Abstract. The reliable prediction of future atmospheric CO₂ concentrations and associated global climate change requires an adequate understanding of the CO₂ sources and sinks. The sparseness of the existing surface measurement network limits current knowledge about the global distribution of CO₂ surface fluxes. The retrieval of CO₂ total vertical columns from satellite observations is predicted to improve this situation. Such an application however requires very high accuracy and precision. We report on retrievals of the column-averaged CO₂ dry air mole fraction, denoted XCO₂, from the near-infrared nadir spectral radiance and solar irradiance measurements of the SCIAMACHY satellite instrument between 2003 and 2005. We focus on northern hemispheric large scale CO₂ features such as the CO₂ seasonal cycle and show - for the first time - that the atmospheric annual increase of CO₂ can be directly observed using satellite measurements of the CO₂ total column. The satellite retrievals are compared with global XCO₂ obtained from NOAA's CO₂ assimilation system CarbonTracker taking into account the spatio-temporal sampling and altitude sensitivity of the satellite data. We show that the measured CO₂ year-to-year increase agrees within about 1 ppm/year with CarbonTracker. We also show that the latitude dependent amplitude of the northern hemispheric CO₂ seasonal cycle agrees with CarbonTracker within about 2 ppm with the retrieved amplitude being systematically larger. The analysis demonstrates that it is possible using satellite measurements of the CO₂ total column to retrieve information on the atmospheric CO₂ on the level of a few parts per million.