Computational aspects of chance-constrained sustained groundwater yield management

A methodology for including chance-constraints in models for maximizing regional sustained-yield groundwater extraction is presented. The procedure utilizes the spatially distributed mean and variance of saturated thickness in a confined aquifer. This data is derivable from kriging, a geostatistical technique. The technique is applied to a scenario of somewhat egalitarian agricultural development dependent on groundwater. Decision makers in this situation know that groundwater availability is insufficient to supply all potential demand. They plan to implement a sustained yield groundwater extraction strategy. They know that because of uncertain knowledge of aquifer parameters, at some point in the future, available groundwater in some locations will exceed the initially allocated amount. They also realize that some locales will probably have less available groundwater than was expected, based on initial knowledge of the aquifer. To avoid future social and political unrest, they prefer that the spatial availability of groundwater remain somewhat consistent with time. Thus, they develop a groundwater extraction strategy using a predetermined confidence level. They are then relatively certain that actual sustainable extraction does not exceed the allocated groundwater use rate for locales in which allocated pumping is at a specified upper limit.