L8 ac of mesotrophic waters

[IsabelBrand]

Dear Quinten,

I am using acolite to retrieve rhow and rrs for a mesotrophic water body with the average turbidity of 5 NTU, suspended solids 3 mg/L and chl-a max 15 ug/l for the period of the L8 image acquisition. My study area is located at the path 224 and row 63 and the image test was acquired on 10-September-2015.
In the months where the data presents higher constituents in the water, the rhow and rrs are retrieved very well, but in the dry season, it seems that the reflectance by water is very low. Then I was wondering if you could give some advice on the choice of the more suitable ac model in this case.

I have a RRC_655, RRC_561, RRC_483 as RGB and left side a RRS_561 image. So far, I've tried many possibilities and better results were accomplished with Swir, user defined fixed epsilon, with USGS TOA reflectance disable.[img]

[quinten]

Dear Isabel

This masking is caused by sun glint increasing the SWIR reflectance above the "non-water" threshold used in ACOLITE.

Due to the low viewing zenith angles of the satellite, many regions are affected by sun glint, especially at lower latitudes and in the eastern viewing direction. At the moment there is no good sun glint correction for coastal and inland waters, although people are investigating the problem (by developing separate glint corrections, or by integrating it into the atmospheric correction - e.g. POLYMER).

You could try increasing the SWIR threshold used in ACOLITE (Advanced Settings > Masking) but the performance of the atmospheric correction will degrade due to the glint signal.

Quinten

[IsabelBrand]

Dear Quinten,

Thank you for your reply. I am using only scenes with low glint effect and it seems to present good Rrs results for now. You did a very good job investing time on ACOLITE. I am very happy with my outputs, and I could not say the same when I was using Flaash results. ACOLITE allows us to perform atmospheric correction in easier steps, with output parameters as a bonus.

Isabel

[IsabelBrand]

Dear Quintem,

I was wondering if Rayleigh corrected reflectance (of the ACOLITE output) is based on the same definition described by Hu et al (2009) in "A novel ocean color index to detect floating algae in the global ocean" on page 2120 equation 3

Where,

Rrc = Pi*Lt/(F0 cosTheta0)-Rr

Lt = is the calibrated sensor radiance after adjustment for ozone and other gaseous absorption;
F0 = is the extraterrestrial solar irradiance at data acquisition time;
Theta0 = solar zenith angle;
Rr = Rayleigh reflectance estimated with 6S

With kind regards,

Isabel

[quinten]

Dear Isabel

That's right, the RRC outputs from ACOLITE are TOA reflectance - Rayleigh reflectance. If you input pressure or elevation the Rayleigh reflectance will be scaled accordingly. In the 6SV LUT used there is a glint factor which could lead to overcorrection in some cases - you could switch to the analytical Rayleigh computation if you experience this.

Note that by default there is no conversion to radiances within ACOLITE (due to Landsat 8 and Sentinel-2 data being provided as TOA reflectances).

Quinten

[IsabelBrand]

Dear Quintem,

Thank you for your very quick and clear answer.

regards,

Isabel

[bbbarnes]

Hi Quinten-
Probably not best to reply to a 2-year old post, but I'm looking for specifics on rhorc calculation in ACOLITE as discussed above.

Isabel's equation above specifies Lt as at-sensor radiance "after adjustment for ozone and other gaseous absorption" (Hu et al 2009 termed this Lt' in the equation she copied). Your response stated "RRC outputs from ACOLITE are TOA reflectance - Rayleigh reflectance." So, has the ozone/gaseous absorption adjustment occurrred? Does it require exogenous input of ozone (DU) and k_oz? [rho_t'= rho_t*e^(k_oz*oz_DU*((1/cosTheta0)+(1/cosTheta0)))]

Additionally, does the rayleigh reflectance term include surface Fresnel reflected light? In SeaDAS, rho_r does include it (so it's removed for Rrc calculation). For most land purposes (e.g., 6S), rho_r does NOT include this term. Since you use 6S, I assume that it's the latter, but I wanted to make sure.

Thanks in advance
-brian

[quinten]

Hi Brian

Note that in those two years the processing environment (IDL->Python) and the default algorithm (EXP->DSF) in ACOLITE have changed!

At the moment the Rayleigh corrected reflectance is computed as: (rhot - rorayl) / tr, with rorayl the Rayleigh reflectance and tr the two-way Rayleigh transmittance.

The current default algorithm (DSF, Vanhellemont & Ruddick, 2018 and Vanhellemont 2019) does not need to calculate the rhorc, so I added this in by popular demand. The gas (ozone, water) and (analytical) sky reflectance corrections are done on the rhot prior to running the DSF, but are currently not applied before computing rhorc. If you are looking to use an index like FAI, I believe that the rhos product (i.e. fully atmospherically corrected) gives good results over non-water pixels.

Hope this helps, and let me know if you have any suggestions.

Quinten

[bbbarnes]

Quinten-
Thanks so much for your prompt reply. Just to clarify on rorayl - this is from 6S, yes? So surface Fresnel isn't included, correct?
-brian

[quinten]

That's correct! I estimate the Fresnel separately, but that can definitely be improved.

Quinten