In the future, dry and wet extremes are projected to intensify, while the time between the extremes is expected to shorten. Therefore, both extremes should be analyzed and modelled together. Antecedent soil moisture strongly shapes catchment response: at low levels, more storage reduces run-off, but once a threshold is exceeded, run-off rises sharply in a non-linear way. Conversely, dry conditions may induce water repellency, leading to increased run-off even with low soil moisture levels. It remains unclear whether this non-linear relationship changes during drought–flood transitions, forming the first knowledge gap of this study. To address this, the study examines how soil moisture correlates with peak discharge during such transitions.

Furthermore, both hydrological and hydrodynamic studies have a modelling gap, as initial drought states are rarely included and modelling of compound extremes remains scarce. To address this, the 3Di hydrodynamic model is used, which has been widely applied for floods but not yet for droughts. This study examines how well 3Di simulates peak flow after drought using effective precipitation as input, and whether incorporating soil moisture conditions through recharge further improves the results.

The research is conducted in the Hupselse Beek as a case study. The standardized streamflow index (SSI) is a drought index that is used to assess the hydrological droughts. Criteria are set in order to find drought events, of which three are selected for further modelling within 3Di: one for calibration and two for validation. The correlation analysis is performed by analyzing run-off against the antecedent soil water content for all drought events.

In addition, a 3Di model of the study area is developed containing the domains of surface water and groundwater. Measurements are compared with model results to test performance. The simulations cover a one-week period that includes the extreme rainfall event. Horton infiltration values, effective porosity and hydraulic conductivity values are calibrated in a sequential way to see if the results can approach the observations, with performance assessed through key metrics.

The selected drought dataset contains 38 events, demonstrating a positive non-linear relationship between soil-water content and run-off. Furthermore, a threshold in soil water content is observed around 0.22 mm3/mm3. It is concluded that the Hupselse Beek remains responsive to soil moisture, even under hydrological drought conditions.

For 3Di, the groundwater results show relatively good key metric performance while the surface water deviates strongly in response to effective precipitation. For the recharge simulations, the results in performance are worse. The Horton infiltration needs to decrease to compensate for the decrease in input. The incorporation of soil moisture conditions, and the effect of not representing them therefore needs to be researched further for 3Di modelling. At the same time, the groundwater results highlight the potential of 3Di for modelling peak flows during hydrological drought.