Mapping Albedo Change Across the South Polar Perennial Cap: A Multi-Year MARCI Analysis

Introduction:

The Southern Polar Perennial Cap (SPPC) is a prominent surface feature composed of CO₂, water-ice, and dust [1]. It can be broken down into two main sections: the large water-ice outlier and the residual cap (Figure 1a). The water-ice outlier exhibits high seasonal and interannual variability, while the residual cap is comparatively stable [2][3]. Its surface ice morphology has been categorized, reflecting the many complex processes shaping this area of the SPPC (Figure 1b) [4]. However, the relationship between these morphological units and interannual albedo variability remains poorly understood.

This work uses newly developed true-colour mosaics from the Mars Color Imager (MARCI) to identify the extent of the SPPC, classify albedo levels, and identify areas with the greatest annual variability. By correlating these with surface morphology groups, we aim to identify which morphologies are most affected by seasonal processes and the underlying mechanisms driving them.

Figure 1: A) True-colour MARCI mosaic (L_s= 310°–319.9°) with the SPPC (white box), residual cap (blue box, [4]), and water-ice outlier (red box). B) Morphological units of the residual cap [4].

Methods

The MARCI mosaics span Mars Years (MYs) 30, 33, and 36, between L_s= 310° and 332°, with a temporal resolution of 2° of L_s. While offering higher temporal resolution than [4], they cannot resolve morphological details directly. To identify distinct albedo levels, we apply a Gaussian Mixture Model (GMM) to cluster pixel values in the CIELAB colour space. The GMM is trained on temporally averaged mosaics for each Ls date (Figure 2a & b), capturing the general spatial extent of the SPPC while providing a stable reference for interannual comparison. MYs 28, 29, 34, and 35 are excluded due to direct or lagged impacts from the MY 28 and MY 34 Global Dust Storm Events (GDS). An elbow analysis indicates an optimal cluster count of four, with a silhouette score of 0.7 (Figure 2c). Cluster maps from individual MYs are then compared against the averaged reference to identify regions of cluster change.

Figure 2: A) Region of interest (white box, Figure 1a). B) Cluster map (K = 4). C) Bayesian Information Criterion (BIC) showing K = 4 as optimal.

Results/Discussion 

For our initial analysis, we focus on the mosaic spanning L_s = 310°–319.9°, which provides the greatest number of overlapping MYs for interannual comparison. Variability is predominantly concentrated on the margins of the SPPC, with the water-ice outlier exhibiting the greatest variability. Within the residual cap, changes are similarly confined to the margins, with little to no variability observed in the interior, a consequence of the coarse spatial resolution of the mosaics.

The largest departure from the average mosaic occurs during MY 28, when the margins of both the residual cap and water-ice outlier show substantial deviation (Figure 3a). The water-ice outlier is reduced to background dust albedo levels, indicating a far more advanced state of sublimation than the multi-year average. Within the residual cap, the most significant change occurs around 83–85° S, 287–303° E, where clusters transition from the brightest to the second-brightest albedo class in MY 28. Notably, MY 28 is the only MY with significant transitions from bright ice to the background dust cluster, indicating a smaller residual cap extent and consistent with previous studies ([2]; Figure 3c).

Figure 3: A) Cluster transition map (average vs. MY 28). B) Transitions overlaid on average mosaic. C) Transition count by MY; MY 28 shows the highest transitions, including the only ice-to-dust transitions.

When examining the surface ice morphologies, we find no consistent trend linking a specific morphology group to the magnitude of cluster transition (Figure 1b & 3a). However, the morphology groups located at the margins of the residual cap, specifically the B group, are observed to change in our analysis. These groups are associated with various forms of pitted morphologies, suggesting that pitted features may be strongly affected by seasonal processes, such as interannual variability in CO₂ ice sublimation.

Conclusion

Applying GMM clustering in CIELAB colour space to MARCI true-colour mosaics provides a quantitative framework for characterizing interannual albedo variability across the SPPC. Variability is concentrated at the margins of the residual cap and the water-ice outlier, with MY 28 representing the most extreme departure from the multi-year average and the only MY in which portions of the residual cap reach background dust albedo levels. Pitted morphologies at the residual cap margins (group B) emerge as the surface types most strongly associated with this variability. Future work will extend the analysis to additional Ls bins to track seasonal evolution and assess correlations with specific atmospheric or surface processes.

References

[1] Cartwright S. F. A. et al. (2023) JGR: Planets, 128(11). [2] Acharya P. et al. (2024) Icarus, 417, 116104. [3] Calvin W. M. et al. (2017) Icarus, 292, 144–153. [4] Thomas P. C. et al. (2016) Icarus, 268, 118–130.