Green roof detection using remote sensing and machine learning (Study case: Rotterdam)

M-GEO
ACQUAL
Staff Involved
Additional Remarks

Suggested elective courses: Radar remote sensing and advanced image analysis

Topic description

Access to safe drinking water, sanitation and hygiene (WASH) services and associated health benefits are widely enjoyed in high-income countries , where centuries of investment in infrastructure of services have reduced the incidence of infectious diseases. WASH access is closely tied to housing, leaving provision, financing, regulation, and maintenance of WASH to the responsibility of the property owners. Those who lack stable housing, including people experiencing homelessness in urban areas may, thus, be excluded from WASH services. For the hundreds of thousands of homeless people in Europe, homeless station missions, accommodation services and public restrooms are critical and sometimes the only option for accessing services, leaving them without year-round access, and exposed to water-related health risks. What is challenging under “normal” circumstances already, gets further complicated during extreme weather events: during heatwaves, more water for hydration and cooling may be needed; during flooding, public restrooms and services may be inaccessible; during extreme colds, water supply may be dysfunctional. People experiencing homelessness suffer particularly from infrastructural failures, and the resulting disease burden.

Despite the high societal relevance, the needs of homeless communities have not only not yet been fully met; WASH inequalities and resulting disease burden remain hidden in official statistics and understudied. And despite increasing frequency, intensity and unpredictability of extreme weather events, and evidence that extreme weather events significantly shape the daily experiences of people at risk of homelessness, implications on WASH infrastructure, and the consequences for people experiencing homelessness and their health have not been comprehensively studied either. This thesis aims to fill the combined knowledge gap in understanding the challenges that homeless communities in urban areas are facing with regard to WASH insecurity overall, specifically with regard to accessibility and usability of WASH during extreme weather events. It is part of the ITC Blue Sky Research project “Winter-resilient, heat-proof, flood-resilient WASH for those left behind in cities”.

Topic objectives and methodology

The main objective of this proposal is to use remote sensing data to detect and identify green roofs in Rotterdam. Given the advantages of synthetic aperture radar (SAR) data, which are weather-independent, some radar-derived vegetation indices can be very useful to identify green regions on the images. However, the spatial resolution of SAR data may limit the detection performance of green roofs. Therefore, very high resolution aerial photogrammetric data can also be used to further improve the detection capability. The idea is to use a set of data that includes SAR images, high-resolution photogrammetric images, and DSM to develop a machine learning method for green roof detection.