Contents

(Search or Access Pages Above and Below)

Crop Agriculture

Dark Objects for Landsat 8 Scatter Values

The images below show points centered on Landsat 8 pixels that have very low relative values in the blue, green, and red bands for entire Landsat 8 scenes. (Google Earth imagery is shown with point location in all graphics below.) These areas of relatively very low radiance are essentially dark objects. Corresponding radiance ratios shown below (as well as power relationships, not shown) are fairly consistent but are similar to a very clear atmosphere (1.777 and 0.518 for blue:green and red:green, respectively) yet should be more similar to a clear atmosphere (1.333 and 0.720) (Chavez, 1988). Other methods, such as using the lowest value not associated with a large break above it in the histogram will result in similar ratios. The Landsat 8 Relative Scatter Values adjust this whereby ratios and power relationships typically correspond to atmospheres at about the midpoint between very clear and clear. Using strictly the lowest DNs as a scatter selection method is erratic and should not be applied - it may result in ratios that are realistic or way out of the realm of possibility.

 

4/17/13: not in shadow in Landsat image

Blue DN value is 8090; 38790925 of the total 38790944 nonzero pixels have higher value; 14th lowest value out of 16582 different values (0.084 %). Green DN value is 6824; 38792060 of the total 38792061 nonzero pixels have higher value; the lowest value out of 17942 different values (0.006 %). Red DN value is 6214; 38791779 of the total 38791785 nonzero pixels have higher value; 6th lowest value out of 19796 different values (0.030 %). Radiance Ratios (converted to radiance based on .MTL file), blue:green = 1.850, red:green = 0.564.  

Dark object for dark object subtraction (DOS) scatter establishment for atmospheric correction and surface reflectance

 

6/20/13: in cloud shadow in Landsat image

Blue DN value is 8231; 38778443 of the total 38778466 nonzero pixels have higher value; 15th lowest value out of 23107 different values (0.065 %). Green DN value is 7035; 38779506 of the total 38779532 nonzero pixels have higher value; 20th lowest value out of 24768 different values (0.081 %). Red DN value is 6120; 38779252 of the total 38779272 nonzero pixels have higher value; 12th lowest value out of 27258 different values (0.044 %). Radiance Ratios (converted to radiance based on .MTL file), blue:green = 1.734, red:green = 0.466.  

Dark object for dark object subtraction (DOS) scatter establishment for atmospheric correction and surface reflectance

 

7/18/13: in cloud shadow in Landsat image

Blue DN value is 8040; 41634923 of the total 41634931 nonzero pixels have higher value; 7th lowest value out of 52089 different values (0.013 %). Green DN value is 6914; 41636179 of the total 41636182 nonzero pixels have higher value; 3rd lowest value out of 52938 different values (0.006 %). Red DN value is 6137; 41635866 of the total 41635891 nonzero pixels have higher value; 16th lowest value out of 55686 different values (0.029 %). Radiance Ratios (converted to radiance based on .MTL file), blue:green = 1.734, red:green = 0.503.  

Dark object for dark object subtraction (DOS) scatter establishment for atmospheric correction and surface reflectanc

 

8/23/13: in cloud shadow in Landsat image

Blue DN value is 8014; 41657020 of the total 41657031 nonzero pixels have higher value; 11th lowest value out of 38579 different values (0.018 %). Green DN value is 6780; 41658217 of the total 41658218 nonzero pixels have higher value; the lowest value out of 40001 different values (0.002 %). Red DN value is 6013; 41657900 of the total 41657920 nonzero pixels have higher value; 19th lowest value out of 43269 different values (0.044 %). Radiance Ratios (converted to radiance based on .MTL file), blue:green = 1.849, red:green = 0.482. 

Dark object for dark object subtraction (DOS) scatter establishment for atmospheric correction and surface reflectanc


 

 

Reference

Chavez, P.S., Jr. 1988. An improved dark-object subtraction technique for atmospheric scattering correction of multispectral data. Remote Sensing of Environment 24: pp.459-479.