HR: 16:15h
AN: H14A-02 INVITED [Abstracts]
TI: Modeling Limited Foresight in Water Management Systems
AU: * Howitt, R
EM: howitt@primal.ucdavis.edu
AF: University of California, One Sheilds ave , Davis, Davis, ca 95616
United States
AB:
The inability to forecast future water supplies means that their management inevitably occurs under situations of limited
foresight. Three modeling problems arise, first what type of objective function is a manager with limited foresight
optimizing? Second how can we measure these objectives? Third can objective functions that incorporate uncertainty be
integrated within the structure of optimizing water management models? The paper reviews the concepts of relative risk
aversion and intertemporal substitution that underlie stochastic dynamic preference functions. Some initial results from the
estimation of such functions for four different dam operations in northern California are presented and discussed. It appears
that the path of previous water decisions and states influences the decision-makers willingness to trade off water supplies
between periods. A compromise modeling approach that incorporates carry-over value functions under limited foresight within a
broader net work optimal water management model is developed. The approach uses annual carry-over value functions derived
from small dimension stochastic dynamic programs embedded within a larger dimension water allocation network. The
disaggregation of the carry-over value functions to the broader network is extended using the space rule concept. Initial
results suggest that the solution of such annual nonlinear network optimizations is comparable to, or faster than, the
solution of linear network problems over long time series.
DE: 1884 Water supply
DE: 6309 Decision making under uncertainty
DE: 6334 Regional planning (1880)
DE: 6344 System operation and management
SC: Hydrology [H]
MN: Fall Meeting 2005