Welcome to the PRUDENCE data distribution page using DODS!

This site is presently under construction, and i urge you to tell me about any problems. Otherwise I cannot fix them. All files are accessible in two ways: Direct download, and processing with DODS. Some data sets (HC HadAM3H and HadRM3H) are only available for PRUDENCE participants.

As gateway to direct download please follow this link

To access the data through DODS, please go here.

Since 20070711 direct download of PRUDENCE-participant-only data is no longer possible!

DODS download of these data is unfortunately not possible.

Seasonal and monthly data are rather small files, which you are very welcome to download directly if you want to. But the daily data are large, up to 1.5 GB per field per experiment. Therefore, unless you need the entire area and the entire time period, please take advantage of DODS. This way you can limit the downloaded data to the points, periods, and frequencies that are relevant to you! You can find more information on DODS download and the PRUDENCE netCDF-based data repository in this page, and on how the kinds of data here.

We are presently manipulating data sets, in that attributes are being redefined in the data sets. This may happen several times for each experiment, so don't feel safe.

Status

The data situation here is still changing at the moment. A summary of the status on 18/5/04 is given below: The timing table and experimental recommendations (Tim Carter's table) have been updated, and links have been added. PLEASE check your own data and see if the table is correct.

All data sets are now without boundary zone. Monthly means on native as well as CRU grids exist for all data.

Since I have generated CRU-grid monthly means it was easy to also do the total seasonal means. So, I have put the original files generated by the individual groups into a "original_seasonal" directory and started putting my own ones into the "seasonal" directory. If anybody would like to compare it would be great. The biggest (known 8-) ) difference is that the new means do not use any RCM boundary points.

Now all directories with netCDF files also contain a file called "md5sums", which contains md5 signatures of all files in that directory. If you are in doubt about the integrity of a downloaded file, please compare the data in "md5sums" with the result of running "md5sum file.nc" in your end. This command exists in linux and hopefully also in other OS'.

A few recently discovered inconsistencies: Some evaporation fields (all MPI and ETH files and CNRM monthly and seasonal files) have had a negative sign on evaporation; this has been corrected now. Some soilw (KNMI, MPI, GKSS, SMHI) files have a wrong unit, m instead of mm (=kg/m^2); this has been corrected. The ERA-15-based DMI calculations F50 had errors in drainage, evap, and LWdown. These have been corrected now, and new monthly and seasonal files have been calculated.

The 1-d arrays of rotated latitude and longitude (rlon and rlat) have not always been correct. These have been corrected now, but you may want to double check. As far as I know the information in the area description page are correct. I have run through all files and homogenized rlon, rlat, variable names and attributes. I hope they are OK now, otherwise tell me about it. All UCM temperatures on native grid were previously in Celsius and not in Kelvin. This has been corrected.

The monthly and seasonal w10dir files have been recalculated with a different algorithm. Previously, I did a simpleminded averaging of the angle, which of course is very wrong (e.g. the average between directions of 1 and 359 degrees...). Now I have done the monthly averages by weighting cos(direction) and sin(direction) with the w10m daily wind speed. Correspondingly for seasonal averages. The interannual seasonal standard deviation probably does not make any sense no matter what is done. At present the file contains the standard deviation of the angle of the particular season for all individual years.

Dave Rowell discovered a few inconsistencies: Some pressure fields are in Pa instead of in hPa. Some seasonal files have erroneous out-of-range values; also SMHI monthly evap and runoff. These have been corrected now. Again, please report any problems with erroneous out-of-range values.

We have received a B2 experiment from the ICTP. Some errors in the CRU-lattice data (between months 258 and 288) have been corrected.

There were errors in the last 9 years of evap.HC.acdhd.1960-1990.nc -They have been corrected now.

Finally the HS3 data are up -The DMI third HC-driven ensemble member.

We have received a new set of experiments from the GKSS called CTLsn and SA2sn. They are reruns of CTL and SA2 with improved TKE and cloud ice schemes and spectral nudging.

DMI: Acronyms: pre-PRUDENCE ECHAM4/OPYC-driven runs are ecctrl, ecscA2, and ecscB2. HadAM3H-driven control runs are HC1, HC2, and HC3 and A2 scenarios are HS1, HS2, and HS3. The common ensemble member is the first. The 25-km version of these is F25 (control) and S25 (A2). A further experiment like HS1 is called HS4; here the Baltic SSTs have been taken from the SMHI RCAO model to avoid the unphysical warming of the Baltic. I have added the MERCURE run called F50, which was driven by the ECMWF reanalysis ERA-15 and covers the years 1979-1993. I have corrected an erroneous snow.DMI.F25.nc (mostly contained unphysical values).
HC: According to the agreement in Wengen we have received the first old ensemble member of the HadAM3H runs (control: acdhd; A2: acftc; no B2) and of HadRM3H (control: achgi; A2: ackda; B2: ackdd (yet to come)). Furthermore we have received HadRM3P data for the three new ensemble members (control: adeha, adehb, and adehc; A2: adhfa, adhfe, adhff; B2: adhfd). NOTE the error until now that I have previously mislabeled adhfd as being A2 and adhff as being B2. It is the other way round. The HC delivers wind data on a staggered grid, so that w10m and w10max are on DIFFERENT GRIDS than the rest of the data, both for global and regional data. The achgi t2min field contained some unphysical values. Daily, monthly, and seasonal data have been corrected. The acftd precipitation was in the wrong units (mm/s) and has been scaled with the factor 86400 for both daily, monthly, and seasonal files.
We have received seasonally averaged t2m, precip, and MSLP for the two HadAM3P experiments driving the first HadRM3P ensemble members: addfa (control; driving adeha) and addja (scenario A2; driving adhfa).
ETH: Now I have received ETH data; HadAM3H-driven control (HC_CTL) and scenario A2 (HC_A2). They have been providing daily, monthly, and seasonal files. Also files flagged ...monthly... in the "seasonal" directory with 12 values: means and stddevs for individual months. The HC_CTL daily t2m file erroneously contained t2max previously. This has been corrected now; the monthly and seasonal files were correct.
GKSS: These data are complete: HadAM3H-driven control (CTL) and scenario A2 (SA2).
MPI: From the MPI I have received 4 sets of data: ERA-15 driven (3002), HadAM3H-driven control (3003), the newest and correct HadAM3H-driven A2 (3006) and also the previous A2 where the GHG and aerosol concentrations corresponded to control (3005) -The latter is only for specific studies of the effects of local radiative forcing. The ERA-driven run is on a different lattice.
SMHI: Here we have 6 experiments, namely control, A2, and B2 both driven by ECHAM4/OPYC and by HadAM3H. I have received high-resolution data corresponding to the runs HCA2 and HCCTL. These are called HCA2_22 and HCCTL_22. The precipitation files have been corrected now. The soilw variable was erroneous for all SMHI experiments and have now been corrected.
UCM: These data are complete. Control is called control, the scenario is a2.
ICTP: We have received control (ref) and scenario (A2 and B2) from ICTP. Note that the ref files miss the last 10 days, and that the scenario files miss the last month (30 days). The latter should be fixed shortly, whereas the former will probably be half a year into the future due to data accessibility problems. This means that the monthly files (native-grid are present) have only 359 months for A2 and B2, whereas the last month for ref is an average of the first 20 days.
met.no: No daily data yet.
KNMI: The experiments are HC1 for control and HA2 for scenario. Some data files have been corrected, as the time ordering was wrong. These are HC1 MSLP and clcov, and HA2 clcov, snow, and soilw.
Météo-France: Regarding daily data we have one control (DA9) with observed SST/ice and two scenario runs (DE6 and DE7) with HadCM3H SST/ice changes; we also have a corresponding B2 run (DE5) and a B2 run with Arpège/OPA SST/ice changes (DC9). Monthly values exist now for several more experiments (see the timing table). The Arpège data correspond to a rotated and reduced grid, so there is no quasi-longitude and quasi-latitude to make the grid rectangular. Please adjust your data massaging programs accordingly.