The Remote Sensing Group (RSG) at the University of Arizona is currently refining an automated system for the absolute radiometric calibration of earth-observing sensors. The Radiometric Calibration Test Site (RadCaTS) relies on semi-permanent instrumentation at the Railroad Valley (RRV) test site to collect data from which surface reflectance and an atmospheric characterization is determined. Multispectral surface reflectance is determined from calibrated ground viewing radiometers and assimilated to determine the hyperspectral reflectance used in radiative transfer calculations. The reflectance retrieval algorithm relies on an accurate determination of the diffuse sky irradiance for the time of interest. Currently, diffuse sky irradiance is modeled using the atmospheric characterization as input into MODTRAN5. This work investigates the accuracy of the diffuse sky modeling by comparing modeled results to measurements made at the test site. Diffuse sky irradiance from several alternative methods are also presented. Surface reflectance is computed and compared to in-situ measurements taken with a portable spectoradiometer.