Chloride-bearing deposits are mineralogical markers of Mars’ dynamic water past. Because they are highly soluble and therefore record the time the surface was last covered with water. Chloride-bearing landscapes have become prime targets for astrobiological exploration because sediments, and especially their depositional environments, can provide optimal conditions for biological activity and preservation. First discovered in 2, chlorides occur as decameter to kilometer-sized deposits, often penetrated by polygonal fractures and ridges, 3,4,5 but local They tend to be located in topographic depressions1,6 and are sometimes associated with channel-like features or valleys. Network 4, 7.
So far, chloride has been identified primarily in the southern highlands of Mars2,4,5,6,8. This has strongly influenced hypotheses regarding their formation and evolution. For example, we suggest that the distribution of chlorides in the south roughly follows the expected maximum precipitation zone for modern Mars, indicating that the chlorides are predominantly Hesperian in age. Regardless of age, most authors agree that chloride is formed as a result of evaporation/precipitation of pond water from near-surface runoff and/or groundwater upwelling3,4,7,10 . Other alternative formation scenarios include diagenetic and/or hydrothermal brines, efflorescent crusts, and deep lakes11,12.
Chlorides are bright-toned and typically have a characteristic pink-to-purple color in color thermal imaging (pseudocolor synthesis using a combination of near-infrared, ~900 nm, red, ~670 nm, and blue channels, ~500 nm). It has a hue. This makes it easily recognizable from orbit (Figure 1). In the absence of color, chloride is easily confused with other light-toned substances such as clays6,8. In short, unambiguous detection and classification of chlorides requires (1) multispectral (color infrared) information; At the same time, (2) high to moderate spatial resolution (<~10 meters per pixel) is required to recognize smaller deposits10. Global studies additionally require (3) images that cover a wide range of the Martian surface; However, currently no image dataset exists that meets all three criteria.
Figure 1
Chloride deposits on Mars imaged by various orbital sensors. Top left; pseudocolor image of Mars Odyssey THEMIS 875 DCS (Decorrelated Enhancement). East Terra sirenum chloride deposit characterized by feature 2 (light blue). Located near the CRISM Terra Sirenum chloride-type locality (31.6°S, 206.4°E), feature 36; TGO (Trace Gas Orbiter) CaSSIS and MRO (Mars Reconnaissance Orbiter) HiRISE image footprints are outlined in white. I am. Top right, bottom right. CaSSIS NPB (near-infrared, panchromatic, blue-green) and HiRISE MIRB (infrared, red, blue-green) acquisitions of the same sediment section (note the limited width of the HiRISE IRB color strip). lower left. DS corrected (dark object subtracted 37,38 spectrum for chloride (C) and one background position (B). See markers in CaSSIS and HiRISE images. Clear similarities between DS corrected CaSSIS chloride spectra and USGS halite Note the reference spectrum, specifically the positive RED-NIR gradient (39, resampled to CaSSIS wavelengths). In this example, CaSSIS color infrared data is used to potentially detect chloride as well as HiRISE MIRB. We show that it is possible to identify the containing terrain, but the remaining ambiguity is acknowledged. Credit: ESA/TGO/CaSSIS CC-BY-SA 3.0 IGO, NASA/JPL/University of Arizona/Arizona State University.
Previous global mapping efforts have relied on THEMIS (Thermal Emission Imaging System), CRISM (Mars Compact Reconnaissance Imaging Spectrometer), OMEGA (Mineralogy Observatory), and/or Ta. HiRISE (High Resolution Imaging Science Experiment) data2,4,6. THEMIS and OMEGA have global coverage but low spatial resolution (THEMIS > ~100 m/pixel, OMEGA > ~300 to ~5,000 m/pixel), while CRISM and HiRISE have moderate to high spatial resolution. (CRISM > ~18 to ~200 m/pixel). m/pixel, HiRISE ~0.3 m/pixel) but lack spatial coverage (less than 3% and only 0.7% of the surface covered by multispectral data, respectively – HiRISE has a central 1.2 km wide Note that only one color strip is provided (for the image) (Example 13).
In addition to data limitations, previous studies have relied on manual mapping tasks (e.g. 2, 4, 6), which can be affected by, for example, operator awareness fatigue, differences in operator expertise, and operator may be affected by various observational biases related to, for example, differences in expertise between researchers. Observer expectation effects come into play and can lead to incomplete or inconsistent mapping results (Figure 2). In particular, 14 built a machine learning-based classification routine for CRISM data focused on spectral variations. However, this routine suffers from the same instrument-related limitations mentioned above (lack of spatial resolution and coverage), while being unable to extract the morphological information that is also encoded in CRISM image data. As a result of limitations in data and mapping techniques, previous efforts may have missed relatively large and small-scale chloride-bearing deposits that are important for conclusions drawn about Mars’ past climate and evolution. may have a negative impact.
Figure 2
Examples of chloride-bearing sediments identified by CaSSIS that were missed in previous surveys. Left; THEMIS 964, 875, and 642 DCS (Decorrelated Enhancement) pseudocolor images of chloride deposits are not reported in the database. The CaSSIS image footprint is outlined in white. In THEMIS DCS products, chloride-bearing terrain appears blue-green in 964, blue in 875, and yellow/orange in 642. Right; CaSSIS NPB (near-infrared, panchromatic, blue-green) acquisition of the same sediment. This example shows that CaSSIS color data can be used to detect potentially chloride-bearing terrain, recognizing the remaining ambiguity (in this particular case, THEMIS DCS data does not detect the presence of chloride). (strongly suggested). Image credit: ESA/TGO/CaSSIS CC-BY-SA 3.0 IGO, NASA/JPL/Arizona State University.
This study addresses some of the limitations associated with data and mapping methods by (1) utilizing new color infrared (BLU – green/blue, PAN – panchromatic, RED – far red, NIR – near infrared); I’ll deal with it. , see Figure 1) and high resolution (~4 m /pixel) Image Data (2) CaSSIS Employs a reliable and rapid few-shot learning-driven detection and mapping workflow that can effectively derive from the spectral and morphological information encoded in images By that. To date, CaSSIS data have not been used to systematically map chloride-bearing deposits on Mars. It is important to note that CaSSIS cannot meet all three major mapping requirements. The data are characterized by high spatial resolution (~4 m/pixel), but so far only cover ~7.1% of the surface. Additionally, its four spectral bands (BLU, PAN, RED, NIR) allow for the detection of potentially chloride-containing terrain, but only in sufficient amounts to allow unambiguous identification of chloride. Spectral resolution and range are not provided. Additionally, THEMIS and OMEGA do not have high enough spatial resolution to characterize most of the potentially chloride-bearing small sediments identified in CaSSIS data, making thorough validation of CaSSIS detections impossible. Throughout this manuscript, we provide qualitative evidence of good agreement between large-scale chloride deposits detected by CaSSIS and chloride deposits detected by THEMIS (Figs. 1, 2). However, due to the remaining ambiguity, we refer to all CaSSIS detections as “chloride deposit candidates.” or “potential chloride-bearing deposits,” pending further verification.
Our dataset includes a total of 965 potentially chloride-bearing deposits on the surface of Mars, identified in a total of 487 CaSSIS images (Figure 3). Of note, approximately 14% (n = 136) of potential chloride deposits are located in highlands north of the equator (up to 35°N latitude), which, if verified by future missions and experiments, could This is the first detection in the Northern Hemisphere. effort. We note that the chloride candidates in the north appear to be less distinct and spatially discontinuous than in the south, which may indicate more degradation (weathering and erosion) in these regions. Masu. The size of identified candidate chloride deposit features ranges from less than 300 meters to more than 3000 meters in diameter, with an average of 1043 meters. Approximately 60% of all candidate sites are located in Noachian terrain and approximately 38% are located in Hesperian terrain. Only 2% are located in Amazonian landscapes (Figure 3). We acknowledge that candidates on the Amazon landscape appear to be larger on average than on other landscapes, but our sample size is small. The majority of identified chloride candidates are located within meandering channels, valleys, and along the edges of local topographic lowlands, but also on surfaces above local topographic highlands, i.e. slopes and valleys. , and a number of chloride candidates have also been identified on the topography above the valley. This suggests a deposit formed before weathering and erosion formed the valley, as also found in previous studies (Example 6). Note that our dataset includes potential chloride deposits that were missed in previous efforts, such as Maurus Valley, Vichada Valley, Nirgal Valley, and Syrtis Major (Fig. 2 , 3).
Figure 3
Distribution of potentially chloride-bearing deposits on Mars identified by CaSSIS. Top: Global distribution of all CaSSIS-derived chloride candidates (purple shades). The color indicates the estimated size of each candidate. White/black crosses indicate locations of THEMIS/OMEGA-derived chloride deposits/hydrous mineral detections4,35. Vikings integrated color mosaics in the background. Add a note mark at the position (black circle) in Figure 1. Bottom: CaSSIS image density coverage heatmap (hexmap) with orbit number 27,816, February 17, 2024.
Our dataset provides new global insights into the distribution and properties of chloride on Mars, expands existing chloride deposit datasets, and explores the presence of near-surface water in Mars’ distant past. enable future modeling efforts towards a better understanding of the – Past habitability of Mars. Additionally, this dataset can be used to directly inform future high-resolution and multi/hyperspectral imaging campaigns as well as concept studies for future lander missions.
