Abstract: Based on the environmental and meteorological data of 6 automatic ambient air quality monitoring stations in the main urban area of Kunming from January to December 2019, the temporal and spatial distribution of PM_2.5 concentration in Kunming was discussed by using the function of ArcGIS spatial analysis and mathematical statistics method, and the prediction model of PM_2.5 concentration and monitoring parameters was established. The results showed that in April, the concentration of PM_2.5 at all stations except Longquan Town reached the maximum value in a year, and the lowest value appeared in July; the concentration at each station was low from June to November and relatively high from December to May. As for season, the concentration of PM_2.5 was lower in autumn and higher in winter, spring and summer. In terms of spatial distribution, the concentration of PM_2.5 in Wuhua District and Xishan District was high, while the concentration in Panlong District, Guandu District and Chenggong District was relatively low. The prediction models of the stations in Jindingshan, Biji Square, Longquan Town and Chenggong New Area have good fitting effect, while the fitting degree of Guandu District Museum and Dongfeng East Road was relatively poor.