Tim Hewison	Jan 12 (1 day ago)
to m_takahashi, me, ashim, Dohyeong, Fangfang, Peter, Na

Hi Masaya, and other GPAT members,

Just an update. Now I have caught up and fixed the outstanding issues mentioned in this email:

1. Fixed the bug with short overlap periods, which caused the zero correction coefficients in the first 7 days.
(This also removed the erroneous uncertainty jump you reported.)

2. Added reference_weight variable.

3. Reprocessed the data with the revised code and uploaded to our GSICS Server:
http://gsics.eumetsat.int/thredds/catalog/msg3-seviri-prime-iasi-demo-rac/catalog.html?dataset=msg3-seviri-prime-iasi-demo-rac/W_XX-EUMETSAT-Darmstadt,SATCAL%2BRAC%2BGEOLEOIR,MSG3%2BSEVIRI-PRIME_C_EUMG_20120812000000_demo_00.nc

4. Revised the ATBD with the updated code sample and uploaded it to the GSICS Wiki:
ATBD for EUMETSAT Demonstration Prime GSICS Corrections for Meteosat-SEVIRI (Candidate for Demonstration Mode)

Please continue your reviews.

I look forward to hearing your confirmation that this product can enter demonstration mode asap.

Best regards,

Tim

From: Tim Hewison
Sent: Thursday, December 17, 2015 6:43 PM
To: 'm_takahashi@met.kishou.go.jp'; Manik Bali - NOAA Affiliate
Cc: ashim mitra; Dohyeong Kim; Fangfang Yu - NOAA Affiliate; Peter Miu; Na Xu; Tim Hewison
Subject: RE: Fwd: GPPA Submission for Meteosat-IASI Prime GSICS Corrections to enter Demo Phase

Hi Masaya,

Thanks for your feedback and careful review \x96 most valuable as always.

A) ATBD

I have corrected the typos you pointed out in the ATBD. With regard to your other comments:

It\x92s not clear how to calculate this quantity (equivalent biases calculated from cumulative Delta Correction). Would you please add a short explanation?

I revised the text a bit:

\x93Figure 3 shows the time series of their double difference, and the delta correction defined cumulatively from it using Equation 4 (after converting to equivalent brightness temperature bias for a standard radiance scene), together with their associated uncertainties. This shows that step changes in either time series increase the uncertainty in the delta correction.\x94

Is it reasonable (to assume no correlation between Delta Correction and GISCS Corrections)?

It would be great if you could add the reason why we can assume that.

There are no reasons to assume we can neglect these terms, but otherwise the equations become unmanageable. So I have just re-written that as: \x93For simplicity, we assume no correlation between Delta Correction and GSICS Corrections:\x94

Similarly, for simplicity, we make the crude approximation that the coefficients\x92 covariance is proportional to their fractional uncertainty, scaled by the covariance of the secondary correction coefficients

Just a comment. IASI-B uncertainty is smaller than that you presented at the web meeting in July 2014: slide 52 ofhttps://gsics.nesdis.noaa.gov/pub/Development/20140723/PrimaryReferences_Examples2.pptx.

Did you change the processing?

Well spotted! Yes \x96 I made several changes since July 2014, including fixing a bug in the way I inflated the uncertainties generated by the Demo GSICS products.

How about referring the ATBD for Operational Phase?

Just as soon as it is published on our website!

B) GPAF

B-2) [RSP/TJH2]: It would be a nice proposal to be discussed at the forthcoming annual meeting. I think we have two ways for the operation/distribution: one is GSICS THREDDS server and the another is operational L1 dataset which contains a part of GSICS Correction.

Agreed \x96 I added it to the topics on the GSICS Wiki:

C) netCDF file

I will keep the \x91000000\x92 term in the filename for the \x91hhmmss\x92 field - at least for now.

# Global Attributes

C-2) How about adding a comment that this is Prime GSICS Correction?

Changed \x93Comment\x94 Global attribute to read:

\x93These Prime GSICS Re-Analysis Corrections allow the data from the infrared channels of this Meteosat SEVIRI to be corrected to be radiometrically consistent with the Primary GSICS Reference (MetopA/IASI). This is achieved by combining inter-calibration results using different IASIs, as Secondary References.

Use the RAC with the time closest to the time of interest. Take great caution when applying it at a date where this difference is greater than the window_period.\x94

C-3) Can we suppose "reference_instrument" represent Primary Reference in this case? It would be good to add Global Attributes to clarify Prime reference and Secondary references.

There is already a variable called \x93reference_name\x94, which is a character array. However, it was not correctly populated in the versions I distributed with the GPPA. I have corrected this, so it now contains

reference_name =

"METOPA+IASI",

"METOPB+IASI" ;

C-4) I would recommend to use window_period = "P-14D+14D". This is the outcome of the discussion through gsics-dev.

Thanks for spotting this too. I have changed it.

C-5) It would be useful if we add a comment on period of data for each reference to new Global Attribute or summary.

Agreed! But I would like to agree a format for this first. I was thinking of adding a new variable 'reference_weight', with dimensions ['chan', ,'ref', 'date']. However, I have not managed to get this working yet.

# Variables

C-6) Does "number_of_collocation" represent total number of the collocation data (e.g. IASI-A + IASI-B) if there is an overwrap period?

This refers to the total number of collocations from all reference instruments used in this window_period (not the total used to define the delta correction).

C-7) Attached two figures are time series of standard TB biases and its standard error (uncertainty) from the Prime Correction netCDF attached in Manik's previous email and IASI-A/-B RAC "demo" files on the EUMETSAT THREDDS server:

IASI-A)

http://gsics.eumetsat.int/thredds/catalog/msg3-seviri-metopa-iasi-demo-rac/catalog.html

;

IASI-B)

http://gsics.eumetsat.int/thredds/catalog/msg3-seviri-metopb-iasi-demo-rac/catalog.html

.

The first 6 standard TB bias in the Prime Correction (12-17 Dec. 2012) are about -282 [K]. Are they expected values? I also found uncertainty jumps around Feb/Mar 2014 even though the uncertainty for IASI-A/-B looks smooth and not so large. Did I use wrong netCDF files for IASI-A/-B?

Thanks for spotting this too. It did not show up in my plotting tools. There must be a bug in the part of my code that deals with short overlap periods. I will check\x85 Also, I don\x92t know what went wrong in Feb 2015. I will recheck after fixing the short overlap bug.

So, to sum up. Thanks again for your feedback. For the official Demo products I will:

1. Update the ATBD as mentioned.

2. Fix the above bugs in the code.

3. Make the changes to the netCDF you suggested.

I had hoped to get all this done this week, but now I\x92m leaving for my Christmas break, so it will have to wait for the new year.

Meanwhile, wishing you all a Merry Christmas and Happy New Year!

Regards,

Tim

-----Original Message-----
From: m_takahashi@met.kishou.go.jp [mailto:m_takahashi@met.kishou.go.jp]
Sent: Wednesday, December 16, 2015 9:45 AM
To: Manik Bali - NOAA Affiliate; Tim Hewison
Cc: ashim mitra; Dohyeong Kim; Fangfang Yu - NOAA Affiliate; Peter Miu; Na Xu
Subject: Re: Fwd: GPPA Submission for Meteosat-IASI Prime GSICS Corrections to enter Demo Phase

Dear Manik and Tim,

I am very pleased to see EUMETSAT submitted new kind of GSICS Correction which will ensure traceability back to single primary reference instrument. It is also useful from users point of view: we can keep using single file even after the end of reference instrument life time. I have the following minor comments and questions. Once my comments are reflected, I would like to encourage SEVIRI Prime GSICS Corrections to enter Demonstration Phase.

A) ATBD

Please have a look of attached docx file.

B) GPAF

B-1) Tim's comment [RSP/TJH1]: I agree with you. Manik, would you please update the point of contact from Bob to you?

B-2) [RSP/TJH2]: It would be a nice proposal to be discussed at the forthcoming annual meeting. I think we have two ways for the

operation/distribution: one is GSICS THREDDS server and the another is operational L1 dataset which contains a part of GSICS Correction.

C) netCDF file

# File naming

C-1) "20120812------" for the filename is fine, but it may be good to keep using "000000" to clarify the starting period of time.

Just for your information, Simon's comments can be found on:

https://groups.google.com/forum/#!topic/gsics-dev/uV-UmMoijgM (11 Nov

2014):

"Just a thought about the use if the naming convention: note that "If a particular date and time stamp field is not specified, it should be replaced by a '-' (minus) character. For example:

------311500-- represents a stamp that specifies only the day (31st), hours (15) and minutes (00)."

So please don't put zeroes in the name unless you really mean that time.

If the seconds or minutes are not relevant, you should use a '-'."

I think the expression using '-' is suitable for the data such as collocation intermediate data. The example is at the bottom of https://gsics.nesdis.noaa.gov/wiki/Development/FilenameConvention.

# Global Attributes

C-2) How about adding a comment that this is Prime GSICS Correction?

C-3) Can we suppose "reference_instrument" represent Primary Reference in this case? It would be good to add Global Attributes to clarify Prime reference and Secondary references.

C-4) I would recommend to use window_period = "P-14D+14D". This is the outcome of the discussion through gsics-dev.

C-5) It would be useful if we add a comment on period of data for each reference to new Global Attribute or summary.

# Variables

C-6) Does "number_of_collocation" represent total number of the collocation data (e.g. IASI-A + IASI-B) if there is an overwrap period?

C-7) Attached two figures are time series of standard TB biases and its standard error (uncertainty) from the Prime Correction netCDF attached in Manik's previous email and IASI-A/-B RAC "demo" files on the EUMETSAT THREDDS server:

IASI-A)

http://gsics.eumetsat.int/thredds/catalog/msg3-seviri-metopa-iasi-demo-rac/catalog.html

;

IASI-B)

http://gsics.eumetsat.int/thredds/catalog/msg3-seviri-metopb-iasi-demo-rac/catalog.html

.

The first 6 standard TB bias in the Prime Correction (12-17 Dec. 2012) are about -282 [K]. Are they expected values? I also found uncertainty jumps around Feb/Mar 2014 even though the uncertainty for IASI-A/-B looks smooth and not so large. Did I use wrong netCDF files for IASI-A/-B?

All the best,

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

+++++++++++++++++++++++++++++++++++++++++++++

Manik Bali - NOAA Affiliate <manik.bali@noaa.gov> wrote on 2015/11/27

10:38:33:

> 送信元: Manik Bali - NOAA Affiliate <manik.bali@noaa.gov>

> 宛先: ashim mitra <ashimmitra@gmail.com>, Fangfang Yu - NOAA Affiliate

> <fangfang.yu@noaa.gov>, Masaya Takahashi

> <m_takahashi@met.kishou.go.jp>, Na Xu <xuna@cma.gov.cn>, Peter Miu

> <Peter.Miu@eumetsat.int>, Tim Hewison <Tim.Hewison@eumetsat.int>,

> Dohyeong Kim <dkim@kma.go.kr>

> 日付: 2015/11/27 10:39

> 件名: Fwd: GPPA Submission for Meteosat-IASI Prime GSICS Corrections to

> enter Demo Phase

>

> Dear GPAT Members,

> Recently EUMETSAT submitted their Prime GSICS Correction product for

> acceptance in Demo Phase.

> This product developed by Tim at EUMETSAT is another product that

> breaks out of the league of classical products that GSICS has so far

> produced

as

> it gives the users the ability to tie their correction to a unique

> reference instrument that GSICS has selected for them after careful

> evaluation.

> GCC needs your help in review of this product.

> I also wish to thank you for reviewing two products in the recent

> months one of which is still underway( Seb's DCC). We that each of

> these

reviews

> are enormous work. However based on a first hand look at the documents

> which appear reader friendly I hope GPAT members would not find it

> difficult to review.

> Looking forward for your review.

> As of today , the product is currently uploaded on the EUMETSAT

> THREDDS server.

> http://gsics.eumetsat.int/thredds/catalog/msg1-seviri-prime-iasi-

> demo-rac/catalog.html

> http://gsics.eumetsat.int/thredds/catalog/msg2-seviri-prime-iasi-

> demo-rac/catalog.html

> http://gsics.eumetsat.int/thredds/catalog/msg3-seviri-prime-iasi-

> demo-rac/catalog.html

> http://gsics.eumetsat.int/thredds/catalog/msg4-seviri-prime-iasi-

> demo-rac/catalog.html

> Regards,

> Tim

> Rgds

> Manik