Browse by author
Lookup NU author(s): Dr Louise RayneORCiD, Dr Filippo BrandoliniORCiD, Jen MakovicsORCiD, Hope Irvine
This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).
Understanding what led to desertification in the long-term is crucial for adaptation to climate change and pressures on resources in North Africa, but existing maps do not accurately show the extent of degraded land or the traditional water systems which underpinned cultivation. These products rely on recent vegetation trends and hindcasted statistical data. Desertification which occurred prior to the later 20th century is poorly represented, if at all. However, large areas of abandoned fields are distinctive in satellite imagery as brightly reflectant and smooth surfaces. We present a new and open-source machine-learning workflow for detecting desertification using satellite data. We used Google Earth Engine and the random forest algorithm to classify five landcover categories including a class representing desertified fields. The input datasets comprised training polygons, a 12-band Sentinel-2 composite and derived tasselled cap components, and a Sentinel-1 VV-polarisation composite. We test our approach for a case study of Skoura oasis in southern Morocco with a resulting accuracy of 74-76% for the desertification class. We used image interpretation and archaeological survey to map the traditional irrigation systems which supply the oasis.
Author(s): Rayne L, Brandolini F, Makovics JL, Hayes-Rich E, Levy J, Irvine H, Assi L, Bokbot Y
Publication type: Article
Publication status: Published
Journal: Scientific Reports
Year: 2023
Volume: 13
Online publication date: 08/11/2023
Acceptance date: 30/10/2023
Date deposited: 10/11/2023
ISSN (electronic): 2045-2322
Publisher: Nature Publishing Group
URL: https://doi.org/10.1038/s41598-023-46319-1
DOI: 10.1038/s41598-023-46319-1
Altmetrics provided by Altmetric
