Riccardo La Grassa
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Riccardo La Grassa has a total of 56 co-authors. Here are the top co-authors by number of shared publications:
- Cristina Re
INAF-Astronomical Observatory of Padova, Padua, Italy
21 shared publications - Ignazio Gallo
Department of Theoretical and Applied Science, University of Insubria, 21100 Varese, Italy
20 shared publications - Adriano Tullo
INAF-Astronomical Observatory of Padova, Padua, Italy
18 shared publications - Nicola Landro
Department of Theoretical and Applied Sciences—DISTA, University of Insubria, Via J.H. Dunant, 3, 21100 Varese, Italy
16 shared publications Gabriele Cremonese
INAF, Italian National Institute for Astrophysics, Italy
15 shared publications- Gabrielle Cremonese
Osservatorio Astronomico di Padova
8 shared publications
Latest Publications
- Open Access
- 21 October 2026
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Planets
- Vol. 131 (10)
Distal Rays and Secondary Craters of the Tycho Impact Event Revealed Through Deep Learning
The nature of secondary impact craters is a matter of ongoing debate, with important implications for understanding the impact cratering process, and for calibration of crater chronology. Rays generated by primary impact craters are characterized by secondary craters, providing identifiable terrains to study the size, morphology and clustering properties of secondary impact processes. The prominent ray system of the young Tycho crater has been extensively studied on the Moon since the Apollo era. In this work, we study the entire crater population of a quadrangle that extends to the northeast of Tycho for ∼3,000 km. We relied on a bespoke application of YOLOLens, a state‐of‐the‐art deep learning model specifically optimized for high‐resolution crater detection. We generated a data set of ∼7.1 × 10 6 craters more than 50 m in diameter when integrated with an existing crater database for completeness. We then used a spatial cluster analysis method to identify clustered (potential secondary) craters. Our analysis identifies rays and clusters associated with Tycho in agreement with previous maps. We identified several distal elongated clusters that belong to the Tycho ray system that have not been previously identified. Also, we interpret additional non‐radial clusters as plausible products of projectiles from Tycho ejected at high angles (∼74°–75°). Our results show that secondary craters <200 m dominate (∼71%) the crater population within the rays and clusters of Tycho. This work shows that the products from cluster analysis applied to large lunar crater data sets hold insightful information about the size and distribution of potential secondary craters.
- 21 October 2026
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Planets
- Vol. 131 (10)
Open AccessDistal Rays and Secondary Craters of the Tycho Impact Event Revealed Through Deep Learning
The nature of secondary impact craters is a matter of ongoing debate, with important implications for understanding the impact cratering process, and for calibration of crater chronology. Rays generated by primary impact craters are characterized by secondary craters, providing identifiable terrains to study the size, morphology and clustering properties of secondary impact processes. The prominent ray system of the young Tycho crater has been extensively studied on the Moon since the Apollo era. In this work, we study the entire crater population of a quadrangle that extends to the northeast of Tycho for ∼3,000 km. We relied on a bespoke application of YOLOLens, a state‐of‐the‐art deep learning model specifically optimized for high‐resolution crater detection. We generated a data set of ∼7.1 × 10 6 craters more than 50 m in diameter when integrated with an existing crater database for completeness. We then used a spatial cluster analysis method to identify clustered (potential secondary) craters. Our analysis identifies rays and clusters associated with Tycho in agreement with previous maps. We identified several distal elongated clusters that belong to the Tycho ray system that have not been previously identified. Also, we interpret additional non‐radial clusters as plausible products of projectiles from Tycho ejected at high angles (∼74°–75°). Our results show that secondary craters <200 m dominate (∼71%) the crater population within the rays and clusters of Tycho. This work shows that the products from cluster analysis applied to large lunar crater data sets hold insightful information about the size and distribution of potential secondary craters.
- Open Access
- 30 July 2026
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLIX-B3-20, 1239-1248
YOLOLens2.0: A Unified Super-Resolution and Detection Framework for High-Fidelity Crater Mapping in Lunar Permanently Shadowed Regions
Accurate crater mapping in lunar permanently shadowed regions (PSRs) is hindered by extreme low-light and low-resolution imagery. We present YOLOLens2.0, a unified, end-to-end deep learning framework designed for high-fidelity crater detection and terrain reconstruction in these challenging environments. The architecture integrates a Dense-Residual-Connected Transformer (DRCT) for multimodal super-resolution (SR) with a YOLO-derived detection module and an affine calibrator to ensure geometric consistency at meter scale. Our framework exploits a bidirectional synergy where SR enhances feature discriminability for detection, while detection-driven supervision refines structural reconstruction. Validation on Kaguya data demonstrates a significant performance leap, achieving a Recall of 89.20% and an mAP@50 of 0.844 an improvement of over 33 percentage points in recall compared to the original YOLOLens. Out-of-distribution validation on ShadowCam imagery, performed without fine-tuning, confirms the model’s robustness and scalability. The framework successfully preserves quantitative elevation fidelity and supports detailed morphometric analyses, including the extraction of the crater size-frequency distributions (SFDs) that align with theoretical lunar production functions. YOLOLens2.0 provides a scalable, high-precision methodology for planetary mapping, offering critical insights for lunar surface evolution studies and future exploration missions.
- 30 July 2026
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLIX-B3-20, 1239-1248
Open AccessYOLOLens2.0: A Unified Super-Resolution and Detection Framework for High-Fidelity Crater Mapping in Lunar Permanently Shadowed Regions
Accurate crater mapping in lunar permanently shadowed regions (PSRs) is hindered by extreme low-light and low-resolution imagery. We present YOLOLens2.0, a unified, end-to-end deep learning framework designed for high-fidelity crater detection and terrain reconstruction in these challenging environments. The architecture integrates a Dense-Residual-Connected Transformer (DRCT) for multimodal super-resolution (SR) with a YOLO-derived detection module and an affine calibrator to ensure geometric consistency at meter scale. Our framework exploits a bidirectional synergy where SR enhances feature discriminability for detection, while detection-driven supervision refines structural reconstruction. Validation on Kaguya data demonstrates a significant performance leap, achieving a Recall of 89.20% and an mAP@50 of 0.844 an improvement of over 33 percentage points in recall compared to the original YOLOLens. Out-of-distribution validation on ShadowCam imagery, performed without fine-tuning, confirms the model’s robustness and scalability. The framework successfully preserves quantitative elevation fidelity and supports detailed morphometric analyses, including the extraction of the crater size-frequency distributions (SFDs) that align with theoretical lunar production functions. YOLOLens2.0 provides a scalable, high-precision methodology for planetary mapping, offering critical insights for lunar surface evolution studies and future exploration missions.
- Open Access
- 9 July 2026
- Published by Springer Nature in Earth, Planets and Space
- Vol. 78 (1) , 136
Enhanced global MESSENGER multispectral mosaics using Gram–Schmidt adaptive (GSA) pansharpening
- 9 July 2026
- Published by Springer Nature in Earth, Planets and Space
- Vol. 78 (1) , 136
Open AccessEnhanced global MESSENGER multispectral mosaics using Gram–Schmidt adaptive (GSA) pansharpening
- Open Access
- 9 February 2026
- Published by Elsevier in ISPRS Open Journal of Photogrammetry and Remote Sensing
- Vol. 19
Transformer-driven monocular high-resolution DTM generation on mars via multimodal integration of CaSSIS imagery and MOLA altimetry
- 9 February 2026
- Published by Elsevier in ISPRS Open Journal of Photogrammetry and Remote Sensing
- Vol. 19
Open AccessTransformer-driven monocular high-resolution DTM generation on mars via multimodal integration of CaSSIS imagery and MOLA altimetry
- Open Access
- 25 September 2025
- Published by MDPI AG in Remote Sensing
- Vol. 17 (19) , 3287
From the Moon to Mercury: Release of Global Crater Catalogs Using Multimodal Deep Learning for Crater Detection and Morphometric Analysis
This study has compiled the first impact-crater dataset for Mercury with diameters greater than 400 m by a multimodal deep-learning pipeline. We present an enhanced deep learning framework for large-scale planetary crater detection, extending the YOLOLens architecture through the integration of multimodal inputs: optical imagery, digital terrain models (DTMs), and hillshade derivatives. By incorporating morphometric data, the model achieves robust detection of impact craters that are often imperceptible in optical imagery alone, especially in regions affected by low contrast, degraded rims, or shadow-dominated illumination. The resulting catalogs LU6M371TGT for the Moon and ME6M300TGT for Mercury constitute the most comprehensive automated crater inventories to date, demonstrating the effectiveness of multimodal learning and cross-planet transfer. This work highlights the critical role of terrain information in planetary object detection and establishes a scalable, high-throughput pipeline for planetary surface analysis using modern deep learning tools. To validate the pipeline, we compare its predictions against the manually annotated catalogs for the Moon, Mercury, and several regional inventories, observing close agreement across the full diameter spectrum, revealing a high level of confidence in our approach. This work presents a spatial density analysis, comparing the spatial density maps of small and large craters highlighting the uneven distribution of crater sizes across Mercury. We explore the prevalence of kilometer-scale (1–5 km range) impact craters, demonstrating that these dominate the crater population in certain regions of Mercury’s surface.
- 25 September 2025
- Published by MDPI AG in Remote Sensing
- Vol. 17 (19) , 3287
Open AccessFrom the Moon to Mercury: Release of Global Crater Catalogs Using Multimodal Deep Learning for Crater Detection and Morphometric Analysis
This study has compiled the first impact-crater dataset for Mercury with diameters greater than 400 m by a multimodal deep-learning pipeline. We present an enhanced deep learning framework for large-scale planetary crater detection, extending the YOLOLens architecture through the integration of multimodal inputs: optical imagery, digital terrain models (DTMs), and hillshade derivatives. By incorporating morphometric data, the model achieves robust detection of impact craters that are often imperceptible in optical imagery alone, especially in regions affected by low contrast, degraded rims, or shadow-dominated illumination. The resulting catalogs LU6M371TGT for the Moon and ME6M300TGT for Mercury constitute the most comprehensive automated crater inventories to date, demonstrating the effectiveness of multimodal learning and cross-planet transfer. This work highlights the critical role of terrain information in planetary object detection and establishes a scalable, high-throughput pipeline for planetary surface analysis using modern deep learning tools. To validate the pipeline, we compare its predictions against the manually annotated catalogs for the Moon, Mercury, and several regional inventories, observing close agreement across the full diameter spectrum, revealing a high level of confidence in our approach. This work presents a spatial density analysis, comparing the spatial density maps of small and large craters highlighting the uneven distribution of crater sizes across Mercury. We explore the prevalence of kilometer-scale (1–5 km range) impact craters, demonstrating that these dominate the crater population in certain regions of Mercury’s surface.
- Open Access
- 1 August 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 1471-1476
The Generation of TGO-CaSSIS Stereo Products after six years of scientific mission: The Complete Stereo Chain of INAF
CaSSIS (Colour and Stereo Surface Imaging System) on board the ExoMars Trace Gas Orbiter, which has been capturing images of Mars since April 2018, is a multispectral stereo camera and its stereo and colour data are used for generating high-resolution, three-dimensional maps and orthorectified images of the Martian surface. The instrument has already taken more than 50.000 images and covering >7.6% of Mars' surface, producing a total of 23.8 Tbit of data, with a small percentage (4.8%) taken using colour filters. These data are processed with the 3DPD software, which creates DTMs and orthorectified images with a ground resolution of 13.5 m per pixel and vertical accuracy of around 8 meters. All the stereo products are available to the scientific community through the INAF-hosted repository. The 3DPD software pipeline includes various stages: pre-processing to remove optical distortions and mosaicking, a matching core and DTM interpolation routines. A recent improvement through the integration of Bundle Adjustment, using the jigsaw tool from the USGS ISIS suite, has enhanced the precision of the data, reducing errors in the final 3D products. Overall, the work on CaSSIS and 3DPD plays a significant role in advancing Mars exploration, producing high-quality topographic maps and spatial data critical for scientific research.
- 1 August 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 1471-1476
Open AccessThe Generation of TGO-CaSSIS Stereo Products after six years of scientific mission: The Complete Stereo Chain of INAF
CaSSIS (Colour and Stereo Surface Imaging System) on board the ExoMars Trace Gas Orbiter, which has been capturing images of Mars since April 2018, is a multispectral stereo camera and its stereo and colour data are used for generating high-resolution, three-dimensional maps and orthorectified images of the Martian surface. The instrument has already taken more than 50.000 images and covering >7.6% of Mars' surface, producing a total of 23.8 Tbit of data, with a small percentage (4.8%) taken using colour filters. These data are processed with the 3DPD software, which creates DTMs and orthorectified images with a ground resolution of 13.5 m per pixel and vertical accuracy of around 8 meters. All the stereo products are available to the scientific community through the INAF-hosted repository. The 3DPD software pipeline includes various stages: pre-processing to remove optical distortions and mosaicking, a matching core and DTM interpolation routines. A recent improvement through the integration of Bundle Adjustment, using the jigsaw tool from the USGS ISIS suite, has enhanced the precision of the data, reducing errors in the final 3D products. Overall, the work on CaSSIS and 3DPD plays a significant role in advancing Mars exploration, producing high-quality topographic maps and spatial data critical for scientific research.
- Open Access
- 9 July 2025
- Published by Copernicus GmbH
Investigating Low-Reflectance Material (LRM) on Mercury using an integrated spectro-morphologic clustering approach
IntroductionLow-Reflectance Material (LRM) represents a widespread surface unit on Mercury, typically occurring as basin ejecta and characterized by the lowest reflectance values among the planet’s surface materials (e.g., Peplowski et al., 2015; Klima et al., 2018). LRM deposits are predominantly of pre-Tolstojan age (>3.9 Ga; Denevi et al., 2018), and their compositional modeling suggests they may contain up to ~4 wt% graphite, a conclusion supported by both reflectance analyses and modeled gamma ray and neutron spectra (Peplowski et al., 2016; Klima et al., 2018). These findings support the hypothesis that LRM may be remnants of an early graphite-rich crust (Vander Kaaden & McCubbin, 2015; 2018), later excavated and redistributed across the surface by large impact events (Lark et al., 2023).Method and Analytical approachPrevious efforts to map and characterize LRM have largely relied on spectral techniques, using combinations of reflectance ratios and principal component analyses to delineate their distribution (e.g., Peplowski et al., 2015; Klima et al., 2018). While these approaches have yielded important constrains on LRM distributions and possible compositional information, they offer limited integration with the morphological and stratigraphical context of the surface.In contrast to terrestrial geology, where lithology, morphology, and in situ measurements are jointly assessed, in general, planetary investigations must rely exclusively on remote sensing data. Given these observational constraints, integrated methodologies that combine compositional (i.e., spectral) and morphological data are essential for inferring geologically meaningful information from remote observations.To address this need, we previously developed an unsupervised learning framework for generating exploratory classification maps of Mercury’s surface, integrating spectral and topographic inputs into multidimensional data cubes (Figure 1, see also Vergara Sassarini et al., 2025). As a case study, we selected the Hokusai quadrangle and used as a starting dataset the 665 m/px DTM (Becker et al., 2016) and the 8-color map (Zambon et al., 2022) to build 3 different cubes: morphological (DEM + TRI-ruggedness), spectral (430-1000nm + 750-1000nm + 430-560nm spectral slopes + 750nm band), and a morpho-spectral cube that contains both spectral and morphology data. These cubes were processed using a Gaussian Mixture Model based approach to derive surface classes that reflect both compositional and morphological characteristics of the surface. Figure 1. A. Workflow for the generation of exploratory classification maps based on unsupervised clustering and B. morpho-spectral predictions for H05 quadrangle (Vergara Sassarini et al., 2025).From the derived morpho-spectral map (Figure 1B), we identified several clusters that spatially correspond to known LRM units, as observed in enhanced color imagery (Figure 2A). Notably, the LRM-related clusters exhibit internal diversity (Figure 2B), suggesting potential compositional heterogeneities.Figure 2. Comparison between LRM (blue colors) in A. Mercury Dual Imaging System (MESSENGER) enhanced color 665 m/pixel global mosaic (Denevi et al., 2016) and B. Isolated clusters from Vergara Sassarini et al., 2025.To investigate these heterogeneities, this ongoing work will include a detailed analysis of the reflectance spectra across the derived LRM-related clusters to test for relatively compositional differences among them. We also aim to refine the input cubes used during fitting by including additional spectral ratio layers or PCA-derived features that may improve the clustering sensitivity to subtle spectral variations. As a future application, we intend to apply this integrated and tailored classification model to selected regions of interest on other regions of the planet different from Hokusai Quadrangle, with the goal of producing a global-scale distribution map of LRM, investigating their possible heterogeneity highlighted by this preliminary investigation. Such automated, multi-dimensional classification will be essential for identifying regions of interest for detailed investigation with the high-resolution data expected from SIMBIO-SYS (Cremonese et al., 2020) data on the BepiColombo's mission (Benkhoff et al., 2021). AcknowledgmentsThis study has been supported from the Italian Space Agency (ASI) under ASI-INAF agreement 2024-18-HH.0.ReferencesBecker, K. J., Robinson, M. S., Becker, T. L., Weller, L. A., Edmundson, K. L., Neumann, G. A., ... & Solomon, S. C. (2016, March). First global digital elevation model of Mercury. In 47th annual lunar and planetary science conference (No. 1903, p. 2959).Benkhoff, J., Murakami, G., Baumjohann, W., Besse, S., Bunce, E., Casale, M., … Zender, J. (2021). BepiColombo—Mission overview and science goals. Space Science Reviews, 217(8), 90.Cremonese, G., Capaccioni, F., Capria, M. T., Doressoundiram, A., Palumbo, P., Vincendon, M., … Turrini, D. (2020). SIMBIO-SYS: Scientific cameras and spectrometer for the BepiColombo mission. Space Science Reviews, 216(5), 1–78.Klima, R. L., Denevi, B. W., Ernst, C. M., Murchie, S. L., & Peplowski, P. N. (2018). Global distribution and spectral properties of low‐reflectance material on Mercury. Geophysical Research Letters, 45(7), 2945-2953.Lark, L. H., Head, J. W., & Huber, C. (2023). Evidence for a carbon-rich Mercury from the distribution of low-reflectance material (LRM) associated with large impact basins. Earth and Planetary Science Letters, 613, 118192.Peplowski, P. N., Lawrence, D. J., Evans, L. G., Klima, R. L., Blewett, D. T., Goldsten, J. O., ...Weider, S. Z. (2015). Constraints on the abundance of carbon in near-surface materials on Mercury: Results from the MESSENGER Gamma-Ray Spectrometer. Planetary and Space Science, 108, 98-107. Vander Kaaden, K. E., & McCubbin, F. M. (2015). Exotic crust formation on Mercury: Consequences of a shallow,...
- 9 July 2025
- Published by Copernicus GmbH
Open AccessInvestigating Low-Reflectance Material (LRM) on Mercury using an integrated spectro-morphologic clustering approach
IntroductionLow-Reflectance Material (LRM) represents a widespread surface unit on Mercury, typically occurring as basin ejecta and characterized by the lowest reflectance values among the planet’s surface materials (e.g., Peplowski et al., 2015; Klima et al., 2018). LRM deposits are predominantly of pre-Tolstojan age (>3.9 Ga; Denevi et al., 2018), and their compositional modeling suggests they may contain up to ~4 wt% graphite, a conclusion supported by both reflectance analyses and modeled gamma ray and neutron spectra (Peplowski et al., 2016; Klima et al., 2018). These findings support the hypothesis that LRM may be remnants of an early graphite-rich crust (Vander Kaaden & McCubbin, 2015; 2018), later excavated and redistributed across the surface by large impact events (Lark et al., 2023).Method and Analytical approachPrevious efforts to map and characterize LRM have largely relied on spectral techniques, using combinations of reflectance ratios and principal component analyses to delineate their distribution (e.g., Peplowski et al., 2015; Klima et al., 2018). While these approaches have yielded important constrains on LRM distributions and possible compositional information, they offer limited integration with the morphological and stratigraphical context of the surface.In contrast to terrestrial geology, where lithology, morphology, and in situ measurements are jointly assessed, in general, planetary investigations must rely exclusively on remote sensing data. Given these observational constraints, integrated methodologies that combine compositional (i.e., spectral) and morphological data are essential for inferring geologically meaningful information from remote observations.To address this need, we previously developed an unsupervised learning framework for generating exploratory classification maps of Mercury’s surface, integrating spectral and topographic inputs into multidimensional data cubes (Figure 1, see also Vergara Sassarini et al., 2025). As a case study, we selected the Hokusai quadrangle and used as a starting dataset the 665 m/px DTM (Becker et al., 2016) and the 8-color map (Zambon et al., 2022) to build 3 different cubes: morphological (DEM + TRI-ruggedness), spectral (430-1000nm + 750-1000nm + 430-560nm spectral slopes + 750nm band), and a morpho-spectral cube that contains both spectral and morphology data. These cubes were processed using a Gaussian Mixture Model based approach to derive surface classes that reflect both compositional and morphological characteristics of the surface. Figure 1. A. Workflow for the generation of exploratory classification maps based on unsupervised clustering and B. morpho-spectral predictions for H05 quadrangle (Vergara Sassarini et al., 2025).From the derived morpho-spectral map (Figure 1B), we identified several clusters that spatially correspond to known LRM units, as observed in enhanced color imagery (Figure 2A). Notably, the LRM-related clusters exhibit internal diversity (Figure 2B), suggesting potential compositional heterogeneities.Figure 2. Comparison between LRM (blue colors) in A. Mercury Dual Imaging System (MESSENGER) enhanced color 665 m/pixel global mosaic (Denevi et al., 2016) and B. Isolated clusters from Vergara Sassarini et al., 2025.To investigate these heterogeneities, this ongoing work will include a detailed analysis of the reflectance spectra across the derived LRM-related clusters to test for relatively compositional differences among them. We also aim to refine the input cubes used during fitting by including additional spectral ratio layers or PCA-derived features that may improve the clustering sensitivity to subtle spectral variations. As a future application, we intend to apply this integrated and tailored classification model to selected regions of interest on other regions of the planet different from Hokusai Quadrangle, with the goal of producing a global-scale distribution map of LRM, investigating their possible heterogeneity highlighted by this preliminary investigation. Such automated, multi-dimensional classification will be essential for identifying regions of interest for detailed investigation with the high-resolution data expected from SIMBIO-SYS (Cremonese et al., 2020) data on the BepiColombo's mission (Benkhoff et al., 2021). AcknowledgmentsThis study has been supported from the Italian Space Agency (ASI) under ASI-INAF agreement 2024-18-HH.0.ReferencesBecker, K. J., Robinson, M. S., Becker, T. L., Weller, L. A., Edmundson, K. L., Neumann, G. A., ... & Solomon, S. C. (2016, March). First global digital elevation model of Mercury. In 47th annual lunar and planetary science conference (No. 1903, p. 2959).Benkhoff, J., Murakami, G., Baumjohann, W., Besse, S., Bunce, E., Casale, M., … Zender, J. (2021). BepiColombo—Mission overview and science goals. Space Science Reviews, 217(8), 90.Cremonese, G., Capaccioni, F., Capria, M. T., Doressoundiram, A., Palumbo, P., Vincendon, M., … Turrini, D. (2020). SIMBIO-SYS: Scientific cameras and spectrometer for the BepiColombo mission. Space Science Reviews, 216(5), 1–78.Klima, R. L., Denevi, B. W., Ernst, C. M., Murchie, S. L., & Peplowski, P. N. (2018). Global distribution and spectral properties of low‐reflectance material on Mercury. Geophysical Research Letters, 45(7), 2945-2953.Lark, L. H., Head, J. W., & Huber, C. (2023). Evidence for a carbon-rich Mercury from the distribution of low-reflectance material (LRM) associated with large impact basins. Earth and Planetary Science Letters, 613, 118192.Peplowski, P. N., Lawrence, D. J., Evans, L. G., Klima, R. L., Blewett, D. T., Goldsten, J. O., ...Weider, S. Z. (2015). Constraints on the abundance of carbon in near-surface materials on Mercury: Results from the MESSENGER Gamma-Ray Spectrometer. Planetary and Space Science, 108, 98-107. Vander Kaaden, K. E., & McCubbin, F. M. (2015). Exotic crust formation on Mercury: Consequences of a shallow,...
- Open Access
- 9 July 2025
- Published by Copernicus GmbH
Controlled DTM and orthoimages mosaics from ExoMars TGO CaSSIS stereo-pairs
IntroductionThe Colour and Stereo Surface Imaging System (CaSSIS) is an advanced stereo camera developed for the ExoMars Trace Gas Orbiter mission, which has been orbiting Mars since April 2018. The CaSSIS camera can produce detailed 3D maps and capture images in four distinct color bands, providing essential data for analyzing the Martian surface and its composition [1].An innovative telescope rotation system creates a convergence angle of approximately 22° between the images of the stereo pairs, making CaSSIS's stereo configuration possible. Operating from a circular orbit approximately 400 km above the Martian surface, CaSSIS's average ground resolution is 4.5 m per pixel, resulting in digital terrain models (DTMs) with a resolution of about 13.5 m per pixel [2].CaSSIS has captured over 50,000 images, covering more than 8% of Mars' surface, including over 2,250 stereo-pairs. The planning of CaSSIS observations, which is guided by requests from the scientific community and the team, has focused on acquiring multiple images in areas of particular scientific interest. This has recently allowed for the acquisition of partially overlapping stereo pairs, enabling a more in-depth and comprehensive analysis of these regions.The work outlines the new methods for creating CaSSIS mosaics of DTMs and orthophoto and their application in various areas of Mars. MethodsThe 3DPD software, developed by the INAF OAPd team [2,3], serves as the foundation for generating Digital Terrain Models (DTMs) and orthorectified images. It is also the official tool for creating CaSSIS DTMs. Since the establishment of the framework, the software has been continuously updated to enhance the performance and quality of stereo data product generation. As a result, over 450 DTMs and orthoimages have been produced and are available in the OAPD-hosted repository (https://cassis.oapd.inaf.it/archive/) and soon in the ASI Space Science Data Center repository (SSDC).The developed pipeline enables the generation of stereo products from radiometrically calibrated CaSSIS frames by following several processing steps. Recent pre-processing improvements include a Bundle Block Adjustment phase and geometric corrections necessary for mosaicking the framelets, which help resolve misalignments that could otherwise lead to step artifacts and projection errors. The stereogrammetric processing of 3DPD, detailed by [3], is based on a two-step approach: a preliminary rough, feature-based step followed by an advanced multiscale refinement (dense-matching) step. According to [2], this process achieves a vertical accuracy of up to 8 meters, while the best horizontal accuracy is estimated at up to 13.5 meters per pixel, corresponding to CaSSIS DTM GSD (Ground Sample Distance) (3 pixel on ground). The resulting point cloud (PC), which has a heterogeneous density due to perspective viewing, is then interpolated onto a regular grid of height values within a reference coordinate system. To enhance geographical projection and absolute elevation, the point cloud is co-registered with the global MOLA-HRSC DTM [4]. This improves the accuracy to a standard deviation generally below 50 meters when compared to MOLA, effectively correcting any residual tilting. The point cloud is then interpolated, and the two original panchromatic images are orthorectified. The co-registration with MOLA simplifies the mosaicking of adjacent DTMs, as their georeferencing often requires minimal or no mutual corrections. These can typically be addressed through simple horizontal and vertical rigid translations. The seamless mosaicking process is based on the "feathering" approach, in which the overlapping areas are averaged using a weighted average that considers the inverse of the distance from the image edges and, in this implementation, from null pixels. Region of interestsThe number of areas potentially interested by multiple contiguous stereo pairs of CaSSIS is constantly being updated. Six areas have been selected for the initial assessment of the potential of CaSSIS DTM mosaics. Two of these are of particular interest for landing site studies: the north-western area of the Jezero crater (Fig. X), the location of NASA Mars 2020 surface mission [5], and Oxia Planum in Arabia Terra, chosen as the landing site for the second ESA ExoMars mission [6], currently postponed to 2028. The areas are currently interested by 14 and 13 contiguous DTMs, respectively.The other mosaic areas include various geological contexts in different regions of Mars: southern Isidis Planitia, well-preserved craters in Utopia and Daedalia Planitiae, and peri-glacial features in the Northern Plains.In addition to being applied in targeted geological studies, the preliminary mosaics will be evaluated globally using lower-resolution data (HRSC and MOLA) and locally using high-resolution HiRISE DTMs to verify vertical accuracy and georeferencing precision.Fig, 1 Mosaic of CaSSIS DTMs at 13.5 m/px of the northwestern part of Jezero Crater. The data is derived from the composition of 14 different DTMs. The display shows the elevation in transparency on a hillshade map. Acknowledgement This work has been developed under the ASI-INAF agreement n. 2024-40-HH.0 References[1] Thomas, N. et al., 2017. Space Sci. Rev., 212, 1897–1944. https://doi.org/10.1007/s11214-017-0421-1[2] Re, C. et al., 2022. Planet. Space Sci., 219, 105515. https://doi.org/10.1016/j.pss.2022.105515[3] Simioni, E. et al., 2021. Planet. Space Sci., 198, 105165. https://doi.org/10.1016/j.pss.2021.105165[4] Fergason, R.L. et al., 2018. USGS Astrogeology PDS Annex.[5] Williford, K.H. et al., 2018. In: Grotzinger, J.-P. & Webster, C.R. (Eds.), From Habitability to Life on Mars, Elsevier, pp. 275–308.[6] Quantin-Nataf, C. et al., 2021. Astrobiology, 21(3), 345–366. https://doi.org/10.1089/ast.2020.2244
- 9 July 2025
- Published by Copernicus GmbH
Open AccessControlled DTM and orthoimages mosaics from ExoMars TGO CaSSIS stereo-pairs
IntroductionThe Colour and Stereo Surface Imaging System (CaSSIS) is an advanced stereo camera developed for the ExoMars Trace Gas Orbiter mission, which has been orbiting Mars since April 2018. The CaSSIS camera can produce detailed 3D maps and capture images in four distinct color bands, providing essential data for analyzing the Martian surface and its composition [1].An innovative telescope rotation system creates a convergence angle of approximately 22° between the images of the stereo pairs, making CaSSIS's stereo configuration possible. Operating from a circular orbit approximately 400 km above the Martian surface, CaSSIS's average ground resolution is 4.5 m per pixel, resulting in digital terrain models (DTMs) with a resolution of about 13.5 m per pixel [2].CaSSIS has captured over 50,000 images, covering more than 8% of Mars' surface, including over 2,250 stereo-pairs. The planning of CaSSIS observations, which is guided by requests from the scientific community and the team, has focused on acquiring multiple images in areas of particular scientific interest. This has recently allowed for the acquisition of partially overlapping stereo pairs, enabling a more in-depth and comprehensive analysis of these regions.The work outlines the new methods for creating CaSSIS mosaics of DTMs and orthophoto and their application in various areas of Mars. MethodsThe 3DPD software, developed by the INAF OAPd team [2,3], serves as the foundation for generating Digital Terrain Models (DTMs) and orthorectified images. It is also the official tool for creating CaSSIS DTMs. Since the establishment of the framework, the software has been continuously updated to enhance the performance and quality of stereo data product generation. As a result, over 450 DTMs and orthoimages have been produced and are available in the OAPD-hosted repository (https://cassis.oapd.inaf.it/archive/) and soon in the ASI Space Science Data Center repository (SSDC).The developed pipeline enables the generation of stereo products from radiometrically calibrated CaSSIS frames by following several processing steps. Recent pre-processing improvements include a Bundle Block Adjustment phase and geometric corrections necessary for mosaicking the framelets, which help resolve misalignments that could otherwise lead to step artifacts and projection errors. The stereogrammetric processing of 3DPD, detailed by [3], is based on a two-step approach: a preliminary rough, feature-based step followed by an advanced multiscale refinement (dense-matching) step. According to [2], this process achieves a vertical accuracy of up to 8 meters, while the best horizontal accuracy is estimated at up to 13.5 meters per pixel, corresponding to CaSSIS DTM GSD (Ground Sample Distance) (3 pixel on ground). The resulting point cloud (PC), which has a heterogeneous density due to perspective viewing, is then interpolated onto a regular grid of height values within a reference coordinate system. To enhance geographical projection and absolute elevation, the point cloud is co-registered with the global MOLA-HRSC DTM [4]. This improves the accuracy to a standard deviation generally below 50 meters when compared to MOLA, effectively correcting any residual tilting. The point cloud is then interpolated, and the two original panchromatic images are orthorectified. The co-registration with MOLA simplifies the mosaicking of adjacent DTMs, as their georeferencing often requires minimal or no mutual corrections. These can typically be addressed through simple horizontal and vertical rigid translations. The seamless mosaicking process is based on the "feathering" approach, in which the overlapping areas are averaged using a weighted average that considers the inverse of the distance from the image edges and, in this implementation, from null pixels. Region of interestsThe number of areas potentially interested by multiple contiguous stereo pairs of CaSSIS is constantly being updated. Six areas have been selected for the initial assessment of the potential of CaSSIS DTM mosaics. Two of these are of particular interest for landing site studies: the north-western area of the Jezero crater (Fig. X), the location of NASA Mars 2020 surface mission [5], and Oxia Planum in Arabia Terra, chosen as the landing site for the second ESA ExoMars mission [6], currently postponed to 2028. The areas are currently interested by 14 and 13 contiguous DTMs, respectively.The other mosaic areas include various geological contexts in different regions of Mars: southern Isidis Planitia, well-preserved craters in Utopia and Daedalia Planitiae, and peri-glacial features in the Northern Plains.In addition to being applied in targeted geological studies, the preliminary mosaics will be evaluated globally using lower-resolution data (HRSC and MOLA) and locally using high-resolution HiRISE DTMs to verify vertical accuracy and georeferencing precision.Fig, 1 Mosaic of CaSSIS DTMs at 13.5 m/px of the northwestern part of Jezero Crater. The data is derived from the composition of 14 different DTMs. The display shows the elevation in transparency on a hillshade map. Acknowledgement This work has been developed under the ASI-INAF agreement n. 2024-40-HH.0 References[1] Thomas, N. et al., 2017. Space Sci. Rev., 212, 1897–1944. https://doi.org/10.1007/s11214-017-0421-1[2] Re, C. et al., 2022. Planet. Space Sci., 219, 105515. https://doi.org/10.1016/j.pss.2022.105515[3] Simioni, E. et al., 2021. Planet. Space Sci., 198, 105165. https://doi.org/10.1016/j.pss.2021.105165[4] Fergason, R.L. et al., 2018. USGS Astrogeology PDS Annex.[5] Williford, K.H. et al., 2018. In: Grotzinger, J.-P. & Webster, C.R. (Eds.), From Habitability to Life on Mars, Elsevier, pp. 275–308.[6] Quantin-Nataf, C. et al., 2021. Astrobiology, 21(3), 345–366. https://doi.org/10.1089/ast.2020.2244
- Open Access
- 9 July 2025
- Published by Copernicus GmbH
Mercury Global Crater Catalog using Multimodal Deep Learning for Crater Detection and Morphometric Analysis
IntroductionImpact cratering shapes the surfaces of airless bodies such as the Moon and Mercury, providing both a record of impacts history and a chronostratigraphic tool for dating geological units. Traditional crater catalogs, compiled manually or via semi-automated algorithms, remain incomplete below ~1 km diameter on the Moon[1] and ~5 km on Mercury[2], limiting studies of small-scale resurfacing, secondary-crater populations, and polar processes.This study has compiled the first impact-crater dataset for Mercury by a multimodal deep-learning pipeline. Our approach adapts YOLOLens single-stage object detector model initially fine-tuned on lunar imagery to 3-bands inputs and through domain-adaptive transfer learning is applied on Mercury.The key innovation is the construction of a three-band cube for each orthographic tile: WAC/MDIS visible spectrum mosaics Digital Terrain Model (DTM) Hillshaded DTM version MethodsEach band is co-registered at 100m/px resolution and normalized to [0,1] to ensure balanced feature learning . Tiles overlap by 128 px to mitigate edge effects, and a triple-scale subsampling captures craters whose diameters exceed individual tile dimensions .To achieve acceptable generalization model capabilities, a balancing function is applied across latitude ranges and FoVs to ensure a robust model, finally, the entire workflow is shown in Fig.1. Fig. 1: Workflow.The YOLOLens model[3], leverages on super-resolution techniques to improve impact crater detection on planetary surfaces. By addressing challenges such as low spatial resolution and variable solar illuminance in satellite imagery, YOLOLens demonstrated enhanced precision compared to state-of-the-art methods.The parameters are modified at the first/last layers such that the model creates super-resolution (SR) images from a single-channel image. Then, the SR output is used by an object-detection model in an end-to-end model.ResultsThe model has been trained across the lunar dataset (Robbins catalogue[1]) to demonstrate the effectiviness of the multimodal approach respect to the catalog derived in[4], thus, we applied a transfer learning step rapidly adapts to Mercury’s distinct albedo contrasts while retaining learned shape priors from the Moon.Validation on multiple quadrangles yields recall rates of 58% – 95.7% across primary/secondary craters, confirming robust cross-body generalization (Tab.1 and Tab.2).TABLE 1: Comparison of Crater Diameter Counts for Mercury. The Herrick catalog reports initial insights derived from the database, while also addressing several limitations that constrain future analyses. Our validation set was compiled through different craters manual counts[5,6,7] see Tab.2 .TABLE 2: Summary of crater detection performance by YOLOLens across different datasets. In the investigated regions, high recall values are obtained, suggesting robust performances of the YOLOLens model to identify impact craters. Since the areas used for comparison are representative of all the latitudes, such robustness could be extended to all the Hermean surfaces. Fig. 2 shows the size-frequency distributions. Fig. 2: Comparison of size-frequency distributions (SFDs) between the ME6M300TGT and [6], [7]. The development of a comprehensive global catalog allows us to investigate the distribution of craters and their morphometric characteristics, and from them observe clues about the evolution of the hermean surface. The crater density information provides us with important insights about the age of the different regions and the rates of degradation or remodeling (Fig.3).Fig. 3: (a) Extract from the Geological Map of the Hokusai quadrangle (H05)[8] and (b) the same region extracted from the ME6M300TGT to derive the density map by diameter range.An important parameter for the surface characterization is the depth-to-diameter (d/D) ratio, which is useful to define the post-impact degradation and modification of the crater. The d/D was measured for the entire catalog. Fig.4 shows the distribution of the d/D of the craters. Fig. 4: The map shows the d/D distribution of Mercury’s craters. A) Diameter > 15km over Mercury, B) The North Pole (> 5km) and C) The South Pole (> 15km). This study presents advancements in planetary surface analysis through the development of a comprehensive crater catalog for Mercury, which exploit images and topographical information simultaneously. The YOLOLens model has demonstrated its capability to detect craters with great accuracy and scale. The datasets also integrate morphometric information, enabling novel insights into crater distributions and the geological evolution of these celestial bodies. In addition, the catalog can function as an important tool in cartographic endeavors currently compiled as a multi-mapper regional-scale project[9]. References [1] Robbins (2019) J. Geophys. Res. Planets, 124(4), 871-892.[2] Herrick et al. (2011) Icarus 215(1), 452-454.[3] La Grassa et al. (2023) Remote Sens. 15(5), 1171.[4] La Grassa et al. (2025) ISPRS J. Photogramm. 220, 75-84.[5] Giacomini et al. (2020) Geosci. Front. 11(3), 855-870.[6] Galluzzi et al. (2016) J. Maps 12(sup1), 227-238.[7] Martellato et al. (2023) Bull. AAS, 55(8), #116.06.[8] Wright et al. (2019) J. Maps 15(2), 509-520.[9] Galluzzi (2019) In: Hargitai (eds) Planetary Cartography and GIS, Springer.Acknowledgements: we gratefully acknowledge funding from the Italian Space Agency (ASI) under ASI-INAF agreement 2024-40.HH.
- 9 July 2025
- Published by Copernicus GmbH
Open AccessMercury Global Crater Catalog using Multimodal Deep Learning for Crater Detection and Morphometric Analysis
IntroductionImpact cratering shapes the surfaces of airless bodies such as the Moon and Mercury, providing both a record of impacts history and a chronostratigraphic tool for dating geological units. Traditional crater catalogs, compiled manually or via semi-automated algorithms, remain incomplete below ~1 km diameter on the Moon[1] and ~5 km on Mercury[2], limiting studies of small-scale resurfacing, secondary-crater populations, and polar processes.This study has compiled the first impact-crater dataset for Mercury by a multimodal deep-learning pipeline. Our approach adapts YOLOLens single-stage object detector model initially fine-tuned on lunar imagery to 3-bands inputs and through domain-adaptive transfer learning is applied on Mercury.The key innovation is the construction of a three-band cube for each orthographic tile: WAC/MDIS visible spectrum mosaics Digital Terrain Model (DTM) Hillshaded DTM version MethodsEach band is co-registered at 100m/px resolution and normalized to [0,1] to ensure balanced feature learning . Tiles overlap by 128 px to mitigate edge effects, and a triple-scale subsampling captures craters whose diameters exceed individual tile dimensions .To achieve acceptable generalization model capabilities, a balancing function is applied across latitude ranges and FoVs to ensure a robust model, finally, the entire workflow is shown in Fig.1. Fig. 1: Workflow.The YOLOLens model[3], leverages on super-resolution techniques to improve impact crater detection on planetary surfaces. By addressing challenges such as low spatial resolution and variable solar illuminance in satellite imagery, YOLOLens demonstrated enhanced precision compared to state-of-the-art methods.The parameters are modified at the first/last layers such that the model creates super-resolution (SR) images from a single-channel image. Then, the SR output is used by an object-detection model in an end-to-end model.ResultsThe model has been trained across the lunar dataset (Robbins catalogue[1]) to demonstrate the effectiviness of the multimodal approach respect to the catalog derived in[4], thus, we applied a transfer learning step rapidly adapts to Mercury’s distinct albedo contrasts while retaining learned shape priors from the Moon.Validation on multiple quadrangles yields recall rates of 58% – 95.7% across primary/secondary craters, confirming robust cross-body generalization (Tab.1 and Tab.2).TABLE 1: Comparison of Crater Diameter Counts for Mercury. The Herrick catalog reports initial insights derived from the database, while also addressing several limitations that constrain future analyses. Our validation set was compiled through different craters manual counts[5,6,7] see Tab.2 .TABLE 2: Summary of crater detection performance by YOLOLens across different datasets. In the investigated regions, high recall values are obtained, suggesting robust performances of the YOLOLens model to identify impact craters. Since the areas used for comparison are representative of all the latitudes, such robustness could be extended to all the Hermean surfaces. Fig. 2 shows the size-frequency distributions. Fig. 2: Comparison of size-frequency distributions (SFDs) between the ME6M300TGT and [6], [7]. The development of a comprehensive global catalog allows us to investigate the distribution of craters and their morphometric characteristics, and from them observe clues about the evolution of the hermean surface. The crater density information provides us with important insights about the age of the different regions and the rates of degradation or remodeling (Fig.3).Fig. 3: (a) Extract from the Geological Map of the Hokusai quadrangle (H05)[8] and (b) the same region extracted from the ME6M300TGT to derive the density map by diameter range.An important parameter for the surface characterization is the depth-to-diameter (d/D) ratio, which is useful to define the post-impact degradation and modification of the crater. The d/D was measured for the entire catalog. Fig.4 shows the distribution of the d/D of the craters. Fig. 4: The map shows the d/D distribution of Mercury’s craters. A) Diameter > 15km over Mercury, B) The North Pole (> 5km) and C) The South Pole (> 15km). This study presents advancements in planetary surface analysis through the development of a comprehensive crater catalog for Mercury, which exploit images and topographical information simultaneously. The YOLOLens model has demonstrated its capability to detect craters with great accuracy and scale. The datasets also integrate morphometric information, enabling novel insights into crater distributions and the geological evolution of these celestial bodies. In addition, the catalog can function as an important tool in cartographic endeavors currently compiled as a multi-mapper regional-scale project[9]. References [1] Robbins (2019) J. Geophys. Res. Planets, 124(4), 871-892.[2] Herrick et al. (2011) Icarus 215(1), 452-454.[3] La Grassa et al. (2023) Remote Sens. 15(5), 1171.[4] La Grassa et al. (2025) ISPRS J. Photogramm. 220, 75-84.[5] Giacomini et al. (2020) Geosci. Front. 11(3), 855-870.[6] Galluzzi et al. (2016) J. Maps 12(sup1), 227-238.[7] Martellato et al. (2023) Bull. AAS, 55(8), #116.06.[8] Wright et al. (2019) J. Maps 15(2), 509-520.[9] Galluzzi (2019) In: Hargitai (eds) Planetary Cartography and GIS, Springer.Acknowledgements: we gratefully acknowledge funding from the Italian Space Agency (ASI) under ASI-INAF agreement 2024-40.HH.
- Open Access
- 9 July 2025
- Published by Copernicus GmbH
Rays and secondary craters of the Tycho impact event revealed through deep learning.
Introduction: The rays of Tycho crater are prominent features of the lunar nearside [1]. Rays and clusters of secondary craters from Tycho have been mapped as far as Taurus Littrow Valley (TLV), the Apollo 17 landing site, more than 2000 km from the impact crater [2]. It has been suggested that some Tycho ejecta hit the summit of the South Massif and triggered the Light Mantle avalanche in TLV [3,4]. One of the lines of evidence in support of Tycho-ejecta-triggered avalanching is the presence of a cluster of secondary craters on the crest of the South Massif, which is part of a more extended cluster that trends in the general direction of Tycho, spanning from the South Massif across the valley floor of TLV and the North Massif into Littrow crater [4,5] (Figure 1). A competing hypothesis suggests that the Light Mantle was instead triggered by seismic activity associated with the Lee-Lincoln lobate scarp in TLV [6,7]. Definitive evidence for the Light Mantle to be an avalanche either impact-related or seismic-related is yet to be found.In this work, we aimed to further the understanding of the distribution of the rays and secondary craters of Tycho. Our main purpose is to further inform the discussion about the origin of the Light Mantle.Methods: We defined a first region of interest (ROI-1) whose longitude and latitude limits are Tycho crater (SW corner) and TLV/Littrow crater (NE corner). This area covers 5,462,043 km2 (Figure 2). We used YOLOLens, an object detection model with performance enhanced by super-resolution methodology [8], on Kaguya Terrain Camera (TC) imagery (~7 m/px resolution) to generate a dataset of craters >100 m in diameter (Ncraters = 6,867,654) within ROI-1. The TC image resolution has been downsampled to 21 m/pixel. To generate the dataset, it took ~90 hours for pre-processing, ~25 hours for testing, ~6-8 hours for post-processing, and ~4 hours for a second post-processing stage. In ArcGIS Pro, we generated crater density diameter maps [e.g., 8] from the generated crater dataset.In a narrower area about 150 km wide that extends between Tycho crater and TLV (ROI-2, Figure 2), we complemented the generated dataset with existing databases of lunar impact craters >1-2 km in diameter [9,10,11], and manual editing.We used the ASCI (Algorithm for the Secondary Crater Identification) cluster analysis method [12] to identify clustered craters (thus potential secondary craters) based on deviations from a calculated random distribution: we defined 11 crater diameter ranges, and computed Voronoï polygon (VP) tesselation over the crater population of each size range; we then plot together the size-frequency distribution (SFD) of VPs of the mapped craters and of 300 simulations of the same crater population with a randomized distribution. The mean and the confidence envelope at ± 1σ of the simulated VPs SFD are computed. The intersection between the VPs SFD from mapped and simulated crater populations at +1σ corresponds to the threshold area below which polygons are described as clustered (thus the craters contained have a likely secondary origin), whereas those above the threshold value are considered random (thus considered primaries) (Figure 3).Results: The dataset generated contains 6,867,654 craters. We generated RGB composite diameter density maps to visualize the crater distribution by size range (100-150 m; 150-250m; 250-400m) (Figure 2). This initial analysis of the crater population shows well the existence of several rays associated with Tycho crater (bottom half of Figure 2). Further away from Tycho crater (upper part of Figure 2), less prominent rays are still visible. Many of such further rays seem to also point back to the Tycho crater location. In particular, a swarm of such rays appear near to TLV.The crater cluster analysis reveals that in ROI-1 2,863,257 craters are clustered, corresponding to 42% of the total population (Figure 4a). In ROI-2, the corridor from Tycho to A17 landing site, the analysis reveals that 240,546 are clustered craters (Figure 4b). When accounting for the expected maximum secondary crater size from Tycho (i.e., 5% of Tycho diameter), the number of clustered craters is 239,865, corresponding to 52% of the population in this region.Discussion and Conclusions: The density maps that we have generated allow us to identify ray segments that would otherwise be hard to detect using optical and spectral imagery alone. In particular, such density maps enable us to identify proximal and distant ray material that could be potentially correlated with Tycho. Maps of clustered craters show that their density decreases with distance from Tycho and that such a crater population (assumed secondaries) is dominated by craters smaller than 1 km. Additionally, the ROI-1 clustered craters map shows local areas of high density around smaller craters.We identify crater clusters aligned with the Tycho direction in TLV and beyond. This observation supports the previously mapped clustered craters at the Apollo 17 landing site as potentially derived from Tycho secondary impacts. Especially, identified clustered craters on the summit of the South Massif suggests that impacts may have destabilized slope material. However, we cannot comment whether these impacts were energetic enough to trigger the Light Mantle.An important observation of this work is that potentially secondary craters from Tycho may represent a large percentage of the total crater population. Therefore, their effect on dating lunar surfaces may not be negligeable if not removed.References: [1] Dundas and McEwen (2007), Icarus, 186(1), 31-40. [2] Lucchitta (1972), USGS Misc. Inv. Map 1-800. 1-800. [3] Muehlberger et al. (1973), USGS PSR, 70042539. [4] Lucchitta (1977), Icarus, 30(1), 80-96. [5] Iqbal et al. (2019), LPSC, Abstract 1005. [6] Schmitt et al (2017), Icarus, 298, 2-33. [7] Magnarini et al. (2023), JGR-Planets, 128(8), e2022JE007726. [8] Lagain et al. (2021a),...
- 9 July 2025
- Published by Copernicus GmbH
Open AccessRays and secondary craters of the Tycho impact event revealed through deep learning.
Introduction: The rays of Tycho crater are prominent features of the lunar nearside [1]. Rays and clusters of secondary craters from Tycho have been mapped as far as Taurus Littrow Valley (TLV), the Apollo 17 landing site, more than 2000 km from the impact crater [2]. It has been suggested that some Tycho ejecta hit the summit of the South Massif and triggered the Light Mantle avalanche in TLV [3,4]. One of the lines of evidence in support of Tycho-ejecta-triggered avalanching is the presence of a cluster of secondary craters on the crest of the South Massif, which is part of a more extended cluster that trends in the general direction of Tycho, spanning from the South Massif across the valley floor of TLV and the North Massif into Littrow crater [4,5] (Figure 1). A competing hypothesis suggests that the Light Mantle was instead triggered by seismic activity associated with the Lee-Lincoln lobate scarp in TLV [6,7]. Definitive evidence for the Light Mantle to be an avalanche either impact-related or seismic-related is yet to be found.In this work, we aimed to further the understanding of the distribution of the rays and secondary craters of Tycho. Our main purpose is to further inform the discussion about the origin of the Light Mantle.Methods: We defined a first region of interest (ROI-1) whose longitude and latitude limits are Tycho crater (SW corner) and TLV/Littrow crater (NE corner). This area covers 5,462,043 km2 (Figure 2). We used YOLOLens, an object detection model with performance enhanced by super-resolution methodology [8], on Kaguya Terrain Camera (TC) imagery (~7 m/px resolution) to generate a dataset of craters >100 m in diameter (Ncraters = 6,867,654) within ROI-1. The TC image resolution has been downsampled to 21 m/pixel. To generate the dataset, it took ~90 hours for pre-processing, ~25 hours for testing, ~6-8 hours for post-processing, and ~4 hours for a second post-processing stage. In ArcGIS Pro, we generated crater density diameter maps [e.g., 8] from the generated crater dataset.In a narrower area about 150 km wide that extends between Tycho crater and TLV (ROI-2, Figure 2), we complemented the generated dataset with existing databases of lunar impact craters >1-2 km in diameter [9,10,11], and manual editing.We used the ASCI (Algorithm for the Secondary Crater Identification) cluster analysis method [12] to identify clustered craters (thus potential secondary craters) based on deviations from a calculated random distribution: we defined 11 crater diameter ranges, and computed Voronoï polygon (VP) tesselation over the crater population of each size range; we then plot together the size-frequency distribution (SFD) of VPs of the mapped craters and of 300 simulations of the same crater population with a randomized distribution. The mean and the confidence envelope at ± 1σ of the simulated VPs SFD are computed. The intersection between the VPs SFD from mapped and simulated crater populations at +1σ corresponds to the threshold area below which polygons are described as clustered (thus the craters contained have a likely secondary origin), whereas those above the threshold value are considered random (thus considered primaries) (Figure 3).Results: The dataset generated contains 6,867,654 craters. We generated RGB composite diameter density maps to visualize the crater distribution by size range (100-150 m; 150-250m; 250-400m) (Figure 2). This initial analysis of the crater population shows well the existence of several rays associated with Tycho crater (bottom half of Figure 2). Further away from Tycho crater (upper part of Figure 2), less prominent rays are still visible. Many of such further rays seem to also point back to the Tycho crater location. In particular, a swarm of such rays appear near to TLV.The crater cluster analysis reveals that in ROI-1 2,863,257 craters are clustered, corresponding to 42% of the total population (Figure 4a). In ROI-2, the corridor from Tycho to A17 landing site, the analysis reveals that 240,546 are clustered craters (Figure 4b). When accounting for the expected maximum secondary crater size from Tycho (i.e., 5% of Tycho diameter), the number of clustered craters is 239,865, corresponding to 52% of the population in this region.Discussion and Conclusions: The density maps that we have generated allow us to identify ray segments that would otherwise be hard to detect using optical and spectral imagery alone. In particular, such density maps enable us to identify proximal and distant ray material that could be potentially correlated with Tycho. Maps of clustered craters show that their density decreases with distance from Tycho and that such a crater population (assumed secondaries) is dominated by craters smaller than 1 km. Additionally, the ROI-1 clustered craters map shows local areas of high density around smaller craters.We identify crater clusters aligned with the Tycho direction in TLV and beyond. This observation supports the previously mapped clustered craters at the Apollo 17 landing site as potentially derived from Tycho secondary impacts. Especially, identified clustered craters on the summit of the South Massif suggests that impacts may have destabilized slope material. However, we cannot comment whether these impacts were energetic enough to trigger the Light Mantle.An important observation of this work is that potentially secondary craters from Tycho may represent a large percentage of the total crater population. Therefore, their effect on dating lunar surfaces may not be negligeable if not removed.References: [1] Dundas and McEwen (2007), Icarus, 186(1), 31-40. [2] Lucchitta (1972), USGS Misc. Inv. Map 1-800. 1-800. [3] Muehlberger et al. (1973), USGS PSR, 70042539. [4] Lucchitta (1977), Icarus, 30(1), 80-96. [5] Iqbal et al. (2019), LPSC, Abstract 1005. [6] Schmitt et al (2017), Icarus, 298, 2-33. [7] Magnarini et al. (2023), JGR-Planets, 128(8), e2022JE007726. [8] Lagain et al. (2021a),...