Validation of open-source science tools and background model construction in γ-ray astronomy

In classical analyses of $\gamma$-ray data from IACTs, such as H.E.S.S., aperture photometry, or photon counting, is applied in a (typically circular) region of interest (RoI) encompassing the source. A key element in the analysis is to estimate the amount of background in the RoI due to residual cosmic ray-induced air showers in the data. Various standard background estimation techniques have been developed in the last decades, most of them rely on a measurement of the background from source-free regions within the observed field of view. However, in particular in the Galactic plane, source analysis and background estimation are hampered by the large number of, sometimes overlapping, $\gamma$-ray sources and large-scale diffuse $\gamma$-ray emission. For complicated fields of view, a three-dimensional (3D) likelihood analysis shows the potential to be superior to classical analysis. In this analysis technique, a spectromorphological model, consisting of one or multiple source components and a background component, is fitted to the data, resulting in a complete spectral and spatial description of the field of view. For the application to IACT data, the major challenge of such an approach is the construction of a robust background model. In this work, we apply the 3D likelihood analysis to various test data recently made public by H.E.S.S., using the open analysis frameworks ctools and Gammapy. First, we show that, when using these tools in a classical analysis approach and comparing to the proprietary H.E.S.S. analysis framework, virtually identical high-level analysis results are obtained. We then describe the construction of a generic background model from data of H.E.S.S. observations, and demonstrate that a 3D likelihood analysis using this background model yields high-level analysis results that are highly compatible with those obtained from the classical analyses. (abridged)