Hazardous chemical gas leakage accidents pose significant threats to public safety and the environment, and how to accurately locate the leakage source strength and location is particularly important. Aiming at the defects of traditional leak source inversion methods such as long duration and low accuracy, this study proposes an efficient inversion method based on a multi-strategy improved Harris Hawk optimization algorithm. Firstly, the position of the Harris Hawk swarm is initialized using Logistic-Tent composite chaotic mapping to improve population diversity and global search capability. Secondly, an elite level strategy is employed to scientifically divide the population, thereby enhancing the local search ability and optimization accuracy of the algorithm. Finally, a logarithmic function is adopted to flexibly adjust the escape energy and achieve a smooth transition in algorithm performance. The results show that the improved Harris Hawk optimization algorithm has achieved significant improvements in both optimization efficiency and inversion accuracy. Even under complex terrain conditions, it can accurately control the relative error of the leakage source intensity and location within a maximum of 1.00%. This method greatly improves the accuracy and response speed of leakage source localization, providing a reliable solution for emergency decision-making in accidents.