EM 1110-2-2907
1 October 2003
Figure 2-16. Atmospheric windows related to the emitted energy supplied by the sun and the
Earth. Notice that the sun's maximum output (shown in yellow) coincides with an atmos-
pheric window in the visible range of the spectrum. This phenomenon is important in optical
remote sensing. Modified from
j. Geometric Effects. Random and non-random error occurs during the acquisition of
radiation data. Error can be attributed to such causes as sun angle, angle of sensor, ele-
vation of sensor, skew distortion from the Earth's rotation, and path length. Malfunctions
in the sensor as it collects data and the motion of the platform are additional sources of
error. As the sensor collects data, it can develop sweep irregularities that result in hun-
dreds of meters of error. The pitch, roll, and yaw of platforms can create hundreds to
thousands of meters of error, depending on the altitude and resolution of the sensor.
Geometric corrections are typically applied by re-sampling an image, a process that shifts
and recalculates the data. The most commonly used re-sampling techniques include the
use of ground control points (see Chapter 5), applying a mathematical model, or re-sam-
pling by nearest neighbor or cubic convolution.
k. Atmospheric and Geometric Corrections. Data correction is required for calculat-
ing reflectance values from radiance values (see Equation 2-5 below) recorded at a sensor
and for reducing positional distortion caused by known sensor error. It is extremely im-
portant to make corrections when comparing one scene with another and when perform-
ing a temporal analysis. Corrected data can then be evaluated in relation to a spectral data
library (see Paragraph 2-6b) to compare an object to its standard. Corrections are not nec-
essary if objects are to be distinguished by relative comparisons within an individual
scene.
2-19
Previous Page
