Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Info
iconfalse
titleTable of Contents

Table of Contents
maxLevel3

History of modifications

Expand


Version

Date

Description of modifications

Chapters / Sections

1.0

23/06/2022

First draft

Whole document

2.0

14/09/2024

Major Update to CUONv.3 

Section 4-5-6

3.0

20/06/2025

Update for public release of dataset version 1.1.0

Whole document


Acronyms

Expand


AMMA

African Monsoon Multidisciplinary Analysis

ATBD

Algorithm Theoretical Basis Document

BUFR

Binary Universal Form for the Representation

CSV

Comma Separated Values (a table interchange format)

CDM

Common Data Model

ECMWF

European Center for Medium-Range Weather Forecasts

ECV

Essential Climate Variable

ERA5

5th Generation European Reanalysis

GIUB

Geographisches Institut Universität Bern

HARA

Historical Arctic Rawinsonde Archive

IGRA2

Integrated Global Radiosonde Archive, version 2

NCAR

National Center for Atmospheric Research

NCEP

US National Centers for Environmental Prediction

NetCDF

Network Common Data Format

WOUDC

World Ozone and Ultraviolet Radiation Data Centre

MAESTRO

Mesoscale organisation of tropical convection (field campaign)

ODB

ECMWF internal representation of observation and observation feedback data

...


Product description

The In situ Comprehensive Upper-Air Observation Network (CUON) serves the most comprehensive harmonised public collection of conventional radiosonde and pilot balloon data available to date. It contains measurements of the ECVs temperature, humidity (relative, specific, dewpoint, dewpoint depression) and wind (speed and direction, u and v components). The network consists of more than 5400 individual upper-air observing stations at fixed location, and ca. 9000 mobile platforms. While the majority of records contain only a limited number of observations, around 1000 observing stations provide data spanning several decades. Besides the measured values and the most basic metadata (position, time), it provides observation minus background and observation minus analysis departure values from ERA5 during assimilation. CUON contains homogeneity adjustments for temperature and several wind as well as humidity variables that can be applied to the raw observations, together with observation error estimates derived from the departure statistics based on ERA5 back to 1940. Metadata regarding the instrumentation type information is also provided. All the information adhere to the Common Data Model for observations https://github.com/ecmwf-projects/cdm-obs/blob/master/ and more specifically to its subset of variables, the CDM-OBS-core .
Data are provided in binary (netCDF) as well as compressed CSV format, which are best read by standard python packages (e.g. xarray or pandas, h5py). Requests can be created via the CDS forms or using simple python scripts. A few example scripts that demonstrate basic usage are given below. 

...


Figure 1: Location of fixed-based station records that contain at least one ascent. Notes: (1) "ECMWF" is the union of several data collections (see section 5, identifiers "ERA5 1" , "ERA5 2", "ERA5 1759" and "ERA5 1761"); (2) Station records may consist of ascents from different sources, and colours are plotted over the previous colours in the order listed in the legend.

Data and Metadata Sources

The following data sorces were used to build the CUON "merged" database. References describing the
sources and the data can be found in the reference section of the download tab.

...

Metadata on the sensors used has been extracted from a database described by S.Schroeder ( https://ams.confex.com/ams/pdfpapers/54950.pdf) up to the year 2013 if no other instrumentation metadata was available. Other instrumentation metadata is retrieved directly from the above sources where it was reported using the WMO manual on codes https://library.wmo.int/doc_num.php?explnum_id=10235). More details on the specific codes used are given in the Appendix. 

Data Processing

The original merged data set contains a total of 5408 merged netCDF files for fixed-position station, and 9634 for mobile stations. The total amount of records exceeds 50 million.

...

For a detailed description how those are calculated, we refer to the ATBDs, which accompany this document.


Access and Examples

Users can retrieve CUON data with requests generated by the web form or by using the CDS API.

...

Please be aware that retrievals of time series at this stage (June 2025) take some time (~5 min) and may fail due to memory limitations if more than 5 years are retrieved. In this case it may be necessary to split requests into shorter time slices and then concatenate the series. 

How to use additional metadata

CUON records contain sensor metadata collected offline by various researchers in the past (e.g. Gaffen, 1994, Schroeder, 2013). Since several years, this information has been reported also in standard TEMP and BUFR messages, using WMO manual on codes 1.2:  https://library.wmo.int/records/item/35625-manual-on-codes-volume-i-2-international-codes. In CUON the instrument codes of Schroeder (2013) and the WMO have been harmonised to be able to provide metadata from early years as well as for recent ascents in one data column (sensor_id). This column can be used to filter stations using certain radiosonde types. 

...

Figure 6: Drift of balloons at station Vienna during the period 1980-1982 (ca. 2000 ascents). Red: All positions reached, Yellow: Position reached at or below the 300 hPa level, Black: Position of launch site. 

How to work with background departures and bias adjustments

Bias adjustments are available for all temperature, humidity and wind variables except wind speed. They are automatically retrieved together with the observation values and other columns. Owing to the method used for calculating them, they are usually set to zero if the available time series for a record is shorter than 6 months. Breakpoints and bias adjustments were calculated through analysis of the variable fg_depar@offline in the case of wind (see the corresponding ATBD document). To get adjusted observation time series one needs to subtract them from the observation_values. Users should do that when calculating statistics such as trends from the observation time series, when comparing observations e.g. with satellite radiances, or when calculating gridded observation products. 

...

For expert users: Please note that it makes no sense to subtract the adjustments from fg_depar@body and an_depar@body for temperature, since these have been calculated AFTER subtracting the bias adjustments used when assimilating ERA5. The situation is different for humidity and wind adjustments, since humidity and wind were not adjusted when assimilating ERA5. 

How to retrieve uncertainty estimates

Observation uncertainty estimates have been calculated for records where the data source was ERA5_1 or ERA5_2 (about 76% of records) from the archived fg_depar@body and an_depar@body values, using the Desroziers (2005) method, using an averaging interval of 30 days. Note that these estimates contain
both measurement and representation errors, and it was assumed that the assimilation process was bias-free. This assumption is justifiable for most of the temperature and wind data, for humidity data it is questionable since these are known to have relatively large unadjusted biases.
Still we consider these estimates a quite useful tool to show the improvement in the insitu upper-air observation input over time. 

...

Figure 8: .Time series of fg_depar@offline, without adjustment (blue) and with adjustmend (fg_depar adj), for radiosonde station 91408 (KOROR, in the tropical Pacific). Left: for relative humidity at 200hPa, right for temperature at 50 hPa. The shift in 1995 is clearly indicated. For details on the calculation of the adjustments, see the ATBD on temperature and humidity adjustments (to be published soon). 


How to request the intercomparison data

CUON contains ascents from many international radiosonde intercomparisons, such as the one in Vacoas, Mauritius, in 2010. This information will be released at a later stage. Please look for updates of this document in the forthcoming months. 

User support

The service team provides tier 2 support with a latency of one week. Queries should be submitted as JIRA requests via the CDS. They will be directed to us. We will try to maintain an FAQ.

Acknowledgements

The authors wish to thank the C3S team and collaborators for their continuous support, particularly Paul Poli, Markel Garcia-Diez and Edward Comyn-Platt.

References (other than data sources)

Desroziers, G., Berre, L., Chapnik, B. and Poli, P. (2005), Diagnosis of observation, background and analysis-error statistics in observation space. Q.J.R. Meteorol. Soc., 131: 3385-3396. https://doi.org/10.1256/qj.05.108

...

Voggenberger, U., Haimberger, L., Ambrogi, F., Poli, P.,: Balloon drift estimation and improved position estimates for radiosondes, GMD, https://doi.org/10.5194/gmd-17-3783-2024 2024

 Appendix

As noted above, the comprehensive in situ upper-air data set relies on a station inventory. It also contains a few special tables or variables, which are either extensions to the CDM, like the ERA5 observation feedback table, or where the CDM-OBS only gives table definitions and the data provider has to fill in the tables. Examples are the header_table or sensor_configuration tables.

Radiosonde type codes


Instrumentation metadata is essential for assessing the uncertainty of observations. Instrument type codes can be retrieved using the optional variable 'type', but in order to interpret the codes, a sensor_configuration table, which resolves the codes, is needed. The upper-air data set distinguishes between 3000 instrumentation codes. Most of them come from S. Schroeder's vapor data base (Schroeder, 2008). From 2013 onwards, WMO BUFR codes (as described in the WMO manual on codes https://library.wmo.int/records/item/35625-manual-on-codes-volume-i-2-international-codes) are used. The whole table can be downloaded not from the CDS API but from https://gitlab.phaidra.org/ulrichv94/CUON/-/blob/master/CUON/public/merge/sensor_configuration_all.csv, where the columns sensor_id and comments contain the relevant information. A few example rows are given below. 

...


The sensor_id code given in the output from the CDS API matches to one row of this table. 

Station inventory

Table A. Links to comma-separated value station inventory files featuring station availability.

...

Info
iconfalse

This document has been produced in the context of the Copernicus Climate Change Service (C3S).

The activities leading to these results have been contracted by the European Centre for Medium-Range Weather Forecasts, operator of C3S on behalf of the European Union (Delegation Agreement signed on 11/11/2014 and Contribution Agreement signed on 22/07/2021). All information in this document is provided "as is" and no guarantee or warranty is given that the information is fit for any particular purpose.

The users thereof use the information at their sole risk and liability. For the avoidance of all doubt , the European Commission and the European Centre for Medium - Range Weather Forecasts have no liability in respect of this document, which is merely representing the author's view.

Related articles

Content by Label
max5
cqllabel in ("in-situ","cuon") and type = "page" and space = "CKB"

...