Climatic Research Unit (CRU) TS (time-series) datasets are month-by-month variations in climate over the last century or so. These are calculated on high-resolution (0.5x0.5 degree) grids, which are based on an archive of monthly average daily maximum and minimum temperatures provided by more than 4000 weather stations distributed around the world. They allow variations in climate to be studied. CRU TS 3.21 variables are cloud cover, diurnal temperature range, frost day frequency, PET, precipitation, daily mean temperature, monthly average daily maximum and minimum temperature, vapour pressure and wet day frequency for the period Jan. 1901 - Dec. 2012. In addition to updating the dataset with 2012 data, the v3.21 release corrects two errors in the v3.20 dataset. This data is citable (DOI: 10.5285/D0E1585D-3417-485F-87AE-4FCECF10A992).
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).
Harris, I., Jones, P.D., Osborn, T.J., and Lister, D.H., 2013: Updated high-resolution grids of monthly climatic observations. In press, Int. J. Climatol., Doi: 10.1002/joc.3711
RELEASE NOTES FOR CRU TS v3.21: 16 July 2013 ************************ Updated 17 September 2013 files added: comparison.cruts321vs321mozfix.PRE.pdf comparison.cruts321vs321mozfix.WET.pdf CRU_Advisory_v3.2x_NE_Africa.txt These should be read by all users ************************ The 3.21 release of the CRU TS dataset covers the period 1901-2012. It supercedes all earlier releases. Differences between CRU TS v3.20 and v3.21 Summary In addition to updating the dataset with 2012 data, the v3.21 release corrects two errors in the v3.20 dataset. These are described in detail below. Areal means based on country boundaries have been constructed to aid comparison: 'comparison.cruts320vs321.PRE.pdf' and 'comparison.cruts320vs321.WET.pdf'. 1. WMO Code Error (All dates, PRE and WET variables only) A bespoke update of the precipitation observations database in December 2011 resulted in the corruption of some European station codes. This resulted in the observations from a number of European stations being relocated to countries in the Equatorial 'band'. The v3.20 release was based on this partially-corrupted database, affecting both PRE and WET (since WET is partially generated using synthetics derived from PRE). Both gridded data files and station files were affected. The relevant observations database was, of course, also affected. The files affected are as follows: All files beginning cru_ts3.20, for PRE and WET only pre.1207171231.clean.dtb In general, the station relocations took place to and from areas already served by numerous stations, absorbing any potential changes. However, there are exceptions: . Where the European station extends back further in time than any of the stations in the 'destination' region. . Where local station density in the 'destination' region is low. . Where local station density in the 'source' region is low. Many countries appear to have been affected in very minor ways, but some precipitation means have been affected in more noticeable (and perhaps significant, depending on context) ways: Colombia: 1901-1910 Venezuela: 1901-1920 Gabon: 1920-1940 Congo: 1901-1940 Tajikistan: 2005-2011 Israel: 2005-2011 Bulgaria: 1901-1940 and 1990-2011 Czech Republic: 2005-2011 Andorra: throughout For wet days, the countries most affected are: Bioko: 1915-1925 Gabon: 1930-1940 Cabinda: 1901-1940 Congo: 1915-1940 Socotra: 1925-1940 Georgia: 1920-1930 The recently-released precipitation database file for v3.20 is: pre.1207171231.clean.dtb This includes the corrupted metadata, and so will remain available to assist in any analysis of the issue. 2. Missing Observations (April-August 2011; PRE and WET variables only) An error in the database update process resulted in four almost- completely missing months in the PRE observation database: April thru August 2011 (inclusive). The dataset production process took the predefined approach for sparse areas and substituted the 1961-1990 monthly climatology for these months, so the error was not immediately apparent. The error propagated to the WET variable because of its dependence on the PRE variable. The files affected are as follows: All files beginning cru_ts3.20, for PRE and WET only pre.1207171231.clean.dtb Ian Harris Climatic Research Unit School of Environmental Sciences University of East Anglia Norwich NR4 7TJ
CRU TS/CY ADVISORY NOTE: 29 August 2013 Scope This advisory relates to the v3.20 and v3.21 releases of the CRU TS dataset, and to the v3.21 release of the CRU CY dataset. It concerns a spatial area of Northeast Africa, for the PRE and WET variables only. Summary An error has been detected in new Mozambigue precipitation observations introduced in 2011 for the v3.20 dataset and continuing to be used in the v3.21 dataset. The new stations had the signs of their latitudes swapped, so that they appeared north of the equator instead of south of it. Thus, the erroneous station data affects cells in an area corresponding to Mozambique if it were 'mirrored' in the equator. The error impacts on both PRE and WET data (as WET is derived from PRE). An examination of the error found 35 new stations that had been incorrectly located. A test run using corrected data was used to construct annual averages for each country, allowing comparison with the published data (v3.21). Comparison plots are available as: comparison.cruts321vs321mozfix.PRE.pdf comparison.cruts321vs321mozfix.WET.pdf A visual inspection identifies Eritrea and Egypt as those countries showing the greatest changes, though some others show small differences in one or two years. Action Users who feel their work may be affected by this error should make contact via BADC. At this time there are no plans to issue corrections, since the next scheduled release (3.22, covering 1901-2013) will incorporate these. Urgent requests for corrected data will, however, be viewed sympathetically. Ian Harris CRU, UEA, Norwich, UK 29 August 2013