2017-02-22 / 23 /-03-02 Per Undén

UERRA MST meeting no 16

22 February 2017



Per Undén, (Manola Brunet), Andrea Kaiser-Weiss, Richard Renshaw, Maarten Plieger, Gé Verver


Agenda approval

Progress reports updates from the WPs and Deliverables per WP

Status and plan of data production and archives

UERRA GA end of November

WP4 and WP8 Workshops in the Autumn

Coordinator's reporting to PO at REA

Copernicus Climate Change services, any news?

Meetings and communication (EMS inter alias)


Next MST meeting

Progress reports in the WPs


Manola was not able to join but provided an extensive report in advance of WP1 as well as about the GA.
Work carried out for Deliverable 1.6, ensuring the provision of a spatially homogeneous dataset, will be finished this week, as we are finishing the data validation task. It has just been delivered.
D1.7 (All the quality assessed sub-daily data made available to WP2 and publicly available through WP4. This will include additional datasets of daily and monthly averages and totals as some will likely be of use in this form in Task 1.3 and in WP3) has already started by exploring with WP2 and WP3 partners (UK-MO, Meteo-France, UBonn and DWD) about the feasibility of using the dataset recovered and gathered in digital format for their current RRAs production. While data subsets (rainfall and dew point temperature) were provided by the end of Jan and extracted by Meteo-France for the 1981-1985 and 2000-2015 to use mainly the gathered data in digital format from Catalonia, Norway and Sweden, in the case of UK-MO and UBonn productions was not feasible at this stage to include additional data. In the case of UK-MO, the assimilation of the new dataset won't be able to take place until March or possibly April although it may be used after some work to include it later on. For DWD they will be happy to count on independent data for their evaluation analysis once the whole dataset is ready, although in this case, DWD did not confine to the end of Jan to have access to the dataset, as proposed by URV, and it's waiting for the delivery of the whole dataset. As stated in D1.7 title, we should provide daily and monthly averages and totals for some of the variables recovered, those to be likely in use in WP3's DWD analysis. Therefore, it'd be important to know if there is in need to compute these daily/monthly averages/amounts for all the variables or only for a set of  them. In the latter case, it'd be good to know which variables and which daily or monthly means/totals are in need for WP3. 
For WP3, the availability in ECA&D is fully sufficient and the timing is end of May. Whatever is available then will be used for the final evaluation. 
(After the meeting it has been confirmed that the hourly data are enough for ECA&D and for WP3 and all averaging can be made on the application side.)
Andrea, please let Manola know what is needed for averaging; if it is enough to make the averages/amounts yourself from the hourly dataset? We (URV is confident that they can  provide D1.7 by the 20th March, or a bit later only if averaging/accumulaton is needed.
Re D1.8 (Inclusion of D1.3, D1.4, D1.5, D1.6 data in the ECA&D system and MARS archive) also steps have been taken to know how long for ECA&D and MARS would take to "include" the recovered datasets in their digital archives. In the case of ECA&D, an extra week of work is envisaged once the datasets are delivered to them, while for MARS we haven't got yet any rough idea if providing the data to them for including them at MARS would be possible or not, due to the fact they only assimilate data from ISPD (only air pressure data) and ICOADS (only marine data) and the extra work required to assimilate data from other variables different to air pressure.  In here, a comment has to be done. URV made a mistake when wording the MoU for WP1, since it was wrongly stated we would "include" the recovered data, instead of stating we would "provide" the recovered data to the databanks, since as data rescue practitioners we can provide, but not include these data into their digital archives. Logically, this function is databank managers responsibility, not ours.  So this should be conveyed to our PO. In addition, the recovered datasets will be provided to other global databanks, such as ISPD (for air pressure), NCEI and UK-MO hourly datasets and ISTI (all the variables data), by mid-April, as well as the data digitised by accessing original data sources from German, Sweden and Spanish NMHSs that provided us access to their historical assets. As in the case of providing the whole datasets to ZENODO, or any other global repository (e.g. PANGAEA) it has to go accompanied by a description paper of the datasets, we envisage that a paper draft along with the datasets could be ready by the end of June, but we can't estimate when both the datasets and the paper will be made publicly available and/or published.  
Finally with regard to deliverable D1.13 (KNMI will both continue to update E-OBS in real time, and also develop the enhanced gridding software into a tool for more widespread use), this was delivered by KNMI at the end of January. The uncertainty estimate will be delivered for end of February (D 1.14).


The Met Office is now running in production and the GRIB2 conversion correct and the archiving is working. The 2008 has been run but there is a severe delay in that 2008-2010 may not be ready until mid March. The three years are run in parallel with one month's overlap/spin up which was checked to be sufficient. It is very slow and the archiving takes a long time. Possibly the 30h long forecasts can be skipped (twice a day) and only keeping the necessary 9 hour ones, but it may be unfortunate to lose that information in case e.g. precipitation or other fluxes between +12 and +24 hours are wanted. (It was checked with some in WP3 afterwards and they don't really need the longer forecasts. It would still be a big deviation from the UERRA archive plan and the Coordinator would not like to do this and also there might be useful data lost in this way and the saving would not be major. So the MO will continue with the forecasts up to 30 hours). The 1981-85 period will be run too, but also not meeting the agreed deadline. A mail discussion will be made to check which years are now of priority in this new delayed situation. (Richard and Andrea and WP3, WP1, WP4 and WP8, Gé)

SMHI has completed the cloud MESAN analysis for 2004-2008 with a well working system and submitted the D2.10. Now, the second cloud mask data set processed by Meteo-Swiss has arrived but it is in a different kind (fraction and not binary cloud mask) and it is not sure if they can do the analysis all the way back to 1983 due to some calibration problems. Maybe somewhere around 1991 is enough but a longer period than at present is needed to at least go closer to the MoU. The HARMONIE is running again and the period up to 2015 done and a re-run of the 2006-10 and also the earlier decades are about two thirds or more done. MARS data are archived for 2005-2010 and 1981-1986 and beginning of other decades, 1990 and 2000-2005 (so 2000-2010 is complete). There was a mixup of geopotentials being archived with the code name of geopotential height and that has been corrected although the existing archives have to be corrected and replaced for that variable (pressure levels).

At UB Lilo was running 2008 but not quite all of it and then the work was suspended due to change of jobs but she did resume the runs and archiving before leaving. A new person will take over the job at UB / DWD and the runs will be done now and 2008 will first be finished.

MF has run the ensembles for 2006-10 but a mistake was found in reading a file needed for SURFEX in the dynamical donwscaling from SMHI so there was no soil water and that caused to warm temperatures. 2008 will be rerun but not other years. The production runs 2006-10 have been run for the deterministic version for T2m/RH and precipitation and 1981-85 will be completed before the end of February. They are also in MARS archive now (test version, not oper).

There is an overestimation in shortwave radiation downward flux due to the averaging of different albedos from SURFEX (which is not used in the operational ARPEGE French model). It gives a problem in drying out and underestimating discharge in TRIP. It has been found that there is a long spin-up time for TRIP so there will be only one continuous stream from the beginning (1971) and thus delaying the production somewhat.

The GANTT diagram will be updated with the plans for finishing the RRAs and Richard will continue to do that. On ECMWF wiki Richard Mladek is keeping a similar diagram updated which shows the progress of the RRA and also one for archiving.


The D 3.5 Preliminary assessment was delivered at the end of last year.

The updated GANTT diagram at ECMWF is really helpful for WP3 members to see where we are. Of course nobody is happy with the delays but we can se what we can do and when.

In this connection the possibility of extending the Project was discussed. It would obviously increase the basis for the evaluation and improve the scientific quality, if it can be considered by our PO and if partners have the flexibility of keeping staff. Per will bring it up with our PO next week. It seems to be helpful e.g. for MS and KNMI who can handle this, but it is unsure for other partners.It would be helpful not only for WP3 but also for WP4, 7 and 8.

Other partners will be asked too (but probably better to wait for the view of our PO next week).

At DWD there have been some staff changes but the work continues. Cristian has started to compare SMHI RA against the Nordic gridded data set and MS against the Alpine data set. DWD will concentrate on winds and satellite data for comparison.

The timing of the Workshop and other UERRA meetings will be discussed under the following points.


The MARS work at ECMWF is going on very actively now and can be followed on the wiki UERRA pages (and in the Mars Catalogue).

The ESGF system is now working at KNMI and there is a node for UERRA data. E-OBS data have been put there and there will be RRA data from MARS there. The five year period of SMHI data will be the first to be put there. KNMI is retrieving the data and converting it to NetCDF. There is a 10 year period possible to get now, from Mars. Data on the node will be published soon. Later there will be ensemble data in Mars too (from the others in WP2) but probably too ambitious to put on the ESGF node for the current work.

Climate indices will be computed from reanalyses when available. DWD and other RRA producers will be informed about the contents and results before publication.


An ETCCDI workshop is or was foreseen in Spring 2017 in Botswana (organised by Omar Baddour from WMO/CCl) with UERRA contribution and outreach. There has been no new information about this from WMO and the communication is slow. Another possibility is to coordinate with MF and DWD for activities in Western Africa and ACMAD and try to organise a WS there in the autumn. KNMI is actively pursuing this but it is a long process. Albert might take a discussion with our PO if there is the case that we can't live up to this deliverable. Per will first ask Monika about this if it will be suitable.


The planning of the second User workshop will need to start. See also below.


Telephone meetings are scheduled for March and later in the year.

Workshops in the autumn

There does not seem to be any opportunity for colocating with C3S after contacts with Carlo Buontempo as there was a GA for SIS last autumn but not this one. Then it seems best to coordinate with the GA in Tarragona in November. planning and if it can be colocated with the the other meetings as well.

Status and plan of data production and archiving and impact

Action: GANTT diagram to be updated

Reporting to our PO at REA

On 2 March Per will have a telephone conference with our PO about the status and delays and how to cope with them.

Copernicus climate change services, any news

There are now two lots, the Arctic and the Europe CORDEX domain and each with a single DA system. UERRA partners are involved and there seems to be competing tenders.

Meetings and communication

C3S GA Toulouse 7-9 March. Per goes there and several from KNMI (and SMHI and more) who are involved in the projects.

EMS meeting 4-8 September. Abstracts to be submitted! (21 April). We will concentrate on the Copernicus session and not on the reanalysis session this time.

EGU meeting probably for several researchers.

HIRLAM/ALADIN Workshop Helsinki 3-6 April.

5th International Conference on Reanalysis (ICR5), 13 - 17 November 2017 in Rome, Italy.

GA eo 2017, 28-29 November

As mentioned above, it may be colocated with the WP3 and WP8 Workshops, to be confirmed. Moreover it may be from afternoon Monday to noon Friday, if it is convenient to travel that way. 
Last, but not less, to let you know that there is already booked a meeting room at the chancellorship building, an old modernist slaughter building in Tarragna's old city (http://www.urv.cat/en/about/directory/how-to-get/1/edifici-rectorat)  that can be reached on foot and sightseeing in the medieval and roman part of Tarrgona when walking to there. There is also many restaurant options around. We haven't yet made any block reservation for the attendees to the last GA, since the hotel that we use to accommodate our guests (Imperial Tarraco Hotel: http://www.husa.es/fotos-hotel-husa-imperial-tarraco.htm)  with beautiful balconies looking at the sea seems to be in poor maintenance conditions these days. Therefore, I've been advised to offer you other also central hotels, such as Hotel Lauria (http://www.hotel-lauria.com/fotos/) or Hotel Plaça La Font (http://hotelpdelafont.com/),  the latter in the old city, and both two cheaper than Hotel Imperial Tarraco (about less than 70 euros is the agreed price for URV, I think). Please, look at the options and let us know whether we should make a block booking in any of them and for how many bedrooms, roughly, or if just sending hotel options to the attendees would be fine.  


Next MST meeting

In about 2 month's time.

Action points

See the statements in bold in the text above and below, exracted

Andrea, please let Manola know what is needed for averaging;

A mail discussion will be made to check which years are now of priority in this new delayed situation. (Richard and Andrea and WP3, WP1, WP4 and WP8, Gé)

Per will bring it up [the possible extension] with our PO next week. (Done now). (Other partners will be asked too.

Richard will update the GANTT diagram and continue to do that.

DWD and other RRA producers will be informed about the contents and results before publication [of CIBs]

Deliverables' status

D1.6 : Homogeneity and homogenisation assessments of station data as they are collected from the NMHSs and from other sources. M36. A delay due to more problems and data formats. Delivered on February 21st .

D1.7 : All the quality assessed sub-daily data made available to WP2 and publicly available through WP4. This will include additional datasets of daily and monthly averages and totals as some will likely be of use in this form in Task 1.3 and in WP3[month 36]. 20 March is the planned delivery date.

D1.8 : Inclusion of D1.3, D1.4, D1.5, D1.6 data in the ECA&D system and MARS archive (URV 2 pm, UEA 1 pm) [month 36] March.

D 1.13 E-OBS updates. Ongoing and was to be done before end of year, now before end of January. It has been delivered.

D 1.14 Gridding uncertainty estimate. Was to be done before end of year, now before end of February. It was shown at the GA in November and it was clear that it needed a higher resolution master grid and involves both KNMI and MS(EDI) and more testing. There are no dependencies and it is better to have the work properly done and arrive at a fully satisfactory code and algorithm(s).

D 2.3 Ensemble variational DA diagnostics delayed but it was delivered end of 2016.

D 2.4 Ensemble diagnostics and documentation. M42. Even if all production is not completed, the report will be written and delivered during June (M42).

D 2.9 MESCAN ensemble. M30. Delayed from September – was to be end of November, now it was almost completed before Christmas but will be finished before mid March. It does not delay anything, but the report needs results from the runs which were delayed.

D4.4 : Implementation of ESGF (Earth System Grid Federation) node for at least one UERRA dataset (KNMI 8pm) [month 36]. The ESGF node at KNMI for UERRA was set up early autumn after many difficulties but it is there. E-OBS (observational gridded data have been put there (since KNMI produces them anyway and partly for / within UERRA. We want to put some of the RRA data there as well, entailing getting e.g. T2m or and precipitation for at least 10 years from MARS at ECMWF and it is beeing filled with data and about 5 years are there now. It will be delivered within the next month or when agreed (no dependencies of this D; it is merely a demonstration for the future and for other projects to take over and is complementary to the main archives in MARS and future Copernicus C3C Climate data store.).

D 7.9. African Workshop. Delayed but will be done in connection with other activities in the late autumn. Now delayed to spring 2017 but communication with WMO is slow. KNMI is pursuing the case for organising in the autumn with others in Western Africa. To be discussed with our PO since ther is a risk that it will fail.

D 8.3 Second Workshops involving external climate service providers. M42 (June), M46? May be colocated with the GA in November. To be confirmed.