REBECCA - Real-time Blending EPS-METEOSAT by Canonical Correlation Analysis

The goal of this project is to develop a standardized procedure of merging potential synergistic combination of satellite data of the geostationary orbit and those of the low earth orbit which at first sight appear rather incompatible in terms of content, spatial and temporal availability.

Short Description

Two different types of satellites are currently used in operational meteorology:

  1. The geostationary satellites of the “Meteosat Second Generation” (MSG) series, which deliver images of 12 spectral (visible and infrared) channels; these images are available every 15 minutes.
  2. The polar-orbiting satellites, like MetOp-A of the EUMETSAT Polar System (EPS) that has been orbiting since 19 October 2006. Albeit the temporal resolution is much lower, EPS can reveal a lot of interesting details of the state of the atmosphere by virtue of (partly unique) instruments that deliver information about the vertical structure of the atmosphere (temperature and humidity profiles), about its chemical composition (trace gases) and about precipitation – all of them parameters which cannot be seen (directly) by MSG.

The question which arose is about a potential synergistic combination of satellite data of the geostationary orbit and those of the low earth orbit. 

With the help of such a system, information of EPS should be extended in time and to such geographical regions where EPS is currently not available. The correlation with MSG channels within the overlapping areas shall provide the information necessary to simulate the EPS parameter.

Meteorological correlations are often dependent on season, time of day, latitude or the general synoptic situation. Especially the last one is hard to catch by conventional regressions. Therefore the employed methods do not search for average relations between meteorological variables, but – analyzing each image individually – are based on the most recently observed interdependencies.

The prominent correlations and common patterns are sought in a more-dimensional data space by the help of eigentechniques, a group of sophisticated statistical methods (comprising principal component analysis, factor analysis, independent component analysis, canonical correlation analysis,...).

Project Partners


Zentralanstalt für Meteorologie und Geodynamik - Alexander Jann

Contact Address

Zentralanstalt für Meteorologie und Geodynamik
Alexander Jann
Hohe Warte 38
A-1190 Wien
Tel.: +43 (1) 360 26 23 16