Taking Nanjing as an example, using the geo-social data combined with optical remote sensing information from the Landsat8 and nighttime light satellites, we design the characterization parameters of the local detection environment within the 2 km buffer around the observation site, such as the anthropogenic factors (population density and anthropogenic heat flux), geometric parameters (area ratio of land use type and altitude), and physical optical parameters (area ratio of impervious surface and vegetation index), and investigate their impacts on the air temperature (AT) observations. The results indicate that under the same weather background, when the vegetation index and area ratio of water body increase and the area ratio of impervious surface decreases, the AT decreases. The quantitative regression model between the vegetation index, area ratio of impervious surface, and area ratio of water body around the observation site and AT is given, and passes over the 99.9% significance level. The proposed method can be used to quantitatively evaluate the air temperature distribution at some region with high resolution.