Latest News

  • 19/8/2015

    OpenDaP functionality is down again! We are working on it. Apologies for any inconvenience!

  • 17/9/2014

    We now have OpenDaP functionality again! Note that the URLs have changed from what they were before. Check the links in the "RCM data portal" page.

  • 2/7/2014
    Finally the new server is up and running. Let me know if you run into any technical glitches.
    Note that OpenDAP is not yet running, so you will have to make do with direct download. On the other hand you now, again, have online access to the ENSEMBLES sub-daily data.
    Sorry to you all for the inconvenience!

  • 14/8/2013
    We seem to be having some hardware problems at the moment, so service will be intermittent. Sorry about that!

  • 29/1/2013
    Happy new year, everyone! This just to inform you that we finally got a new disk installed. I am currently uploading the data from tape, which will take several days. Enjoy!

    -By the way, we have moved this server; the IP address has changed to Sorry for any hangs due to this.

  • 21/12/2012
    We seem to have continuing problems with this disk. We shall have to wait for a new disk, if the world still exists tomorrow.

  • 12/12/2012
    At 23 minutes past 12-12-12 12:12 we got our disk system fully running again.

  • 11/12/2012
    In order to test the system we shall have to take the server down tomorrow mid-morning, possibly for the rest of the day. Sorry for your inconvenience!

  • 10/12/2012
    We seem to be having a problem with one of our disks. This means that the A1B European data sets DMI_ECHAM5, KNMI_ECHAM5-r1, KNMI_ECHAM5-r2, KNMI_MIROC, SMHI_ECHAM5, SMHI_HadCM3Q3 and VMGO are currently offline, unfortunately. We are working on the problem.

  • 12/4/2012
    We recently discovered an inconsistency in the last 40 years of th DMIHIRHAM5-ARPEGE simulation. The effect is (at least) that there is a sudden increase in snow cover. The reason is that we inadvertently used a different (debugged) executable for this period.

  • 10/4/2012
    We are slowly replacing the corrupted almost 1000 files; expect the job to complete during this week. Keep an eye on the date of the file you need.

  • 6/4/2012
    It turns out that there has indeed been problems with some files, around 300 of them. The upload has not worked, so the files are corrupted. I am currently scanning the archive and will start uploading files from backup as soon as possible.I will also put up a list of the files as soon as the scan has completed.

  • 29/3/2012
    Finally we (think we) are up again with the full archive, minus the parts that have been removed on purpose.

    PLEASE let me know if some of the things don't work -There could easily be links that have broken. Anyway, good luck with the data.

  • 23/3/2012
    Time passes... Luckily we found a new and much more efficient way to restore our files, which are scattered across tapes in the DMI mass storage system. This means that we ought to be able to put the full data collection online by mid next week, if nothing unexpected happens. Thanks for your patience!

    To the least patient among you, who have been writing mails asking for specific datasets, I have to apologize and decline. We do not have the manpower to do this. I realise this is frustrating, but we just cannot do it. Of course, if you are looking for 6H datasets, which are to be taken offline according to plan, you should write me about it; I expect rather few of that kind of requests.

  • 5/3/2012
    The machine is up, but still with only half the data. My apologies! The problem has turned out to be a bit complicated than we thought. We will probably have to reformat the relevant disks, which means a restore from tape backup in the end. In other workds, it will be many days before we are up again with the full data sets.

    I cannot guarantee either that we will not have further downtime in the near future.

    To the people who have sent me mails: You are not alone. In fact, you are so many that I can not get back to you all by mail. Sorry about that, but allow me to wish you good luck from here!

  • 22/2/2012
    ATTENTION! As part of the system checking connected to the hardware replacement, we shall have to take down the system one or more times this morning. So, it is not wise to start major downloads right now.

  • 15/2/2012
    -And we are down again. Seems like a bad disk cabinet, I am afraid. The good news is that the warranty is valid for another 12 days... You will, however, have to live with a lot of the ENSEMBLES data being out of reach for several days. Sorry for the inconvenience!

  • 15/2/2012
    We are up again! But the 6H data are being deleted at the moment.

  • 14/2/2012
    We are having file system problems at the moment. Something unexpected happened during file deletion (see below), and we are left with only one of our two disk arrays online. Since we do a lot of cross linking, it is probably even less than that, which is in fact accessible at the moment. We will have a look tomorrow!

  • 14/2/2012
    We are up again! But the 6H data are being deleted at the moment.

  • 14/2/2012
    We are having file system problems at the moment. Something unexpected happened during file deletion (see below), and we are left with only one of our two disk arrays online. Since we do a lot of cross linking, it is probably even less than that, which is in fact accessible at the moment. We will have a look tomorrow!

  • 14/2/2012
    We are up again! But the 6H data are being deleted at the moment.

  • 14/2/2012
    We are having file system problems at the moment. Something unexpected happened during file deletion (see below), and we are left with only one of our two disk arrays online. Since we do a lot of cross linking, it is probably even less than that, which is in fact accessible at the moment. We will have a look tomorrow!

  • 14/2/2012
    In the future, the CORDEX project data will be available through an interface based on the ESG interface used for CMIP5 global data. This will, however, take some time to implement and get up. In the mean time we need a place to make CORDEX data accessible for users.

    At the moment, the only available place is this server. As the disks are full, we have decided to take ENSEMBLES data offline as CORDEX data come in. These data are not lost; they exist on our tape backup system and can be reestablished in a few days.

    For now, I am removing all 6H data from the online archive. Sorry for the inconvenience!

  • 1/2/2012
    Our correction of DMI BCM-driven daily evaporation turned out to be incomplete. The period 1961-2000 was still erroneous. These files have now been corrected.

  • 9/11/2011
    We are now hosting a bit of CORDEX data, as we are waiting for the proper software to be installed at http:/

  • 6/10/2011
    A stupid mistake may have affected you: The METNO BCM and ERA40 simulations at one time had an error in the hydrological variables, a factor of 30 too big. This has been corrected in 2009. However, the new files had a wrong name (underscore instead of dash between years) and therefore coexisted with the original wrong files. These have been replaced now -And the METNO HadCM3Q0 MM file names have been harmonised at the same time. Sorry!

  • 29/9/2011
    One file from the KNMI ERA40 simulation had errors and has been replaced: Furthermore, I have received updated model descriptions for the 50km and 25km RACMO simulations. Please note that the proper reference to this set of simulations is van Meijgaard et al., 2008.

  • 27/9/2011
    It took some time with the interpolated HIRHAM5-BCM files, but they are finally there.

    Please take care with the HIRHAM5-ARPEGE simulation also: Time series of T and Precip look OK, but when e.g. snow cover is investigated, there is unfortunately some discontinuity around 2060 (It seems to deviate at 105906). This looks like an actual model problem and not postprocessing. We have chosen to leave the data in the archive, but please be careful.

  • 26/7/2011
    The more we look, the more errors we find... I am very sorry to say that we have found errors in the HIRHAM5-BCM simulation: Monthly average files from the variables stored as instantaneous values (in directories 6H, 12H and 24H) are wrong. Erroneous extra months have entered the file, with the consequence that the annual cycle has shifted. New files are being uploaded now.

    If you wonder what happened: We used the CDO "monavg" command on the 10-year files with (sub-)daily values. Unfortunately, the time stamp of the very last value actually belongs to the following month. So, when running "cdo monavg <10-year-file>" we obtained 121 values and not 120 -The last value being identical to the very last value in the (sub-)daily file. When all these decadal monthly files were concatenated, we did not notice the mistake.

    Note that the variables we normally look at all come from DM-files, which are not affected by this mistake.

  • 6/7/2011
    The last 3 years of were erroneous. The file has been replaced.

  • 7/3/2011
    There are now new ensemble-mean files for tas, pr, and evspsbl. The two former have very small differences to the previous versions, whereas the latter should now be OK.

  • 4/3/2011
    It is apparently only the CRU-interpolated HC monthly files that are upside down.

    I have now successfully calculated and uploaded the corrected monthly-mean DMI_BCM files!

  • 3/3/2011
    Dear all, please be careful with evaporation. I have looked at the evspsbl files recently, and I found that there are different sign conventions in use: Apparently MPI, ETHZ and CNRM have it positive downwards, i.e., mostly negative numbers. ICTP, KNMI, DMI, SMHI and HC use the opposite convention, with mostly positive numbers. I have not looked at the remaining institutes.

    Note also a potential issue, depending on your software: NetCDF CRU-interpolated files from the HC start with high latitude and go south, in contrast to everybody else. The "lat" array is correct, and the files are consistent. But, if you blindly read data from several models, you may get errors!

    These issues should of course be corrected. But, I am afraid that we do not have many human resources available, so I can guarantee that it will take some time. Good luck!

  • 24/2/2011
    The last 15 years of the DMI_BCM simulation have now been repaired, and all files should be updated by the end of today. The ensemble mean files will be done shortly.

  • 21/2/2011
    The units of evaporation, evspsbl, are wrong in the DMI BCM simulations. Possibly the same applies to runoff etc. We are checking this out and will replace the files asap. The missing period 2085-2099 has been rerun, and we are postprocessing at the moment. This also means that the ensemble-mean evspsbl file is very wrong; it will be replaced shortly. Apologies to everybody affected!

  • 4/2/2011
    In the left menu I have added links to what came before us, the upcoming data archive of the CORDEX project. Once it grows, there will be an unprecedented amount of data with many combinations of GCMs and RCMs, several RCP scenarios and a patchwork of areas covering all land masses of the Earth.

  • 3/2/2011
    I have now finally removed the original CNRM A1B and KNMI ERA40 simulations, which have been replaced by newer versions years ago. In case you need these data they still exist on tape. 25/1/2011
    I have made the quick-look analyses public for your information. I have also uploaded a document describing the climate change signal of the ENSEMBLES RT3 simulations which reach 2099. Note that the error-ridden HIRHAM5-BCM simulation is part of the note.

  • 15/12/2010
    A serious error has been found in the DMI-HIRHAM5_BCM run: The period 208512-209912 cannot be used, since the insolation has been shifted by one month; boundary conditions have not, so there is no alternative to a rerun. Our apologies!

  • 18/8/2010
    There seems to be an error in the 50km-file (gzip corrupted). It may take a while to regenerate it at the SMHI. 24/6/2010 11:40
    For my own, and possibly your, convenience I have constructed two files with interpolated monthly temperature and precipitation for the period 196101-209911. They contain pointwise monthly data for the ensemble mean of ICTP-REGCM3_ECHAM5, MPI-M-REMO_ECHAM5, DMI-HIRHAM5_ARPEGE, KNMI-RACMO2_ECHAM5, SMHIRCA_BCM, DMI-HIRHAM5_ECHAM5, METO-HC_HadRM3Q0_HadCM3Q0, SMHIRCA_ECHAM5, ETHZ-CLM_HadCM3Q0, CNRM-RM5.1_ARPEGE, DMI-HIRHAM5_BCM in the area covered by all simulations. These are the simulations going to the end of the century. I have excluded the simulations driven by HadCM3Q3 and HadCM3Q16, which are not very realistic. The data for 209912 are bogus; this is a copy of 206912, since not all simulations included 209912 and I wanted an array with only full years for practical purposes. Let me know if you find problems.

    Be aware that the ECHAM5 GCM has driven the majority of these simulations, so the averaging is not "fair". Also, remember that each of the 4 GCMs have followed their own trajectory from year to year. These fields only make any kind of sense if you look at longer-time averages.

    Finally, not all md5sums files have been up to date. I am currently running a script that ought to take care of this, so from now on it really should be possible to check whether your download was successful. If you forgot: Each and every data directory contains a small text file called "md5sums" with "fingerprints" of the data files. At a linux system you can run "md5sum -c md5sums" to check all files, or you can run "md5sum downloaded_file" and compare the resulting string with the content of my md5sums-file. This can be done under Windows as well, though I have no experience. Google came up with

  • 6/4/2010
    In case you are not aware of what the Climate Explorer is: You can access not just ENSEMBLES and other model data, but more importantly lots and lots of historical station data. And the user interface is several levels above what you find here.

  • 30/03/2010
    Some good news for once: We have received a 3rd party contribution from Geert Jan van Oldenborgh of KNMI Climate Explorer fame: CRU-interpolated MM files for the important fields psl, tas and pr for the two METNO A1B simulations where we have been waiting for interpolated files for a while. Thank you!!

  • 14/01/2010
    Take care with CDO. I have been told that there is a serious problem when using the MPI tool CDO to handle monthly mean (MM) files, and have just realised how serious it is.

    Frequently, the files in this archive have their time unit as "months since 1950-01" with type "float" or "double"; and frequently e.g. the first month of 1950 would have the value 0.5 to indicate that it is centered in the middle of the month. However, CDO assumes that months would be of type integer and "solves" this problem by rounding. This means that a value of 0.5 would be rounded to 1, and it being interpreted as February 1950. This is only a problem (as far as I know) for the CDO tool and for files with floating point time coordinates with half-integer values. I am currently about to correct this for CNRM, but there are lots of these files around, so it will take a long time to convert the time coordinate of monthly files to integer type.

  • 13/01/2010
    Do you know by heart what a Rockel-box is? If not, look here. These European sub-areas have been used both in PRUDENCE and in ENSEMBLES for model intercomparison.

  • Incomplete timeseries: tdps
  • Wrong value of mdi mrsos (all levels), tso (all levels), evspsblpot, hfso, mrso, mrros, mrross, sic, tsi, as, clisccp (all levels).
  • 16/4/2007:
    Erik van Meijgaard has found inconsistencies in several data sets. The SMHI data are being replaced by ones where the time variable is the same as for the other institutes. I am behind schedule in fixing flaws in other data sets; I apologize for this, and I will try to catch up. Some issues are:


    Ensembles RT3

    Research theme 3 of the ENSEMBLES project has the responsibility for providing improved climate model tools developed in the context of regional models, first at spatial scales of 50 km at a European-wide scale and later also at as high a resolution as 20 km for specified sub-regions ... RT3 will produce a multi-model ensemble based system for regional climate prediction from seasonal to decadal and longer time scales... (quoted from the description of work)

    Maintained by Ole Bøssing Christensen