5.17.1.B. Atmospheric Correction

Atmospheric correction is done to modify the DN values to remove noise, i.e., contributions to the DN due to intervening atmosphere. Lower wavelengths of visible spectrum are more subject to haze, which falsely increases the DN values.

Atmospheric correction is a technique used to adjust the digital numbers (DN values) of an image to account for atmospheric effects. These effects can distort the actual measurements captured in remote sensing images. One major atmospheric effect is haze, which tends to obscure visibility, particularly at low wavelengths of the visible spectrum. When haze is present in an image, the DN values can be misleadingly increased, suggesting that an object reflects more light than it truly does. Therefore, applying atmospheric correction helps to retrieve more accurate data from the images.

Imagine trying to take a photo of a beautiful landscape on a foggy day. The fog makes everything look blurred, and colors appear washed out. If you were to look at the photo later, you might think the colors of the trees are less vibrant than they actually are because of the haze. Now, if you were able to apply techniques to make the fog 'disappear' from the photo, you would see the landscape as it truly is. This is similar to what atmospheric correction does for images obtained from remote sensing.

The simplest approach for its correction is known as dark object subtraction method, which assumes that if there was no haze present then the dark pixel will have zero DN value, e.g., deep water in near infrared will have complete absorption, and therefore those pixels will have zero DN values. But in reality, we won’t find any dark pixel with zero values.

One of the most common methods for atmospheric correction is the dark object subtraction method. This method operates on the assumption that the darkest objects in a scene (such as deep water) would ideally have a DN value of zero if there were no atmospheric interference. However, in practical terms, every observed pixel will have some non-zero DN value due to noise and haze. To correct this, the lowest DN value found in the image is identified. This value is then subtracted from all other DN values in the image, effectively adjusting the dark areas to reflect their true absence of reflected light. By calibrating the DN values this way, clearer and more accurate representations of the scene can be obtained.

Consider trying to determine the depth of a lake. If there is a layer of dirt or pollution causing the water to appear darker than it actually is, your measurements might tell you the lake is shallower than it is. To get a true reading, you would first need to subtract that dirt layer's darkness from your measurements, just like the dark object subtraction method does with DN values in image correction.

--