Technique is presented for producing estimates of near-surface air temperature (Ta) in complex terrain based on biweekly composite data from the National Oceanic and Atmospheric Administration (NOAA) advanced very high resolution radiometer (AVHRR). Results are tested against independently derived DAYMET gridded meteorological data. The model utilizes radiant surface temperature (Ts) extrapolations as a surrogate for Ta, and digital terrain information is used to adjust temperatures as a function of environmental adiabatic lapse rates. The Earth–Sun–sensor geometry information is used to constrain air temperature estimates in three different implementations. Increasing the complexity of geometric constraints on the model resulted in increased accuracies of the predictions, and yet caused a decrease in the number of days Taestimates could be made. When comparing DAYMET surfaces to the satellite estimates of Ta, correlation coefficients (R) as high as 0.742 and standard errors of the estimate of 2.73 were observed in the final model implementation. The logic presented here serves as a useful technique for relatively simple derivation of near-surface air temperatures in a variety of remote sensing applications.