Enset (Ensete ventricosum) as a Nature-Based Hydrological Intervention: A Process-Based Framework for Rainfall Partitioning, Direct-Runoff Regulation, and Flood Mitigation in the Upper Awash Basin, Ethiopia.

Abstract Nature-based solutions (NBS) are increasingly recognized as complements to conventional flood-control, yet their hydrological effectiveness depends strongly on how vegetation modifies rainfall partitioning, storage, infiltration, and runoff generation. In the Ethiopian highlands, Enset (Ensete ventricosum) a perennial food-security crop-possessed a distinctive architecture that may influence rainfall-runoff processes, but its hydrological function has rarely been represented explicitly in catchment-scale models. This study develops process-oriented ecohydrological framework that represents Enset as a hydraulic component of the rainfall-runoff system rather than solely as a land-cover class. The model explicitly accounts for leaf storage, midrib conveyance, stemflow concentration, and enhanced infiltration. The remaining non-Enset area is represented using an Analytic Hierarchy Process (AHP) loss formulation, expressed as L=0.40I+0.25D+0.15S+0.12E+0.08INT. Enset and non-Enset runoff pathways are subsequently combined through area weighting. We implemented the framework using observations from the Holeta catchment, integrating HEC-HMS with machine learning validation. For the 0% Enset baseline, the model achieved NSE = 0.652, KGE = 0.758, (R^2) = 0.655, RMSE = 2.138 m³ s⁻¹, PBIAS = 5.46%, and MAE = 0.938 m³ s⁻¹ during calibration. Independent validation produced NSE = 0.611, KGE = 0.718, (R^2) = 0.616, RMSE = 2.362 m³ s⁻¹, PBIAS = 9.76%, and MAE = 1.004 m³ s⁻¹. Increasing Enset cover from 0% to 75% of the cropped area reduced modeled direct runoff by 13.5-67.5% Q95 flows by 13.5–67.6%, indicating substantial flood-regulation potential. The results indicate that Enset cover can substantially modify catchment direct runoff through the combined effects of rainfall interception, hydraulic redistribution, stem-zone concentration, and enhanced infiltration. However, the modeled dynamic ring-storage filling ratio was very low, indicating that storage activation remains an important uncertainty. The study demonstrates the value of explicitly representing plant architecture in ecohydrological models and provides a basis for evaluating Enset-based land management as a complementary nature-based approach to flood-risk reduction in the Ethiopian highlands