramatic decreasing phases identified (1988-2008 and 2009-present). The variation in the sediment load was mainly related to the seasonal and inter-annual hydrological conditions in the basin, as well as the damreservoir impact. Before 1988, river discharge and rainfall were well correlated to the first EOF mode, indicating that the natural forces were the main driving factor for the sediment load. However, anthropogenic pressure (dam-reservoir impact) emerged as an important driver to the declining of sediment load since the end of 1988 and became more severe towards the end of 2008. We conclude that the modified EOF method developed in this study successfully demonstrates a simple and efficient means of reconstructing and evaluating the sediment load variability over time in data-poor regions, influenced by complex natural and anthropogenic - drivers. It can, thus, be useful for water resources management in ungauged basins in the region.