Abstract: The shortterm spatial distribution of City’s respiratable particulate matter (PM10) is mainly controlled by weather conditions, while the distribution over one year or longer is mainly dependdent on the sources of emissions. These sources of emissions are closely related to the distribution of urban traffic, industrial zones, urban builtup areas, and developing zones. While surface temperature difference within a year can integrate these features of underlying surface. So we can use this correlation and set up a model to estimate the spatial distribution of annual average PM10. Taking Wuhan as an example, we firstly used Landsat 8 thermal remote sensing data to inverse the summer and winter surface temperature of 2013 and 2014, and calculated the surface temperature difference of every year. Then, according to the principles of impact attenuation with distance, we used inverse distance weighting (IDW) to get the weighted value of every year surface temperature difference. After that, we formulated a linear regression between the weighted value of every year surface temperature difference and the PM10 annual average of 2013 and 2014, with an aim to find the best weighted value by comparing the accuracy, and to get the spatial distribution of the estimated model; its R2 reached 0655 and 0752, respectively. Finally, we got the spatial distribution of annual average PM10 in 2013 and 2014 based on the model. The spatial distribution of PM10 showed that the high value of 2013 and 2014 PM10 annual average mainly concentrated in the main city compared to suburban area. Low value is located mainly in the outskirts of towns, remote and mountainous areas or large bodies of water. Compared with the Kriging interpolation, the new method takes into account the influence of the underlying surface, so it can finely reflect the distribution characteristics and laws of PM10. It is more simple and effective. Key words：PM10 spatial distribution; the underlying surface; annual surface temperature difference; IDW; linear regression