Managing Forecast Uncertainty and Risk in Multireservoir, Multiobjective, and Multistakeholder Systems
Abstract
Hydroclimatic forecasts are often issued in probabilistic form to capture the uncertainties that remain in forecasting inflow magnitude and timing. These forecasts can then be used as input to stochastic management models to help operate reservoir systems and provide stakeholders with probabilistic forecasts of system variables, such as reservoir storages, releases, hydropower production, water withdrawals, and water quality parameters. However, due to high computational requirements, many stochastic management models only evaluate one management policy and are therefore not capable of exploring risk management options. Furthermore, such models usually neglect elements of the real-life water resources systems that they are supposed to represent. This presentation focuses on several advances designed to address these shortcoming and shows their application to a real-life reservoir system in California's Central Valley. This first part of the study focuses on exploring and managing the risks and uncertainties in multi-objective and multi-stakeholder systems. Traditional management models only find management policies that optimize the expected values of system benefits or costs, thereby not allowing operators and stakeholders to explicitly consider issues related to uncertainty and risk management. A technique that can be used to explore different risk management options was developed. The technique allows users to derive policies that produce desired probabilistic distributions of reservoir system outputs reflecting stakeholder preferences by imposing variance constraints on relevant system variables. The second part of this study highlights the expansion of an existing management model, previously only considering monthly water quantity fluxes, to incorporate water quality and flood control objectives. An existing water temperature model is analyzed and used to produce a reduced order model while the flood control objectives are considered through the development of routing models. Both of these models are directly incorporated into the existing management model and therefore allow for the exploration of management policies that can explicitly meet water quality and quantity objectives on several different time scales.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2012
- Bibcode:
- 2012AGUFM.H21C1186K
- Keywords:
-
- 1816 HYDROLOGY / Estimation and forecasting;
- 1880 HYDROLOGY / Water management;
- 6309 POLICY SCIENCES / Decision making under uncertainty;
- 6344 POLICY SCIENCES / System operation and management