Suspended sediment fluxes in humid tropical highlands: process-oriented and multi-scale hydro-sedimentological modelling and statistical sedigraph reconstruction

Sediment-flux quantification is essential for erosion and reservoir-sedimentation management. This study couples the process-based Water Availability and Sediment Analysis – Semi-Distributed (WASA-SED) model with Quantile Regression Forest (QRF)-based sedigraph reconstruction to simulate streamflow and sediment dynamics (1990–2020) in two Lake Tana tributaries, Gilgel Abay and Gumara, Upper Blue Nile Basin, Ethiopia. Monthly calibration/validation performance was good (streamflow NSE = 0.55–0.82; sediment NSE = 0.74–0.86), with strong accuracy for peak events but underestimation at low-to-mid conditions. Decadal trend and change-point analyses revealed contrasting responses: Gilgel Abay showed rainfall intensification (2005) followed by delayed streamflow increase (2012) with stable, tree-cover-buffered sediment loads, while Gumara showed stable rainfall but synchronized abrupt increases in streamflow and sediment load (2006), linked to 63% wetland loss and land-use change. These results show that WASA-SED combined with statistical sedigraph reconstruction enables integrated hydro-sedimentological assessment in data-scarce tropical highlands, improving sediment budgets for Lake Tana and downstream infrastructure planning, including the Grand Ethiopian Renaissance Dam.