Atmospheric Compensation and Reflectance Retrieval

Research Team: Brent Bartlett (Ph.D. student), John Schott

Task Scope: This task explores the incorporation of ground-based measurements into existing atmospheric inversion algorithms. These measurements are then used to account for the variability produced by partial cloud cover.

Task Status: Many algorithms exist to convert imagery from units of either radiance or sensor specific digital counts to units of reflectance. This conversion removes unwanted atmospheric variability allowing objects on the ground to be analyzed. These algorithms perform with relatively low error levels in homogenous atmospheric conditions. In many cases however, clouds are present in the atmosphere, which introduce errors into reflectance retrieval. For example the relationship that is defined between sensor reaching radiance and ground reflectance in direct sun will not be the same as in a cloud shadow.

The location of each cloud is therefore found using a fisheye lens and a radiometric model is created of the hemisphere. Creation of this model is accomplished by looking from the ground into space using the radiative transfer code MODTRAN. The model is then used to simulate the radiance at different locations on the ground. Figure 5.1-1 shows both a two and three-dimensional visualization of a hemispherical model which has partial cloudy conditions.

Figure 3.5.1-1: Grey scale model showing the hemisphere broken into 1224 discrete parts, or quads. a) Two-dimensional projection. b) Three-dimensional view. c) Plot showing total radiance obtained by numerical integration of all quads

By predicting the spatial variation in downwelled radiance from the sky dome we expect to improve the inversion of image forming radiance to reflectance for scenes with scattered clouds. This is part of a long term effort to develop improved remote sensing analysis tools suitable for use over a wider range of operating conditions (i.e. to move beyond the severe clear only situation that is common today).