Mapování záplav z roku 2024 v Ostravě

Abstract

This bachelor's thesis maps and quantifies the extent of floods that affected Ostrava and its surroundings in September 2024. Satellite data from the Copernicus programme are used as the primary data source – specifically optical imagery from the Sentinel-2 mission and radar data from the Sentinel-1 mission. The methodology focuses on the application and comparison of spectral water indices (NDWI, SFMI, AWEIsh) applied to atmospherically corrected data from two processors: the standard Sen2Cor and the specialized ACOLITE correction. A thresholding method was used for the classification of water and flooded areas, and its accuracy was verified using statistical indicators derived from manually digitized reference areas. Based on the most appropriate approach, the flooding is then quantified. The results revealed redundancy between the SFMI and NDWI indices and demonstrated a significant advantage of the AWEIsh index in combination with the ACOLITE correction, which achieved the highest F1 score of 86.11 % in the overall classification. Sentinel-1 radar data proved suitable for detecting permanent water bodies but are considerably less reliable for flood mapping than optical methods. The total area of floods and floodwater detected within the city of Ostrava to the date of 18.9.2024 was 4.08 km², representing an 83.5 % increase relative to the permanent water area.

Description

Delayed publication

Available after

Subject(s)

floods, Ostrava, Sentinel-2, Sentinel-1, spectral indices, remote sensing, thresholding, flood mapping, ACOLITE

Citation