Author(s): Cibele Patricia Damasceno De Oliveira; Ana Carolina Pires Pereira; Rafaela Alves De Carvalho; Hugo Henrique Cardoso De Salis; Talita Fernanda Das Gracas Silva

Linked Author(s):

Keywords: Remote sensing; Sentinel-2; Social media and news; Urban flash floods; Google street view

Abstract: Floods have caused severe economic, social and environmental impacts in recent decades all over the world, especially in densely populated urban areas. It is expected that throughout the 21st century, the expansion of the urbanization process along with climate change will increase flooding economic damage, as well as the number of affected people. In Latin America, floods are a recurrent problem, especially in Brazil, which ranks among the 15 countries with the largest population exposed to the risk of flooding in the world. Advances in orbital remote sensing represent new opportunities for mapping flood damages thanks to the wide spatial coverage provided by satellites, satisfactory temporal and spectral resolution, in addition to the free availability of images from several missions. The launch of Sentinel-2 satellite in 2015 by the European Space Agency brought new perspectives for mapping urban areas impacted by floods thanks to its high spatial resolution (up to 10 m in the visible band), 13 multispectral bands and revisit frequency of 5 days. This study aims to investigate changes in the spectral response of urban areas affected by floods in Belo Horizonte (Brazil), where floods are characterized by their short duration, and by high impacts due to dense population and urban infrastructure. The methodology consisted in (1) news survey in digital media to identify flooding episodes that caused damage in different parts of the city; (2) definition of a specific region to study within the city based on the damages reported in the news; (3) download and processing of Sentinel-2 satellite images from before (12/18/2019) and after the flooding episodes (01/22/2020, 02/01/2020 and 03/12/2020) and assessment of cloud coverage in the study area; (4) comparison between spectral response of targets before the flooding, using a reference image, and after flood events using selected cloud-free images; (5) the validation of the analysis’ results by comparison with photos and videos picturing damages caused by flooding collected on the news and provided by the municipality of Belo Horizonte. The chosen area of study is Teresa Cristina Avenue, one of the main traffic axes in the city. Preliminary results have shown an increase in the mean reflectance value after the flooding for some tranches of the avenue. Photos, videos and Google Street View images exhibit mud traces and asphalt damages on these tranches after flooding episodes. The increase in the reflectance registered by the satellite in some segments of the avenue is related to soil transport and damage to the asphalt coverage caused by the flooding since unpaved roads and bare soil have higher reflectance than paved roads. Using Sentinel-2 imagery for identifying urban areas damaged by floods is a promising approach that may contribute for improving flood risk management in cities.

DOI: https://doi.org/10.3850/IAHR-39WC252171192022951

Year: 2022