Hello Manik and Sebastien,

First of all, I would like to congratulate EUMETSAT for their great
progress of the first VNIR product! Attached here please find my comments
(only minor comments) on the ATBD and GPAF. Once my comments are reflected
in the documents, I would encourage SEVIRI-MODIS VNIR GSICS Correction(s)
to enter Demonstration Phase.

I also have the following comments on the netCDF filename and its
contents. They are not necessarily corrected before entering Demonstration
Phase, but it would be great if you could take into account them. Some of
them would be discussed via gsics-dev.

----------------------------------------
A) Filenaming
----------------------------------------

Original:
W_XX-EUMETSAT-Darmstadt,SATCAL+RAC+GEOLEOVNIR,MSG2+SEVIRI-AQUA+MODIS_C_EUMG_20121231000000_demo_blend_v1.nc

A-1) GEOLEOVNIR
At present, "GEOLEOIR" and "LEOLEOIR" are only accepted by GSICS:
https://gsics.nesdis.noaa.gov/wiki/Development/FilenameConvention. I think
GEOLEOVNIR is suitable For VIS/NIR product, but GEOLEOVISNIR would also be
acceptable. I would like to post this issue to gsics-dev.

A-2) AQUA
It is recommended to use the name used by the owner agency or used on the
WMO-OSCAR even though file naming is not case sensitive. So "Aqua" is
expected.

A-3) _blend
This part should be removed

A-4) v1
Only the major version number is used in the two-digit format for
filenaming. Namely, "v1" should be "01".

Finally, expected filename is
W_XX-EUMETSAT-Darmstadt,SATCAL+RAC+GEOLEOVNIR,MSG2+SEVIRI-Aqua+MODIS_C_EUMG_20121231000000_demo_01.nc


----------------------------------------
B) Global attributes
----------------------------------------

B-1) id
This should be the filename, but different name is included.

B-2) reference_instrument
"Aqua MODIS" is expected.

B-3) wmo_monitored_instrument, wmo_reference_instrument
We got agreement on these attributes:
https://gsics.nesdis.noaa.gov/wiki/Development/NetcdfConvention.
They should be modified to: monitored_instrument_code = "56 207" and
reference_instrument_code = "784 389".

B-4) processing_level
"demonstration/vv1.0.0" should be changed to "demonstration/v1.0.0" (there
is extra "v").

B-5) window_period
Sorry, there were two descriptions about window_period on the GSICS wiki
(I have just removed the wrong one). As you may know, we discussed this
issue a few month ago and got the consensus to express like "P-14D+14D"
for Re-Analysis Correction and "P-14D+0D" for Near Real Time Correction.

In the case of your product (31 days window_period), this would be
"P-15D+15D".


----------------------------------------
C) Dimentions
----------------------------------------

C-1) method and blend+methods
We may be able to remove blend+methods because "BLEND" is one of the
inter-calibration methods. In such a case, method = 2 (BELND and DCC for
the moment).


----------------------------------------
D) Variables
----------------------------------------

D-1) method_name:standard_name = "method_name" ;
As for attribute: standard_name, names accepted by netCDF community are
only usable. Here is the list:
http://cfconventions.org/Data/cf-standard-names/29/build/cf-standard-name-table.html
. At present, "method_name" is not accepted, so please remove this
attribute.

D-2) central_wavelength:units = "um" ;
standard_name is associated with units. In the case of standard_name:
"sensor_band_central_radiation_wavelength", "m" must be used (not "um").

D-3) name of "gain" and "slope" used in "mon_gain" and "official_slope"
Both words represent same quantity, so how about using either one of the
two? It would also be good to use the same words for IR and VNIR products.
GEO-LEO-IR adopts "slope".

D-4) attributes: valid_min, valid_max, and _FillValue
We cannot use the same value for valid_min and _FillValue (e.g.
-999999.f). In addition, we should allow negative value for offset and
slope.

D-5) variable regarding inter-calibration coefficient: offset
I think some inter-calibration methods might compute the offset, so it
would be better to add this variable (e.g. "mon_offset", 0 can be put for
DCC method).

D-6) variables regarding uncertainty
My understanding is that we have no consensus on what kinds of
uncertainties we should put in the GSICS Correction. But if we have any
idea, it would be great to add this variable.





Best regards,
Masaya

+++++++++++++++++++++++++++++++++++++++++++++
(Mr.) Masaya Takahashi
Meteorological Satellite Center
Japan Meteorological Agency
3-235 Nakakiyoto, Kiyose, Tokyo 204-0012, JAPAN
Tel: +81-42-493-7452 / Fax: +81-42-492-2433
E-mail: m_takahashi@met.kishou.go.jp
+++++++++++++++++++++++++++++++++++++++++++++