Fwd: RE: VIIRS geolocation

[Adam Dybbroe]

Thomas et al.,

Time is flying, and the launch is now only one week away!
I never commented your reply Thomas concerning reflectances and sun-elevation correction.

I had a statement at the end:
"This implies that for the shortwave channels the reflectances have been corrected (normalised) by the solar zenith angle. I am rather used to a practise where you
apply the normalisation by sun-zenith yourself afterwards. I would prefer it that way, but apparently this is not how it is for NPP/VIIRS."

Nigel you said you agreed.

I have to admit I haven't yet come any further with this, but I intend to do the coming days/weeks if possible.
But maybe someone else have?

I agree of course with your comments Thomas that a reflectance needs to take into account the solar zenith angle since it defines how many photons are hitting the footprint.
But, we still need to understand what is done at the terminator, don't we? Division by zero!
And I am not up till now been trying to think about what is exactly meant by a "terrain corrected solar zenith angle".

Is it so that data are masked out near the terminator, and if you still want to use the measurement in the solar channels you go for the radiance instead, and derive your own pseudo reflectance assuming sun in zenith?

Data are still useful I think at the terminator. You may use the spatial coherrence to give you a hint of you have clear surface or cloud for instance. Even with the uncertainty of atmospheric refraction, and the average surface slope over the FOV. A terrain model is still only a model, and associated with inaccuracies.

I would appreciate input if you have some, but otherwise I will continue digging into the documentation and the sample data.
However, the sample data doesn't help a lot here because they are far from terminator conditions.

Best
Adam



-------- Original Message --------

Subject:	RE: VIIRS geolocation
Date:	Mon, 19 Sep 2011 10:08:40 +0200
From:	Thomas Heinemann <Thomas.H...@eumetsat.int>
To:	'Adam Dybbroe' <Adam.D...@smhi.se>, Anders Meier Soerensen <Anders.S...@eumetsat.int>, Nigel Atkinson <nigel.a...@metoffice.gov.uk>, Trygve Aspenes <Trygve....@met.no>, Esben Stigård Nielsen <e...@dmi.dk>

Hi Adam,
Thanks for your analysis of the document!
As far as the reflectance is concerned, the correction of the incoming radiation by the sun incidence angle is part of the definition of REFLECTANCE and if we want to deliver reflectance instead of radiance, we need to define which sun-zenith angle should be used. If you use the terrain-corrected value, you will get wrong result for clouds, again. For land surface observations these terrain-corrections are nice, but for atmospheric applications, we probably should just try to switch them off wherever possible.  

A discussion group is probably a good idea. 
 
Thomas

----------------------------------------------------------
Thomas Heinemann
Meteorological Product Implementation Expert 
MOD
Tel. 4910
----------------------------------------------------------


-----Original Message-----
From: Adam Dybbroe [mailto:Adam.D...@smhi.se] 
Sent: Monday, September 19, 2011 9:27 AM
To: Anders Meier Soerensen; Nigel Atkinson; Trygve Aspenes; Esben Stigård Nielsen; Thomas Heinemann
Subject: VIIRS geolocation

Hi,

As Trygve I also "enjoyed" myself a little on my way home looking into the sample data and documentation.

I stumbled over a few writings which I think are relevant to our discussions last week:

----------------------------------------------------------------------------------
D34862-03E_CDFCB-X_Volume_III_PR.pdf
Page 287

Document page number = 300.

"""
While the geolocation angles included with VIIRS SDRs are with respect to the Geoid, the calibrated TOA reflectance is calculated using the terrain corrected solar zenith angle. Note that radiances are not corrected for out of band contributions.
"""
----------------------------------------------------------------------------------

D34862-01F_CDFCB-X_Volume_I_PR.pdf
Page 235-236

Document page number = 249-250

"""
This column is provided to indicate which terrain corrected GEO corresponds to a particular SDR. SDR Geolocations delivered with SDRs are not the terrain corrected geolocations. However, users who would like the terrain corrected version of a Geolocation may order a corresponding granule of any EDR listed in this column.
"""


In same document on page 328:

"Table A-8, Geolocation Identifiers"



GIMGO : VIIRS Image Bands SDR Ellipsoid Geolocation Data GITCO : VIIRS Image Bands SDR Terrain Corrected Geolocation Data GMODO : VIIRS Moderate Bands SDR Geolocation Data Provided in Degrees
       (D34862-03: NPOESS CDFCB-X Volume III) GMTCO : VIIRS Moderate Bands SDR Terrain Corrected Geolocation Data
----------------------------------------------------------------------------------


The plot attatched shows the differences in geolocation between the Ellipsoid and the terrain corrected geolocation for the VIIRS M-bands.
I simply read the two sample files GMODO and GMTCO and plotted the differences.


So, to me it looks like we will get both (maybe dependent on the configuration of the CSPP) and can choose to take the one which is not terrain corrected.

Then, as Trygve also concluded, we may need to better understand the statement "the calibrated TOA reflectance is calculated using the terrain corrected solar zenith angle". This implies that for the shortwave channels the reflectances have been corrected (normalised) by the solar zenith angle. I am rather used to a practise where you apply the normalisation by sun-zenith yourself afterwards. I would prefer it that way, but apparently this is not how it is for NPP/VIIRS.
Does anybody know more or have any comments?


I was considering if we should create a google-group on NPP (NPP preprocessing and understanding the data).
Are you in for that idea?

-Adam