A green GDP accounting system based on the evaluation indicators provided by the existing literature is established. After determining the indicator system, the AHPEWM model is used to determine the weight of each subindicator Then, the Spearman correlation coefficient is utilized to find that this GGDP accounting system has a measurable impact on climate mitigation. Adversarial generative neural network is used to generate data for the three-level evaluation indicators, and then these data are put into the input layer of the BP neural network together with the original data. Then Total Carbon Dioxide Emissions (TCDE), Population Density (PD) and Forest Resource Value (FRV) are used as the output layer, which influence the climate mitigation, and train the BP neural network. It is found that when the data in the input layer changes, the indicators in the output layer also change, so the introduction of GGDP will have an impact on climate mitigation.