Site selection problem is one of the most important research fields of modern geographic information resource distribution. Genetic algorithm with strong universality and robustness can solve this problem. A common approach is to use a binarycoded genetic algorithm to locate sites on a grid map. To deal with the early maturity of the standard binarycoded genetic algorithm, this paper studies two methods that use different operators and observation concepts to solve early maturity problem of the standard binarycoded genetic algorithm, chooses to introduce the diversity measure and niche technology to improve the genetic algorithm, further explores the improvement of accuracy, online performance function and offline performance function of genetic algorithm in solving the site selection problem, and proposes a method to improve the niche technique in order to make every individual in the genetic group involved in genetic operation and avoid the situation where two identical individuals are involved in crossover operation. Finally, in the site selection experiment, the improved niche genetic algorithm is combined with the diversity measure to improve the performance of genetic algorithm successfully.