2016-02-04                                                                                                                Per Undén


Minutes from the UERRA General Assembly (GA) 1-3 February, 2016, Météo-France, Toulouse.


The meeting was held at the MF Conference Centre in Toulouse.

Consortium meeting

The GA started with the formal and administrative consortium matters. All partners were present.

The Financial Officer, Madeleine Benderyd, gave an overview of the use of resources by Work Package and partner for the first two years of the Project (estimated for year 2). The figures were as expected with overall about half of the budget used up and some WP s (1) and some partners nearly at or around the budget for the whole Project. That was about as expected since their tasks were planned to be carried out in the first half. Then the Deliverables were presented with some explanations of delays (see more later). The Reporting period is up to 1 July this time and the reports should be received by the Coordinator for 25 June. A more detailed time table will follow later.

MESCAN/MESAN cloud analysis (D2.9) is delayed until the end of Summer due to earlier lack off staff resources, which has now been resolved. The science has been developed, tested and proven.

The Data plan, D4.2 is just being completed or complemented with the archive plan and there have been extensive consultation of users, earlier projects and reference material for ECVs and survey in CORE-CLIMAX. It will be completed in two weeks as also the D2.5, the HARMONIE two physics 5 year runs report. It was delayed due to the problems in setting it up and the work of assembling data and analysing them.

D1.10, 12 will be completed in April.

The Advisor said that station lists like those from NCAR are valuable.

The observation dissemination of the rescued and processed observations should be shown.

The Coordinator went trough the points that the partners might want to bring up under the terms of the Consortium Agreement. There were no concerns or comments from the partners.

A list of coming meetings, in the project and in associated projects was shown. Next GA will be arranged last week of November, 21-23 November, noon-noon, at ECMWF. Then the Showcase event will take place noon-noon 23-24 November. (Deliverable M32).

An overview of the objectives and achievements in the Project was shown.

EEA and Advisor, Blaz Kurnik listed many indicators and examples of CIBs. A new SOER report is being prepared for 2020. Extreme events are more common but there is a lack of reliable data before 1951. A showcase event should involve selected users.


WP1 is involved in many data rescue efforts like ACRE, ERA-CLIM2. Post 1950 data from several European countries have been processed like Slovenia, Catalonia, and already digitized data from Sweden.

Data will be available in month 30 and sent to international data banks at the end of the year plus put in ecfs in ascii format at that time and can be converted to BUFR at this time too, when it may be put into MARS as input to at least future RA. It is unclear to what extent they can be used in the SMHI or MO RA since they will be in production before that time.

The digitization effort was described in more detail and a number of checks that were done according to a template. All the data in UERRA will be sent to URV for inclusion in the big data set for further dissemination.so automatic QC and checks against neighbouring data.

The Romanian digitized data in UERRA will be sent to URV for inclusion in the big data set for further dissemination.so automatic QC and checks against neighbouring data.


The E-OBS data sets have been improved through addition of much more station data.

Interpolation improvements have been developed e.g. through Kriging of monthly means.

Uncertainty estimations by having perturbed interpolators and perturbations of the Gaussian distribution.

Copernicus Discussion

Notes on the C3S regional RA call: A workshop will be aranged in April. We should focus on developing operational systems. The ITT is expected in 2017.

Discussion ensued about the QA aspects which should be by users and the the QA project which is unsure who got this. A set of questions will be compiled and sent around.

WP1 (cont.)

The E-OBS data set has now also a daily update using GTS data up to the current date and climatology for the rest of the month and this has shown to be useful even if there are some relatively big differences to the final product. GTS data are used to fill in missing time series but only up to 10 years. (France is missing e.g.)

A new CRU data set has been released with more observations.

With 100 realisations one can describe the uncertainty but most users may use the 50 percentile of the distribution.

The Alpine data set has been extended and with different realisations. It is improved by interpolating Tx, Tn and the daily range separately. One can ask whether users can deal with different realisations, and probably they cannot. One can validate ensembles by holding stations back and this is the best way. For uncertainty at the grid scale one needs to do statistical modelling. There is an uncertainty estimate pdf depending on station density and area unit. The precipitation interpolation has been done for different scales, from 400 (hydrological) units and then on larger scales. The Delivarables (1.10 and 12) will be completed in April.


The work in WP2 was described. MO has a working system now and testing included variational bias correction and observation rejection list. There is an ensemble of static 4DVARs which will be used for uncertainties and as input for the deterministic higher resolution 4DVAR. The production will start in April with 4 parallel streams. The first test in the autumn had something wrong in the ensemble assimilation and number of possible reasons have been identified.

1 year can be run in less than a month. (and 50 M SBU (billing units),

SMHI has run the 5 years 2006-2010 and some years of the long RA but an error in the large scale constraint was found so that the RA will probably be rerun. It also for the purpose of providing also the correct radiative fluxes for the MF downscaling.

The results of the 5 years RA and intercomparisons with ERA-Interim, observations and the two model versions was shown and some aspects warrants further investigation. It is what is also in the D2.5 report.

The UB work had been developed with both LETKF (Kalman Filter) and ensemble nudging but it didn't work well for technical and possible scientific reasons in the first trials. The ensemble nudging works well and the research work will be done with this system in order to produce and be ready in 2017. The test can be run with the available resources and 2006-2010.

For intercomparison purposes it was decided to concentrated on 3-2 years, 2007/2008-2009 (when we have the Alpine data set (-2008)). MO will start with 2009 due to the availability of ERA5.

The cloud analysis is developed and works but the report based on the SEVIRI period will be in M32. It will still not have much, if any, impact on WP3.

The MF downscaling does not provide more information at the smallest scales but dynamical is better than static (interpolation). The perturbed T2m observations work quite well for the spread even if underdispersive but precipitation is more difficult. All different precipitation observations are combined for different accumulation times. Verification is done on a 0.1 degree grid.

V8 of SURFEX and ISBA-ES will be used for the SURFEX-TRIP hydrological downscaling and at higher DEM resolution ( 1/12 deg)

The 5 year period will be done in July so then the Deliverable D2.9 will be for M33 (delayed).

It was advised that one did not need extreme events to prove the RA and 2006-2010 works well. More model or low levels are needed for e.g. rare events like the Slovenian ice storm.


Validation has been made using testbed archives from EURO4M and the COSMO RA and observation data from surface and masts. Comparisons showed the COSMO 6 km to be very good and best.

The Nordic NGCD exists now for 1981-2010 for the Fenoscandia region and with daily precipitation at 2 km.

New methods have been applied with an intensity scaling.

The Romanian gridded data set showed to agree well with observations. The UERRA digitized ones (6 stations) will be submitted to URV.

Station records and reanalysis comparisons showed that the ERA-20C RA does not show the warming trend and is not recommended to be used.


The Data Plan in WP4 is delayed and was discussed. It is in final preparation and will be submitted within 2 weeks. In spite of this the Project has made a detailed archiving parameter document after several and extensive consultations in the beginning of the Project. It will be appended to the Data plan and it has had no delay impact on UERRA.

It was asked about wider data dissemination but the UERRA contacts are well known. Moreover, the data services will be integrated with the C3S CDS and interaction will be needed.

The MARS archive of both test data in GRIB1 and the UERRA common archive in GRIB2 was described. The latter will be with the opendata server and WebAPI.

It was suggested that one needed to use the full ensemble rather than a member ad to test the impact. The toolbox is in another C3S call. Uncertainty info and resolution impact can also be interacted with the 6 SIS projects.

The Work plan (DoW) is best to adhere to as far as possible but maybe with some changes if needed. The 6 or 12 km resolution for the German RA was discussed and 6 km looks to provide better quality but the time period would not be long enough.

The Advisor reminded about the extensive consultation process that had taken place after ERA15 for ERA40 and stressed how important it is. (UERRA had had an extensive process in the beginning, see above).

Visualisation with the ADAGUC tools was demonstrated and NetCDF data can be displayed there with or without ESGF node and own data can be uploaded. The Climate4impact portal is used.

The hydrological modelling with E-HYPE is a validation tool with some sinks though, in terms of evaporation. Subdaily forcing and actual model evaporation will be tested.

There will also be CIBs from the UERRA archives.

WP8 and others and evaluation and review

The WP8 involves the User interaction and particularly the following on workshop now. (first User WS). Early access to archives is important and to get user requests.

WP9 was also discussed and the overarching telephone conferences are being resumed. The Lessons learned has been written from us and EUCLEIA but not the others as far as known. It will be coordinated by the WP leader.

WP7 has dissemination and a lot through the web site. It was advised to promote and spread the site in various ways so that people become aware (it is also of course a function of the data being there, when they are). Also social media can be used for spreading information.

The WP has also three EU briefs which are hard to define and find a venue for. It was recommended and agreed that we write a short document where we put out the message and then use a suitable venue. Possible a GEO WS in June but it may be more a number of projects showing what they do at that WS. Then REA and DG-GROW plan a co-location WS with the 5 Space projects in the autumn.

We had not been so successful reaching JPI-CLIMATE but our reviewer pointed out that it is an important group for coordinating funding opportunities and more.

The ESAB and reviewer presented their recommendations. It is important to document all the observation sources that go in (The MO has in D2.1, Nb.). Data coverage and what goes in was a prominent output in ERA-CLIM. Also snow and ice cover inputs should be documented.

Now, the extreme events should be selected and show how the RA s perform.

Simple tables and flowcharts should be displayed to show what has and is being done. The archive document can be extended and serve as documentation.

Long time series of ECVs and more than the normal ones are good. Also at more levels should be provided (it is so in UERRA !).

There should be links to web sites and CLIPC for tools since UERRA does not provide derived parameters more than the CIBs.

For the reporting, it is good that we tell what we do but it should be more of how we do it and consequences if they don't work or of delays.

The Deliverables are needed now (in two weeks and before and at the end of the summer).

Consequences of the not working ESGF nodes need to be addressed.

We need to focus on users.

Contacts with other DARE are maintained in WP1. We need to have discussion among the WP s like at this GA).