Advances in homogenisation methods of climate series: an integrated approach HOME

  • Increase font size
  • Default font size
  • Decrease font size


The action is over, the main conclusions are:

1. Homogenisation improves climate data and does not cause artificial trends. Because the test was blind and because of the realism of the data, this can now be stated with confidence.
2. Modern algorithms, which are designed to also work with an inhomogeneous reference, are clearly better than traditional ones. It needed a realistic benchmark dataset with surrogate climate networks to see this difference clearly.

3. Two new software packages containing some of the methods recommended by HOME are now avalible. The code has been produced by Olivier Mestre, École Nationale de la Météorologie, Météo France, Tolouse". HOMER (for monthly data) and HOM/SPLIDHOM (for daily data)


The main article of HOME has been published in Climate of the Past. In this paper homogenisation methods were blindly tested on a dataset with unprecedented realism. All the most used and best algorithms have participated.

To view the article: "Benchmarking homogenization algorithms for monthly data"


Please, download the Monthly Benchmark Dataset .

Please, check our literature review on homogenisation .


Introduction to the Action Cost-ES0601

Long instrumental climate records are the basis of climate research. However, these series are usually affected by inhomogeneities (artificial shifts), due to changes in the measurement conditions (relocations, instrumentation and others).As the artificial shifts often have the same magnitude as the climate signal, such as long-term variations, trends or cycles, a direct analysis of the raw data series can lead to wrong conclusions about climate change.

In order to deal with this crucial problem many statistical homogenisation procedures have been developed for detection and  correction of these inhomogeneities. At present only a limited number of publications intercompare some common methods and their impact on the climate record.The large number of different methods could be seen as a weakness in the science and is a challenge for the climatological  community to address.

There is therefore a need for a coordinated European initiative in order to produce standard methods designed to facilitate such comparisons and promote the most efficient methods of homogenisation.

The Action's main objective is to achieve a general method for homogenising climate and environmental datasets.The method will be derived from the most adapted statistical procedures for detection and correction of varying parameters at different space and time scales.

See Memorandum of Understanding for more information.

Picture from the MC meeting Bucarest (19-20th of May, 2010).

The action is structured in 5 working groups:



Installation requirements

E-mail Print PDF
For series with data gaps:
1. Download First Version of "cghseg" (this is 03 Aug 2011 version)
2. Install the package "cghseg" downloaded:
  • under Windows: packages => Install package(s) from local zip files...
  • under LINUX: type:  " install.packages('cghseg_0.0.1.tar.gz') "
3. Install packages "maps" and "mapproj"
  • under Windows: packages => Install package(s) => select your CRAN miror => select the package "maps" | then do the same for "mapproj"
  • under LINUX: type: install.packages('maps') | install.packages('mapproj')

Windows users have to work with version R 2.15.0



COST Benchmark Workshop Zurich

E-mail Print PDF

Sept. 14, 2010.
Participants: Victor Venema, Petr Štepánek, Gregor Vertacnik, Tamás Szentimrey, Jose A Guijarro, Barbara Chimani, Renate Kocen, Michele Brunetti, Lars Andresen.

The Report


Expert meeting in Oslo 25-26 November 2009

E-mail Print PDF

Homogeneity testing of early instrumental series. Methods of testing long-term series from northern Europe in particular. The Report