A fundamental goal of hyperspectral remote sensing is to extract information related to the intrinsic spectral properties of remotely imaged objects. In a laboratory environment, where sensing environments can be carefully controlled, the spectral reflectance and transmission properties discussed previously can be accurately measured, and such intrinsic properties inferred from them. In a remote sensing context, however, the situation can be quite complex due to the uncontrolled nature of illumination sources, atmospheric properties, and other environmental variables. To deal with these issues, it is of paramount importance to at least understand the impact of these variables on remotely sensed spectral radiance, and ultimately compensate the collected data for such influences.

This chapter describes the principles of radiative transfer modeling, which is the methodology for understanding how radiation from a scene propagates to a sensor and is influenced by the environment. The first section focuses explicitly on characteristics of the atmosphere and natural illumination sources. The spectral nature of these environmental influences is explored based on a sophisticated radiative transfer model, which is grounded in empirical data of the intrinsic spectral properties of atmospheric constituents. The second section provides observation models for the received spectral radiance at the sensor, based on scene and environmental characteristics for solid, gaseous, and liquid objects. The remotely sensed observable is called pupil-plane spectral radiance, as it relates to the measureable quantity at an entrance pupil of a hyperspectral sensor.