Desert landform detection and mapping using a semi-automated object-based image analysis approach

Mohammad Kazemi Garajeh, Bakhtiar Feizizadeh, Qihao Weng, Mohammad Hossein Rezaei Moghaddam, Ali Kazemi Garajeh

Research output: Journal article publicationJournal articleAcademic researchpeer-review

10 Citations (Scopus)

Abstract

Traditional landform modeling approaches are labor-intensive and time-consuming. We proposed and developed a semi-automated object-based image analysis (OBIA) rule set approach for desert landforms detection and mapping. Sentinel-2 image and digital elevation model (DEM) were acquired for the study area. The multi-resolution segmentation algorithm was employed on the datasets to select relevant features to define appropriate segmentation scales for all landform categories. Object-based rule sets were then employed using spatial (DEM and its derivatives, e.g., slope, aspect, and hillshade) and spectral information for semi-automated classification of the desert landforms. Desert landforms are detected and classified into four classes: saline dome, barchan, playa, and dune. The Fuzzy Synthetic Evaluation (FSE) technique was applied in concert with the error matrix to validate the accuracy of the classification results based on field data, Google Earth, and geological maps. Our findings demonstrated the highest confidence of overall accuracy (OA) 96.21%, 92.58%, 95.99%, and 95.05% respectively, for the saline dome, barchan, playa, and dune. Results showed the strong potential of the rule-based OBIA remote sensing approach for desert landform detection and delineation. Results further demonstrated the efficiency of spatial and spectral features for desert landforms detection and delineation.

Original languageEnglish
Article number104721
JournalJournal of Arid Environments
Volume199
DOIs
Publication statusPublished - Apr 2022

Keywords

  • Desert landforms
  • Earth's landforms
  • Fuzzy rule-based classification
  • Geomorphology
  • Object-based image analysis (OBIA)

ASJC Scopus subject areas

  • Ecology, Evolution, Behavior and Systematics
  • Ecology
  • Earth-Surface Processes

Fingerprint

Dive into the research topics of 'Desert landform detection and mapping using a semi-automated object-based image analysis approach'. Together they form a unique fingerprint.

Cite this