Using orthogonal array sampling to cope with uncertainty in ground water problems

Uncertainty in ground water hydrology originates from different sources. Neglecting uncertainty in ground water problems can lead to incorrect results and misleading output. Several approaches have been developed to cope with uncertainty in ground water problems. The most widely used methods in uncertainty analysis are Monte Carlo simulation (MCS) and Latin hypercube sampling (LHS), developed from MCS. Despite the simplicity of MCS, many runs are required to achieve a reliable result. This paper presents orthogonal array (OA) sampling as a means to cope with uncertainty in ground water problems. The method was applied to an analytical stream depletion problem. To examine the convergence rate of the OA sampling, the results were compared to MCS and LHS. This study shows that OA can be applied to ground water problems. Results reveal that the convergence rate of the OA sampling is faster than MCS and LHS, with a smaller error of estimate when applied to a stream depletion problem.