As part of an interdisciplinary effort to describe the hydrologic and surface energy balance of a semiarid rangeland watershed, remote and ground radiometric surface temperature measurements were collected at the Walnut Gulch experimental watershed near Tombstone, Arizona. To correct for atmospheric effects on the remotely sensed data, radiosondes were launched on and off the watershed, and the LOWTRAN7 radiative transfer model was used to evaluate these observed as well as two standard atmospheric profiles. Comparison of the results for five atmospheric profiles, for each of three days, showed no correlation between radiosonde location or launch time and resulting temperature corrections. Comparison of corrected temperatures using five different atmospheric profiles on each of two days with the corresponding ground-based radiometric temperatures demonstrated significant errors (greater than 2.0¿C) as the atmospheric profiles generally yielded underestimates of ground radiometric temperatures.