Bragg Scattering on SAR Imagery
A study on detecting Bragg Scattering on Sentinel-1 SAR Imagery over Water Reservoirs in Ghana using Google Earth Engine and Machine Learning
C.S.H. van Meurs (TU Delft - Civil Engineering & Geosciences)
N.C. van de Giesen – Mentor (TU Delft - Water Systems Monitoring & Modelling)
S.C. Steele-Dunne – Graduation committee member (TU Delft - Mathematical Geodesy and Positioning)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Small and medium reservoirs are of vital importance to the local communities in Ghana. Since these smaller reservoirs are especially sensitive to problems such as evaporation and sedimentation, and these reservoirs react strongly to seasonal fluctuations in temperature and precipitation patterns, continuous monitoring is of great importance. Sentinel-1 SAR imagery offers a reliable manner in which these reservoirs can be delineated and monitored, but the wind-induced phenomenon of Bragg scattering poses problems when we want to further increase automation of the monitoring process. In this thesis, a classifier has been developed that uses Sentinel-1 imagery, Gray Level Co-Occurance Matrix (GLCM) features and a Simple Non-Iterative Clustering (SNIC) to differentiate between surrounding shore, unaffected water and Bragg scattering patches. The combination of these features and methods, together with a Random Forest classifier showed to be accurate in differentiating between the different classes and delineating the final reservoirs, even when Bragg scattering is present on the SAR imagery.