hello
when I calculate NDVI using landsat OLI sensor (after atmospheric correction using FLAASH algorithm), i see some of pixels (not water bodies) has negative NDVI and when i check those pixels location with google earth i see they are industrial places roofs.
i did this for many places and cities and this result was repeated.
is this true or there is a problem in my FLAASH algorithm parameters?
Hi Amir,
What sort of values are you getting? For paved surface/ built up areas, the NIR reflectance and red band reflectance values are very close. So if you're getting a negative NDVI but the magnitude is close to zero I would say this is realistic. But if the values are close to -1.0 than there is a problem.
Regardless, you can try using the QUAC algorithm for atmospheric correction instead of the FLAASH algorithm to see if the problem lies in the atmospheric correction process. FLAASH contains a lot of approximations, which might sometimes lead to poor results.
Good Luck!
Dear Fahad Khan Khadim , Taha Sadeghi Chorsi , Luís Flávio Pereira
Thank you very much for your attention. the negative NDVI values i mentioned is In the range from 0 to -.2 and most probably belongs to industrial places with galvanized sheet roofs. i checked NDVI values with QUAC atmospheric correction and the result was repeated. i attached the spectral curve of galvanized sheet to this message(from spectral library of USGS), It is clear the reflectance in the RED is greater than NIR and I thinck this leads to negative NDVI.
Best regards,
Amir Tahooni
Yes if it’s in the range of 0~-0.2 then to me it looks realistic. This can happen, in paved surfaces the red band sometimes can have high reflectance than NIR. As you got similar result with the QUAC algorithm then I don’t think FLAASH was the problem.
Hi,
i think that you dont use FLAASH method. in multispectral images its best to do atmospheric correction with Log-Residual method.
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing