Model Inputs and Data Requirements for Process-Based Stream Temperature Modeling in Regulated Peri-Alpine Rivers

Abstract

Regulated rivers can experience sharp temperature variations induced by intermittent hydropower production (thermopeaking). To mitigate ecological impacts, dam operators need to assess the impacts of hydropeaking on stream temperature, and to test scenarios that might reduce them. While stream temperature modeling has been investigated in numerous studies, few have systematically assessed how integrated processes and their representation affect model performance, and models capable of capturing both sub-hourly variations and long-term thermal dynamics remain a challenge. Herein, a stream temperature model within the HEC-RAS platform was used to model the thermal regime of a regulated river in Switzerland, with a 10-min timestep over the annual time-scale and for a 22-km long reach; and for which we had installed a network of stream temperature sensors. While the initial model demonstrated an acceptable performance at the yearly scale (Mean Absolute Error: 0.78–2.10°C and Kling-Gupta Efficiency: 0.55–0.85), this was not the case at the daily or seasonal time-scales. Two model corrections were found to be crucial; (a) the correction of potential incoming solar radiation for local shading; and (b) the representation of the heat flux linked to water-sediment exchanges. With these two corrections, the annual performance improved (MAE: 0.48–0.83°C and KGE: 0.85–0.93) as did the daily and seasonal performance. Although physically based, the model required calibration, underscoring the importance of high-quality in situ temperature data. The resulting model proves effective for practical applications in hydropower mitigation and river temperature management under complex flow regimes.