• 01.07.2008 - Switch of Arpège "observation" E-suite tomorrow 

FISCHER Claude wrote:

 

Dear colleagues,

 the present Arpège E-suite, which mostly focuses on observations, is
 scheduled to switch tomorrow, July 1st.
 The switch will occur for the 00 UTC assimilation and 06 UTC
 production (the 18 UTC assimilation from June 30th and the 00 UTC
 production still will run with the "old" operational Arpège).

The switch remains pending to any last minute veto from MF's
Forecasting Department.

 best regards
 Claude Fischer
 tass
 

 

 

Dear all,

the aforementioned switch is confirmed ...

best regards
Claude Fischer
tass

 

 

• 23.06.2008 - next Arpège E-suite switch to operations: July 1st

Dear all,

the next , "observations", Arpège E-suite is now scheduled to switch to operations on Tuesday, July 1st
(06 UTC prod and 00 UTC assim runs).

The switch remains pending to any possible veto from the Forecaster Department (in case of risks of severe weather
on that day).
We will give you further confirmation of the switch as days go by.

best regards
Claude Fischer
tass

 

 

• 04.06.2008 - Arpège E-suite launch

Dear colleagues,

Arpège's new E-suite has just been installed in our Operations (actually, on May 28th and re-launched after yesterday's
problems on our NEC).

The E-suite description follows:

*

Arpège and Aladin-France E-suite number 1 for 2008 :

• CY33T0
• assimilation of new AQUA/AIRS channels (~54 channels),
• assimilation of Metop/IASI channels (~60 channels), Metop/HIRS,
• assimilation of MSG/SEVIRI Clear Sky Radiances (the 2 so-called "water vapor channels"),
• assimilation of clear-sky microwave radiances over land (DMSP F14 SSM/I),
• assimilation of microwave radiances (AMSU-A/B, MHS from NOAA and Metop) over land, in clear-sky, using improved surface emissivity maps,
• assimilation of AMSU-A channel 13
• couple the assimilation with variances derived from the ensemble assimilation,
• ALADIN-France : new obs like in Arpège, remove RH2/T2 obs in night time runs (spurious surface/PBL forcing via B matrix, flag will be on real solar time), VarBC for SEVIRI, monitoring of some radar radial winds from the French ARAMIS
• network (to be confirmed).
• ARPEGE physics: new GWD, revised surface turbulent exchange coefficients, corrected snow melting reaching the ground
• modified timestep for Aladin-FR (450 s) to have an even number of iterations for 1h, modified post-processing for
• isolated lakes Note that the presence of VarBC in Arpège induces a change in the strategy for starting an E-suite containing new observation types. Since VarBC is a bias correction scheme where all observations are inter-dependent (plus the analysis) inside VarBC, one needs to proceed to a warm-up of the E-suite prior to the actual start. This warm-up consists in introducing progressively the new observations, over about 2-4 weeks. For Aladin-France, the Arpège VarBC file is read in and the coefficients are merged with those adaptively computed for SEVIRI by the
• Aladin/VarBC.

A possible preliminary date for a switch to operations could be June 24th. Preliminary runs
have already been performed under OLIVE in R&D context.

We do not plan for the time being to produce Aladin test coupling files from this E-suite.

 

best regards
Claude Fischer
tass

 

 

• 06.02.2008 - Switch of Arpege: today Wednesday Feb 6th confirmed

Dear colleagues,

The switch to operations of the Arpège high resolution suite (T538L60) is confirmed for today, Wednesday February 6th,
on the 06 UTC production run and on the 00 UTC assimilation run.
Aladin-France will switch simultaneously to 60 levels (+IDFI).
LACE deterministic LBC files will switch simultaneously to the new coupling domain prepared by M. Janoušek.

Claude Fischer
tass

• 05.02.2008 - Switch of Arpege: now tomorrow Wednesday Feb 6th

Dear colleagues,

The switch to operations of the Arpège high resolution suite (T538L60) is now planned for Wednesday February 6th,
on the 06 UTC production run and on the 00 UTC assimilation run.
Aladin-France will switch simultaneously to 60 levels (+IDFI).
LACE deterministic LBC files will switch simultaneously to the new coupling domain prepared by M. Janoušek.

As usually, the new date will remain pending to any last minute veto from the MF Forecasting Department.

Claude Fischer
tass
 

• 01.02.2008 - Arpege HR E-suite switch to oper: postponed to after Monday

Dear colleagues,

The switch to operations of the Arpège high resolution suite (T538L60) o far planned for Monday February 4th, n the 06 UTC production run and on the 00 UTC assimilation run,has again to be postponed, due to remaining internal technical tests at MF Operations independently from the E-suite itself)

Accordingly, Aladin-FR and LACE/LBC switches also are delayed.

We will only give you next week a new date for the switch, which has not yet been fixed.

As usually, the new date will remain pending to any last minute veto from the MF Forecasting Department.

We apologize for this further delay,
Claude Fischer
tass

• 30.1.2008 - Arpege HR E-suite switch to oper: Monday Feb 4th

Dear colleagues,

The switch to operations of the Arpège high resolution suite (T538L60) is now planned for Monday February 4th,
on the 06 UTC production run and on the 00 UTC assimilation run.
Aladin-France will switch simultaneously to 60 levels (+IDFI).
LACE deterministic LBC files will switch simultaneously to the new coupling domain prepared by M. Janoušek.
As usually, the new date will remain pending to any last minute veto
from the MF Forecasting Department.

Claude Fischer

• 29.1.2008 - Arpege HR E-suite switch to oper: postponed

Dear colleagues,

The switch to operations of the Arpège high resolution suite (T538L60) , which was planned for tomorrow , Wednesday January 30th, is postponed.
Possible future dates for the switch are Thursday January 31st, or early next week. I will let you know as soon as
I do have a new schedule from our Operations. As usually, the new date will remain pending to any last minute veto
from the MF Forecasting Department.

we apologize for this delay, but obviously be prepared for a
soon-to-come change.
Claude Fischer
tass

• 29.1.2008 - Arpege HR E-suite switch to oper: tomorrow

Dear colleagues,

After recent confirmations here in Toulouse, I remind everybody about the switch to operations of the Arpège
high resolution suite (T538L60) tomorrow , Wednesday January 30th.
The switch will occur on the 06 UTC production run and on the 00 UTC assimilation run.
Aladin-France will switch simultaneously to 60 levels (+IDFI).
LACE deterministic LBC files will switch simultaneously to the new coupling domain prepared by M. Janoušek.
Yesterday, Monday 28th, on the 18 UTC run, PEARP has been switched to 55 levels and a generalized set of initial perturbations valid over the globe.

best regards,
Claude Fischer
tass

• 27.12.2007 - Increased-resolution LACE coupling files from ARPEGE

Dear Colleagues,

As I was informed Meteo-France has started production of increased-resolution LACE coupling files (15.4km60L) from ARPEGE high-resolution e-suite. The files can be found in cougar.meteo.fr:/chaine/mxpt/mxpt001/radmila/$DD/ directories.

I wish you peaceful Christmas and I am looking for working with you in 2008.
Best regards,
Martin

• 09.11.2007 - Arpege HR E-suite launched three days ago

Dear colleagues,
We announce the launch of the Arpège High-Resolution E-suite , with November 5th assimilation run of 18 UTC.
The E-suite contains the following:
_Arpège_:
- base cycle 32T0; 32T0.op2.10
- TL538C2,4L60, timestep 900s
- within 4D-VAR: TL107C1L60, 25 iterations; TL224C1L60, 30 iterations; timestep 1800s;
- finite vertical elements for vertical discretization
- increased vertical diffusion for momentum, especially near the tropopause
- PDF-based sedimentation scheme in APCS microphysics
- assimilation of ASCAT ; correction of bias correction for ERS/AMI
- variational bias correction for all satellite radiances
- monitoring of 314 IASI channels
- B matrix statistics for 60 levels are extrapolated values from the previous (operational) B matrix (46 levels). A native 60 level small ensemble is currently computed in order to derive true 60 level statistics.
- optimization features for the NEC activated

_Aladin-France_ will inherit some of the modifications above:
- 60 levels, ASCAT assimilation, vertical finite elements, changes in the physics
- LAM B matrix also is an extrapolated matrix (from the LAM 46 level statistics)
- bias correction coefficients are those stored from the Arpège VarBC preliminary experiments, and are used in the LAM in static (non adaptive) way like usual bc terms. SEVIRI bias correction is therefore untouched.

- Initialization: an important change occurs with DFI :
-- incremental DFI formulation with retuned parameters (stop-band edge 2h and time-span 1h40)
-- use LAM analysis file as initial coupling file (instead of Arpège interpolated C0)
-- tune background total error variances down (REDNMC=1.2 )
-- non-linear and omega balances activated

The Aladin-France E-suite should be launched next week.

Also next week, we'll start the production of E-suite LBC files for the LACE and Belgium domain, daily on the R00 production runs.

best regards
Claude
tass

 

• 05.09.2007 - Arpege switch confirmed

Dear colleagues,

the Arpège E-suite has indeed been switched to operations this morning, on the 00 UTC assim and 06 UTC production runs.

Claude Fischer
tass

• 30.08.2007 - Provisional date for Arpege E-suite switch: September 5th

Dear colleagues,

the provisional date for the switch of the present Arpège E-suite is September 5th. The switch remains as usual subject to any last minute veto from our Operations Department.
Further info next week
Claude F
tass

• 30.08.2007 - Likely switch of the Arpege E-suite ~ Sept 4th

Dear all,

The current Arpège E-suite, containing:
*********************************************************************************************************
- base cycle 32T0;
- assimilation of GPS radio-occultations in the troposphere and stratosphere (COSMIC);
- about 20-30 % more ground based GPS ZTD assimilated;
- assimilation of AMSU-A (12 channels) and MHS (all channels) from METOP-A;
- assimilation of ERS-2 scatterometer (AMI), monitoring of ASCAT scatterometer onboard METOP-A;
- modified cloud detection scheme for IR radiances (detection moved from after to before screening, which allows for more clear-sky radiances to be retained);
- stop assimilating some AMSU-A channels from NOAA 16;
- reduced evaporation rate of precipitations during their fall, in order to avoid spurious very intense and localized flow patterns to appear (also called "Aladinades" in MF's jargon);
- use NESDIS SST at 1/12 degree resolution;
- correction of a bug in the treatment of soil water content in the surface CANARI analysis (Françoise and Alena Trojakova)

In Aladin-France, additionally to the above list:
- assimilation of 10 m wind from the French surface mesonet and GTS in 3D-VAR;
*********************************************************************************************************
On the period of evaluation, Arpège produces positive scores on about all parameters in the lower stratosphere and in the Southern Hemisphere. Over Northern America or Europe, the
impact is neutral. The synoptic behavior of Arpège remains unchanged.
In ALADIN-France, the introduction of 10m winds over land improves the low level winds in the first forecast ranges.
The next E-suite for Arpège and Aladin-France will be the high resolution one. Technical preparations for this launch will probably last until mid-September.

best regards
Claude Fischer
tass

• 06.07.2007 - Evolutions of the Arpege E-suite

Dear colleagues,
here is a brief summary of our E-suite evolution at MF:

* the current E-suite, described in my May 23rd email has produced poor scores over Europe.
In order not to postpone the high resolution change later this year, MF has decided to change the content of this E-suite last week, with the following:
- removal of some AMSU-A/NOAA16 channels
- removal of DMSP-F14/SSMI (thus only 1 SSMI sensor remaining in the assimilation)
- addition of 12 AMSU-A and all MSH (formerly AMSU-B) from METOP in assimilation

we expect the first statistically relevant results next week, and these might be followed by a rather quick decision to switch. Basis cycle for the E-suite remains CY32T0.

* By week 30, MF intends to launch its high resolution Arpège and Aladin-France E-suite:
- 60 levels in Arpège and Aladin-France
- Arpège resolution over France about 15 km
- vertical finite elements
- several physics adaptation due to the change of resolution.
The latter will remain for about at least 3 months in double suite.

Further modifications in the Aladin-France assimilation might be added (10m wind assimilation, non-linear and Omega balances, incremental DFI).

best regards,
Claude Fischer
tass

• 31.05.2007 - Arpege E-suite "GPS and METOP"

Dear colleagues,

The next Arpège E-suite has been launched, starting with the R18 assim run for May 29th:

* CY32T0
* assimilation of GPS radio-occultation (8 satellites, based on COSMIC constellation)
* modified pre-treatment and data selection for ground based GPS delays, leading to the assimilation of more GPS zenital delays
* assimilation of SSM/I microwave radiances from the DMSP14 satellite
* monitoring of AMSU-A, MHS (formerly AMSU-B) and HIRS sensor radiances from METOP
* assimilation of scatterometer data from ERS-2; monitoring of scatterometer data from ASCAT onboard METOP
* modified clear sky IR radiance selection (placed before sampling/thinning, rather than after). This change allows for about 10 % more clear-sky IR radiance data to pass screening
* reduction of the rate of evaporation of precipitations in subsidences, in order to avoid spurious too strong local circulations in the Aladin-France forecasts
* in the Aladin-FR assimilation: assimilation of 10 m SYNOP and MESONET wind observations

A further change might occur after the start of this E-suite, namely the use of new NESDIS SST data (on a 1/12 deg grid resolution).

Aladin-FR shall follow within a few days time.

Please note that this E-suite does not include the change of resolution in Arpège, which is rather planned to start on late June, after the present E-suite has switched.

best regards
Claude Fischer
tass

• 23.05.2007 - Next Arpege E-suite

Dear colleagues,

this is a pre-announcement for the next Arpège and Aladin-France E-suite, to be launched within the next week:

* CY32T0
* assimilation of GPS radio-occultation (8 satellites, based on COSMIC constellation)
* modified pre-treatment and data selection for ground based GPS delays, leading to the assimilation of more GPS zenital delays
* assimilation of SSM/I microwave radiances from the DMSP14 satellite
* monitoring of AMSU-A, MHS (formerly AMSU-B) and HIRS sensor radiances from METOP
* assimilation of scatterometer data from ERS-2; monitoring of scatterometer data from ASCAT onboard METOP
* modified clear sky IR radiance selection (placed before sampling/thinning, rather than after). This change allows for about 10 % more clear-sky IR radiance data to pass screening
* reduction of the rate of evaporation of precipitations in subsidences, in order to avoid spurious too strong local circulations in the Aladin-France forecasts
* in the Aladin-FR assimilation: assimilation of 10 m SYNOP and MESONET observations

A further change might occur after the start of this E-suite, namely the use of new NESDIS SST data (on a 1/12 deg grid resolution).

We'll keep you further informed about the exact date of launch and possible changes.

best regards
Claude Fischer
tass

• 09.05.2007 - News from MF computers and operations

Dear colleagues,

the technical switch of Arpège operations to NEC has been performed as announced. So since the 0 UTC assim and the 6 UTC production (not 12 UTC, as I probably wrongly typed this morning), our NWP operations run on NEC and so do your coupling file productions. So far, I have been informed of no problem for any remote Aladin application.

We need until tomorrow to make sure that all the , rather complex, sequence of operational tasks have switched successfully, but there should be no emergency move back to VPP (unless some crucial problem is detected in the go of our NEC suite in the next hours).

Furthermore, Internet application seems fine again, as is the access to cougar

best regards
Claude Fischer
tass

• 07.05.2007 - Switch to NEC and other stuff ... for today

Dear colleagues,

The technical activities for switching to NEC our Arpège/Aladin-FR operations have started, as anticipated.
The first run on NEC would thus be the assimilation run for 00 UTC, followed then by the subsequent assim steps as well as the 12 UTC production.

Furthermore, the change of our Internet provider also is confirmed, which will breakdown our usual emails between 10 and 12 MEST (8 and 10 UTC).
I refer to previous mails; an emergency email address can be used for this exceptional circumstance: meteo.france1++#++wanadoo.fr

Finally, please note that our Archiving platform ("cougar") is down , due to technical reasons, for the time being.
We hope that cougar will be fine again as soon as possible.

best regards
Claude Fischer
tass

• 07.05.2007 - Last informations for MF/partner communication on Wednesday 9th

Dear colleagues,

MF confirms the switch of Arpège operations to NEC, along with the changes of its Internet provider, on Wednesday May 9th (the whole operation being now only conditioned to the absence of any severe weather event).
As announced previously, this situation potentially leaves a gap of about 2 hours , between 10 and 12 MEST, where we would have difficulties to communicate between MF and the partner Centers.
To circumvent this problem, we propose that any message of problem with coupling files shall be sent to the usual MF address:

pupreso++#++meteo.fr with copy to me

and , additionally, in copy to the following address:

meteo.france1++#++wanadoo.fr

this latter address will be reachable during the Internet change.

Quite importantly, PLEASE CONSIDER THIS ADDRESS ONLY FOR YOUR COMMUNICATION WITH MF ON WEDNESDAY 9TH.

best regards,
Claude Fischer, Dominique Birman
tass

• 04.05.2007 - Changement d'operateur sur les liaisons Internet de Meteo-France

Dear colleagues,

as already pre-announced, May 9th shall be a busy day for changes at Météo-France.
Indeed, in addition to the switch of Arpège operations to NEC (see mail from yesterday), MF will also change its Internet provider. This change will cause a total breakdown of
Internet access to MF from 8 to 10 UTC (10 to 12 MEST).

We invite all partners to therefore keep a careful eye on the fetch of coupling files; and use the RMDCN back-up for those who have this facility in case of problems, or straight access
via IAA/firewall.

We also will have a disrupted email service during these two hours. Therefore, don't hesitate to phone us for any information on Wednesday morning.

As for the switch to NEC, this change remains conditioned to no severe weather occurrence on May 9th.

MF apolodgizes for any inconvenience these changes might cause,
Claude Fischer
tass

• 03.05.2007 - Switch to NEC ... May 9th

Dear colleagues,

This mail simply to re-state that Météo-France keeps its plan to switch its operations to the NEC on May 9th.
The first run on NEC will be the assimilation run for 0 UTC (not the production run), which would take place at about 10 MEST.

The switch remains as usual conditioned to the weather situation (severe weather might cause our Forecaster Department to cancel the change).

best regards
Claude Fischer
tass

• 26.04.2007 - New date for the switch to NEC

Dear colleagues
The date of the switch to NEC was postponed to May 9th.
A confirmation mail will be sent before.
Best regards
Bernard Chapnik

• 24.04.2007 - New date for the switch to NEC : may 2d

Dear colleagues

The envisaged date for the switch to the NEC computer is now May 2d

Of course, a confirmation mail will be sent before.

Best regards

Bernard Chapnik

• 24.04.2007 - Switch to NEC delay

Dear alacoord and operala user.
We tried to send the following message an hour ago concerning a delay for the switch to the NEC computer
A problem with the e-mail lists management tool prevented this message from reaching the alacoord and operala list.
Please find below the message
Regards
Bernard Chapnik
--------------------------------------------------------------------------------------------------------
Dear colleagues

Due to a computing problem (the file management system crashed last night) the switch to NEC was delayed.

It may be expected that the switch will be made tomorrow. Anyhow a confirmation mail will be sent before.

We are sorry for any convenience this delay may cause.

Regards

Bernard

• 23.04.2007 - Switch to NEC

Dear colleagues

This e-mail is to confirm that the switch to the NEC computer will occur tomorrow, April 24th.

Please, take note that this switch remains open to a Forecaster Department veto, in case of hazardous weather events.

Best regards

Bernard

• 20.04.2007 - Short notice on the switch to NEC

Dear colleagues,

as a short notice on the switch to the NEC computer at MF, I may here simply state that this switch still is envisaged for April 24th.

A firmer confirmation will be sent on Monday, 23rd.

As usual, the switch remains open to a Forecaster Department veto, in case of hazardous weather events.

best regards
CF
tass

• 17.04.2007 - Reports of tests of new clim and coupling files made on new NEC

Dear Colleagues,

Let me summarize results of your testing of new climate and coupling files made on the new Meteo-France computer NEC SX-8. All pieces of information which has reached me so far can be found on http://www.chmi.cz/meteo/ov/aladin/climtest07/


1) New coupling files.

The new coupling files were produced by the mirror ARPEGE suite running the same model like the operational version but on a different computer. The results of ARPEGE on Fujitsu and NEC are different, I guess mainly due to
different ways how both compilers reorganize the code during the optimization process. All testers report visible differences in the ALADIN integrations using operational and new coupling files, both in terms of field norms and maps.
None of the tests claim non-realistic or clearly wrong results obtained by the new coupling files. However, it is vital to assess if those differences stay within limits of "the numerical noise" also for a longer testing period or if they indicate systematic deterioration/improvement of the forecast when going from Fujitsu to NEC (e.g. due to a hidden bug active on NEC only). The 28-days verification scores for ALADIN/France (http://www.chmi.cz/meteo/ov/aladin/climtest07/ALADINFr.html, courtesy of Claude) indicate that the differences are insignificant in longer time.
Further, the 9-days blending test (see http://www.chmi.cz/meteo/ov/aladin/climtest07/ClimCouplTest_CHMI.pdf, test C) I run at CHMI shows less close evolution of scores of operational ALADIN/CHMI to the ones driven by new coupling files but the period is shorter than in the ALADIN/France test and the few systematic features in the scores differences seem to be similar to those seen on ALADIN/France plots where they have been smoothed by the longer testing period.

Therefore I propose the conclusion the tests done so far do not show any systematic deterioration of ALADIN applications if switching the operational ARPEGE to NEC at Meteo-France.

2) New climate files

The new climate files corresponding to the telecommunication coupling files (produced at Meteo-France) were recreated on NEC. They should be quite close to the current ones but their usage in ee927 causes some differences in results due to the orography gradient changes etc. I run 9-day blending test and found no systematic difference in terms of verification scores of forecasts run using the current operational clims or the new ones.

So my conclusion is that you can quite choose when the new climate files will be installed in your local ALADIN operational suite and that it makes no harm if you install them before/after the switch in Meteo-France. It is
however vital that every Partner makes test usage of all kinds of new clim files to check there are technically OK.

Let me stress that the conclusions I offer here are subjective ones. You might have other feelings or findings I am not aware of or I have overseen.
So please let me know any new facts or your opinion which would speak against the switch at Meteo-France (currently scheduled to 24 April).

Best regards,

Martin

PS. I would like to apologize to all of you who have sent me your test reports for not being able to reply individually to all messages. Your input was very welcome and used.

PPS. And double appologies to those who will have got this message multiple times.

• 17.04.2007 - Some info about switch to NEC at MF

Dear colleagues,

As announced previously, MF is still targeting as earliest date for the switch of Arpège operations to NEC, April 24th.

A more firm decision is expected by Thursday 19th, this week.

In any case , the switch to NEC remains subject to a last minute veto from our Forecasting Department in case of severe weather events.

Most Aladin partners now have tested the NEC coupling files, and the Prague Center pursues a double suite evaluation with blending. As noticed by several testers, the NEC suite indeed has numerically diverged from the VPP one, which makes any "infinitesimal" or at least small amplitude comparison irrelevant. The technical tests have been , to our present understanding, positive for everybody.
As a statistical additional element, I simply attach to my email the (yesterday's) results for Aladin-France, on objective scores from our COMPAS Department. This illustrates the neutrality of the NEC suite in terms of scores.

so, more info by the end of the week,
best regards
Claude Fischer
tass

PS: one missing country for the testing on Aladin.

• 06.04.2007 - News about NEC coupling tests

Dear colleagues,

thanks for the quite many , yet not all, operators who have replied to the inquiry for NEC coupling tests (previous mails by Martin Janousek and myself).
I would like here simply to give some precision about the level of differences one might expect in the test 1 (comparing two runs with VPP versus NEC LBC files).

The two Arpège suites now have significantly diverged from one another. The wording 'mirror' suite probably is not reflecting the actual situation , as it suggests that the two suites should be "very" close numerically (say, differences of about 10^-3/10^-5 times a given field).

What we see is that NEC/Arpège differs locally in terms of intensity and positioning of gradient areas, so that when computing mean spectral norms or RMS, or when plotting fields, one can indeed obtain differences of the order
of magnitude of the field (or at least 10^-1 times a given field).

As an example, I attach plots of 500 hPa U, V, T differences on the Arpège analysis of April 4th, 0 UTC. You will notice differences of about 1.5 to 5 m/s locally on wind, and about 0.8 K on temperature.


Of course, we continue to monitor the NEC suite carefully in Toulouse, both for Arpège and Aladin-France. So far, the local situation-dependent divergences prove to be statistically neutral. I have understood further that
no detrimental systematical bias was noticed during subjective verification.

I think that I can still confirm a possible date for switch as early as April 23rd/24th.

best wishes for your further tests, happy Easter to all

Claude
tss

 

• 26.03.2007 - Test coupling and clim files from NEC: now ready

Dear colleagues,

thanks to the efficient preparation by our GCO team, we now have test coupling and clim files produced on NEC for all partner applications coupling with Toulouse model data.

I refer to my yesterday, March 25th, email for the expected testing and feedback which Martin Janousek and I expect from you.

So please, start testing and let us know about it !

Provisional dates for the switch of our operations to NEC are: April 23rd, May 9th or May 21st.


I would also , at this stage, ask all of you to simply reply to this email, so that I can clearly see that all of you
have received this message. No answer from any of the partners by April 6th would suggest me that I have to re-send my messages.
I of course understand that some of you are , at my time of writing, in Radostovice at the ALARO training course ...
but not all are ...

best regards
Claude
tss

• 25.03.2007 - Switch to NEC: proposed strategy for testing coupling files

Dear all,

In this mail we are going to describe explicitly the strategy which we will set on for testing the coupling files and the robustness of local applications with respect to the change from Fujitsu VPP to NEC SX-8 for MF operations.

The tests are proposed primarily in order to verify that NEC-generated coupling files can be properly read by your local ee927 procedures. Indeed, there will be for sure a few digits of differences in the numerical outputs, and we simply wish to make sure that nobody runs into an abort due to some lack of tolerance in specific threshold in the file geometry consistency checks.

The following tests are proposed:

1. Every Toulouse-generated coupling domain and associated Center will receive a set of NEC-generated coupling files, for a given date and network, and for a long production forecast. We will ask every partner coupling with Toulouse to run his local ee927+forecast on this set of coupling files, and REPORT about completeness of this test. This test requires NO CHANGE in your local ee927 procedures (no change in namelist or climate files).
-> The aim of the test 1 is to check that the switch of Météo-France operations from Fujitsu to NEC will be technically transparent to the ALADIN applications.

2. Further, Météo-France will provide to all partners (coupling with Toulouse) the coupling domain climatological files generated on NEC. Those climate files will be generated with the same algorithm and from the same databases like the current operational climate files and the only difference will be that they will be created on NEC. Every partner will be asked to redo the test 1 with the new climate files.
Again, every partner is asked to REPORT from that test, and possibly give a short notice on numerical differences. We expect no meteorological difference!
-> The aim of the test 2 is to check that the new NEC-made climate files cause no technical problem in Partners' ee927 procedures and hence that every Partner can install them at any convenient time at the time of or after the computer switch in Météo-France.

3. Once the ARPEGE e-suite on NEC starts (early-to-mid March), LACE coupling files will be generated from this e-suite as well. CHMI intents to run ALADIN blending assimilation cycle on the ARPEGE e-suite coupling files. The results of the test will be reported. LACE coupling files will be available to other LACE Members as well and everybody will be welcome to join the test.
-> The aim of the test 3 is to assess the meteorological impact of the switch of ARPEGE operations to NEC on a cycled ALADIN application.

4. Additionally, MF can produce Belgium E-suite coupling files on the NEC E-suite, so that Belgium could also test its application on a regular basis (to be further decided with our Brussels colleagues).
-> The aim of the test 4 would be to check the impact of the switch of ARPEGE and ALADIN operations to NEC on an ALADIN application coupled to ALADIN/France.


Test failures will be treated on a case by case basis. The main contact person for the test reporting and troubleshooting is Martin.Janousek###chmi.cz, with Claude.Fischer###meteo.fr in CC:. We stress that an early test is vital for the smooth adaptation to the Météo-France computer change especially in case it occurs during the Oslo workshop.


Once files for your tests are ready, we will inform you.


Best regards
C. Fischer, M. Janousek
tass

• 06.03.2007 - Start of the NEC mirror E-suite

Dear colleagues,

MF has started its mirror test suite for operations on NEC. This E-suite has no meteorological change with respect to its VPP counterpart.
The mirror suite has been started on the assim run of 18 UTC March 4th, and the first production run is 00 UTC on March 6th. The suite is scheduled for until end of April.
We will come back later to you in order to resume some basic tests for coupling files (I awe some pending messages to you on this matter: strategy for tests as established with Martin Janousek, input information).

best regards
CF
tss

• 26.02.2007 - Confirmation of switch

Dear colleagues,

as announced earlier, Arpège operations have been switched to CY31T1 this morning, on the 00 UTC assim and 06 UTC production runs.
Additionally, this new operational suite contains the modifications in the surface analysis in order to speed up the melting of frozen soil in late winter, early spring.

best regards
Claude Fischer
tss

• 13.02.2007 - Arpege E-suite switch delayed

Dear all,
due to hazardous weather over part of France, the planned E-suite switch of Arpège and Aladin-France has been discarded this morning.
The next proposed date will be February 26th.

Claude
tos

 

• 12.02.2007 - Arpege E-switch still scheduled for tomorrow

Dear all,

MF confirms that the present E-suite (CY31T1 + increased frozen soil water melting) still is scheduled for switching to operations tomorrow.
Some active weather events this night and tomorrow may provoke a last minute veto from our forecasters department, which we can only confirm or discard ... tomorrow.

Claude Fischer
tos

 

• 06.02.2007 - Preliminary possible date for Arpege E-suite switch

Dear colleagues,

I was informed that a first date for switching the present Arpège E-suite (see previous mails) would be next week, on Tuesday 13th Feb.
This date shall be further confirmed. Results of the E-suite are neutral to positive, and an early switch would free the path towards the parallel testing of NEC/VPP and additional porting efforts to be prepared by GCO.
I hope I can give you more info soon.

best regards
Claude Fischer
tss

 

 

• 25.01.2007 - New Arpege E-suite

Dear colleagues,

as previously announced, MF has launched a new E-suite, with a more technical oriented set of modifications (see below), based on CY31T1. The E-suite will start with the R18 run today in assimilation and with the R00 for 26/01 for production.
The changes are now the following:

1/ Optimisation of the TL/AD of simplified physics, giving several percents gain in computational time in the 2 minimizations of 4D-VAR.
2/ Modified formulation of the reduction of frozen soil water with positive T2m in the analysis, leading to an enhancement of the melting process when T2m switches back to above 0 deg. C. in late winter/spring conditions.
3/ Retrait en namelist des modifications scientifiques principales . As a consequence of earlier decisions, all scientific changes tested in the previous E-suite have been discarded, including the semi-envelope orography. MF has switched back to full envelope clim files.
4/ Introduction of code optimizations and adaptations in RTTOV, in preparation for the next mirror E-suite between VPP and NEC in Toulouse.


best regards
Claude Fischer
TSS

 

• 24.01.2007 - Arpege/Aladin-FR E-suite ... switched off and follow-on operations

Dear colleagues,

the present Arpège and Aladin-France E-suite, which has been under investigation since November 2006, has now been stopped (today, with the R00 network).
As a consequence, changes like semi-envelope orography and SLHD will need to be re-assessed in a future E-suite, on the NEC platform.
There will be more detailed descriptions later on about the E-suite results , most possibly for the coming newsletter.One major shortcoming was a too big dissipative effect of the SLHD scheme in Arpège on the relatively strong stormy events of the last two weeks. Of course, relevant Aladin scientists will be kept informed about this issue .
MF will now launch a more technical E-suite with:
* all adaptations in order to make the observational pre-treatment and input of data files more flexible in the Arpège and Aladin-FR assimilations
* a change in the surface analysis, consisting in an amplified correction of the deep soil ice reservoir for melting (*)

We intend to test this new E-suite until the second part of February.
After this, our major operational change will be focused on the move to NEC, for which we will give further information in due time. The expected calendar would be to switch to NEC as soon as possible after about April 20th, but we cannot state a very precise time period so far. However, note that the switch might take place during the Aladin/Hirlam workshop .

Claude Fischer
TSS

(*) this modification in Arpège has been also motivated thanks to the data and analysis provided by Maria Derkova, on behalf of the Aladin-2 transversal information passing. Several partners have announced unrealistic 2m temperature in spring time, during the heavy ice melting period. The change under investigation consists in a more drastic correction of the ice soil reservoir with the analyzed 2m temperature value (when it is above 0 deg. C.).

 

• 14.11.2006 - Start of new Meteo-France parallel e-suite cy31t1

 

message from Francois Bouttier to operala and alacoord email lists
------------------------------------------

Dear ALADIN partners,

A new ARPEGE/ALADIN-France parallel e-suite has just started in Météo-France, Toulouse using software version cy31t1_op1 . The first ARPEGE forecast is on 00UTC 14/11/2006. This will be the last e-suite on Météo-France's Fujitsu VPP. Its operational implementation is planned for early February 2007.

The next big change will be the purely technical switchover of the Météo-France NWP production to the new NEC SX8R supercomputer, in early Spring 2007, using the same cy31t1_op1 software version as this e-suite.

The e-suite mainly contains physics improvements, which are all applied to both ARPEGE and ALADIN-France.

SUMMARY OF THE E-SUITE CHANGES: (a more detailed documentation will be provided later)

- activation of the SLHD horizontal diffusion in ARPEGE and ALADIN-France.
(with thanks to our partners for this beneficial development !)

- modification of the subgrid orographic drag scheme, combined with a 50% reduction in the envelope orography

- increase from 2 to 6 of the number of visible spectral bands in the Morcrette radiation scheme

- several modifications to the microphysics scheme (critical humidity profile, evaporation of precipitation, introduction of statistical sedimentation algorithm, modifications to the microphysics adjustment steps sequence)

- the ALADIN-3DVar observation preprocessing (the ODB preparation) is technically split from the ARPEGE-4DVar preprocessing

- activation of the non-linear balance in the background term of ALADIN-France 3D-Var

EXPECTED IMPACT ON ALADIN PARTNERS:

- the overall CPU cost increase of the French production is about 10%. No change to the production timing of the ALADIN coupling files is anticipated.

- the modification to the envelope orography will be visible in the coupling files, but it does not mean that urgent action is required on the partners' side. In particular, SIMULTANEOUS SWITCHING OF ALADIN REMOTE CONFIGURATIONS IS NOT RECOMMENDED. CNRM/GMAP (C. Fischer, F. Bouyssel) will further check this point with our correspondents (starting with M. Derkova and J.-F. Geleyn) in the next few weeks. The ALADIN partners will be informed as early as possible if any problem is anticipated, or if the above recommendation changes.

The exact date of the operational switch of this e-suite will be broadcast by mail a few days in advance.

Please forward any question or concern you may have to ....

Francois Bouttier (on behalf of Claude Fischer and Francois Bouyssel)

 

• 15.09.2006 - Announcement of Arpege switch on 18/09/06

Dear colleagues,

as already pre-announced, the Arpège and Aladin E-suite from July to September 2006 is scheduled for switch to operations on Monday September 18th, for 00 UTC long cut-off (assimilation) and 06 UTC short cut-off (production).

the final decision is pending the usual possible veto from the Forecasters' department in case of hazardous weather.

the next E-suite could start middle of October, but its exact content is not yet fixed. More about this later.

Claude Fischer

PS: a short doc file for the present E-suite.

 

• 08.09.2006 - Pre-announcement of Arpege switch

Dear colleagues,

MF intend to switch its present E-suite:

OBS:

- Ground-based GPS over Europe assimilated in Arpège and Aladin
- SSM/I assimilated over sea in Arpège and Aladin
- 20 AIRS channels assimilated, about a hundred monitored in Arpège
- extended vertical usage of wind profiler data
- modified "bator" task in Aladin => bator+merge+LAMFLAG+reduce blended into one single task "bator"

Assimilation:

- number of inner loops in 4D-VAR: 25-25
- Sigma_b's of the day used in screening (obtained through a Monte-Carlo estimate using the minimization B) for Arpège


The switch would take place on week 38, most likely on the 18th of September.

The final decision for switch will be taken next week. The switch itself would then , as usually, still remain subject to a last-minute veto from the Production Department (due to hazardous weather, a severe technical problem etc...)

We keep you further informed about the E-suite decision.

No impact on coupling data, in terms of content, size or delivery time is expected.

Claude F.