Roughness Effects on Sub-pixel Radiant Temperatures in Kinetically Isothermal Surfaces

Temperature/emissivity estimation from remotely measured radiances generally assumes that scene elements represented by pixels in fact have a single emissivity spectrum and are isothermal. Thus, estimated temperatures and emissivities are effective values that would be found if these simplified assumptions were met. In reality, the physical scene is neither homogeneous nor isothermal, and the effective values are not strictly representative of the scene. How much in error are they? In this study we report on the dispersion of radiant temperature from the unresolved scene elements comprising a pixel due to roughness for the simple case when the scene actually is isothermal: i.e., the kinetic (but not radiant) temperature is everywhere the same. We use a radiosity model adapted for thermal infrared and driven by cm-scale digital terrain models (DTMs) measured by LiDAR. The DTMs cover small (0.5-10 m) areas of natural surfaces in the Mojave Desert, California (USA). Also we use high-resolution FLIR images of the same natural surfaces to test the model predictions for the homogeneous scenes. The differences between effective and kinetic temperatures and the variance of the calculated radiant temperature distributions are reported as functions of root-mean-squared (RMS) elevations within the modeled terrain.