Original paper

Upper-air observations from the German Atlantic Expedition (1925–27) and comparison with the Twentieth Century and ERA-20C reanalyses

Stickler, Alexander; Storz, Samuel; Wartenburger, Richard; Hersbach, Hans; Compo, Gilbert P.; Poli, Paul; Dee, Dick; Brönnimann, Stefan

Meteorologische Zeitschrift Vol. 24 No. 5 (2015), p. 525 - 544

38 references

published: Aug 31, 2015
published online: Aug 31, 2015
manuscript accepted: Jul 7, 2015
manuscript revision received: Jul 3, 2015
manuscript revision requested: Jun 29, 2015
manuscript received: Mar 23, 2015

DOI: 10.1127/metz/2015/0683

BibTeX file


Open Access (paper can be downloaded for free)

Download paper for free


Between April 1925 and June 1927, the research vessel Meteor cruised the tropical and southern Atlantic Ocean in the framework of the German Atlantic Expedition. One purpose was to systematically explore the vertical structure of the atmosphere. To this end, the ocean was crossed in 14 profiles across parallels of latitude. 801 pilot balloons and 217 kites were launched. The resulting data have been digitised in the framework of the European project ERA-CLIM. Here, they are compared to the Twentieth Century (20CR) and ERA-20C reanalyses, independent datasets based on the assimilation of synoptic pressure and hurricane pressure records, and marine surface winds for the latter, using monthly sea surface temperature and sea ice as boundary conditions. Both reanalyses display similar patterns of systematic differences relative to the observations for temperature, specific humidity, and wind. Furthermore, 20CR and ERA-20C show generally comparable anomaly correlations for all parameters, with the highest values found for pressure and temperature. In the southern extratropics, high (> 0.75) anomaly correlations are found for pressure in both 20CR and ERA-20C, and for temperature in 20CR. Medium (> 0.5) anomaly correlations are found for specific humidity in 20CR. Moderate anomaly correlations (> 0.44) are found for meridional wind in both 20CR and ERA-20C, and for temperature in ERA-20C. In contrast, low anomaly correlations (< 0.44) are found for zonal wind both in 20CR and ERA-20C, and for specific humidity in ERA-20C. In the Tropics, low anomaly correlations are found for all parameters except for pressure, which shows medium anomaly correlations for both 20CR and ERA-20C, and for meridional wind, which shows moderate anomaly correlations for 20CR. In all regions, both reanalyses strongly underestimate the observed range of zonal and particularly meridional wind variability. Even though remaining errors in the observational data cannot be excluded, we estimate that the inherent observational uncertainties do not alter our conclusions. Vice versa, two pieces of evidence support the credibility of the early upper-air data: the robust regressions of both 20CR and ERA-20C against observed pressure and temperature over a large spatial and temporal range, and the similarity between the uncertainties predicted by 20CR and the actual uncertainties determined from the root mean square difference of reanalysis and observation values.


Twentieth Century ReanalysisERA-20Cupper-air dataERA-CLIMCHUANkitepilot balloonGerman Atlantic ExpeditionMeteorAtlantictemperaturepressurewindcirculationhumidityTropicsextratropics