Drought dynamics across the hydrological cycle – an extensive validation of the National Hydrological Model of Denmark

Abstract

Abstract. Droughts are gaining attention in temperate regions, as underscored by the severe European droughts of 2018 and 2022. In Denmark, these events caused widespread agricultural losses, degradation of surface waters and ecosystems, and infrastructure damage from soil subsidence. Hydrological drought propagation from precipitation deficit to soil moisture, streamflow and groundwater is shaped by topography, soil, vegetation, hydrogeology, and human activity. While streamflow and soil moisture droughts have been widely studied, groundwater droughts remain underexplored. In Denmark, where groundwater and surface water are closely linked, and groundwater resources are heavily relied upon, an integrated approach to drought assessment is essential. In this study, we compile a high-quality observational dataset, including soil moisture, streamflow, and groundwater levels, to systematically evaluate model-simulated drought and its propagation throughout all hydrological compartments by the National Hydrological Model of Denmark (DK-model), an integrated, distributed hydrological model. The DK-model's nationwide coverage, combined with Denmark's dense monitoring network for streamflow and groundwater, enables a detailed assessment of the model's ability to simulate drought events over up to 34 years evaluation period (10 years for soil moisture). This includes model skill in reproducing observed anomalies, drought event detection and response times, and propagation dynamics. The DK-model was found to reproduce drought indices well for groundwater levels and streamflow compared to respective observational time series (median Pearson correlation coefficient of 0.76 and 0.79, and F1 scores for moderate drought event detection of 0.59 and 0.58, respectively). For soil moisture, model performance was lower. Drought propagation, evaluated by accumulation periods for precipitation with optimal correlation to hydrological drought, is likewise reproduced well for streamflow and groundwater. In contrast, the model struggles with the soil moisture signal. By evaluating the DK-model's performance in simulating drought propagation, this study contributes to improving large-scale hydrological drought modelling and enhances the understanding of the strengths and weaknesses of this approach, while increasing its potential for drought analysis, monitoring, and forecasting. The findings provide critical insights into drought dynamics in temperate regions and support sustainable water resource management in a changing climate.