Object-based Image Analysis for Lava Flow Morphology Classification using Optical and SAR Satellite Imagery

Daniel Hölbling, Muhammad Aufaristama, Zahra Dabiri

Research output: Contribution to conferenceAbstractAcademic

Abstract

Lava morphology is related to the surface characteristics of lava flows after solidification. Knowledge about lava morphology can be beneficial for lava flow modelling and can provide insights into the emplacement conditions. Earth observation (EO) data and advanced image classification algorithms offer suitable possibilities for lava morphology classification, while being more efficient than time-consuming and expensive field mapping.

Object-based image analysis (OBIA) is a well-established and recognized method for semi-automated geomorphological mapping. The main advantage of OBIA over pixel-based approaches is its ability to consider, next to spectral information, also spatial, contextual and hierarchical properties during the classification process. This makes OBIA particularly suitable for mapping complex natural phenomena. Various studies demonstrated the applicability OBIA for the classification of landforms, landslides or glaciers, but only little research has been done for semi-automated object-based mapping of volcanic deposits.

In this study, we assess the applicability of optical, i.e. Sentinel-2, and synthetic aperture radar (SAR), i.e. Sentinel-1, satellite imagery for semi-automated lava morphology mapping within an OBIA framework. As study area, we selected the Krafla lava field in northeast Iceland. The machine learning algorithm random forest (RF) is used for lava flow morphology classification, whereby we differentiate between five different lava morphology classes. The classification accuracy is assessed by comparison to a reference dataset, which is created by a combination of visual interpretation of very high resolution (VHR) imagery and information from existing maps and literature.
Original languageEnglish
Number of pages1
Publication statusPublished - 8 Jul 2019
Externally publishedYes
Event27th IUGG General Assembly 2019 - Palais des Congrès, Montreal, Canada
Duration: 8 Jul 201918 Jul 2019
Conference number: 27
http://iugg2019montreal.com/

Conference

Conference27th IUGG General Assembly 2019
CountryCanada
CityMontreal
Period8/07/1918/07/19
Internet address

Keywords

  • ITC-CV

Fingerprint Dive into the research topics of 'Object-based Image Analysis for Lava Flow Morphology Classification using Optical and SAR Satellite Imagery'. Together they form a unique fingerprint.

Cite this