Back to Existing Special Sessions >>
Optimization Problems in Hydrology
Convener: Stefan Finsterle, Lawrence Berkeley National Laboratory
This session will focus on computational methods and their application to various hydrological optimization problems, including (1) parameter estimation by automatic model calibration (inverse modeling), (2) optimal design of testing and monitoring systems, and (3) optimization of water resources management and remediation operations. Common to all these optimization problems is the need to minimize some objective function by adjusting certain model parameters that represent hydrologic properties, the system layout, or operational variables. The objective function may measure the misfit between calculated and observed data, actual (and/or penalty) costs, and other quantifiable objectives; it may also include various regularization terms. A large number of global and local algorithms have been proposed to map out or minimize potentially nonconvex, nonlinear, or noncontinuous response surfaces in high-dimensional, usually constrained parameter spaces. Successful solution of a hydrological optimization problem requires careful parameterization of the problem, choosing an appropriate objective function, and selecting a robust and efficient minimization strategy. We seek contributions that address these optimization challenges by discussing the overall approach as well as algorithms and computational issues. We also welcome case studies that demonstrate the usefulness and limitations of mathematical optimization in hydrology.
