Event-Based Design-Flood estimation in small catchments: how catchment storage controls sensitivity to antecedent moisture

Design-flood estimation for small fluvial catchments in the Benelux plus Germany region relies predominantly on simplified event-based methods, reflecting limited data availability and the absence of consistent regional guidelines. While such approaches are widely applied in practice, key modelling choices are often unguided, leading to large uncertainty and inconsistent protection levels across regions. This study investigates how basin similarity, catchment storage, and geology can be used to improve the robustness of event-based flood estimation in data-fragmented settings. This study integrates high-resolution observations from a network of experimental and operational basins with hydro-meteorological records. Event-scale hydrological signatures relevant to event-based modelling (time to peak, rising-limb characteristics, peak discharge, and runoff coefficients) are derived and analysed across simple antecedent moisture conditions (AMC) classes. Basin similarity is assessed using a combination of physical descriptors and observed event responses, and basins are grouped into response classes using clustering techniques. Leave-one-basin-out testing is applied to evaluate the transferability of response characteristics within and across classes. A parsimonious event-based rainfall–runoff model, representative of methods commonly used in design-flood studies, is then applied. Model parameters are constrained using empirical ranges derived from observed events. Model experiments systematically vary response-time formulations, AMC assumptions, and parameter sources (regional data, proxy basin, or class-based values) to quantify their influence on simulated peak discharges. Sensitivity and robustness are evaluated across storage and geology classes. Results show that uncertainty in event-based design floods is dominated by response-time and AMC assumptions, and that similarity-based parameter transfer can reduce uncertainty in storage-controlled catchments but performs poorly where storage contrasts are large or data are sparse. The findings provide empirical guidance on when event-based regionalization is best suited and highlight structural limitations of current practice. This work supports the development of more consistent, evidence-based design-flood guidelines for small basins in the Benelux region.