Status of the Integrated Global Radiosonde Archive
(Additional information on IGRA is available at 
www.ncdc.noaa.gov/oa/cab/igra.)


January 6, 2011, new IGRA-derived data:
Version 2 of the IGRA-derived sounding parameters is now available, 
containing a significant number of additional parameters. For 
details, see the readme.txt file in subdirectory derived-v2. The 
related webpage will be updated shortly. For the benefit of regular 
users, version 1 will continue to be available through March 31, 2011, as well.

January 6, 2011, known uncorrected problem:
Several users have identified stations in IGRA where wind speeds during 
January and February of the years 1975-1978 and 1981-1982 are an order 
of magnitude too large when compared with other data sources. Stations in 
Texas, Alaska, and the Canadian Arctic are known to be affected. Unfortunately,
we are unable to address this issue at this time. It will, however, be 
rectified in the next version of IGRA. In the meantime, users are advised to 
look out for this issue at any IGRA station during the affected years and 
months and handle it as they see it.

October 5, 2010:
Thanks to Liangying Zhang of NCAR, the metadata for U.S. and Australian stations
Have been updated. The 475 added or edited records are
identified with an update date of 06/2010 and a reference of "Updated by
NCAR/EOL". Updates are based on information received by NCAR from the U.S.
National Weather Service and the Australian Bureau of Meteorology.


April 29, 2010:

An updated IGRA metadata file has been placed online.

The main changes compared to the previous version include:
- the addition of several thousand records;
- a change in the format to accommodate instrument names up to 40 
  characters in length; 
- standardization of instrument names;
- elimination of various format issues; and
- enhanced documentation.

For details, including a description of the new format, see
http://www1.ncdc.noaa.gov/pub/data/igra/igra-metadata-description.txt. 


November 18, 2009, change:

An error in The calculation of relative humidity in the IGRA-derived data files
was fixed in the version that became available on 18 November 2009. The error
was due to inadvertent premature rounding of the actual and saturation vapor
pressures to the nearest 0.1 mb and the unintentional rounding of the relative
humidity itself to the nearest whole percent rather than the nearest 0.1%. the
impact of the error ranged from less than 1% under warm, moist conditions to a
few tens of a percent when the saturation vapor pressure was less than 1 mb
(i.e., under cold and dry conditions under which radiosonde humidity
measurements are also not very reliable). At the same time, the precision of
both the actual end saturation vapor pressures written to the output files has
been increased from mb*10 to mb*1000 in order to achieve improved consistency
between the vapor pressure and relative humidity values written. Users are
cautioned, however, that this precision is not necessarily a reflection of the
actual measurement precision of the instrument. Since the entire data set is
re-created every day, both changes apply to all IGRA-derived data effective
immediately, and returning users who may have been affected by the error can
refresh their own holdings. Thank you to Penny Rowe and her colleagues for
pointing out the issues.


January 3, 2008, change:

The following changes have been made to the format of the IGRA-derived 
data:
1. The missing value code for all variables has been changed from -9999 to 
-99999, so that it falls outside the range of possible values in all 
fields. Accordingly, calculated values that fall outside the range -99998 
and 99998 are set to missing.
2. The fields for actual vapor pressure and saturation vapor pressure have 
been moved, such that they now immediately precede relative humidity. See 
the readme file in the subdirectory "derived" for specifics on the format 
of the derived data files.
3. When the saturation vapor pressure is 0, relative humidity is now set 
to missing rather than to 0.

August 23, 2007, fix:

1. The pre-2006 monthly means, which had been lost during the last monthly 
   update on August 5, have been restored and, as of August 23,
   they are based on IGRA version 1.2.
2. The files in the monthly-por and monthly-upd directories are now being
   compressed into .gz format as the sounding files have been since July. 
   They can be uncompressed using a Windows unzip utility or the Linux/UNIX 
   gunzip command.

August 20, 2007, Problem:

Due to an oversight, the changes that were made to the online sounding files
in July were not applied to the monthly-mean files. As a result, the monthly
files are currently still being compressed into .Z rather than .gz format, and
the period-of-record monthly-mean files currently only contain data for 2006
and 2007. We will get the problem rectified as soon as possible and update and
notify users of the fix via the status file.

August 9, 2007, update:
The igra-composites.txt file has now been updated to reflect the compositing
performed as part of the last period-of-record reprocessing. See the
readme.txt file for further details on the composites list.


August 6, 2007, Fix:
During the last period-of-record update in mid-July, the 1949-1973 data 
for 72249 had been lost. These data have now been restored.

July 30, 2007, addition:
Some parameters that have been derived from IGRA are now available in the
subdirectory "derived" on the FTP site of IGRA. Parameters include, for
example, potential temperature, relative humidity, and temperature gradient
between consecutive levels and are computed for each suitable level available
in IGRA. For details, see
http://www1.ncdc.noaa.gov/pub/data/igra/derived/readme.txt. 
July 30, 2007, addition:
Some parameters that have been derived from IGRA are now available in the
subdirectory "derived" on the FTP site of IGRA. Parameters include, for
example, potential temperature, relative humidity, and temperature gradient
between consecutive levels and are computed for each suitable level available
in IGRA. For details, see
http://www1.ncdc.noaa.gov/pub/data/igra/derived/readme.txt. 

July 16, 2007, changes to create IGRA Version 1.2:
1. Data files are now compressed in GZIP format (creating files with the 
   extension .gz) rather than by using the Unix "compress" command (which 
   generates .Z files). The .gz files can be uncompressed, for example, using 
   WINZIP or the Linux "gunzip" command.
2. The data through the end of 2006 that are contained in the 
   period-of-record files in the data-por directory have been processed through 
   the entire IGRA processing system, including those checks that cannot be 
   run during the daily processing (see Durre et al. 2006). Accordingly, the 
   year-to-date files in the data-y2d directory now begin with January 1, 2007.
3. The dates in the igra-composites.txt files have yet to be updated
   accordingly.

November 2006 correction:
The period-of-record monthly-mean files uploaded this past summer were 
found to be missing data for all inactive stations. The problem has been 
rectified, and the files now again contain data for all qualifying IGRA 
stations. See the readme file for more information on these files.

July 2006, addition:
The directory monthly-upd was added to the IGRA FTP directory. It contains
files of monthly means for only the last full month of data available and is
intended for users interested in only those monthly means that have newly
become available in the past month.

July 2006, change to version 1.10:
1. The entire period of record through December 31, 2005, has been reprocessed
through the entire suite of quality control checks. The reprocessing permitted
the application of checks for runs-in-time and temporal inconsistencies that
cannot be applied during the daily updates and allowed for the use of
climatologies and monthly median elevation statistics that have been updated
using data from 2004 in 2005. As a result of this reprocessing, some values
previously retained are now removed, and vice versa. The number of values
deleted experiences a net change by approximately 7000, or 0.2% of the total
number of values deleted.
2. The year-to-date files in directory data-y2d have been changed to begin on
January 1, 2006, rather than January 1, 2004. 
3. Monthly means have been recomputed using the reprocessed data. 
4. The list of IGRA composites has been updated to reflect the period of
record through December 31, 2005, and the last year of data in the IGRA
station list has been updated to reflect data holdings as of July 2006.

February 6, 2006:
The paper describing IGRA has appeared in the January 1, 2006, issue of the
Journal of Climate. See the references link on the IGRA homepage.

January 27, 2006, addition:
Station history information for IGRA stations is now available at the 
"metadata" link on the IGRA homepage.

November 10, 2005, change:
Inconsistencies between the country codes in the IGRA station list and those
in the country code list were discovered and fixed today. 

August 2005, change:
The files of monthly means are now sorted properly by station number and 
year. This fixes the issue of 2004+ data being appended to the bottom of files
that was discovered in January 2005.

July 18, 2005, addition:
A paper providing an overview of IGRa (accepted by J. Climate) is now
available in the FTP directory as igra.overview.pdf.

July 18, 2005, change:
The files containing lists of IGRA stations, country codes, and composite
stations have been renamed to igra-stations.txt, igra-countries.txt, and
igra-composites.txt in order to adhere to NCDC standards.

June 29, 2005, changes (creating v1.02):
1. Reorganized FTP directory to provide all documentation in the main igra
   directory and streamline the data directories (see readme.txt).
2. Added stations 43599 (Gan, Maldive Islands) and 83779 (Marte, Brazil)
   to the IGRA station list. Processing of the two new stations began in 
   February, and their data extend back to 2004.
3. In the station list, GUAN identifiers and periods of record were updated,
   and the 3-digit country codes were replaced with 2-character FIPS codes
   (see readme.txt) to conform with NCDC convention.

May 21, 2005, discovery and fix:
Fourteen extraneous station files were discovered in the data-por subdirectory
under the IGRA FTP directory. These files have been removed.

November 2004 addition:
Beginning of near-realtime daily updates to the sounding data.


October 2004 discovery:
The Thai stations 48327 and 48407 have a tendency for reporting
soundings whose launch time differs by 5-6 hours from their observation
hour and, therefore, are elim`nated by the quality assurance process. The
issue may be related to launch times being reported as local time rather
than as UTC and will be addressed when the next version of IGRA is created.
(During IGRA development, an analogous problem was noted at the 
South Pole Station (89009) for 2000 and 2001, prompting the exclusion 
of that station from the check for consistency between launch time and 
observation hour.)

September 2004 addition:
Beginning of monthly updates to the monthly-mean data.

May 2004: 
Completion of the base version of IGRA 1.0 (through 2003)