EM 1110-2-2907
1 October 2003
l. Atmospheric Correction Techniques. Data can be corrected by re-sampling with the
use of image processing software such as ERDAS Imagine or ENVI, or by the use of
specialty software. In many of the image processing software packages, atmospheric cor-
rection models are included as a component of an import process. Also, data may have
some corrections applied by the vendor. When acquiring data, it is important to be aware
of any corrections that may have been applied to the data (see Chapter 4). Correction
models can be mathematically or empirically derived.
m. Empirical Modeling Corrections. Measured or empirical data collected on the
ground at the time the sensor passes overhead allows for a comparison between ground
spectral reflectance measurements and sensor radiation reflectance measurements. Typi-
cal data collection includes spectral measurements of selected objects within a scene as
well as a sampling of the atmospheric properties that prevailed during sensor acquisition.
The empirical data are then compared with image data to interpolate an appropriate cor-
rection. Empirical corrections have many limitations, including cost, spectral equipment
availability, site accessibility, and advanced preparation. It is critical to time the field
spectral data collection to coincide with the same day and time the satellite collects ra-
diation data. This requires knowledge of the satellite's path and revisit schedule. For ar-
chived data it is impossible to collect the field spectral measurements needed for devel-
oping an empirical model that will correct atmospheric error. In such a case, a
mathematical model using an estimate of the field parameters must complete the correc-
tion.
n. Mathematical Modeling Corrections. Alternatively, corrections that are mathe-
matically derived rely on estimated atmospheric parameters from the scene. These pa-
rameters include visibility, humidity, and the percent and type of aerosols present in the
atmosphere. Data values or ratios are used to determine the atmospheric parameters.
Subsequently a mathematical model is extracted and applied to the data for re-sampling.
This type of modeling can be completed with the aid of software programs such as 6S,
MODTRAN, and ATREM (see http://atol.ucsd.edu/~pflatau/rtelib/ for a list and descrip-
tion of correction modeling software).
2-6 Component 3: Electromagnetic Energy Interacts with Surface and Near
Surface Objects.
a. Energy Interactions with the Earth's Surface. Electromagnetic energy that reaches
a target will be absorbed, transmitted, and reflected. The proportion of each depends on
the composition and texture of the target's surface. Figure 2-17 illustrates these three in-
teractions. Much of remote sensing is concerned with reflected energy.
2-20
Previous Page
