pyFRESCO: An open-source Python framework for hyperspectral data-driven discovery in planetary geology

Hyperspectral data are essential for investigating the composition of planetary surfaces and have been a pivotal tool to understand the geological history of many planetary bodies, such as the Moon, Mars, Ceres and many others. In order to analyze this type of data, proprietary software are commonly employed, but often presents some scientific, economic and ethical complications. We here introduce pyFRESCO (Flexible RGB Extraction and Spectra Comparison for Observations), an open-source Python library designed to process and model hyperspectral datacubes from the CRISM instrument (Mars Reconnaissance Orbiter). Even if other open-source softwares exists for hyperspectral data analysis (e.g. JCAT, PyHAT, Hyperspy), PyFRESCO is tailored on planetary hyperspectral datacubes, offering a wide range of capabilities, from basic statistical analysis and spectral preprocessing, to advanced analyses for entire hyperspectral datacubes or specific Regions of Interest (ROIs). For instance, the library includes automatic absorption line detection and the development of spectral unmixing by Modified Gaussian Models and georeferenced RGB map generation by interaction with a Graphical User Interface (GUI). The library also facilitates the comparison of observed spectra with laboratory or reference datasets and incorporates rigorous error propagation via Taylor expansion and bootstrapping. To demonstrate pyFRESCO's capabilities, we report the case study based on the CRISM target FRT00009B5A, which covers the northern portion of Kai crater in Meridiani Planum (Mars). Our results showcase the library’s effectiveness in supporting detailed geological interpretations in complex planetary environments.