Groundwater pollution sources inversion based on local-global hybrid adaptive surrogate model

A local-global hybrid adaptive surrogate model was proposed based on optimal solution criterion and cross validation-Voronoi (CV-Voronoi) criterion to improve the inversion accuracy and computational efficiency of groundwater pollution source. The local-global hybrid adaptive surrogate model combined with genetic algorithm was applied to groundwater pollution sources inversion case. The inversion results were compared with those of the local and the global adaptive surrogate model. The comparison results reveal that the local-global hybrid adaptive surrogate model combined genetic algorithm had the highest inversion accuracy and the lowest computational cost. The pollution sources inversion results can identify the actual pollution source characteristics, and the maximum relative error was only 3.51%. The results of the paper prove the robustness of the proposed local-global hybrid adaptive surrogate model in improving the accuracy and computational efficiency of groundwater pollution sources inversion. © 2023 Chinese Society for Environmental Sciences. All rights reserved.