We analyze a field experiment where ambient air is injected into the soil during the summer and extracted again during the winter. A multiphase model accounting for the conductive transport as well as the convective transport with the moving liquid and gas phases is used along with a more simplified single-phase model where the convective transport is due to the gas alone. The latter model also accounts for subzero wintertime temperatures. The multiphase model captures well both the seasonal variations and the actual test sequence, the main calibration being in the adjustment of medium permeabilities based on the observed pressure responses. The effect of the injection pump on the temperature and humidity of the injection air needs to be known accurately. Taking into account the humidity of the injection air explicitly instead of using humidity-corrected enthalpy values also has an effect. The effect of various humidity and specific enthalpy assumptions is of the order of 1-1.5degreesC, while ignoring the wintertime subzero temperatures has an effect of 1-2degreesC. These differences are of the same order of magnitude as the heterogeneity-introduced differences in field data. Using the simplified single-phase model typically appears to cause a difference of 1-2degreesC, but can yield an even higher deviation of the order of 3-4degreesC.