Unsupervised classification of lava flows in Harrat Lunayyir using remote sensing and GIS

Abstract

Mapping of lava flows based on remote sensing data of high accuracy has become a common tool for exploring volcanic eruptions in greater detail. Mapping data based on remote sensing data provides information on the location and flow direction of lava flows as well as their areas and enables the localisation of volcanic vents—valuable knowledge for pre- and post-eruption volcanic activity and behaviour estimations. The current research seeks to expand the understanding of the volcanic and tectonic processes in Harrat Lunayyir, Saudi Arabia. In view of this, remote sensing data was classified using iterative self-organising data analysis technique (ISODATA) algorithms. The aim of the classification is to identify spectral top-of-atmosphere (TOA) reflectance values to distinguish layers of old and recent lava flows based on differences. As a result, three distinct basaltic units were identified to have the following ages: 15.1 ± 6.1 ka (4%), 15.0 ± 8.4 ka (6%), and 14.6 ± 23.1 ka (10%). The differences in the calculated areas of the lava flows as mapped based on the remote sensing data and earlier modelling are captured and identified in relation to the local geomorphologic and geologic structures. The differences can be partly explained as being related to a considerable weathering of the observed geological formations at 800–1600 nm, with reflectance of 12%. The limitation of the methodology relates to the lack of accurate geochronological timeline, that is, an inability to identify accurate age of the samples.