A comparison of Sentinel 1 and Sentinel 2 Image Classification for Detection of Water Bodies in Turkiye and World-Wide

Date

2024-9

Author

Forsyth, Ben

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

92
views

73
downloads

Cite This

This thesis explores the use of Synthetic Aperture Radar (SAR) and optical remote sensing technologies to monitor and analyze the water extents of various lakes, with a primary focus on Lake Tuz in Turkiye. Employing the Random Forest Classifier, the study classifies land cover types and detects water extents, demonstrating that machine learning techniques maintain consistently high classification accuracy and water monitoring capabilities for Sentinel-1 data. However, the same techniques did not prove as robust when applied to Sentinel-2 data, showing the advantages of SAR over optical data in detecting water bodies under varying atmospheric conditions. The research shows the significant impact of seasonal changes in water extent on global water resources, with notable seasonal declines and increases in water extent. The Random Forest Classifier was trained iteratively on multiple regions of interest to enhance classification accuracy and adapt to diverse geographical conditions. The results indicate that the integration of SAR and optical data, coupled with machine learning algorithms, presents a powerful tool for environmental monitoring. Sentinel-1 data showed high overall accuracy in water classification, while Sentinel-2 data faced challenges due to cloud cover and spectral similarities between different land cover types, resulting in lower overall accuracy. These findings emphasize the potential for remote sensing technologies to enhance our understanding of climate change impacts and aid in the development of effective water resource management strategies.

Subject Keywords

Sentinel-1, Sentinel-2, Lakes, Random Forest Algorithm, Remote Sensing

URI

https://hdl.handle.net/11511/111338

Collections

Graduate School of Natural and Applied Sciences, Thesis