Evaluation of a typical hydrological model in relation to environmental flows

• A Critical-Flow-Calendar is produced linking ecology and hydrology.
• The hydrological models are tested for the period where flow criteria are critical.
• Flow criteria should be selected so they can be simulated by the hydrological model.
• The hydrological model should be tailored to simulate the relevant flows.

SummaryHydrological models and environmental flow criteria (EFC) are important in water management, but there are potential complications of combining the two. This study aimed to establish and demonstrate a framework for assessing a suitable combination of EFC and hydrological modelling. For this purpose, local expert knowledge on stream ecology is first used to establish a Critical-Flow-Calendar (CFC) depicting the typical periods for different types of stream flow being most critical to the selected ecological indicator. Then EFC are quantified for the period most critical to the specific case. Finally, the EFC are used to build objective functions for calibration of the hydrological model, and the hydrological model’s capability to simulate the critical stream flows are evaluated. The purpose is to tailor the hydrological model to simulate the critical stream flows and, if necessary, to select the EFC so that they can be simulated by the hydrological model.The utility of the method was tested for hydrological modelling of EFC in 49 catchments 25–763 km2 located at the island of Zealand in Denmark. Establishment of a CFC facilitated a constructive dialogue between ecologists and hydrologists which caused the hydrological modelling to focus on the most critical flows (low flows) and the summer low flow period. Seven different methods were applied for quantification of EFC using observations from 49 stream flow gauging stations. The Danish National Water Resource Model was then applied for hydrological modelling using (a) conventional model calibration focusing on annual water balance and (b) model calibration based on objective functions that were related to the CFC and EFC. Results showed that the 20% lowest stream flows were substantially overestimated (>100%) by the conventional hydrological model, whereas the low flow calibrated model performed better only substantially overestimating the 10% lowest stream flows. The relatively high simulation error for the lowest stream flows meant that the highest EFC was most suited to be combined with a hydrological model. To our knowledge this is the first study that tests a large scale hydrological model against EFC specifically for the period where the flow criteria are critical to the ecological indicator in question. In this case it is therefore recommended that EFC should be selected so that they can be simulated by a hydrological model and the hydrological model should be tailored to simulate the relevant flows.