pyFRESCO, a Python open source tool to democratize CRISM spectral data management, analysis and mapping

Remotely sensed hyperspectral data provide essential information on the composition of rocks on planetary surfaces. On Mars, these data types are provided by the CRISM instrument (Compact Reconnaissance Imaging Spectrometer for Mars) [1], a hyperspectral camera that operated onboard the MRO (Mars Reconnaissance Orbiter) probe that collected more than 10 Tb of data. CRISM covers a spectral range going from ca. 400 to ca. 4000 nm, with a spectral resolution of 6.55 nm/channel and a spatial resolution of up to 18.4 m/px. The most advanced CRISM data products are the MTRDRs (Map-Projected Target Reduced Data Records) [2]. These data are re-projected onto the Martian surface and are cleaned from the so-called ”bad bands” (noisy stripes).

The two main CRISM MTRDR subproducts are the hyperspectral datacube and the spectral parameter datacube. The first contains the detected reflectance spectra, while the second is composed of 60 different spectral parameters, as defined in [3], usually to produce RGB maps emphasizing specific minerals within the scene. Usually, the creation of the RGB maps and the spectral analysis are conducted using commercial software like ENVI^©, which are usually not specifically meant for these dataset, inherently implies some costs for the user,  and does not always meet some of the fundamentals principles of Open Science [4].

Here we present a free open-source tool completely written in Python called pyFRESCO (Flexible RGB Extraction and Spectra Comparison for Observations), which allows the use of CRISM MTRDR data. The software can create RGB maps and select Regions Of Interest (ROI) from which the user can extract and further analyze the spectral data available from the chosen CRISM scene.

The possible types of analysis that pyFRESCO supports range from simple descriptive statistics (to obtain, for example, mean spectra of a ROI), and spectra-related pre-processing methods like continuum removal and/or smoothing, to analysis through more complex techniques. One of those techniques being a direct spectral comparison tool that, given a guess taken from the minerals that are present in the MICA files [5], infer the nearest absorption features to the tabulated ones. Another technique, applicable only to mafic minerals, is the calculation of band parameters given in [6] for mafic mineral discrimination (e.g. between pyroxenes and silicate glasses). The last one involves a machine learning driven Modified Gaussian Model [7] for spectral unmixing with either skewed or non-skewed normal distributions.

Moreover, with pyFRESCO it is possible to georeference the RGB maps to export them to ArcGIS^©/QGIS software.

In support of the effectiveness of pyFRESCO, we present a case study conducted on the CRISM target FRT00009B5A that covers the northern portion of Kai crater, located in Meridiani Planum. The choice of this particular case study is motivated by the morphological and compositional complexity shown by this specific target [8], demonstrating that pyFRESCO can be useful especially in complex scenarios that require a detailed analysis of the geological context, evolution, and processes.

 

Bibliography:

[1] Murchie, S. 2007. doi:10.1029/2006JE002682.

[2] Seelos, F.P. 2016. URL: https://api.semanticscholar.org/CorpusID:217998642

[3] Viviano-Beck, C.E. 2014. doi:10.1002/2014JE004627.

[4] Barker, M. 2022. doi:10.1038/s41597-022-01710-x.

[5] Viviano-Beck, C. E., 2015. URL: https://crismtypespectra.rsl.wustl.edu/.

[6] Horgan, B.H. 2014. doi:https://doi.org/10.1016/j.icarus.2014.02.031.

[7] Sunshine, J.M. 1990. URL: https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/JB095iB05p06955

[8] Baschetti, B. 2025. doi: https://doi.org/10.1029/2024JE008564