Band-filtering applied to Rrs

Use this Forum to find information on, or ask a question about, NASA Earth Science data.
Post Reply
tihicha
Posts: 6
Joined: Wed Jun 05, 2013 4:47 pm America/New_York
Answers: 0

Band-filtering applied to Rrs

by tihicha » Thu Oct 08, 2020 10:49 pm America/New_York

Hello,

I know this was discussed and answered a long time ago, but with the current Rrs NASA OBPG data, is all Rrs NASA data from all sensors processed to the equivalency of a 11-nm tophat band-average filter, centered on each nominal sensors wavelength? I.e. if I wanted to compare to modeled or measured hyperspectral Rrs, I would ideally want to do band-pass integration with such 11-nm wide tophat filter first, before comparison with NASA satellite Rrs data?

If this is not applied to all sensors, which ones is it applied to and which ones - not?

Thanks a lot for your time!
Tiho

Tags:

daurin
Posts: 39
Joined: Thu Apr 22, 2010 8:06 am America/New_York
Answers: 0
Has thanked: 1 time

Band-filtering applied to Rrs

by daurin » Fri Oct 09, 2020 8:41 am America/New_York

Tiho,

If you want to compare satellite Rrs to hyperspectral Rrs, you should convolve the hyperspectral data using the Relative Spectral Response (RSR) functions for the given satellite. Also worth mentioning that the convolution should be conducted on the principal (ir)radiances prior to calculation of the Rrs. (Burggraaff, 2020; https://doi.org/10.1364/OE.391470). Spectral response functions can be found here: https://oceancolor.gsfc.nasa.gov/docs/rsr/rsr_tables/

Dirk

tihicha
Posts: 6
Joined: Wed Jun 05, 2013 4:47 pm America/New_York
Answers: 0

Band-filtering applied to Rrs

by tihicha » Mon Oct 12, 2020 4:05 pm America/New_York

Hi Dirk,

Thanks for the fast reply, and pointing me to that reference. I also thought about convolving with each sensor's SRF, but I remember a prior discussion here stating that Rrs has been corrected to a uniform 11nm top hat bandpass, except MERIS:
https://oceancolor.gsfc.nasa.gov/forum/oceancolor/topic_show.pl?pid=22152

Is this still the case? And for which sensors is it done now?

I was also thinking that using the individual sensor convolution would be hard on a merged data set like OC CCI.

Take care,
Tiho

OB.DAAC - SeanBailey
User Services
User Services
Posts: 1447
Joined: Wed Sep 18, 2019 6:15 pm America/New_York
Answers: 1
Been thanked: 2 times

Band-filtering applied to Rrs

by OB.DAAC - SeanBailey » Tue Oct 13, 2020 8:31 am America/New_York

Tiho,

The information in that older post is still accurate.  The Rrs output by l2gen has been adjusted to an 11nm wide square bandpass for *most* sensors.  Some are spectrometers, like MERIS, with narrow spectral response functions.  For these we do not adjust the data.  The full list of sensors with unadulterated Rrs (i.e. outband_opt=0):
MOS, MERIS, OLCI, OCM2, HICO, and OSMI  (we only distribute the MERIS and HICO data (soon OLCI L2s, but not there yet);  the others are capable of being processed with l2gen)

If you have hyperspectral in situ data, you could convolve those to the spectral response function of the satellite sensor(s) as DIrk suggests.  You would then want to process the data through l2gen with outband_opt=0 to turn off the "out of band" correction.

Sean

Post Reply