How morphologic variability, a novel metric derived from supervised landform classification, can offer insights into the processes governing fault scarp morphology in jointed rock

Fault scarps are steps in the landscape created by surface-rupturing faults. Study of their morphology can supply paleoseismic information, such as the timing and size of past events, but the form of a single fault scarp can show great variability along the length of the scarp, complicating its interpretation. We developed a methodology to quantify this variability: scarp-normal profiles are extracted from point clouds, their shape is automatically classified using a supervised learning algorithm, and these classification results are used to calculate the morphologic variability metric, a measure of the frequency and degree of change in profile form along strike.

Using point clouds derived from structure-from-motion photogrammetry, we computed the morphologic variability along thirteen jointed-bedrock fault scarps from four field sites, located in southwestern Iceland, northern and central California, and southeastern Hawai’i. Quantifiable characteristics such as climate, vegetation, lithology, fault throw, and fracture spacing change either internally along a single scarp or between these four sites.

In an individual scarp, we make pairwise comparisons between measurements of a characteristic and the morphologic variability: a strong correlation between the two indicates that this characteristic is an important driver of scarp form. For example, in the young Icelandic scarps, scarp throw is correlated with morphologic variability, suggesting that the initial slip distribution along a fault contributes to the variability in the profile forms of younger scarps. We also compute the fracture intensity and orientation along the scarps and hypothesize that increased fracture spacing leads to decreased morphologic variability.

To understand variation between sites, we make pairwise comparisons between the average values of morphologic variability of the scarps and site-specific characteristics. For example, maximum scarp throw is negatively correlated with average morphologic variability, suggesting that scarp profile form evolves towards a common morphology as scarps mature.

We show that the morphologic variability metric is a useful tool to understand the agents responsible for changes in scarp form, an essential step in accurately interpreting any paleoseismic information that might be present. Morphologic variability is a metric that can be determined for any other type of linear landform (e.g. coastal bluffs or cross-channel profiles) and our approach of linking morphologic variability to process is applicable to a wide array of geomorphic questions.