Simulation of long-term storage dynamics of headwater reservoirs across the globe using public cloud computing infrastructure.

Reservoirs play an important role in relation to water security, flood risk, hydropower and natural flow regime. This study derives a novel dataset with a long-term daily water-balance (reservoir volume, inflow, outflow, evaporation and precipitation) of headwater reservoirs and storage dynamics across the globe. The data is generated using cloud computing infrastructure and a high resolution distributed hydrological model wflow_sbm. Model results are validated against earth observed surface water area and in-situ measured reservoir volume and show an overall good model performance. Simulated headwater reservoir storage indicate that 19.4-24.4 % of the reservoirs had a significant decrease in storage. This change is mainly driven by a decrease in reservoir inflow and increase in evaporation. Deployment on a kubernetes cloud environment and using reproducible workflows shows that these kind of simulations and analyses can be conducted in less than a day.