Water resources management problems can be computationally intensive and improved methods are needed to allow solution of more complex applications. In this paper, we study a numerical algorithm designed to efficiently solve water resources management applications such as groundwater management problems. The algorithm is a combination of a simple genetic algorithm and a local search method and is called a self-adaptive hybrid genetic algorithm (SAHGA). The paper presents new ways to improve performance of this algorithm together with an analysis of different alternative local search algorithms. The paper also includes an analysis of the reduction in population size that is possible when using SAHGA relative to a simple genetic algorithm (SGA). The results show that the improved algorithm is more reliable and effective in solving the proposed problem, with average savings of 68% with respect to the SGA.