Automated identification of lava flow structures using local Fourier spectrum of digital elevation data

Abstract

The rheological parameters of a lava flow can be inferred by studying structural features on the solidified lava surface. We have developed a technique using the S‐transform (a localized Fourier transform) to analyze and interpret digital elevation data. We examined structural features of a silicic lava flow, the Medicine Lake dacite flow in California. The S‐transform accurately identified and located both repeating and single‐instance structures. Several classes of structures were identified in the data, including the following: localized short wavelength features (1.3–3.6 m in wavelength) identified as large blocks and repeating medium wavelength features (5–8, 10–16, 18–26, 30–36, and 56–67 m in wavelength) identified as multiple generations of crustal folding. Fold wavelengths determined by the S‐transform and a model of compressional folding associated with near surface viscosity gradients were used to evaluate the rheology of the Medicine Lake dacite flow. The similarity of the amount of strain during folding events and the wavelength ratio of successive generations of folding suggest that the flow rheology remained relatively constant during folding of the flow surface. Repeating long wavelength features (67–70‐ and 105–140‐m wavelengths) were identified as the spacing of crease structures. Crease structures were recognized by the absence of short wavelength features at the location of their smooth flanks and by a central spike corresponding to the central trough. The limitations and applicability of this approach lie in the accuracy and resolution of the digital elevation data used.