Hey,
ACOLITE provides the following SR products. Could you explain what the difference is between the products below? Which is the best for monitoring chlorophyll A in lakes? I am looking to use the SR data to develop a blue green band ratio algorithm.
1) RHOW: Water-leaving radiance reflectance
2) RRS: Remote sensing reflectance
3) RTOA: Top-of-Atmosphere reflectance
4) RRC: Rayleigh corrected reflectance
5) RHOAM: Multi-scattering aerosol reflectance
What ACOLITE reflectance product is best for monitoring chlorophyll A in lakes?
Re: What ACOLITE reflectance product is best for monitoring chlorophyll A in lakes?
Hi Amanda
You will probably want to use the water-leaving radiance reflectance (RHOW) or Remote sensing reflectance (RRS = RHOW / pi). These have been corrected for Rayleigh and aerosol contributions, and should represent the water signal.
You should be careful in using the blue/green ratio as the uncertainty on the aerosol correction is quite large for these bands. Furthermore, the absorption by CDOM and/or suspended sediments might influence the blue/green ratio and hence your chlorophyll retrieval.
Quinten
You will probably want to use the water-leaving radiance reflectance (RHOW) or Remote sensing reflectance (RRS = RHOW / pi). These have been corrected for Rayleigh and aerosol contributions, and should represent the water signal.
You should be careful in using the blue/green ratio as the uncertainty on the aerosol correction is quite large for these bands. Furthermore, the absorption by CDOM and/or suspended sediments might influence the blue/green ratio and hence your chlorophyll retrieval.
Quinten
Re: What ACOLITE reflectance product is best for monitoring chlorophyll A in lakes?
Thank you Quinten for your help. I am now analyzing the RHOW results for each band. Is there a reason why these value are mostly negative over water bodies? Should negative values be excluded?
Re: What ACOLITE reflectance product is best for monitoring chlorophyll A in lakes?
Hi Amanda
Negative values show that the atmospheric correction is wrong, that it is subtracting too much from the total signal. So you should exclude these values from your analysis. However, maybe it is possible to adjust the atmospheric correction.
It would be useful to have some more information on the lake and which images you are processing, as there could be several causes:
- water surface effects: sun or sky glint, foam or slicks
- adjacency effects: NIR/SWIR light scattering from adjacent bright surfaces
- the assumption of an exponential shape of the aerosol reflectance is not appropriate
- the lake elevation is significantly different from sea level and is not taken into account in the Rayleigh correction
Quinten
Negative values show that the atmospheric correction is wrong, that it is subtracting too much from the total signal. So you should exclude these values from your analysis. However, maybe it is possible to adjust the atmospheric correction.
It would be useful to have some more information on the lake and which images you are processing, as there could be several causes:
- water surface effects: sun or sky glint, foam or slicks
- adjacency effects: NIR/SWIR light scattering from adjacent bright surfaces
- the assumption of an exponential shape of the aerosol reflectance is not appropriate
- the lake elevation is significantly different from sea level and is not taken into account in the Rayleigh correction
Quinten
Re: What ACOLITE reflectance product is best for monitoring chlorophyll A in lakes?
Hey,
I'm using LS8 L1T imagery over Falls Lake in North Carolina. Sun glint could be a possibility. I tried to select open water body pixels to avoid influence of NIR/SWIR scattering but the whole lake is surrounded by trees which could have an influence. As for lake elevation, the lake is about 100 meters above sea level. Is it ok to use the standard rayleigh correction properties for a lake at this elevation or should I include a DEM?
I'm using LS8 L1T imagery over Falls Lake in North Carolina. Sun glint could be a possibility. I tried to select open water body pixels to avoid influence of NIR/SWIR scattering but the whole lake is surrounded by trees which could have an influence. As for lake elevation, the lake is about 100 meters above sea level. Is it ok to use the standard rayleigh correction properties for a lake at this elevation or should I include a DEM?
Re: What ACOLITE reflectance product is best for monitoring chlorophyll A in lakes?
Hi Amanda
Wikipedia lists a lake elevation of 76 m, which will only have a small impact on the Rayleigh scattering, but it is best to specify the elevation or use a DEM.
I have processed some images using the SWIR with variable epsilon, and specifying the elevation of 76 m for the Rayleigh correction. For some scenes I mainly find negatives on the shoreline (could be mixed pixels) and in the eastern, very dark part of the lake. Atmospheric correction for the blue bands is in general difficult, but in places with high CDOM absorption and strong adjacency effect it is even more challenging.
This typically happens for path/row 15/35, where the lake is in the western part of the OLI swath. In path/row 16/35 the lake is in the eastern part of the swath which is affected by glint, depending on season and region latitude.
The image below shows an RGB for scene LC80150352015293LGN00, with masked in red the pixels with RHOW_443 < 0. Here the negative values are mainly on the shoreline (mixed pixels?) and some speckle in the centre of the lake.
The image below shows the RHOAM for scene LC80150352015293LGN00, scaled between 0 and 3%. There is some speckle in this product as well as a slight increase of RHOAM towards the shoreline. (Clearer when playing with the colour stretch in the SNAP toolbox.)
The image below shows an RGB for scene LC80150352016104LGN00, with masked in red the pixels with RHOW_443 < 0. Here the negative values are along the shoreline but also in large parts of the eastern part of the lake.
The image below shows the RHOAM for scene LC80150352016104LGN00, scaled between 0 and 3%. Here the increase of RHOAM towards the shore is stronger. (Again, clearer when playing with the colour stretch in the SNAP toolbox.)
The image below shows an RGB for scene LC80160352016095LGN00, with masked in red the pixels with RHOW_443 < 0. Here almost the entire lake has negative reflectance at 443 nm, due to glint on the surface.
The image below shows the RHOAM for scene LC80160352016095LGN00, scaled between 0 and 3%. Here the RHOAM is much higher, showing glint that is erroneously included in the aerosol correction. I should probably include an automatic masking of the scenes affected by glint.
I hope this helps you understanding the issues!
Quinten
Wikipedia lists a lake elevation of 76 m, which will only have a small impact on the Rayleigh scattering, but it is best to specify the elevation or use a DEM.
I have processed some images using the SWIR with variable epsilon, and specifying the elevation of 76 m for the Rayleigh correction. For some scenes I mainly find negatives on the shoreline (could be mixed pixels) and in the eastern, very dark part of the lake. Atmospheric correction for the blue bands is in general difficult, but in places with high CDOM absorption and strong adjacency effect it is even more challenging.
This typically happens for path/row 15/35, where the lake is in the western part of the OLI swath. In path/row 16/35 the lake is in the eastern part of the swath which is affected by glint, depending on season and region latitude.
The image below shows an RGB for scene LC80150352015293LGN00, with masked in red the pixels with RHOW_443 < 0. Here the negative values are mainly on the shoreline (mixed pixels?) and some speckle in the centre of the lake.
- LC80150352015293LGN00_L2_RHOAM_865.png (66.36 KiB) Viewed 131357 times
- LC80150352015293LGN00_L2_RGB.png (637.84 KiB) Viewed 131357 times
- LC80160352016095LGN00_L2_RGB.png (761.05 KiB) Viewed 131357 times
- LC80160352016095LGN00_L2_RHOAM_865.png (72.48 KiB) Viewed 131357 times
- LC80150352016104LGN00_L2_RGB.png (734.88 KiB) Viewed 131357 times
- LC80150352016104LGN00_L2_RHOAM_865.png (69.9 KiB) Viewed 131357 times
Quinten
Re: What ACOLITE reflectance product is best for monitoring chlorophyll A in lakes?
Here I have stretched the RHOAM to arbitrary ranges to show some of the speckle and increase towards the shore.
Scene LC80150352015293LGN00.
Scene LC80150352016104LGN00.
Scene LC80150352015293LGN00.
- LC80150352016104LGN00_L2_RHOAM_865.png (74.99 KiB) Viewed 131357 times
- LC80150352015293LGN00_L2_RHOAM_865.png (68 KiB) Viewed 131357 times
Re: What ACOLITE reflectance product is best for monitoring chlorophyll A in lakes?
Here I have processed the images using a smoothing of the SWIR bands (see ACOLITE advanced settings; 16 pixel kernel size, 8 pixel dilation and fill option). This will get rid of the speckle and some problems near the shore; but the negatives in the darkest part of the April 2016 image remain. Of course this will not help the severely glinted image (not shown.)
LC80150352015293LGN00, RGB, negatives masked in red. No more negatives on this image.
LC80150352015293LGN00, RHOAM, smoothed, scaled between 0 and 3%.
LC80150352016104LGN00, RGB, negatives masked in red. On this image, negatives only remain in the dark eastern part of the lake. Note that the RHOAM is higher than in scene LC80150352015293LGN00.
LC80150352016104LGN00, RHOAM, smoothed, scaled between 0 and 3%.
LC80150352015293LGN00, RGB, negatives masked in red. No more negatives on this image.
- LC80150352016104LGN00_L2_RGB.png (726.13 KiB) Viewed 131357 times
- LC80150352016104LGN00_L2_RHOAM_865.png (16.77 KiB) Viewed 131357 times
- LC80150352015293LGN00_L2_RGB.png (626.98 KiB) Viewed 131357 times
- LC80150352015293LGN00_L2_RHOAM_865.png (14.65 KiB) Viewed 131357 times