A Geographical Environmental–Driven Method for Predicting Aqueous Mineral Distribution on Mars
Photogrammetric Engineering & Remote Sensing · 2026
DOI: 10.14358/pers.25-00161r2 ↗Kurz gefasst
Fachartikel in der Zeitschrift Photogrammetric Engineering & Remote Sensing, 2026. Verfasst von Leilei Jiao und 7 weiteren Autor:innen.
Abstract (Originalsprache)
Aqueous minerals, specifically referring to minerals formed through interactions with liquid water, record the history of Mars’s transformation under the influence of internal and external forces. Their types, abundances, and spatial distribution provide clues for studying the ancient environmental conditions and water–rock interactions on Mars. To overcome the limitations of traditional binary classification approaches and better capture the transitional nature of geological processes, this study treats mineral detection as a spatial regression task aimed at predicting the continuous spatial probability distribution of aqueous minerals. To achieve this, we propose a novel geographical environment–driven (GED) prediction method for detecting aqueous minerals. Unlike traditional geostatistical methods, the proposed method uses the spatial locations of known mineral occurrences and integrates multisource geoenvironmental variables from their surrounding neighborhoods to construct sets of variables based on second-dimension association. By applying Lasso regularization, the method identifies and selects key geoenvironmental variables that exhibit strong synergistic effects, thereby establishing direct mapping from multidimensional geoenvironmental variables to the spatial probability distribution of aqueous minerals that reduces the reliance on dense sampling and outperforms traditional interpolation methods, especially in sparsely sampled areas. Comparative experiments against the traditional geostatistical interpolation method demonstrate that the GED method achieves significantly higher predictive accuracy in multiple Martian hydrological landform regions. The GED model significantly improves R² values across all study areas, with the most pronounced enhancement of Fe/Mg phyllosilicates in the Nili Fossae region, where the value rises from 0.10 to 0.57. Uncertainty analysis further indicates that the GED method produces more concentrated prediction errors, achieving a 31.8% reduction in average mean squared error compared to the first dimension association method from 0.2519 to 0.1718. Moreover, the results of the bivariate Moran’s I analysis demonstrate that the proposed GED method, incorporating geographical environmental variables, exhibits higher bivariate Moran’s I values with respect to aqueous mineral distribution than spatial location–only variables. This finding confirms the GED method’s superior capability in characterizing the synergistic correlations between geoenvironmental variables and aqueous minerals under complex and heterogeneous conditions.
Zitieren
APA
Jiao, L., Huang, R., Zheng, Q., Tang, H., Xu, Y., Ye, Z., Liu, S., & Tong, X. (2026). A Geographical Environmental–Driven Method for Predicting Aqueous Mineral Distribution on Mars. Photogrammetric Engineering & Remote Sensing. https://doi.org/10.14358/pers.25-00161r2
Vancouver
Jiao L, Huang R, Zheng Q, Tang H, Xu Y, Ye Z, et al. A Geographical Environmental–Driven Method for Predicting Aqueous Mineral Distribution on Mars. Photogrammetric Engineering & Remote Sensing. 2026. doi:10.14358/pers.25-00161r2
BibTeX
@article{Jiao2026,
author = {Jiao, Leilei and Huang, Rong and Zheng, Qifang and Tang, Hong and Xu, Yusheng and Ye, Zhen and Liu, Sicong and Tong, Xiaohua},
title = {A Geographical Environmental–Driven Method for Predicting Aqueous Mineral Distribution on Mars},
journal = {Photogrammetric Engineering & Remote Sensing},
year = {2026},
doi = {10.14358/pers.25-00161r2}
}
EUVIMED