Changing storage: A global perspective on reservoirs in a changing climate

Humans are an integral component of the terrestrial water cycle. Global water resources are fundamentally altered by anthropogenic climate change and direct human management.Dams and reservoirs are a key example, as they regulate the river flow and store freshwater.Worldwide, more than 45,000 reservoirs are built since the start of the 20thcentury. The creation of these new open-water surfaces impacts the interactions with the atmosphere. At the same time, anthropogenic climate change causes changes in the hydrological cycle and affects global freshwater resources. To assess the role of reservoirs within the Earth system and under a changing climate, they need to be represented in integrated modelling frame-works, like Earth system models. This thesis aims both to implement the role of reservoirs in the state-of-the-art Community Earth System Model (CESM) framework and advance our understanding of reservoirs in the Earth system across different spatial and temporal scales.

To start, we estimate the potential consequences of climate change and dam management for future water level fluctuations of Lake Victoria, located in East-Africa. Lake Victoria is the second largest freshwater lake in the world and controlled by two dams for hydropower.Using a water balance model forced with lake precipitation, evaporation and inflow projections based on simulations from the Coordinated Regional Climate Downscaling Experiment ensemble, lake level projections are conducted under various climate change and idealized dam management scenarios. The results reveal that the operating strategies at the dam are the main controlling factors of future lake levels, and that regional climate simulations used in the projections encompass large uncertainties. This case study therefore highlights the importance of dam operations and reservoir management when modelling future river flow and water resources.

In the following study, we zoom out to the global scale to provide the first estimate of the global heat uptake by inland waters. Mapping the different components of the heat inventory is key to understand the Earth system response to anthropogenic greenhouse gas forcing. By employing a combination of global lake models, global hydrological models and Earth system models, we quantify the energy stored in lakes, reservoirs and rivers from1900 to 2020. The total heat uptake amounts up to 2.6·1020J, corresponding to 3.6% of the continental heat uptake. Most energy is used to warm natural lakes (111.7%), followed by reservoirs (2.3%). Rivers contribute negatively (-14%) mainly due to decreasing water volumes. Further, dam construction and subsequent reservoir creation leads to a redistribution of heat contained in the water, and thereby increases the potential of inland water heat uptake by warming of reservoir waters, due to the high heat capacity of water.

Then, global reservoir expansion is implemented in the Community Land Model (CLM5),the land model of CESM, as dynamically changing lake area to account for the large in-crease in open-water area following reservoir construction. Land-only simulations covering 20thcentury highlight that reservoir expansion increases in terrestrial water storage and decreases the surface albedo, matching the increase in open water area. In addition,atmosphere-land coupled CESM simulations indicate that globally, reservoirs dampen the diurnal temperature range and mute temperature extremes in the present-day climate. The responses scale with reservoir extent and can be substantial locally, but the influence on global climate is limited.

Finally, we implement and evaluate a widely-used dam parametrisation of Hanasaki et al.(2006) in the global routing model mizuRoute to be coupled to CESM. As this dam parametrisation requires irrigation water demand to determine releases of irrigation reservoirs, we develop a new irrigation topology to integrate the irrigation demands of the vector-based catchments to specific reservoirs. Using a local model setup with observations from individual reservoirs as a benchmark, the reservoir parametrisation outperforms the natural lake scheme, highlighting a clear added value of our model development for river flow modelling. In the global application, using a vector-based river network and simulated runoff from CLM5, the reservoir parametrisation outperforms simulations without lakes for river flow, but shows a similar performance compared to the natural lake scheme. This could be attributed to biases in inflow seasonality and amount, originating from the CLM5runoff and detail of the river network.

Overall, this thesis advances the current understanding on the role of reservoirs in the changing climate and provides important steps towards better representing human water management in Earth system models. Globally, the effects of reservoir expansion on the global climate are small, but locally the influence can be substantial. Future work may build on this research by coupling the routing model mizuRoute into the CESM framework, and thereby represent a two-way coupling between land surface processes and surface water transport. Finally, this study is a direct contribution towards the next generation of Earth system models that fully integrate human management and climate change scenarios to investigate potential mitigation and adaptation strategies.