Large-scale groundwater management problems pose great computational challenges for decision making because of the spatial complexity and heterogeneity. This study describes a modeling framework to solve large-scale groundwater management problems using a newly developed spatial evolutionary algorithm (SEA). This method incorporates spatial patterns of the hydrological conditions to facilitate the optimal search of spatial decision variables. The SEA employs a hierarchical tree structure to represent spatial variables in a more efficient way than the data structure used by a regular EA. Furthermore, special crossover, mutation and selection operators are designed in accordance with the tree representation. In this paper, the SEA was applied to searching for the maximum vegetation coverage associated with a distributed groundwater system in an arid region. Computational experiments demonstrate the efficiency of SEA for large-scale spatial optimization problems. The extension of this algorithm for other water resources management problems.