Urban land spatial optimization is one of the important issues in urban planning and land resource management. As the speed advancement of urbanization and the continuous increase of population, the rational use of land resources has become the key to sustainable urban development. Based on this, the study adopts the optimization goals of maximizing gross domestic product (GDP), reducing aerosol optical thickness and non-point source pollution (NPSP) load, and reducing land use change costs and incongruity. Three constraints are set simultaneously, including minimum construction land, water body, and cultivated land area. In addition, a fast non dominated sorting genetic algorithm (NSGA2) with elite strategy is used to address it. The outcomes denoted that the iterative distance of the proposed algorithm on the Bin and Cohen functions was only 0.048%, which was 0.522% lower than that of the NSGA2. Meanwhile, the reverse iteration distance value of this algorithm was only 4.14%, which was 22.76% lower than the adaptive weighted genetic algorithm. In addition, the algorithm's Spacing value was only 4.28%, and the hypervolume index value was as high as 78.66%. This indicated that the research method had a good optimization effect on the optimal allocation (OA) of land space in urban agglomerations, providing scientific decision-making support for sustainable urban development.