Abstract:In recent years, air environmental pollution is becoming more and more serious. Aiming at the PM2.5 pollution in Guangxi, using the data of Guangxi weather station, the meteorological data of Guangxi air quality monitoring station were obtained by inverse distance weighted interpolation. Then, combining the data of air quality monitoring station and its results of meteorological data interpolation, the model of geographically weighted regression tension spline function (GWR-TSF) interpolation was established and used for PM2.5 concentration interpolation analysis in Guangxi. The results show that the GWR-TSF model has a good effect of PM2.5 concentration interpolation, and its root mean square error is 2.34μg/m3, which is 20.68% and 25.71% higher than that of ordinary Kriging (OK) model and geographically weighted regression (GWR) model, respectively. The mean absolute error of GWR-TSF is 2.13μg/m3, which is 20.22% and 11.62% higher than that of OK and GWR model, respectively; which has certain reference value for regional PM2.5 monitoring and early warning.