Science and Exploration

Planet Four: A Neural Network’s Search For Polar Spring-time Fans On Mars

By Keith Cowing
Press Release
October 19, 2022
Filed under
Planet Four: A Neural Network’s Search For Polar Spring-time Fans On Mars
Planet Four pipeline for image tile APF0000q3k, using a density-based clustering pipeline as described in Aye et al. (2019). The min_samples parameter indicates how many markings need to fall within a set of given pixel distances, and it was found that a value of 3 suppresses false positives efficiently, which was the goal of that pipeline. The parameter n_(blotch|fan) classif indicates how many review submissions actually contained actual markings. In this case, 15 submissions out of the standard 30 required reviews contained markings, indicating that the complexity of the image tile has made 50% of the volunteers skip trying Upper left: The HiRISE input tile to be marked by volunteers; upper middle: the fan markings of 30 volunteers; upper right: the blotch markings of 30 volunteers; lower right: resulting blotches after applying density-based clustering and averaging; lower middle: resulting fans after applying density-based clustering and averaging; lower left: the markings entering the catalog after a location-base >50%-voting between coinciding fan and blotch markings. – astro-ph.EP

Dark deposits visible from orbit appear in the Martian south polar region during the springtime.

These are thought to form from explosive jets of carbon dioxide gas breaking through the thawing seasonal ice cap, carrying dust and dirt which is then deposited onto the ice as dark ‘blotches’, or blown by the surface winds into streaks or ‘fans’.

We investigate machine learning (ML) methods for automatically identifying these seasonal features in High Resolution Imaging Science Experiment (HiRISE) satellite imagery. We designed deep Convolutional Neural Networks (CNNs) that were trained and tested using the catalog generated by Planet Four, an online citizen science project mapping the south polar seasonal deposits.

We validated the CNNs by comparing their results with those of ISODATA (Iterative Self-Organizing Data Analysis Technique) clustering and as expected, the CNNs were significantly better at predicting the results found by Planet Four, in both the area of predicted seasonal deposits and in delineating their boundaries. We found neither the CNNs or ISODATA were suited to predicting the source point and directions of seasonal fans, which is a strength of the citizen science approach.

The CNNs showed good agreement with Planet Four in cross-validation metrics and detected some seasonal deposits in the HiRISE images missed in the Planet Four catalog; the total area of seasonal deposits predicted by the CNNs was 27% larger than that of the Planet Four catalog, but this aspect varied considerably on a per-image basis.

Mark D. McDonnell, Eriita Jones, Megan E. Schwamb, K-Michael Aye, Ganna Portyankina, Candice J. Hansen

Comments: Accepted to Icarus
Subjects: Earth and Planetary Astrophysics (astro-ph.EP)
Cite as: arXiv:2210.09152 [astro-ph.EP] (or arXiv:2210.09152v1 [astro-ph.EP] for this version)
Focus to learn more
Submission history
From: Megan Schwamb
[v1] Mon, 17 Oct 2022 14:57:45 UTC (20,797 KB)

SpaceRef co-founder, Explorers Club Fellow, ex-NASA, Away Teams, Journalist, Space & Astrobiology, Lapsed climber.