I would like to know the relationship (means whether reflectance increases with depth or reflectance decreases with depth) in the regression equation relating reflectance with in situ depth values for bathymetry estimation using landsat 7 data. I have followed the stumpf method to take the ratio of atmospherically corrected blue band divided by green band.
My regression is generating an inverse linear relation as shown in the attachment. Can anyone please provide your suggestions on what i have done is correct or wrong.
Thanks in advance...
vincent
Dear
Vincent Ferrer
From your graph it seems that you have made ratio of Blue to Green band and then using these index values, you are trying to correlate Bathymetry data... If it is so, your graph looks erroneous. However in terms of reflectance values, they (reflectance) will decrease with the increase in depth because light penetration decreases.
You can see this paper... They have directly mention what will happen with reflectance, if depth increases.
http://www.isprs.org/proceedings/XXXVII/congress/8_pdf/13_ThS-19/05.pdf
or
http://oceancolor.gsfc.nasa.gov/REFERENCE/LimnolOceanogr2003_Stumpf_et_al.pdf
Dear Dharmendra Singh
Thank you so much for your suggestions. I have followed the second link (the stumpf et. al. 2003) for bathymetry estimation and as said, I have made ratio of Blue to Green band and then using these index values, I have correlated the field measured Bathymetry data. I would be thankful if I can receive any further suggestions on how to improve on it.
Dear,
Ferrer
Since you are using bands with highest atmospheric attenuation, you may not getting good results. you can try with green and red band ratio. You may also go through correct atmospheric correction process. you can take also a reference of published paper by Roy, P.S., et al. in ISRS journal. The correct year i don't remember, but if you want i can arrange that paper.
good luck,
don't hesitate to ask .......
with regard
Dharmendra Singh
SAC/ISRO
Dear sir
Thank you so much for the repeated responses. To elaborate on the atmospheric correction i have implemented, I have estimated the L1% radiance and subtracted from the at sensor radiance. Then used the path radiance (t0 - upwelling and t1 - downwelling) calculated from the online facility for atmospheric correction of landsat data (http://atmcorr.gsfc.nasa.gov/) and used the formula mentioned in the attachment for estimating the water surface reflectance.
Yet I will be happy to look into the paper mentioned by you provided if you can arrange for the same and I shall do the process once again so that i may get some good results. Regarding green and red bands, I feel that most literature talk about only the blue and green band ration and also our study region being a inland water body with very high chl-a concentration, i have a doubt on how the ratio of green with red shall respond to bathymetry estimation.
Thank you once again for responding patiently to my queries....
Regards
Vincent
Vincent,
To me it makes sense that the blue/green ratio increases as the depth increases as shown in your graph. Keep in mind that the reflectance of BOTH the blue and green spectral bands decrease as the depth increases, however, what often happens is that the amount of green light decreases faster than the blue light as the depth increases. Therefore, the blue/green ratio will increase because the green reflectance values is getting smaller faster than the blue reflectance values. One thing that does surprise me is how small of a range the ratio values have (1.11 to 1.17); it seems like this could cause problems with noise.
Because the amount of water penetration decreases with increasing wave length the usefulness of the red band will be quite limited.
By the way, if you are interested look as some of my papers under the research gate profile; I developed the 'Improved DOS' and 'COST' atmospheric correction models.
Pat
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping