Data derived reservoir operations simulated in a global hydrologic model

Abstract

Globally, there are over 24 000 storage structures (e.g. dams and reservoirs) that contribute over 7000 km3 of storage, yet most reservoir data is not openly accessible. As a result, many studies rely on generalized assumptions about reservoir storage dynamics to create generalized operational policies. With the creation of remotely sensed reservoir storage datasets such as RealSat and GloLakes and localized datasets such as ResOpsUS for the contiguous United States, and the Mekong Data Monitor for the Mekong River basin, the inference of reservoir operations using data derived techniques has become much more ubiquitous for regional studies. Yet to our knowledge, there has been no global application of data-derived methods due to data limitations and model complexities. Our analysis aims to fills this gap by providing a workflow for implementing data derived reservoir operations in large scale hydrologic models. This methodology uses global satellite altimetry data from GloLakes, a parameterization methodology developed by , and a random forest model to extrapolate operational bounds. Our results demonstrate that our random forest algorithm can capture storage dynamics and that the associated errors are propagated from the type of data used. Additionally, we observe that deriving operational bounds from historical reservoir time series only directly impacts streamflow directly downstream of dams and has minimal impacts at the basin outlets. We do, however, observe that the data-derived methodology increases the accuracy of simulated global reservoir storage when compared to remotely sensed and observed storage observations. These derived storages are much lower than in generic operation schemes which suggests that current operational schemes are overestimating the amount of reservoir storage and potentially overestimating water availability. We also evaluated the sensitivity of our modelling framework to different downstream operating areas (i.e. 0 to 1100 km) and found that there were slight improvements when including downstream demands. Ultimately, our workflow allows global hydrologic models to capitalize on recent data acquisition by remote sensing to provide more accurate reservoir storage and global water security.