A new approach describing the behavior of the infrared (IR) images of natural terrain is presented. The model takes into account the nonisotropic nature of the terrain and the image diurnal changes due to sun motion. The model constructs the spatial autocorrelation function (AF) of terrain images, its nonisotropic nature, and the diurnal variations of the parameters describing this function. The proposed model assumes that the AF is composed of a superposition of two terms, one representing the shadowing effect, while the second represents the clutter inherent physical properties. Both are represented by an elliptical AF that is characterized by three parameters: (1) the power coefficient (the relative strength of each process), (2) the axis orientation in the horizontal plane, and (3) the aspect ratio between the two major axes. An analysis of measured and simulated images shows that the shadowing term follows the diurnal sun motion. At sunrise its power coefficient is very high (it "turns on" in a very short time), and its major axis orientation correlates very well with the sun azimuth. The aspect ratio of the two axes is influenced by the sun elevation angle. On the other hand, the inherent term has almost a constant aspect ratio and orientation angle.