Change detecting land use/land cover helps assess and quantify its impact on water resources. This study aims to assess the impacts of land-use/land-cover change on water resources using the SWAT model in the Tana sub-basin, Ethiopia. The research detects and presents the changes between three LULC maps (1986, 2000, and 2014). The results suggest that over the last 28 years, the water body is the least disturbed and sub-afroalpine vegetation is the most transformed in terms of coverage. Cultivated lands gained a large area of cover from the other types. Most of the vegetation cover showed a decreased trend. Forest land and grassland decreased continuously while wetland showed a small variation compared to the other cover types. On the other hand, bush and shrubland recorded about a 1% increase in the total area and an unexpected fast decline in the second period. LULC changes would have an impact on water resources in the study area. The average annual water yield increased by 14.88 and 12.6%, baseflow increased by 18.4% and decreased by 7.16%, surface runoff increased by 12 and 16.16%, evapotranspiration decreased by 18.39 and 13.49%, for 2000 and 2014, respectively, compared to baseline 1986.

The study focused on analyzing the variability and trends of climate parameters in the Tana sub-basin. Various statistical methods and indices were employed to assess precipitation and temperature patterns in the region. The findings indicated a statistically non-significant increasing trend in rainfall across the sub-basin, with values ranging from 1.64 to 5.37 mm/year. In terms of temperature, there was an increasing trend observed, but it was also not statistically significant. The seasonality index ranged between 0.87 and 1.03, indicating different rainfall distribution patterns. In 36.69% of the sub-basin, rainfall occurs in marked seasonal patterns with a long dry season, and the remaining (63.31%) is concentrated in 3 or fewer months, indicating a different rainfall distribution pattern. In addition, the study assessed the precipitation concentration and found that 57.5% of the rainfall data exhibited a strong irregular concentration, 41.5% showed an irregular concentration, and 1% exhibited a moderate concentration. The study underscores the presence of climate variability and trends in the Tana sub-basin, emphasizing the need to align agricultural and water resource management practices with the observed climate variability.

Flood occurrence in tropical regions is intensifying due to climate variability and land-use change, increasing the need for reliable flood response time estimation. Accurate prediction of flood lag time (TL)—the interval between the centroid of excess rainfall and peak runoff—is critical for flood early warning and water resource planning. However, TL estimation remains challenging in data-scarce regions because of complex interactions among watershed morphology, rainfall characteristics, and runoff generation processes. This study evaluates four machine learning (ML) algorithms—Random Forest (RF), Artificial Neural Network (ANN), Extreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM)—for predicting TL across twenty gauged watersheds in the Blue Nile Basin of Ethiopia. Fourteen physiographic and hydro-climatic watershed characteristics were used as predictors. Among the tested models, XGBoost achieved the highest training performance (R² = 0.98, NSE = 0.96), while RF showed better generalization in the test dataset (R² = 0.77, NSE = 0.70, KGE = 0.71). SVM produced the lowest prediction errors (MAE = 0.95; RMSE = 2.25) but had lower explanatory power (R² = 0.49). To enhance interpretability and practical applicability, ML-based feature importance was used to develop a parsimonious empirical model: TL = 0.8 + 0.011A − 0.023RI, where A is watershed area and RI is rainfall intensity. This model explained 51% of TL variability and retained much of the predictive skill of more complex ML models. The proposed hybrid ML–empirical framework provides a transparent and operational approach for flood response time estimation in tropical highland watersheds. Its broader applicability remains subject to additional watershed-level validation and regional calibration.