The necessary ingredients for deterministic modelling of hydropower management and climate change impacts on stream temperature in peri-Alpine streams

The thermal regime of regulated rivers is altered by rapid discharge variation downstream of hydropower plants (hydropeaking). This strongly modifies the thermal regime of such rivers, due to both the upstream water storage and the associated regular release to downstream, resulting in a thermal wave as the water temperature is different from the river temperature (thermopeaking). This temperature alteration needs to be considered when managing hydropower installations because of its influence on the health of aquatic ecosystems. On a longer timescale, global climate change is also influencing the natural thermal regime of rivers through changes in air temperature, vegetation, hydrology, etc. Thus, the assessment of the effects of hydropower on streams needs also to consider the extent to which changing climate will modify existing hydropower impacts, and also the mitigation methods that have been developed for current climate situations.

To evaluate the evolution of river temperature under different scenarios, deterministic coupled, hydrodynamic and temperature modelling can be used. Such models have been used previously to replicate the thermal regime of rivers or evaluate the impact of climate change on river temperature. However, there is a growing realisation that external drivers of stream temperature are more complex than hitherto thought, especially in per-Alpine streams. For instance, such streams can have significant shading whose importance changes as a function of time within the year. Equally, between the zone of water off-take and return, the residual flow may not occupy the full channel perimeter meaning that it is also necessary to consider the energy balance effects of water-adjacent sediments.

To address this challenge this paper identifies the necessary ingredients of deterministic coupled hydrodynamic and temperature modelling for hydropower impacted streams. This is supported by a unique and very high-quality stream temperature dataset which we use to identify the minimum process representation required for such models. In order to reproduce such data, we show that such models need to have a spatially-explicit and time-dependent correction of two key processes: (1) solar shading; and (2) stream bed sediment effects.