20020416: GOES Imager ch.1 calibration

------- Forwarded Message

Date: Tue, 16 Apr 2002 11:49:35 -0400
To: address@hidden
From: "Peter Romanov" <address@hidden>
Subject: GOES Imager ch.1 calibration
Mime-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by ssec.wisc.edu id 
g3GFsX006824
Sender: address@hidden
Precedence: bulk

Hello,

Does anybody know what calibration coefficients are used in McIDAS to
convert GOES Imager ch.1 count into albedo value ? It looks like this
conversion is slightly different from what is recommended by NOAA.

The NOAA procedure to convert raw count to albedo is as follows
(http://www.oso.noaa.gov/goes/goes-calibration/goes-vis-ch-calibration.htm):
R=m(X-Xsp)  
A=kR
where R is reflectance, X is a raw count value, Xsp is a space count and A
is albedo. 

In particular, for GOES -8 Imager 
m = 0.5501873  [W/(m2-sr-m-count)]
k =  1.92979 x 10-3  [(m2-sr-µm)/W]
Xsp= 29 

hence,   A=1.06(X-29)

However with this formula I am getting albedo values, which are 0.3 to
0.4% less than the ones calculated in McIDAS. It seems, that the reason
for the difference is in the lower value of the space count (around 26)
used in mcidas. Of course the 3 counts difference is small considering
the noise and sensor degradation with time, but nevertheless, what is
the right way to do the conversion ?

Thanks,

Peter  

================================================
Peter Romanov
NOAA/NESDIS/ORA, WWB 712
5200 Auth Road
Camp Springs, MD 20746
USA
tel: (301) 763 8042
fax: (301) 763 8108
================================================

------- End of Forwarded Message

From: John Benson <address@hidden>
Date: Tue, 16 Apr 2002 11:59:54 -0500 (CDT)
To: address@hidden
Subject: Re: GOES Imager ch.1 calibration
In-Reply-To: <address@hidden>


Warning: this is not a definitive answer.  It is merely a
discussion of where to look.

The module kbxgvar.dlm in the McIDAS package is where the
calibration is being performed.  A quick reading of the code
suggests (but does not prove) the following:

     1. Tables which originate in the block0 record are being used
        unless they produce strange results, in which case pre-launch
        tables would be used instead.

     2.  It appears to me that the tables are expressed separately
         for each sensor, but that McIDAS is averaging them and using
         the average values regardless of which sensor is in use.

     3.  Early revision data on this module reveals the names Dave
         Santek and Bob Krauss.  One of these folks might yet remember
         what is going on in the visible.

I have actually spent some time in the internals of kbxgvar,
but only to add new satelites.  And most of the in-house work which
has vetted the calibrations were done by people with much more interest
in the correctness of the IR results than in any small discrepancies
in the visible.  Especially since McIDAS does nothing to correct for
path length through the atmosphere or even solar angle.

    --johnb