The Fermi bubbles (Su et al 2010) are two large structures in the Fermi-LAT gamma-ray data extending about 55 degrees above and below the Galactic center. We use two independent data analysis methods to study the spatial distribution and the energy spectrum of the Fermi bubbles. The first method is a well established analysis procedure based on fitting templates to the gamma-ray data using gamma-ray emission model maps generated by the GALPROP code as the templates.

The second method is a novel method in diffuse gamma-ray data analysis. It is a generalization of the principal components analysis and the internal linear combinations method developed by the WMAP collaboration. In this method, the templates for the emission components are composed by taking linear combinations of the gamma-ray data in different energy bins. As a result, one avoids the uncertainties related to the use of external data for modeling the templates of the emission components. The consistency between the results obtained with the two methods strengthens their robustness. We compare our results with recent works.