Data Resolution on the Analysis of Planetary Compressional Structures

Planetary geology studies are mainly based on remote sensing data, among which Digital Elevation Models (DEMs) are fundamental to understand the geological evolution of planetary bodies. DEMs are used to analyze the morphological expression of geological surface features, among which compressional tectonic structures are fundamental to understand the interior of planets and their evolution. Their morphological expression are directly linked to geometry of subsurface faults, their depth and shortening using numerical and forward kinematic modelling [1]. However, the morphological expression is strongly dependent on the spatial resolution of the topographic data.

A systematic investigation to quantify how resolution and data type affect the comprehension of fault parameters and shortening is still missing. This work evaluates the effect of data resolution in the analysis of compressional ridges.

The analysis is based on remote sensing structural mapping, statistic and kinematic analysis of topographic profiles including line length combined with faulting linked to topographic offset and trishear (TS) combined with fault parallel flow (FPF) kinematic modelling. The analysis is carried out on Mars, selecting five different wrinkle ridges covered by MOLA ,HRSC, CTX and HiRISE data to guarantee a gradual increase of resolution from ~463 m/px, ~100 m/px, ~6 m/px to ~0.3 m/px, respectively.

The analysis is based on topographic profiles, equally spaced by 1–2 km (depending on the ridge length), orthogonally oriented to the mean direction of the tectonic structures (Fig. 1). Along each section, and per each DEM resolution, the ridge width and relief (W_R and R_R, respectively) are measured to define the ridge size based on a morphological harmonic index obtained from HiRISE data; a lower index represents a wide but low ridge or a high but narrow ridge; a higher index represents a wide and high ridge.

Fig. 1. The five selected study areas showing the topography through HiRISE data, along with the location and orientation of the topographic profiles.

Along each section and per each DEM resolution, the ridge line length (l₀) and horizontal length (l₁) are measured to obtain values of folding through the line length method in combination with faulting from the relation of elevation offset and assumed fault dip of 30° [2]. Along two representative selected topographic profiles per each study area and DEM resolution, the TS-FPF is applied to obtain values of shortening and depth to detachment [1].

The morphological and structural analysis results are integrated in a statistical analysis to characterize the differences between data resolutions of MOLA, HRSC, CTX respect to HiRISE data; the HiRISE analysis is assumed to give the best representation of the natural case. The analysis comprises the absolute and percentage difference (Bias), the coefficient of determination (R²) and the normalized root mean square deviation (NRMSE). Box-whisker plots are used to show the bias between different parameters.

The effect of resolution on the morphological analysis demonstrated how the MOLA, can generally approximate the overall ridge geometry, but underestimates down to 42% with emphasis on the relief. The HRSC can better resolve structures complexities but it is characterized by higher standard deviations. The CTX is resolves the majority of the morphological characteristics.

The effect of resolution on the structural analysis suggests how the methodology by [2] is subject to high deviations; it can lead to shortening overestimations up to 500% on MOLA and 250% on HRSC data. Its application on CTX data gives results comparable with HiRISE data. Kinematic forward modelling delivers reliable estimations of shortening and detachment depth when applied to CTX data and HRSC, at a lesser extent, while yielding less reliable results when applied to MOLA. The latter can underestimate or overestimate shortening up to -80% and 60%.

Spatial resolution affects the possibility to resolve objects of different sizes: bigger objects could be resolved also by lower resolution data, while smaller object might be hidden. The plots (Fig. 2a,b) show how the ridge size is handled by different spatial resolutions: while smaller structures (lower WRi) lead to more scattering solutions, bigger structures (higher WRi) can be more coherently characterized even by MOLA. The distribution of CTX data is concentrated around 0% deviation from HiRISE, the distribution of MOLA and HRSC data is strongly scattered in the left of the plot, while more concentrated around 0% in the right. A WRi of ~225m marks the shift from poorly resolvable sizes to relatively properly resolvable sized. Below  this threshold, only high resolution data can be considered reliable (e.g., CTX), while above it, even lower resolution data (i.e., MOLA and HRSC) can give overall reliable results.

Fig. 2. Plots showing the relationships between the morphological harmonic index WRi in respect to (a) the combined folding and faulting shortening, (b) the kinematic forward model slip and (c) depth to detachment.

The analyzed data differs in terms of acquisition methods. MOLA is obtained through the interpolation of sampling points acquired along a subpolar orbit with N-S point separations of ~300m and E-W of ~1km, with a total elevation uncertainty of ± 3m; it tends to smooth out steep ridges. HRSC is created by processing two dedicated stereo channels acquired in a single pass with a ~20m vertical accuracy, being affected by artefacts displayed as spikes and lows. CTX and HiRISE requires the stereo processing of separate stereo images acquired at different times and emission angles, with a vertical accuracy of ~3–5m. They are dependent on illumination conditions and emission angles during acquisitions and might be affected by acquisition and processing noise.

[1] Carboni et al., 2025, Icarus 425, 116330. [2] Golombek et al., 1991, 21^st LPSC 21, 679–693.