Search

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., Comment: Accepted to Icarus

The High Resolution Imaging Science Experiment (HiRISE) on the Mars Reconnaissance Orbiter (MRO) routinely observes the surface of martian polar regions during seasonal observational campaigns. The images reveal a variety of seasonal features related to CO2 jet eruptions. The amount and quality of the data collected over 8 martian years provides abundant information on seasonal and inter-annual development of these transient surface features – fans and blotches – considered to be dust and sand deposits from the CO2 jet eruptions. Directions and sizes of the fan deposits showcase information about eruption physics as well as about the state of the martian atmosphere at the time of eruption. To analyze the HiRISE dataset numerically we have created a citizen science project Planet Four (P4). Participants of P4 mark seasonal deposits in sub-frames of HiRISE images with provided tools that have specific geometry. Locations, directions, and sizes of the markings are stored in a database. We have created a pipeline that works with the database, removes mistaken markings, and reduces multiple markings of the same object to the statistically best marking per object. The result of the pipeline is a catalog of fan and blotch markings that provide direct information about wind directions and speeds in the locations observed. The fan directions shift considerably during spring which is indicative of the volatile nature of weather in the southern polar regions in spring as the polar vortex develops. We will present derived wind speeds and directions in multiple regions of interest (ROIs) in the southern polar regions. We have run the Mars Regional Atmospheric Modeling System (MRAMS) at the same locations and portions of spring to compare its results to values derived from P4. The MRAMS model is generally consistent with P4 wind directions and speeds derived from P4 fan sizes. MRAMS winds are most consistent with the observed directions in early spring, while after approximately Ls = 200° the model tends to be a bit less consistent with the P4 observations. This can be explained in part by the relative proximity of the sublimating edge of the seasonal polar cap at these times, which tends to increase the wind direction variability. MRAMS winds are calculated every 10 Mars-minutes for multiple complete martian days, providing an opportunity to determine the timing of jet eruptions. In some of the ROIs we observe a correlation of P4-determined winds with modeled winds at specific times of day. This indicates that CO2 jet eruptions happen during those specific times. The most probable time of jet eruptions determined from these correlations varies through spring. This is consistent with CO2 jet modeling that suggested that timing of jet eruptions is connected to the amount of available solar energy. In the early spring those CO2 jet models expect the eruptions to occur close to local noon or during the afternoon, while later in spring they are expected to shift to earlier hours of the day as the seasonal ice thins. Not all P4 ROIs exhibit this pattern when compared to the MRAMS simulations. Local wind patterns and the near-surface atmospheric environment (where eruptions happen) have to be examined in detail to understand and decouple the sporadic nature of CO2 jet eruptions from the dynamic atmospheric behavior modeled by MRAMS. Conclusions: HiRISE and consequently P4 data created several new ways to investigate seasonal processes and atmospheric dynamics on Mars with the potential for quantitative analysis.

Orbital data from the Colour and Stereo Surface Imaging System (CaSSIS) onboard the ExoMars Trace Gas Orbiter showed interesting images of the circumpolar areas during spring. The winter-formed dusty CO2 ice cap goes under self-cleaning processes producing a translucent slab. With grazing spring sunlight, it starts to sublimate at the base and from overpressure, cold jets erupt leaving erosion marks in the underlying substrate and dust/sand deposits at the surface. This model, proposed by Kieffer, is commonly accepted to explain dark spots and fans deposits as well as spiders.To test different aspects of this model, we combine observational data from CaSSIS with experimental work for which Martian temperature and pressure at high latitudes could be reached in a simulation chamber. Preliminary results on sinking analogous dust material (MGS-1) on a CO2 ice slab have been promising. We aim to quantify in better details the sinking ratio, colour variations and frost (H2O and/or CO2) depositions on CO2/MGS-1 samples under various setup conditions (illumination, material distribution). Using a hyperspectral device, we can measure the reflectance and simulate the CaSSIS signal in the different filters (PAN, NIR, RED, BLU) to compare to actual images that have been acquired during southern spring.

The southern polar area of Mars is home to various seasonal activity commonly explained by the Kieffer model. During southern spring, the ice covering the polar area sublimates and leaves distinct features (spiders, spots, fans) observable from orbit. The Colour and Stereo Surface Imaging System (CaSSIS) onboard the ExoMars Trace Gas Orbiter (TGO), provides high-resolution multi-filter images of the Martian surface offering high sensitivity to colour contrasts. Its stereo capability is pivotal for momentary processes and offers a unique perspective for studying surface sublimation processes and their relation to atmospheric features. For the first time, we identify clouds well correlated with surface features (araneiforms and spots at southern circum-polar latitudes) hence motivating a new campaign to refine these observations over time periods where CO2 sublimation processes occur. We focus here on the structure of spot deposits and their evolution through time. We identify and describe seven structures: dark spot, bright-haloed spot, ringed spot, inverted spot, dark-haloed spot, banded spot, and bright spot. By morphological and spectral analyses, we hypothesize a new chronology of events that characterise the origin, formation and evolution of these features.

In the first 20 orbits of the Juno spacecraft around Jupiter, we have identified a variety of wave‐like features in images made by its public‐outreach camera, JunoCam. Because of Juno's unprecedented and repeated proximity to Jupiter's cloud tops during its close approaches, JunoCam has detected more wave structures than any previous surveys. Most of the waves appear in long wave packets, oriented east‐west and populated by narrow wave crests. Spacing between crests were measured as small as ~30 km, shorter than any previously measured. Some waves are associated with atmospheric features, but others are not ostensibly associated with any visible cloud phenomena and thus may be generated by dynamical forcing below the visible cloud tops. Some waves also appear to be converging, and others appear to be overlapping, possibly at different atmospheric levels. Another type of wave has a series of fronts that appear to be radiating outward from the center of a cyclone. Most of these waves appear within 5° of latitude from the equator, but we have detected waves covering planetocentric latitudes between 20°S and 45°N. The great majority of the waves appear in regions associated with prograde motions of the mean zonal flow. Juno was unable to measure the velocity of wave features to diagnose the wave types due to its close and rapid flybys. However, both by our own upper limits on wave motions and by analogy with previous measurements, we expect that the waves JunoCam detected near the equator are inertia‐gravity waves.