Spectral bands detector tool for future MIRS/MMX data

Martian Moon eXploration (MMX) mission will launch in October 2026 to investigate the Martian system, notably the moons, Phobos and Deimos, in order to uncover their origins. Among the instrumentation, the MMX InfraRed Spectrometer (MIRS) is a spectro-imager which covers the 0.9 - 3.6 μm range. It will permit to detect several minerals and compounds including hydrated minerals between 2.6 and 2.8 m and organics in the 3.3 - 3.5 μm region, with high Signal to Noise Ratio (SNR). MIRS will play a key role in selecting the landing site, as well as improving our understanding of the composition of Phobos and Deimos. It is expected to measure thousands of spectra. To facilitate the processing of these data, we adapted and improved an existing code called Spindex, which was originally developed in IRL for the OVIRS instrument on the OSIRIS-REX mission [1].

The original code used a catalogue comprising over 100 potential spectral bands, each of which was associated with  spectral parameters that are used in a set of 21 different equations (depending on the band form) to compute the spectral band depth (see Fig. 1). If this depth was above 5%, then the spectral band was considered to be detected. The code also computes the spectral slope over seven different regions of interest.

          Figure 1: Example of the spectral parameterls, band depth and error associated equations used by Spinmirs. 

In our version, called Spinmirs, we have slightly reduced the number of potential bands, grouping the different types of olivine bands in the 0.9 - 1.3 μm region together, as this lead to several false detections. This also reduced the number of equations to 19. Furthermore, we have introduced an additional quality criterion. Each band depth is associated with an error, if the band depth is less than twice the error, the band is considered ‘potentially detected’ rather than ‘detected’, as this indicate a low SNR in the area where it is locate. The final change we made was adjusting the threshold of 5% to each mineral if it were over- or under-detected.

          Figure 2: Example of Spinmirs output on a terrestrial sample available on RELAB. 

Spinmirs has been tested on the RELAB dataset, as well as Bennu spectra taken by OVIRS and gives promising results (see Fig. 2), as most of the band detected are indeed appearing on the spectra. There are a few case of false detection, notably with the biotite at 1.30 μm. The average computation time for one spectra is of 0.31 seconds. It will be a valuable tool to treat and interpret future MIRS spectra.

Acknowledgement : This work has received support from France 2030 through the project named Académie Spatiale d'Île-de-France (https://academiespatiale.fr/) managed by the National Research Agency under bearing the reference ANR-23-CMAS-0041, as well as the Centre National d’Etude Spatial (CNES).

References :

[1] Kaplan et al. (2020), M&PS, 55, 4, 744-765