Soil erosion is a serious and ongoing environmental issue in Ethiopia, mainly caused by intensive farming, overgrazing, deforestation, and poor water and wetland resource management. The lack of clear land management planning, which fails to consider topography, erosion severity, land use, and land cover (LULC), as well as the suitability of management measures, has exacerbated the problem. Sustainable land management (SLM) encompasses a diverse range of strategies and integrated approaches aimed at maintaining or enhancing land productivity and ecosystem services. It involves integrating land use and the topographical features of the watershed with environmental and ecological factors, as well as stakeholder participation and the adoption of best practices. This study focused on modeling an integrated and sustainable watershed management for the Gilgel Abay watershed in Northwestern Ethiopia. Key data sources included Landsat (5 TM and 7 ETM+), Sentinel-2 images, and ALOS PALSAR DEM with a 12.5-meter spatial resolution, along with 44 years (1981–2024) of rainfall data from the Climate Hazards Group Infrared Precipitation Version Two (CHIRPS-v2) dataset and soil information for the watershed. Changes in the historical LULC of the watershed were analyzed using the maximum likelihood classification algorithm in ERDAS Imagine 2022. The Revised Universal Soil Loss Equation (RUSLE) model, adapted for the Ethiopian highlands, was used to estimate the spatiotemporal dynamics of soil erosion and sediment yield caused by historical LULC changes. The influence of land management practices on soil erosion was assessed under two scenarios: business-as-usual and sustainable land management (SLM). Future LULC changes were also predicted using the CA–Markov chain model. Results showed significant shifts in land use within the Gilgel Abay watershed, with agricultural land increasing from 34.78% in 1985 to 68.31% in 2024, mostly at the expense of natural landscapes. Forest cover decreased sharply from 15.05% to 4.62%, while grazing land, shrub land, and wetlands also declined notably. Meanwhile, average soil erosion and sediment yield rose from 13.72 t ha⁻¹ yr⁻¹ and 9.6 t ha⁻¹ yr⁻¹ in 1985 to 24.38 t ha⁻¹ yr⁻¹ and 11.71 t ha⁻¹ yr⁻¹ in 2024. In 2024 alone, about 8.9 million tons of soil were eroded across the entire watershed, with 0.9 million tons transported downstream to Lake Tana. Erosion heavily impacted upstream sub-watersheds, specifically the central and southeastern, and southern sub watersheds, where erosion rates are estimated at 41.72 t ha⁻¹ yr⁻¹ and 30.80 t ha⁻¹ yr⁻¹ in 2024, respectively. The northern sub-watershed experienced minimal erosion at 5.49 t ha⁻¹ yr⁻¹, accounting for just 5.61% of the total soil erosion in that year. Under the business-as-usual scenario, the study watersheds' erosion and sediment yield are projected to increase to 30.69 t ha⁻¹ yr⁻¹ and 14.58 t ha⁻¹ yr⁻¹ by 2040. Conversely, the sustainable land management (SLM) scenario reduces these rates to tolerable levels of 7.78 t ha⁻¹ yr⁻¹ and 5.15 t ha⁻¹ yr⁻¹. Overall, total erosion and sediment yield decrease sharply to approximately 2,643,581 tons per year and 388,758 tons per year, representing reductions of about 70.38% and 57.16% from 2024 levels. These findings indicate that applying SLM practices of the SLM-Scenario could greatly reduce soil erosion and sedimentation, helping to protect Lake Tana and maintain watershed sustainability.