CRU Datasets - CRU TS Time-Series

These pages describe the gridded CRU TS datasets produced by the Climatic Research Unit (CRU) at the University of East Anglia.

At present, the BADC holds the preliminary CRU TS3.0 datasets for the period 1901-2006 as well as the final CRU TS3.1 datasets for the period 1901-2009. TS (time-series) datasets are month-by-month variation in climate over the last century or so. These are on high-resolution grids. They allow the comparison of variations in climate with variations in other phenomena. Variables include cloud cover, diurnal temperature range, frost day frequency, precipitation, daily mean temperature, monthly average daily maximum temperature, vapour pressure and wet day frequency. The CRU TS3.0 data are monthly gridded fields based on daily values -hence the ASCII and netcdf files both contain monthly mean values for the various parameters.

CRU TS3.0 data were originally produced and held by the CRU, however in 2007 it was decided that the BADC could provide dedicated long term support for these data and work began to port the software to the BADC systems where they could run on an operational basis. The system has now been ported to the BADC, and the BADC are responsible for updating these data on a regular (6 monthly) basis.

A paper describing the production of the CRU TS3.1 dataset is now in the final stages of preparation at CRU.

Please Note:

The CRU TS3.0 files were produced by CRU and include data up to the end of 2006. However, only data up to June 2006 should be used. For dates later than this, any subsequent updates/corrections to the original data have not been included, so the actual values may be misleading.

This gridded dataset is based on an archive of monthly mean temperatures provided by more than 4000 weather stations distributed around the world. Each station temperature is converted to an anomaly from the 1961-90 average temperature for that station, and each grid-box value is the mean of all the station anomalies within that grid box.

CRU TS data are unrestricted and are available for download to all BADC users. If you are not a BADC user, then please register as a new BADC user.

CRU Conditions of use states that acknowledgement should preferably be by citing one or more of the papers referenced (see Links and References section below).

• Where to find the data:

  The data and metadata (xml files) are now available at: http://badc.nerc.ac.uk/browse/badc/cru

  A CRU TS knowledgebase (FAQ) is regularly updated with questions and answers from the CRU TS data users.

• File Format:

  The CRU TS data are stored in both ASCII and NetCDF formats:

  • ASCII data: The 360-lat x 720-long grid is presented exactly as that, with 720 columns, and 360 rows per timestep. The first row in each grid is the southernmost (centred on 89.75S). The first column is the westernmost (centred on 179.75W).

  • NetCDF data: CRU3.0 - This is experimental, not all attributes are correctly set yet. The data themselves should be identical to those in the ASCII versions.
    CRU 3.1 - These data files should now have the correct attributes set.
    Please see the BADC NetCDF pages at http://badc.nerc.ac.uk/help/formats/netcdf/ for more information.

• How to read the data: Please Note: The CRU 3.1 files themselves are >2GB in size when unzipped, which may cause problems on the computer systems of some users.

  To read the CRU TS 3.0/3.1 ASCII data, users have so far been writing their own scripts as these are fairly easy to sparse. If you would like to share your script to read the CRU TS ASCII data with other users, then please email BADC Support.

  To read the CRU TS3.0/3.1 NetCDF data, you may use any NetCDF enabled software, such as Xconv, CDAT or FERRET).

Documentation:

• Background information on the CRU datasets is available from the CRU web pages at : http://www.cru.uea.ac.uk/cru/data/
  
  Note that additional information can also be found in the README files within the data directories themselves.

  CRU_TS_3.00/Read_Me_Pre_Release_TS_3_0.txt
  NOTES ON PRE-RELEASE DATASET CRU TS 3.0
  1. The status of this dataset is PRE-RELEASE. Access is granted on a per-user basis, please do not allow datasets to be used beyond recall range. Please report any issues to Ian Harris (i.harris@uea.ac.uk).
  2. The data is only complete and safe to use from January 1901 to June 2006. Some parameters will have partial support for months later in 2006 but this data should not be used.
  3. There are two output formats containing identical datasets:
  3.1 ASCII text. The 360-lat x 720-lon grid is presented exactly as that, with 720 columns, and 360 rows per timestep. The first row in each grid is the southernmost (centred on 89.75S). The first column is the westernmost (centred on 179.75W).
  3.2 NetCDF. This is experimental, not all attributes are right yet. The data should be fine, though. Any comments and suggestions will be gratefully received
  3.3 NetCDF missing vlaue issue. The headers of the netcdf tmp files state that the missing value should be -9999, however, the values stored are -999, which could be confused with actual values after the scale_factor is applied.
  If you have NCO, then something like this command works to fix files
  ncatted -a missing_value,tmp,o,i,-999 -a _FillValue,tmp,o,i,-999 cru_ts_*.nc
  (thanks to Cathy Smith of NOAA for providing the fix)
  The new (3.10) NetCDF files will be substantially different, of course, and this problem vanishes as the missing values are default ones and there are no scale factors because the data is FLOAT instead of INT.
  4. There are now two kinds of station file. Both types contain one value for every value in the data file.
  4.1 Regular '.stn.' files. The values in these represent, for each cell and timestep, the number of station that could have influenced the data value for that cell and timestep. The sphere of influence is the Correlation Decay Distance, which is 450 km for precipitation, 750 km for diurnal temperature range, and 1200 km for mean temperature (New et al, 2000).
  4.2 Cell station '.cstn.' files. These new files give the actual number of stations reporting in that cell at that timestep.
  Ian Harris, CRU, July 2008 (clarified October 2009, December 2010 by BADC)
  

• For the CRU TS ASCII (*.dat) data, note that a scaling factor of 10 or 100 is used to allow integers to be used to store the data (see table below for details).

  Abbreviations used for CRU TS3.0 and CRU TS3.1 variables

  label	variable	units	scaling factor
  cld	cloud cover	percentage	10
  dtr	diurnal temperature range	degrees Celsius	10
  frs	frost day frequency	days	100
  pre	precipitation	millimetres	10
  tmp	daily mean temperature	degrees Celsius	10
  tmn	monthly average daily minimum temperature	degrees Celsius	10
  tmx	monthly average daily maximum temperature	degrees Celsius	10
  vap	vapour pressure	hecta-Pascals	10
  wet	wet day frequency	days	100
  pet	Potential Evapo-Transpiration (CRU TS3.1 only)	mm	10

• A grid elevation file (cru_ts_3_00.halfdesg.elv.grid.data.gz) is also available from the CRU data archive.

/badc/cru/data/cru_ts_3.10/00README
CRU TS 3.1 data are produced used the same methodology as for the 3.0 dataset. The main differences is that the 3.1 dataset extends from 1901-2009, and all of the data in this period can now be used.
Slight differences may be noticed between the results for a given time/location between the 3.0 and 3.1 versions, due to additional data now being available. CRU have examined the 3.1 dataset in detail and are confident that such differences are not significant.
The station count files are all integer, because they are counts.
'stn' is all stations reporting within the CDD of the variable in question.
'st0' is all stations reporting within the cell itself.
*************************************************
NOTE:
(updated 6/6/11)
Due to an encoding error, the station count files ('st0' and 'stn') will need to be multiplied by a scale factor to obtain their true integer values, as follows:
* For 'wet', multiply by 100.
* For all other station count files, multiply by 10.
The station counts are scaled by the parameter scale factor:
tmp 0.100
dtr 0.100 \
tmn 0.100 > ('tmpdtr' st0 & stn counts)
tmx 0.100 /
pre 0.100
vap 0.100
wet 0.010
cld 0.100 (only 'stn' files)
frs 0.010 (no station counts)
pet 0.100 (no station counts)
Also note that a scaling factor of 10 should be applied to the PET netcdf data.
**************************************************
It is planned that the dataset will be updated every 6 months, with the first 3.1 update scheduled for July 2011.
Note: The files themselves are >2GB when unzipped, which may cause problems on the computer systems of some users.
A paper describing the CRU dataset is in the final stages of preparation.

Calculation of Potential Evapotraspiration (PET)
The method used is the FAO (Food and Agricultural Organization) grass reference evapotranspiration equation (Ekstrom et al., 2007, which is based on Allen et al., 1994). It is a variant of the Penman Monteith method using the gridded TMP, TMN, TMX, VAP and CLD.
Note that PET values are mean mm/day for each month (with a scaling factor of 10 applied to the PET ascii (*.dat) files, but not the PET netcdf files (*.nc).
The pet values in the datafiles therefore need to be muliplied by the number of days for each month to get the mean pet for that month

Background
At present, the BADC holds the preliminary CRU TS3.0 datasets for the period 1901-2006 as well as the final CRU TS3.1 datasets for the period 1901-2009 (which entirely supersede CRU TS3.0). TS (time-series) datasets are month-by-month variation in climate over the last century or so.
These data are on high-resolution grids. They allow the comparison of variations in climate with variations in other phenomena. Variables include cloud cover, diurnal temperature range, frost day frequency, precipitation, daily mean temperature, monthly average daily maximum temperature, vapour pressure and wet day frequency.
The CRU TS3.0 data are monthly gridded fields based on monthly station values -hence the ASCII and netcdf files both contain monthly mean values for the various parameters.
CRU TS3.0 data were originally produced and held by the CRU, however in 2007 it was decided that the BADC could provide dedicated long term support for these data and work began to port the software to the BADC systems where they could run on an operational basis. The system has now been ported to the BADC, and BADC are responsible for updating these (3.1) data on a regular basis.
BADC Team
6/6/11

Links:

• The Climatic Research Unit at the University of East Anglia .
• The Hadley Centre observations web pages

CRU TS References:

• For CRU TS3.0: In preparation. But in the meantime, Mitchell and Jones, 2005 can be used as background information for TS3.0. The major difference between the CRU TS2.1 and CRU TS3.0 processes is that no new homogenization is explicitly performed in the latter. Existing homogenizations in the underlying datasets, and homogenizations performed by national meteorological agencies prior to releasing their station data, are incorporated.

• For CRU TS2.1: Mitchell and Jones, 2005: An improved method of constructing a database of monthly climate observations and associated high-resolution grids. Int. J. Climatology, 25, 693-712, Doi: 10.1002/joc.1181.

• For CRU TS1.2 and CRU TS2.0: Mitchell, T. D., et al, 2003: A comprehensive set of climate scenarios for Europe and the globe. Tyndall Centre Working Paper 55.

• For CRU TS1.0 and CRU TS1.1: New, M., Hulme, M. and Jones, P.D., 2000: Representing twentieth century space-time climate variability. Part 2: development of 1901-96 monthly grids of terrestrial surface climate.Journal of Climate 13, 2217-2238, DOI:10.1175/1520-0442(2000)013<2217:RTCSTC>2.0.CO;2

University of East Anglia Climatic Research Unit (CRU). [Phil Jones, Ian Harris]. CRU Time Series (TS) high resolution gridded datasets, [Internet]. NCAS British Atmospheric Data Centre, 2008, Date of citation. Available from http://badc.nerc.ac.uk/view/badc.nerc.ac.uk__ATOM__dataent_1256223773328276

General queries about these pages and access to the data should be directed to the BADC support line. Your query should be answered within one working day. When follow-up work is required, the BADC support will carry out the work as quickly and efficiently as possible, and in any case, the user will be kept informed of progress.