Creating a visually-realistic yet radiometrically-accurate simulation of thermal infrared (TIR) imagery is a challenge that has plagued members of industry and academia alike. The goal of imagery simulation is to provide a practical alternative to the often staggering effort required to collect actual data. Previous attempts at simulating TIR imagery have suffered from a lack of texture---the simulated scenes generally failed to reproduce the natural variability seen in actual TIR images. Realistic synthetic TIR imagery requires modeling sources of variability including surface effects such as solar insolation and convective heat exchange as well as sub-surface effects such as density and water content.;This research effort utilized the Digital Imaging and Remote Sensing Image Generation (DIRSIG) model, developed at the Rochester Institute of Technology, to investigate how these additional sources of variability could be modeled to correctly and accurately provide simulated TIR imagery. Actual thermal data were collected, analyzed, and exploited to determine the underlying thermodynamic phenomena and ascertain how these phenomena are best modeled. The underlying task was to determine how to apply texture in the thermal region to attain radiometrically-correct, visually-appealing simulated imagery. Three natural desert scenes were used to test the methodologies that were developed for estimating per-pixel thermal parameters which could then be used for TIR image simulation by DIRSIG. Additional metrics were devised and applied to the synthetic images to further quantify the success of this research. The resulting imagery demonstrated that these new methodologies for modeling TIR phenomena and the utilization of an improved DIRSIG tool improved the root mean-squared error (RMSE) of our synthetic TIR imagery by up to 88%.