View the Abstract for NDP-032
Access all data files from NDP-032
*Present affiliation: Bureau of Meteorology, Melbourne, Australia
Date Published: October 2001
Prepared by the
Carbon Dioxide Information Analysis Center
OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37831-6335
managed by
UT-BATTELLE, LLC
for the
U.S. DEPARTMENT OF ENERGY
under contract DE-AC05-00OR22725
1. Introduction
2. Data Sources
3. Quality Assurance Work and Error Checking
4. Analysis of the Revised Dataset
5. File Descriptions
1. Locations of the twenty-nine Antarctic
stations contributing to the databank
2. Annual Antarctic average temperature
series for 1957-99
1. Annual and mean-monthly trends in surface
temperatures over Antarctica
2. Mean monthly Antarctic surface temperature (C)
presented as anomaly with
respect to 1961-90
P. D. Jones and P. A. Reid. 2001. A Databank of Antarctic Surface Temperature
and Pressure Data.
ORNL/CDIAC-27, NDP-032. Carbon Dioxide Information Analysis Center,
Oak Ridge National Laboratory, U.S. Department of Energy,
Oak Ridge, Tennessee.
This database contains monthly mean surface temperature and mean sea level pressure data from twenty-nine meteorological stations within the Antarctic region. The first version of this database was compiled at the Climatic Research Unit (CRU) of University of East Anglia, Norwich, United Kingdom. The database extended through 1988 and was made available in 1989 by the Carbon Dioxide Information Analysis Center (CDIAC) as a Numeric Data Package (NDP), NDP-032. This update of the database includes data through early 1999 for most stations (through 2000 for a few), and also includes all available mean monthly maximum and minimum temperature data. For many stations this means that over 40 years of data are now available, enough for many of the trends associated with recent warming to be more thoroughly examined. Much of the original version of this dataset was obtained from the World Weather Records (WWR) volumes (1951-1970), Monthly Climatic Data for the World (since 1961), and several other sources. Updating the station surface data involved requesting data from countries who have weather stations on Antarctica. Of particular importance within this study are the additional data obtained from Australia, Britain and New Zealand.
Recording Antarctic station data is particularly prone to errors. This is mostly due to climatic extremes, the nature of Antarctic science, and the variability of meteorological staff at Antarctic stations (high turnover and sometimes untrained meteorological staff). For this compilation, as many sources as possible were contacted in order to obtain as close to official `source' data as possible. Some error checking has been undertaken and hopefully the final result is as close to a definitive database as possible.
This NDP consists of this html documentation file, an ASCII text version of this file, six temperature files (three original CRU files for monthly maximum, monthly minimum, and monthly mean temperature and three equivalent files slightly reformatted at CDIAC), two monthly mean pressure data files (one original CRU file and one slightly reformatted CDIAC version of the file), four graphics files that describe the station network and the nature of temperature and pressure trends, a file summarizing annual and mean-monthly trends in surface temperatures over Antarctica, a file summarizing monthly Antarctic surface temperature anomalies with respect to the period 1961-90, a station inventory file, and 3 FORTRAN and 3 SAS routines for reading the data that may be incorporated into analysis programs that users may devise. These 23 files have a total size of approximately 2 megabytes and are available via the Internet through CDIAC's Web site or anonymous FTP (File Transfer Protocol) server, and, upon request, various magnetic media.
Within this study we have assembled a database of surface temperature and mean sea level pressure data from meteorological stations within the Antarctic region. It draws heavily on the study by Jones and Limbert (1987) (hereafter JL87), and differs from that study in including data up until the early part of 1999, almost 12 extra years of monthly mean temperature and pressure data. Also included here are all the available mean monthly maximum and minimum temperature series. For many stations this means that over 40 years of data are now available, enough for many of the trends associated with recent warming to be more thoroughly examined.
Recording Antarctic station data is particularly prone to errors. This is mostly due to climatic extremes, the nature of Antarctic science, and the variability of meteorological staff at Antarctic stations (high turnover and sometimes untrained meteorological staff). For this compilation, as many sources as possible were contacted in order to obtain as close to official `source' data as possible. Some error checking has been undertaken and hopefully the final result is as close to a definitive database as possible.
Large amounts of data were obtained from the web site of Jo Jacka, hosted by the Antarctic Cooperative Research Centre (Jacka and Budd 1998). Data for the British bases were obtained from the British Antarctic Survey (BAS) web site, British Antarctic Survey (BAS), contact: Steve Harangozo. Data on this site also included those from Bellingshausen (Russia), Orcadas (Argentina), and the UK stations Faraday (now Ukraine and renamed Vernadsky), Halley, Rothera, and Signy (Harangozo and Colwell 1995). New Zealand data were obtained through Jim Salinger at the National Institute of Water and Atmospheric Research Ltd. (NIWA), Wellington, New Zealand.
Despite several sources of Antarctic station data there are frequent missing data and errors. For example, station temperature can be mistakenly recorded in Fahrenheit, rather than Celsius, and transposing data from written media to digital can lead to further errors. In most cases, pressure is expressed here as mean sea level (mslp); however, due to the variability in calculating mslp over the plateau region, pressure data for four high-elevation stations are expressed as surface pressure. These stations are Amundson-Scott (No. 89009), Vostok (No. 89606), Byrd (No. 89125), and Siple (No. 89083). Data availability closely followed JL87. Stations providing data are shown in Fig. 1.
3.1 CRU Error Checking
Considerable error checking had already been undertaken in compiling the JL87 dataset, and this forms the basis of the current compilation. The error checking procedure of the CRU can be summarized as follows:
Amalgamation of Datasets
In some cases there were four sources of data, but more often two or three. For each station, the sources were cross checked against each other. Where data sets overlapped, these were automatically checked for consistency, which gave an initial indication of where problem data may exist. Some large discrepancies became apparent--some from obvious typos--while others (often smaller but more numerous) were not immediately obvious.
Near Neighbor Checks
This involved checking the errors (discrepancies) found in (1) above. Yearly cross sections of the data (Jan through Dec) along with cross sections of nearest neighbors were plotted one over the other. In some cases, due to spatial resolution, there is very little consistency between near neighbors. Often this procedure was able to discern where gross errors existed but sometimes only gave a indication of support for one source of data over another. However, after this process, most major discrepancies, and a few other obvious errors, were either resolved or support was given for one particular value over another.
Mean Test
A check using the formula (max+min)/2=mean temperature was also made. Errors outside 2C were flagged and checked (over 170 such values were found over the entire database!). Also, errors of the kind where max < min and where the monthly mean figure was outside of the max to min window (i.e., where min > mean or max < mean) were checked. This check enabled many of the initial overlap errors (as described in (2) above) to be resolved. In several cases where the error in the mean value was not supported by another source, but a discrepancy existed between the (max+min)/2, the mean was replaced with the value of (max+min)/2.
Anomaly Check
Here data were converted to monthly anomalies from long-term monthly means. Each of these anomalies was plotted and compared to the anomalies from near neighbors (as a second near neighbor check). Few errors were found and where possible these were replaced with station data for that month from another source. After this stage all of the inconsistencies between datasets were resolved.
Hard copy check
Some datasets were originally transposed from hard copy. Where necessary these have been checked against the final dataset.
3.2 CDIAC Quality Assurance Checks and Data Reformatting
The Antarctic temperature and pressure data arrived in very fine shape from the CRU; probably as clean as any database CDIAC has ever received, quality assured, documented, and archived. After conducting basic checks pertaining to things such as consistent formatting of the files, properly sorted records, consistent units, and clear missing data indicators, only one actual correction took place: one station identification record in a data file happened to have a different record length than the others, and was edited. Given the cleanness of the database, CDIAC only chose to concentrate further on offering the data files in an additional format that may be easier for some users to work with: files with a station identification field in every record of the database (the typical CDIAC method), rather than data files with identification and data records interleaved in the same file (the format received from the CRU). Both types of files are offered with this database and are described in detail in the "File Descriptions" section of this documentation.
Surface temperature trends are seen to rise quite steadily up until 1991. In that year Mount Pinatubo (Philippines) and Mount Hudson (Chile) erupted, and it has been suggested (Jacka & Budd, 1998) that these eruptions have had a major influence on the general lowering of temperatures after that year.
Second, the contributors examine time series of temperature trends on a local and monthly basis. The continent-wide analysis masks considerable spatial detail and variability across the large region. The reasons for the variability are a complex mix of atmospheric dynamics and sea ice conditions. In an effort to establish some order in observed temperature trends at different stations, we define four groupings of stations based on a function of their closeness, similarity in temperature trends (on a long-term monthly basis) and long-term circulation patterns. The four categories are:
A - East Antarctic coastal stations: Mawson, Davis, Mirny, Casey and Dumont d'Urville
B - Dronning Maud Land: Novolazarevskaja, Molodeznaja and Syowa
C - Northern Antarctic Peninsula stations: Esperenza (including Hope Bay), Signy Island, Islas Orcadas, B. A. Arturo Prat and B. A. Bernado O'Higgins
D - Western Antarctic Peninsula stations: Rothera Point, Faraday (including Port Lockroy) and Bellingshausen
Each of these stations have long-term records of 30 years or longer.
Temperature trends over the available time period are calculated for each month for the stations within the categories. These trends are plotted against months in Fig. 3. Islas Orcadas has a record stretching back to 1903, so to compare trends with other stations for this analysis its record has been truncated to 1947, matching it with that of Signy Island. Within each category, the monthly cycle of temperature trends from each station correlate at least at a level of 0.5, and in most cases by greater than 0.6 (although little statistical significance can be placed on such a small sample size). The overall trend is for increased temperature within each of the categories, however there are a number of distinct features:
All Category A stations show a marked temperature decrease during May and a general temperature increase during June through September, although some show a local minimum during August, and October.
Category B stations show above average temperature increases during July, while Category C shows a decrease (or distinct local minimum) during July.
For Category B stations there is a local minimum in surface temperature trends during May.
Category C shows a decrease in temperatures during October, and a local minimum in July.
Category D shows temperature increases throughout the year, although the seasonal distribution is inversely proportional to monthly mean temperatures for those stations. This category also shows a local minimum during July.
Clearly, while temperature trends suggest a continent-wide warming, we find variations in temperature trends on a monthly and spatial basis. Van den Broeke (1998) found that a weakening of the Southern Hemisphere semiannual oscillation has occurred since the mid 1970s--associated with a decrease of meridional air exchange--that has caused significant May/June cooling in coastal East-Antarctica. This is reflected to some degree in the long-term cooling during May in Category A stations ( Fig. 3).
As with temperature trends, long-term pressure trends are plotted against months for the stations within the same categories (Fig. 4). These show a general decrease in pressure for each of the stations within each category (noting that these are all coastal stations), although points of note are:
All Category A stations show a greater surface pressure decrease during May and August.
For Category B stations there is a local minimum in surface pressure decrease during May.
Category C stations show above average surface pressure decreases during July-August and a local minimum during March.
Category D shows a clear local minimum in surface pressure decreases during August.
While these observations suggest there may be some link between regional temperature trends and decreases in station surface pressures, it is rather too simplistic to present as a clear argument. Some localized changes in surface temperature do coincide with changes in surface pressure and vice versa. In short, variability of trends in surface temperature, spatially and throughout the year, cannot be simplistically attributed to circulation changes, but more likely result from a mix of dynamic and physical factors involving complex feedback effects. Localized sea ice extent and concentration have been shown to be closely linked with surface temperatures. Simmonds & Budd (1991) found that increased leads in sea ice are shown to increase temperatures in those areas. Both King (1994) and Weatherly et al. (1991) found that the relationship between sea ice extent and surface temperatures are stronger around the west coast of the Antarctic Peninsula (Category D) than the rest of Antarctica. Sea ice anomalies (retreat) lag increases in temperature, although temperature increases persist, suggesting sea ice is to some degree controlled by temperature with some feedback effect.
This file is an ASCII version of the HTML documentation that may be found for NDP-032 on CDIAC's World Wide Web site. It exists for the benefit of those downloading the database directly from CDIAC's file transfer protocol (FTP) area without the use of a web browser.
These files are provided for the benefit of users with FORTRAN or SAS on their systems, enabling them to read any of the data files in this database using these software packages. The program files are:
Each record of the station inventory file contains a station's WMO number, latitude and longitude (decimal degrees X 100), and station name. The file may be read using invent.for or invent.sas. Stated in tabular form, the contents include:
Variable Variable Starting Ending Variable type width column column STANUM Numeric 5 1 5 LAT Numeric 5 8 12 LON Numeric 6 14 19 NAME Character 22 24 45 where STANUM is the WMO number of the station; LAT is the latitude of the station in decimal degrees (times 100); LON is the longitude of the station in decimal degrees (times 100). Positive longitudes are east of Greenwich; negative longitudes are west of Greenwich; NAME is the name of the station.STATION DATA FILES
There are eight data files: four original data files provided by the CRU with station identification records interleaved with the data records, and four analogous files slightly reformatted by CDIAC so that they contain only data records, with the station identification number in the first column of each record. The four files of each type contain monthly mean maximum temperature, monthly mean minimum temperature, monthly mean temperature, and monthly mean sea level pressure (mslp; 25 stations) or station surface pressure (4 stations; detailed in Sect. 2). The file descriptions below discuss measurement units.
Original CRU Data Files
The four files are:
Identification Records
Variable Variable Starting Ending Variable type width column column STANUM Numeric 5 2 6 LAT Numeric 5 9 13 LON Numeric 6 14 19 NAME Character 22 24 45 POR Character 8 56 63 where STANUM is the WMO number of the station; LAT is the latitude of the station in decimal degrees (times 100); LON is the longitude of the station in decimal degrees (times 100). Positive longitudes are east of Greenwich; negative longitudes are west of Greenwich; NAME is the name of the station; POR is the period of record represented in the data file for each station. This character variable is "19012000" for all stations, but most stations have far shorter records. The POR is given this way for consistency, since for all stations, 100 data records (1901-2000) are given even though many years' data are filled with the missing data indicator, "-999".
Data Records
Variable Variable Starting Ending Variable type width column column YEAR Numeric 4 1 4 JAN Numeric 5 5 9 FEB Numeric 5 10 14 MAR Numeric 5 15 19 APR Numeric 5 20 24 MAY Numeric 5 25 29 JUN Numeric 5 30 34 JUL Numeric 5 35 39 AUG Numeric 5 40 44 SEP Numeric 5 45 49 OCT Numeric 5 50 54 NOV Numeric 5 55 59 DEC Numeric 5 60 64 ANN Numeric 5 65 69 where YEAR is the year of the data record; JAN-DEC are the monthly mean maximum temperatures, monthly mean minimum temperatures, monthly mean temperatures, or monthly mean sea level pressures, depending on the data file. Temperatures are given in degrees Celsius (times 10), e.g., a temperature of -125 would indicate -12.5C. Sea level pressure or station pressure is given in hectopascals (times 10), e.g., a pressure of 9876 would indicate 987.6hPa. Missing monthly data are indicated by the value -999; ANN is the annual mean of the 12 monthly means, given in the same units as the monthly values. Missing annual data are also indicated by the value -999.
Data Files Reformatted by CDIAC
These four data files are formatted slightly differently from the original CRU data files and contain only data records, with the station identification number in the first column of each record. The four files are:
Variable Variable Starting Ending Variable type width column column STANUM Numeric 5 1 5 YEAR Numeric 4 7 10 JAN Numeric 5 12 16 FEB Numeric 5 18 22 MAR Numeric 5 24 28 APR Numeric 5 30 34 MAY Numeric 5 36 40 JUN Numeric 5 42 46 JUL Numeric 5 48 52 AUG Numeric 5 54 58 SEP Numeric 5 60 64 OCT Numeric 5 66 70 NOV Numeric 5 72 76 DEC Numeric 5 78 82 ANN Numeric 5 84 88 where STANUM is the WMO number of the station; YEAR is the year of the data record; JAN-DEC are the monthly mean maximum temperatures, monthly mean minimum temperatures, monthly mean temperatures, or monthly mean sea level pressures, depending on the data file. Temperatures are given in degrees Celsius (times 10), e.g., a temperature of -125 would indicate -12.5C. Sea level pressure or station pressure is given in hectopascals (times 10), e.g., a pressure of 9876 would indicate 987.6hPa. Missing monthly data are indicated by the value -999; ANN is the annual mean of the 12 monthly means, given in the same units as the monthly values. Missing annual data are also indicated by the value -999.
The data and an HTML version of the documentation may also be obtained from CDIAC's web site at http://cdiac.esd.ornl.gov/.
For non-internet data acquisitions (e.g., floppy disk, 8mm tape, CD-ROM, etc.), users should contact CDIAC directly.
NOTE: When using these climate data in a presentation or publication, PLEASE acknowledge the principal investigators, P. D. Jones and P. A. Reid of the Climatic Research Unit, University of East Anglia, Norwich, U.K.
Jacka, T. H. and W. F. Budd. 1998. Detection of temperature and sea ice extent changes in the antarctic and southern ocean. Ann. Glaciol. 27:553-559.
Jones, P. D. 1990. Antarctic temperatures over the present century - A study of the early expedition records. J. Clim. 3(11):1193-1203.
Jones, P. D. 1995. Recent variations in mean temperature and the diurnal temperature range in the Antarctic. Geophy. Res. Lett. 22(11):1345-1332.
Jones, P. D. and D. W. S. Limbert. 1987. A databank of Antarctic surface temperature and pressure data. U.S. Dept. of Energy Technical Report, TR038, U.S. Dept. of Energy, Washington D.C.
Jones, P. D. and T. M. L. Wigley. 1988. Antarctic gridded sea level pressure data: An analysis and reconstruction back to 1957. J. Clim. 1(12):1199-1220.
King, J. C. 1994. Recent climate variability in the vicinity of the Antarctic Peninsula. Int. J. Clim. 14:14357-14369.
King, J. C. and S. A. Harangozo. 1998. Climate change in the western Antarctic Peninsula since 1945: observations and possible causes. Ann. Glaciol. 27:571-575.
Raper, S. C. B., T. M. L. Wigley, P. R. Mayes, P. D. Jones and M. J. Salinger. 1984. Variations in surface air temperatures, part 3: The Antarctic 1957-1982. Mon. Weather Rev. 112:1341-1353.
Simmonds, I. and W. F. Budd. 1991. Sensitivity of the Southern Hemisphere circulation to leads in the Antarctic pack ice. Q. J. R. Meteorol. Soc. 117:1003-1024.
Skvarga, P., W. Rack, H. Rott, and T. I. Donangelo. 1998. Evidence of recent climatic warming on the eastern Antarctic Peninsula. Ann. Glaciol. 27:628-632.
Van den Broeke, M. R. 1998. The semi annual oscillation and Antarctic climate, part 1: Influence on near-surface temperatures (1957-1979). Antarctic Science 10(2):175-183.
Weatherly, J. W., J. E. Walsh, and H. J. Zwally. 1991. Antarctic sea ice variations and seasonal air temperature relationships. J. Geophys. Res. 96:15119-15130.
Weller, G. 1998. Regional impacts of climate change in the Arctic and Antarctic. Ann. Glaciol. 27:543-552.