Using Experimental Models to Guide Data Gathering

In constructing a plausible mathematical model of a ground-water basin, including the flow equations and the data describing the physical parameters of the basin, use of a series of experimental models may be found to be the minimal-cost approach to guiding data-gathering programs while simultaneously developing a plausible model to be used for predicting the consequences of management decisions such as increasing or decreasing the amount of water to be withdrawn in various parts of the basin. The basis for experimentation is a sensitivity analysis, for which triangular probability distributions are suggested as appropriate methods for perturbing known values of parameters into erroneous values. The worth of improving the accuracy of measurements or estimates can be evaluated empirically by use of these techniques. Automated techniques for deducing basin parameters from histories of water level were developed but were not fully successful. Use of linear programming shows some promise but require further development before being routinely applicable.