Cloud-Based Modelling of Soil Erosion and Sediment Yield Using an Integrated RUSLE–Slope Connectivity SDR Framework in a Data-Scarce Transboundary Basin

Soil erosion and sediment transport are major drivers of land degradation, reservoir siltation, and declining water quality, especially in data-scarce river basins where field observations are limited. While the Revised Universal Soil Loss Equation (RUSLE) is widely used to estimate soil erosion, it does not quantify the proportion of eroded material that reaches river channels. This limitation, this study integrated RUSLE with a slope-connectivity-based sediment delivery ratio model in Google Earth Engine to assess soil erosion, sediment delivery, and sediment yield across the Upper Zambezi River basin. Freely available geospatial datasets on rainfall, vegetation, land cover, elevation, and soils were used to derive the RUSLE factors. Mean soil erosion rates ranged from 13.52 to 26.42 t ha⁻¹ yr⁻¹ across the six sub-basins, exceeding globally accepted soil-loss tolerance thresholds. Mean sediment delivery ratios were low, varying from 0.037 to 0.052, which indicated substantial sediment retention. Gross sediment yield showed strong spatial differences, with Cuando Chobe exporting the highest annual sediment load of 10.26 million t yr⁻¹, while Lungue Bungo recorded the highest localised sediment production. Model validation using field-measured total suspended solids showed a strong positive relationship with sediment-yield classes (Spearman’s ρ = 0.778, p < 0.0001), and the Kruskal-Wallis test confirmed classes' significant differences (H = 18.998, p = 0.0008). Generally, the integrated RUSLE-SC-SDR framework provides a scalable, reproducible, and computationally efficient approach for basin-scale soil erosion and sediment-yield assessment in data-scarce environments, supporting the identification of critical sediment source areas and evidence-based watershed management and conservation planning.