Optical remote sensing systems have found use in many applications including land use analysis, climate studies, and weather forecasting. In many instances, the instruments have been designed as compromises between available technology and dominant scene phenomenology. As technology develops and provides more options to instrument design and operation, it becomes increasingly relevant that choices be made in a quantifiable manner. By viewing the design of a remote sensing system as an end-to-end process, including the scene, the sensor, and the processing algorithms, one can better justify the selection of instrument parameters based on their impact on the ultimate performance in a given application.

Examples of this end-to-end modeling perspective are presented for two types of remote sensing systems. The first involves systems for atmospheric sounding, i. e., the retrieval of atmospheric temperature and water vapor vertical profiles from a satellite. These profiles are used in numerical weather prediction models, as well as to calculate stability indexes in clear air. The National Oceanic and Atmospheric Administration (NOAA) is in the process of developing requirements for its next generation sounders aboard the Geosynchronous Operational Environmental Satellites (GOES). Results of simulation-based tradeoffs are presented showing the impact of spectral resolution and instrument noise on the accuracy of retrieved profiles.

The second example presented describes an end-to-end analytical model of hyperspectral imaging (HSI) systems. HSI systems have been demonstrated in several applications including surface characterization, material identification, and sub-pixel object detection and linear unmixing applications. The theory and approach behind this model is presented. Examples show the sensitivity of system performance to various parameters that span the remote sensing system. validation of the model is addressed. Comparisons between the model predictions and airborne HYDICE as well as spaceborne EO-1 Hyperion data are shown.