Kd from Landsat 5 and 7

[hnthang]

Thank you for your explanation. It's clear to me now.

May I ask you about k_d? Currently, ACOLITE produce k_d only from L8 and S2. Will it be available for Landsat 5 and Landsat 7 in the next release?

Many thanks!

[quinten]

The Kd from L8/S2 is computed using the QAA, with work on the band shifting and weighting between QAA v5 and v6 done by Dimitry Van der Zande. There are currently no plans to extend this to L5/7. L5/7 also miss a band around 443 nm which is needed in the current implementation.

If you have any suggestions on Kd calculation from L5/7, please let me know!

Quinten

[hnthang]

Can we estimate the k_d at 490nm?

This paper presented the estimation of k_d from landsat and ALOS AVINIR-2:

Estimating the diffuse attenuation coefficient from moderatespatial resolution, multi-spectral satellite data in a seagrass environment - https://ieeexplore.ieee.org/document/6721154

but I don't know whether it can be combined to ACOLITE for further processing.

Thanks,
Thang

[hnthang]

Another paper on the estimation of K_d from Landsat 5/7:

Remote sensing of diffuse attenuation coefficient patterns from Landsat 8 OLI imagery of turbid inland waters: A case study of Dongting Lake

Is it confidence enough to use the empirical model from the paper for the estimation of K_d?

[hnthang]

Hi colleges,

Regard my research, a water column correction is necessary after the step of atmospheric correction. Therefore, a K_d for Landsat 5/7 data is very important to reduce the impact of water column on image pixel values.

For this reason, I attempted to find a reliable solution for the estimaion of K_d from L5/7 itself . There is not equipment for me to measure the in situ K_d, threfore I would like to apply the quasi-analytical algorithm as this paper.

My question is, in stead of using rrs at the wavelength 440 and 555, can I use the band at 486/571 for L5 and 479/561 for L7?

Any sharing experiences would be appreciated!

Thanks,
Thang

[serge]

Thang,

here are the band combinations and coefficients for empirical Kd(490) https://oceancolor.gsfc.nasa.gov/atbd/kd_490/
Wavelength you propose are closed to KD2O and KD2L algorythms.

[quinten]

Hi all

I have split these last few posts into a new thread on a separate topic, to keep things a bit more clear.

Quinten

[Gringo_Perdido]

Hello:

Thank you to everyone for posting such useful information. Acolite is a wonderful program and this online community is an excellent resource.

I'm new to using the Kd algorithms, and I hope the community can clarify the appropriateness of Kd_490 for Landsat-5...so, to ask bluntly?

Yes or no, are Acolite's L2W qaa_v5_Kd_490 and qaa_v6_Kd_490 algorithms appropriate for Landsat-5 image processing? It is obvious the algorithms are well-suited to Landsat-8 and Sentinel-2. But Landsat-5 is not so clear to me.

I'm comparing time-series imagery spanning the Landsat-5 era through to Landsat-8, and I'd like to incorporate Kd_490 in the study. From what I've read above, it seems the Landsat-8 band wavelengths (and other image characteristics) are appropriate for the QAA Kd_490 algorithm to produce realistic results. However, I suspect the Landsat-5 results may not be so reliable. I've looked at the other papers and read the coefficient information, yet I don't seem to find any useful information about Landsat-5 other than its bands are similar to Sentinel-2 (http://www.coastcolour.org/hires_study.html). So does this similarity mean the QAA Kd_490 algorithm in Acolite is appropriate for Landsat-5 image analysis?

This thread is a couple years old and things change. Any clarification would be much appreciated. I'm using Acolite 20211124.0, if that matters. Apologies if I've overlooked an important detail in the thread links above.

Cheers
GP

[Gringo_Perdido]

Well, it seems that I've answered my own question. I went ahead and tried to batch process a bunch of Landsat-5 files with qaa_v5_Kd_490 and qaa_v6_Kd_490. Acolite returns a message:

QAA not configured for L5_TM.

Thus, it seems you can't force Acolite to do the default qaa_v5_Kd_490 and qaa_v6_Kd_490 anyway. I had wrongly presumed that Acolite would process image files fine and offer an answer...although the result may be inappropriate to use.

Hopefully, this clarification will help others. But I would still find it interesting to learn of any new work acquiring qaa_Kd_490 from Landsat-5.

Kinds regards,
GP

[quinten]

Dear GP

Thanks for your kind words and interest in ACOLITE. We have a recent paper out on three band QAA: https://doi.org/10.1016/j.rse.2021.112667

You can output Kd for this version of the QAA using the "p3qaa_Kd_*" parameter, or all available parameters with "p3qaa". There are some details in the ACOLITE manual: https://github.com/acolite/acolite/rele ... 1124.0.pdf

Let me know how it goes!

Quinten