Fairy circles (FCs) are a unique phenomenon in southwestern Africa and western Australia. The fairy circles are characterized by circular bare soil or sand gaps, 4-10 m in diameter, sterilized from higher or lower vegetation (biocrusts), and appearing in an almost uniform and homogeneous distribution in arid landscapes. Their extremely organized shape raised three main theories on the formation of the circles, including (1) the social insect (termites and ants); (2) the euphorbia species allelopathy; and (3) the vegetation self-arrangement. The current study does not aim to justify or reject either of these theories but to explore and characterize the circles’ spatial, spectral, and temporal dynamics using machine learning models over remote sensing retrievals. Several vegetation, biocrusts, and mineralogy indices will be applied for the spectral pattern. A convolution neural network (CNN) will be used to detect the fairy circles. This method will emphasize the correlation between the fairy circles’ different attributes and their locations, such as texture, size, shape, edges, and more. The Girbes Plain in northern-western Namibia is used as a case study since the unique matrix around the circles is characterized by a thin layer of biological soil crusts, and part of the fairy circles have an elliptical shape that is slightly different from the rest of the uniform circles. The use of remote sensing in this work will deal with a large area and give a comprehensive overview of the fairy circles’ spatial distribution and their unique spectral characteristics.

Ms. Klil Noy
Spatial and Spectral Analysis of Fairy Circles Pattern using Satellite Image
Ben Gurion University of the Negev, Israel