Different soils requires different temperature in-situ. Equatorial regions where there is high temperature and high rainfall, humification is rapid. There soils there have define structure and high organic content. While the reverse is the case in sahara desert or arid environment. Here temperature is high and low or not rainfall hence soils have low organic content. However, you can obtain soil data from USGS, for all region in their database. But reliably, go to the field and measure the soil temperature and compare the two data ie (observed data and simulated data) Good luck.
Whether you can reliably predict soil T from air T heavily depends on the spatial diversity of the terrain. For example, in a mountain setting with a very diverse topographical diversity on a short scale, soil T can deviate immensely (!) from air T. For example, imagine soil under a scarcely vegegated patch exposed towards the sun; that might reach >40°C while air T could be as low as 5 °C. On a flat plane or low spatial resolution, estimating soil T from air T might be more succesful. It all depends on the area and scale you want to use such estimates for, but generally I think it's a difficult task. For a global perspective, maybe check this: Article Global maps of soil temperature