2016-11-26                                                                                                                                   Per Undén


Report from the UERRA General Assembly IV, 21-23 November 2016.


The meeting was held at ECMWF in Shinfield Park.

Consortium meeting and administrative matters

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

The Financial Officer, Madeleine Benderyd, had prepared an overview of the use of resources by Work Package and partner for the first 2.5 years of the Project as from the Periodic Report 2. It was presented by Per Undén. The figures were on the average as expected with about 61 % of personmonths used for 62 % of the Project time. In WP s (1) and some partners nearly at or above the budget for the whole Project. That was about as expected since their tasks were planned to be carried out in the first half or that lower salaried staff have been employed. Then the Deliverables were presented with some explanations of delays and more definite dates for delayed ones were presented with the WP s. The most delayed or problematic one is the African Workshop which depends on a WMO workshop to be arranged. Action: Gé Verver and KNMI will check on the progress of those plans or an alternative venue since there is another workshop in March and DWD likes to be informed or involved since they have activities for South West Africa.

The Consortium possible issues were presented but no changes or problems have occurred.

There was the Periodic Report for up to M13 – M30 which has taken a lot of time and effort, both with formalities in the financial reporting and for the delayed Deliverables and particularly the delays in the production and archiving of the WP2 reanalyses. A GANTT diagram has been constructed and we explained at the Review meeting at REA on 7 September how to cope with the delays in WP3.

The ESAB members are Adrian Simmons and Blaz Kurnik who were present at this meeting. A third one has been difficult to recruit or to attend the meetings and now it is not meaningful anymore since the Project is nearing its end.

Some budget changes have been made in agreement with the Partners involved and our PO at REA.

A formal amendment was not needed. ECMWF needs and uses more resources coming out of somewhat reduced KNMI work in WP4 and a bit of SMHI. SMHI has also taken over more responsibility in WP 7 and there is a small transfer from KNMI here. The subcontracting costs and some travel for invited or ESAB members have been much lesss than budgeted so there is a transfer of some 29,000 Euro to Management costs at SMHI since this was under financed originally at the Negotiation phase and has used a lot of its budget.

The UERRA web site has been maintained and updated with a lot more reports and meeting material and list of publications and meeting presentations inter alias.

For the UERRA Reanalysis products the proposed licence terms were shown. They follow the Copernicus rules and are very open but also state that there is no warranty. A html version will be circulated to the producers in WP2 (the almost same version has been circulated before) for final approval. It needs to be on the ECMWF web for the data services to become public.

(Action ).

Recent meetings and coming meetings for 2017 were listed. The next General Assembly should be the first week in December (around 5-6 December) and the Concluding User Workshop (M42 although June) may be co-located with the Evaluation Synthesis Workshop (M45 October) in mid October 2017, around 16-19 October. A possible co-location or coordination with SIS projects will be investigated.

Action: Gé Verver will check with Copernicus about their plans and possibilities for co-location.

Action: Partners to inform the Coordinator if there are problems with these dates.

The location of the GA may be in Norrköping or an alternative is in Tarragona. The workshop(s) should be in a central location like Amsterdam. There is funding in the Project for conference facilities and caterering (and possible for inviting speakers).

The progress with respect to the objectives is generally good but the risks of delays and problems with computing resources have materialised but they are manageable. For e.g. SMHI spare computing resources have been provided by the Met Office. The RRA systems are all specially developed for this purpose and do not correspond to any of the operational systems. The delayed production and archiving has made it difficult for WP3 in that limited UERRA RA material was available. The definitions and final descriptions of the common archiving parameters and the actual conversion to GRIB2 took much longer time than anticipated. It involved a lot of work both for ECMWF and the partners.


EEA and Advisor, Blaz Kurnik related the EEA work to important policies and agreements. (Disaster Risk reduction framework, Sustainable Development goals and the Paris agreement). He listed many indicators and examples of CIBs. They derive many products and indices out of the material provided from climate scenarios etc. A new SOER report is being prepared for 2020. The work is in order to have a better informed decision-making. Climate-ADAPT is being updated and the 2016 report published. Changes in climate is described and in relation to natural hazards like forest fires and floods and storm surges.

Both Advisors called for more derived or modelled parameters and hail is very much requested from the Commission since the damage can be huge. It is however difficult and not directly modelled or postprocessed as such, but it would be good to think of this. Hail correlations from RA with observations are needed. Post processing tools are needed.



WP1 DARE efforts have been carried out both through scanned original data and through already digitised data from Sweden, Norway and Catalonia. Some 8.7 M data have been digitised from some 15 countries in Europe and the Mediterranean area and most from after 1950. In total 178 M data have been processed. An extensive quality control has been done on the digitised data, Visual Quality Control and Automated QC. Then there is the Spatial consistency check that can handle even more cases. A common problem is skewed values when the dates have slipped in the process. Almost all errors are detected and some can be corrected.

The hail climatology in Romania based on 105 stations was shown. It is very much tied to high orography and for cases, on CAPE. It has been published in MWR.

E-OBS has evolved and particularly much more stations have been added from several countries and this risks to give a more inhomogeneous product. The interpolation methods have been developed with a gamma transform for precipitation. There is also a probabilistic product with 100 realisations that was shown in D1.11. The underlying grid has been changed and to a high resolution to avoid boundaries inuser products. More parameters and high resolution orography are being brought in. The regressions will be done for more parameters. The dissemination of 100 members can be a problem technically and conceptually

There is a stochastic precipitation multi-year data set for the Alps based on observation density and representativity and for catchments of varying sizes and it seems to be very realistic in its ensemble spread.



The work in WP2 was described. The MO is using ERA-Interim and not ERA5 and is was shown not to have very much impact. The Ensemble assimilation has started from 1979. It includes uncertainties in observations, boundaries and model. The suite encompasses 450 tasks per cycle. The archiving and conversion to GRIB2 has been solved so the archiving can start. Observations include the conventional and Satellite radiances and reprocessed satwinds and scatterometer winds and GPS RO and ground based. The ensemble spread and error and frequencies seem to be good. The hybrid system needs some more development (tuning). A lot of work has gone into quality control of SYNOPs and rejecting suspicious ones. Outliers and suspicious ones are rejected. There is also a variational bias correctioin of Satellites. There will be a number of data denial experiments (particularly the Satellite one is important and demanded in the DoW.

The SMHI HARMONIE has 5 years ALADIN and ALARO and 2000-2015 with ALADIN is almost ready. 1961-1999 has between 4 and 5 years done per decade. Archiving of 2008 and 2006-2007 are also done. The runs and the archives will be prioritised and optimised as far as possible as it takes up to two weeks to get the data out of ecfs for one year. See the prioritised plan below under WP3. Observation monitoring and verification of forecasts showed good performance and a lot better temperatures than for ERA-Interim.There are many different level types archived and the height levels are new and interesting for the wind energy.

The downscaling reanalysis for 2006-2010 with ensembles has been done but part of it will have to be rerun due to a problem recently discovered. The number of surface observations has increased a lot for the 90:s and it may give a misleading signal. It has also been tested with low and high density networks. The downscaling production has started for the other periods.

The SMHI cloud reanalysis has been developed to a good state and run for 2004-2008. It will continued depending on usability of other data sets.

The ensemble nudging reanalysis has been run for almost all 2008 but stopped due to postprocessing and archiving issues. A range of probabilistic scores showed good performance and a paper has been published by Lilo Bach and now there is the thesis work giving less time for the other issues of continuing the work but DWD have worked on the parameters and GRIB2 issues.

A GANTT diagram and time line will be updated and the wiki at ECMW will have updated tables.



The Work Package has managed the Deliverables but without the proper archives so far and this has been time consuming. Validation has been done against observations (satellite derived and in-situ, towers for wind profiles) and gridded data sets at high resolution. Precipitation RA s from the MO and MF were verified against the Alpine and Nordic gridded data. WP3 is still waiting for the archives of WP2 and a priority list with deadlines was agreed, see below:

by the end of February 2017:



archived in MARS from SMHI and MO

by the end of May 2017:



MO and SMHI agreed to adapt their production plans and have their RA archived in MARS by then.


There are developments of interpolation methods and a proposal to de-trend the data was presented and results for the Nordic gridded data set improved a lot. A good selection in local regions for predictors is beneficial. Earlier RA:s and the HARMONIE RA were compared with the E-OBS data set and the HARMONIE ones seemed to be cooler and drier in this first comparison and less rainy days.

Drought indices were investigated for SE Europe and high resolution RA will help.

Action: The link to the R software package will be put on the web.


WP 4 is about Data service and a lot of work has been carried out at ECMWF with the WP2 RA producers. The model outputs took a long time to understand and how to define them in GRIB2 for the conversion to take place. Most is ready and SMHI has archived operationally for some time.

There is also a strand on ESGF where KNMI has a UERRA node and E-OBS has been put there. Then a RA data set will be transferred and it is to be decided which one. It is probably most interesting to choose temperature and precipitation only but for a long time scale (decades). There is also work on User oriented products, hydrological models driven by the RA:s, at MF with SURFEX too. Finally there will be indices calculated from the RA.


WP8 and others and evaluation

The WP8 involves the User interaction and particularly the following on workshop now. We still target educated users and can build on the first User WS). Access to archives is important and to get user requests. We should also use and build on experiences from other projects.

It shall be about : how to access UERRA, qualities and deficiencies , using uncertainty information, visualisation and evaluation tools.

It should be coordinated with SIS projects and Carlo Bonatempo will be contacted. (Gé Verver).

Action: Contact C3S for the SIS projects and the WS

It was discussed when to release the RA:s and it is important to be quite quick and not wait more than needed for basic checks.

WP9 was also discussed and the overarching telephone conferences are being resumed. The Lessons learned has been submitted. There was an uncertainty and co-location Workshop at the REA with the 5 Space FP7 projects and 2 Horizon 2020 ones. The uncertainties are very important for climate change and the emissions is of course the one big factor.

The ESAB members presented their recommendations. It is of course quite late in the project to affect the Project to make any big changes but in general it is running well. It is typical in all RA:s that it takes longer time and problems are found that require re-runs and cause delays. We should not worry too much about the dissemination as we will be well known when the data is there and we should open them for the public without much delay. Additional postprocessed parameters like hail are important to try to get and they are requested. It was asked how the models do it now and it is not so advanced yet, only very indirect. It is unfortunate that not long periods will be available for the evaluation but it is recommended that we come back in some year's time and do a longer evaluation of the whole period.

Many variables are required with long time series. EEA derives a lot of indices. It is essential to continue the time series for 2015-16 and beyond.


A Showcase Workshop was held after the GA with a lot of more applied results or further discussions of the needs and use of UERRA products. Several tools were displayed like Metview and WepAPI with which users can easily use the tools and data at home.