Constrained Optimization In Designing Groundwater Monitoring Network: Analysis Of Non-Linearity Between The Desired Accuracy And The Data Network Density

摘要

A water level monitoring network in a small watershed of about 60 km~2 in a crystalline fractured aquifer under semi-confined condition has been optimized. Initially a network of about 56 wells fairly distributed in the area and based on the drainage system was selected. The monthly water-level measurement has been analysed geostatistically. The standard deviations of the measured water levels have been on average 12 meters. There have been a number of constraints on measuring the water-levels in the shortest period of time and it was decided to reduce the number of wells in the network without loosing the benefit of measurements. Thus a monitoring network to have only 40 wells and providing a maximum standard deviation of the estimation error (σ_c) of 8 meters was designed. In this paper, it has been studied that if the σ_c varied from the value of 8 meters what could be the effect on various norms of optimization and the final number of measurement points to make the network optimal. It has been found that for the same change in σ_c when σ_c has lower value the number of measurement points are much more compared to the same change in σ_c for the greater value of σ_c. The variation of other norms of optimization has also been studied with different values of σ_c and a non-linear relation was always found. Such an exercise provides the estimated water levels for example to calibrate an aquifer model in an unbiased way.