I have converted the DN value into radiance value of band 4 of 2001 and 2011. When I compare the radiance value between the the band of two time period it shows a difference. For example a pixel of 2001 image has value 23.019 and same pixel of 2011 image 54.556. Will the difference make any impact in further analysis? How can I reduce the difference?
Hi Sushanta,
Beside the fact that radiance changes due to terrain feature and atmospheric changes, I suggest you to check the spectral resolution of the sensors you are using, particularly for b4, which is a NIR band in both L5 and L7. However, I don't think there is much difference.What corrections you applied to the DN?
Maybe I misunderstood the question, but it seems logic that radiance change in a 10yr period.
When you say, further analysis, what do you mean? If you compute NDVI maps or classification and change detection algorithms, it does have an effect, especially if the two values fall into different classification classes.
If you provided us with more details about your issue we could answer you better.
Best regards,
Nic
I have used Landsat TM in this case and resolution is 30m for both cases.
I do supervised classification, PCA, and unsupervised classification after converting DN value into Radiance value. I have applied the radiometric correction procedure proposed by Markham and Barkan [1987].
As you said
"If you compute NDVI maps or classification and change detection algorithms, it does have an effect, especially if the two values fall into different classification classes."
Then how can i reduce this effect?
If you wanna do supervised classification, the radiance level should be the same. You can perform absolute atmospheric correction, or perform relative radiometric normalization. Transforming DN to radiance is not enough.
Thanks Xingwang for your answer. Do you mean DOS by radiometric normalization?
- Badges
- Science topic
- Similar topics
- Computer Science
- Images