Dataset Downloads

A selection of data files can be downloaded from the IRI Data Library for use as either predictors or predictands. Three types of data are available:

Selecting one of the datasets will open a window browser in your default browser. All the datasets accessible from within CPT consist of monthly data, or daily data converted to monthly, but it is possible to calculate seasonal averages or totals. The datasets available for download are detailed in the file download_IRIDL.txt, which is stored in the data subdirectory of CPT. Data can be downloaded as:

  • Specific month : enter the corresponding month number (1 for January, 2 for February, etc.) as the First month of data to download , and 1 for the subsequent prompts.
  • Multiple months : to download, for example, April, May and June as separate months (i.e., lagged fields), then the responses should be 4 (for April as the first month), 1 (for no seasonal averaging) and 3 (three months separately).
  • Specific season : enter the number of the month for the first month of the season, and then the number of months in the season, leaving the Number of seasons as 1. For example, for a April - June (AMJ) average, enter 4 (for April), 3 (for a 3-month average) and 1 (one season only).
  • Sequential seasons : for example, entries of 4 (first month April), 2 (2-month seasons), and 3 (3 consecutive seasons) will download data for (AM, JJ, and AS). It is not possible to download seasonally overlapping data (e.g., JFM, FMA, MAM) using this feature.

If you want to combine datasets, download the individual datasets first, and then use the Data ~ Merge menu item.

It is possible to modify this file in order to update datasets or add new or remove old ones, but the file needs to be formatted carefully. The file is structured as follows, and all the tags are case-sensitive:

  • cpt:downloads : lists all the available variables and all the available data sources for each of the three data types. Each variable has an identifier, which in most cases corresponds to the GRIB code for that variable. In CPT this identifier is indicated by the tag cpt:ivar. After each variable the available data sources are listed. The following variables are available. This listing cannot be changed.
    • cpt:ddt=obs for Observations
      • cpt:var=ENSO for ENSO indices (cpt:ivar=300)
      • cpt:var=sst for sea-surface temperatures (cpt:ivar=80)
      • cpt:var=Tmean for mean air temperatures (cpt:ivar=11)
      • cpt:var=Tmax for mean maximum air temperatures (cpt:ivar=15)
      • cpt:var=Tmin for mean minimum air temperatures (cpt:ivar=16)
      • cpt:var=prcp for precipitation (cpt:ivar=61)
      • cpt:var=wetdays for number of wet days (cpt:ivar=193)
    • cpt:ddt=reanalyses for Reanalyses
      • cpt:var=mslp for mean sea-level pressure (cpt:ivar=2)
      • cpt:var=gph for geopotential heights (cpt:ivar=7)
      • cpt:var=u for zonal wind components (cpt:ivar=33)
      • cpt:var=v for meridional wind components (cpt:ivar=34)
      • cpt:var=uv for both wind components (cpt:ivar=32)
    • cpt:ddt=gcmouts for GCM Outputs
      • cpt:var=Tmean for mean air temperatures (cpt:ivar=11)
      • cpt:var=Tmax for mean maximum air temperatures (cpt:ivar=15)
      • cpt:var=Tmin for mean minimum air temperatures (cpt:ivar=16)
      • cpt:var=prcp for precipitation (cpt:ivar=61)
      • cpt:var=sst for sea-surface temperatures (cpt:ivar=80)
      • cpt:var=gph for geopotential heights (cpt:ivar=7)
      • cpt:var=u for zonal wind components (cpt:ivar=33)
      • cpt:var=v for meridional wind components (cpt:ivar=34)
    For each variable there are two additional compulsory tags:
    • cpt:sumvar is set to 1 if a seasonal total should be calculated (e.g., for precipitation or wetdays), and is set to 0 if a seasonal average should be calculated. This number should not be changed.
    • cpt:nsource should be set to the number of data sources available. This number should be changed if sources are added or removed. The list of sources immediately follows, and it is important that the correct number of sources is listed.
    The sources are then listed, each source taking one line, and using the following compulsory tags:
    • cpt:csource is set to the name of the dataset, and must contain no spaces or commas;
    • cpt:isource is set to a unique identifier for that data type and variable (this number is not used in the Windows version of CPT, but must be set);
    and the following optional tags:
    • cpt:ioldsource is set to a formerly used, but unique identifier for that data type and variable (this number is not used in the Windows version of CPT);
    • cpt:password should set to 1 if the dataset requires password access;
    • cpt:project indicates to which group or project the dataset belongs;
    All tags are separated by commas.

  • cpt:sources : lists details of all the available data sources. The source details are structured as follows, and must appear in the following order:
    • cpt:csource : the name of the dataset exactly as it appears in the cpt:downloads section. The name needs to be unique and if it is a subset of another name the shorter name must appear first (for example, CFS2 appears before CFS2_(original))
    • structure : the data structure, either "index" or "grid" (station data cannot be downloaded by CPT)
    • portal : the dataset portal, which is usually iridl.ldeo.columbia.edu/ but other portals can be used
    • sourceURL : the URL for the source of the dataset
    The remaining part of the section depends on whether the dataset is an index or gridded dataset.
    • For index data:
      • nIndices : the number of indices
      • year1 : the first year of data in dataset
      • tRes : the temporal resolution of the data - either "monthly" or "daily" (if data are daily, the Data Library will calculate monthly averages before downloading the data)
      • cIndices : the names of the available indices - they must appear on one line between square brackets and with spaces separating the indices from the square brackets
      Then follows the name of the available variables using the cpt:var tag, exactly as it appears in the cpt:downloads section, together with the URL for the variable, using the cpt:varurl tag. The actual URL for the index is constructed by appending the value of cpt:varurl to the name of the selected index.
    • For gridded data:
      • nDatasets : the number of datasets - will usually be 1, but should be set to 2 in the case where there are separate hindcast and forecast datasets for GCM outputs
      • year1 : the first year of data in dataset
      • date2 : the first date of the second (i.e., forecast) dataset, must be in the format YYYY MM DD
      • tRes : the temporal resolution of the data - either "monthly" or "daily" (if data are daily, the Data Library will calculate monthly averages before downloading the data)
      • nEns1 : the number of ensemble members in the first dataset - this line should only be included if the dataset contains GCM outputs
      • nEns2 : the number of ensemble members in the second dataset - this line should only be included if the dataset contains GCM outputs and if there are separate hindcast and forecasts datasets (i.e., if nDatasets=2)
      • nLevels : the number of pressure levels - must be set to 0 if pressure levels are not explicitly included in the dataset, otherwise it must be set to the correct number of levels
      • P or P2 : the line must begin with the Data Library grid for pressure level (typically P or P2), and then the list of available pressure levels should follow - they must appear on one line between square brackets
      • nLat : number of latitudes
      • nLon : number of longitudes
      • limitN : northernmost latitude
      • limitS : southernmost latitude
      • limitW : westernmost longitude
      • limitE : easternmost longitude
      • N2S : north-south orientation, must be set to 1 if the first latitude is northernmost, and to 0 otherwise
      • dataURL : the URL for the first dataset
      • dataURL : the URL for the second (i.e., forecast) dataset - this line should be included only if there are separate hindcast and forecast datasets
      Then follows a list of the available variables using the cpt:var tag, exactly as it appears in the cpt:downloads section, together with the URL for the variable, using the cpt:varurl tag. If a data conversion is required, the code for the conversion can be listed using the cpt:unitconv tag. This list should correspond with the list in the cpt:downloads section. If a variable is listed here, but not in the cpt:downloads section it will be ignored. If a source is listed in the cpt:downloads section but not in the corresponding cpt:csource , an error will occur.

The indentations in the file serve only to ease readability, and are not part of the structure of the document.

Last modified: