Exploring urban flooding incidence through spatial information

Abstract

Cities are vulnerable to local floods due to heavy rainfall. Urban flooding causes damage to buildings and contents, and also disturbs daily city activities as it entails drainage, transportation, and electricity interruptions. Urban flooding is expected to increase as climate change drives heavier rainfall events. Population and assets densification, as well as infrastructure aging, increasingly hamper cities from tackling pluvial flooding. Climate adaptation measures can help cities to face the challenge of heavier weather and urban flooding. Examples of those measures are: smart drainage maintenance and emergency responses, urban climate-proofing and retrofitting, and provision of real-time flooding information to citizens and government officials, among others. To plan and perform such measures it is required to know, and even predict before a heavy storm is onset, where, when, and why urban flooding occurs. This knowledge is not always available though. Required knowledge to design and implement adaptation measures against urban flooding is insufficient in cases such as Amsterdam and Rotterdam. In these cities, urban drainage models are limited to certain districts or uncalibrated; they cannot validly predict where or when the drainage system will surcharge or flood, and thus, they cannot be used for flood damage modeling. Moreover, urban flooding may not only depend on hydraulic parameters of underground drainage systems; other physical and socioeconomic
characteristics of the urban fabric may also influence the flooding likelihood at a particular urban location. Urban flooding can be better understood by using non-hydraulic and unconventional sources of information. Available public data, curated by statistics, cadastral, or municipal call-center services, can provide insights about urban flooding damage. Using mainstream technology, such as web, traffic, and smart-phone cameras, can also afford for valuable data about urban flooding impacts, which contributes to the development of climate adaptation measures in lowland cities. This dissertation aimed to determine the potential of such alternative data sources in better explaining urban flooding incidents. Employed methods combined techniques from geographic information systems, graph theory, community ecology, and computer vision. The exploration done in this research follows three main steps: testing previously proposed models, exploring currently available data sources, and evaluating the usefulness of attainable and affordable technology to gather key, nonexistent data about the timing, location, and extent of urban flooding incidents.