Semantic segmentation of light-toned veins in multimodal ChemCam data

Ana Lomashvili; Kristin Rammelkamp; Protim Bhattacharjee; Olivier Gasnault; Elise Clavé; Christoph H. Egerland; Susanne Schröder; Travis S.J. Gabriel; Ari Essunfeld; Stéphane Le Mouélic; Begüm Demir

doi:10.1038/s41598-026-47207-0

USGS · 70275071

Semantic segmentation of light-toned veins in multimodal ChemCam data

Ana Lomashvili·Kristin Rammelkamp·Protim Bhattacharjee·Olivier Gasnault·Elise Clavé·Christoph H. Egerland·Susanne Schröder·Travis S.J. Gabriel·Ari Essunfeld·Stéphane Le Mouélic·Begüm Demir

View original source ↗Export metadata ↓

Abstract

Since the Mars Science Laboratory landed in 2012, the ChemCam instrument aboard the rover has collected in-situ laser-induced breakdown spectroscopy (LIBS) data and context images along more than 35 km of the Gale Crater traverse, providing valuable observations including diagenetic features such as light-toned veins. These veins are of particular scientific interest because they are interpreted as indicators of past fluid circulation on Mars and provide insights into the evolution of habitability on Mars. Their identification, however, currently relies on manual visual inspection of Remote Micro Imager (RMI) images, a process that is time-consuming and sensitive to differences in human interpretation. To address this issue, in this paper we introduce a novel pixel-level labeled, multimodal dataset of ChemCam observations specifically tailored for vein detection, along with customized U-Net models to integrate both textural (RMI) and chemical (LIBS) modalities. To further ensure trustworthy scientific use, we incorporate the Learn-Then-Test (LTT) framework to provide statistical control of the false discovery rate without requiring model retraining. The experimental results demonstrate that the proposed customized U-Net models trained on the developed dataset, combined with risk-controlled prediction, increases the efficiency of pixel-level vein identification through automation and produces statistically reliable predictions for multimodal ChemCam data.

Explore related subjects

Scientific Reports Article

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Ana Lomashvili, Kristin Rammelkamp, Protim Bhattacharjee, Olivier Gasnault, Elise Clavé, Christoph H. Egerland, Susanne Schröder, Travis S.J. Gabriel, Ari Essunfeld, Stéphane Le Mouélic, Begüm Demir. 2026-04-09. Semantic segmentation of light-toned veins in multimodal ChemCam data. https://doi.org/10.1038/s41598-026-47207-0

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related USGS reports

USGS2026

Spontaneous combustion of metal sulfide minerals and implications for seafloor massive sulfide mining

Metal sulfide minerals are abundant phases in many environments on earth, including seafloor massive sulfide (SMS) deposits, which occur at hydrothermal vents throughout the global ocean. These deep-sea settings are currently undergoing exploration for potential future mining. Here, we describe rocks from Escanaba Trough on the southern Gorda Ridge that underwent spontaneous combustion during laboratory processing. We compared the mineralogy, composition, and thermogravimetric behavior of the combusting samples to rocks from the same location that did not combust. We then used thermodynamic analyses to evaluate the exothermic nature of different metal sulfide oxidation reactions. The combusting rocks are primarily composed of nanocrystalline marcasite pseudomorphing blades of primary pyrrhotite. These altered, nanocrystalline rocks are less stable than their primary predecessors. Typically, recrystallization decreases reactivity as minerals trend towards steady state; however, both of the combusting samples we characterized are secondary, and demonstrate the instability of this metastable intermediate phase. Further, thermodynamic calculations indicate that metal sulfide oxidation is more exothermic when oxygen serves as the oxidant, as occurs in seawater or air. The distinct mineralogy, texture, and setting of SMS deposits in the modern oceans thus suggests a greater potential for self-heating in comparison to their ancient analogues.

Scientific Reports↗

USGS2026

Rewilding the upper Klamath River Basin: Rapid recolonization of Chinook salmon following the world’s largest dam removal

Habitat fragmentation is a global driver of decline in biodiversity, yet evaluations of ecological response following re-connection remain rare. In 2024, four hydroelectric dams on the Klamath River were removed, reconnecting over 640 km of anadromous habitat and marking the largest dam removal project completed to date. We used a stationary high-resolution multi-beam imaging SONAR to quantify adult Chinook salmon ( Oncorhynchus tshawytscha ) passage into newly reopened reaches during the first two migratory seasons following dam removal. An estimated 7,742 (95% CI: 7,702–7,778) Chinook salmon migrated upstream in 2024 and 13,310 (95% CI: 12,876–13,733) in 2025, representing 18–19% of fish returning to the Klamath River Basin. Within this period, Chinook salmon recolonized 88% of their documented historical distribution, expanding into 345 km of reconnected habitats and reaching elevations of up to 1,250 m. Recolonization required fish to swim through dam removal sites, complex hydraulic corridors and engineered fishways, demonstrating the propensity for rapid recolonization at basin scale despite these challenges. These findings provide the first quantitative assessment of salmon response to the world’s largest dam removal and demonstrate how reconnection can catalyze large-scale recolonization in historically fragmented river networks.

California, Oregon↗

USGS2026

Rapid intensification of Karenia brevis red tides after hurricane Milton revealed by PACE-OCI hyperspectral unmixing

Estuarine and coastal systems, shaped by transient physical–biological interactions and recurrent harmful algal blooms, often undergo episodic shifts in phytoplankton community composition (PCC). We present PhytoUnmix , a deep learning spectral unmixing framework applied to NASA’s hyperspectral PACE (Plankton, Aerosol, Cloud, Ocean Ecosystem) mission, to resolve PCC dynamics during a red tide bloom following Hurricane Milton (2024) on the West Florida Shelf, USA. PACE-derived PCC, together with upper-ocean physical observations, revealed contrasting phytoplankton community patterns across distinct post-hurricane hydrographic conditions: offshore wind-curl induced upwelling exhibited higher relative abundance of diatoms, whereas nearshore regions affected by northerly winds and Ekman transport along with enhanced upwelling were associated with coastal Karenia brevis bloom. For the first time, synergistic observations using PACE and Surface Water and Ocean Topography (SWOT) showed a rapid rise in dinoflagellates nearshore, with concurrent declines in diatoms and chlorophytes, while cryptophytes co-increased. These transient shifts likely reflected not only physical stirring but also biological responses, as freshwater inputs increased nutrient availability, and coincided with intensified red tides and spatially heterogeneous phytoplankton responses, such as, initial increases in diatoms and cyanobacteria followed by Karenia brevis dominance. Together, PACE and SWOT provide unprecedented capacity to resolve physical–biological linkages in coastal ecosystems.

Florida↗