Improving Hydrologic and Energy Demand Forecasts for Hydropower Operations with Climate Change

Hydropower is an integral part of supplying clean electricity to the state’s electric grid. Besides providing baseload generation, hydropower is increasingly used to mediate load variability in the electrical grid due to the intermittent nature of wind and solar generation. Given the increasing potential of hydropower to meet energy demands, especially in the modernized electric grid, accurate and timely precipitation estimates are critical for optimizing hydropower scheduling. Despite having high-resolution satellite information, precipitation estimation for determining hydrologic flows from remotely sensed data suffer from methodological limita-tions. State-of-the-art deep learning algorithms, renowned for their skill in learning accurate patterns within large and complex datasets, appear well suited for precipitation estimation.

Advanced machine and deep learning mechanisms were developed to improve the accuracy of precipitation forecasts of an existing real-time Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) algorithm. The precipitation estimates delivered by the improved PERSIANN were used as the main input to a hydrological model to generate daily streamflow information.

Improving the accuracy and time resolution of streamflow data contributes to an increase in confidence and higher efficiency of hydropower scheduling decisions generated by both the reservoir and the hydropower dispatch models used by facility operators. Hydropower release decision making relies on multisource information such as climate conditions, downstream water quality, inflow and storage, regulation, and engineering constraints. To improve this decision making, this study developed meta-heuristic generalized reservoir releases and simu-lation algorithms for optimizing hydropower operations. A case study of a major operational hydropower facility serving California was presented to demonstrate the improvement of the streamflow simulation and forecast accuracy, based on improved precipitation estimates in PERSIANN products.