Urban Flood Vulnerability at Street Level

Access to high-resolution information on vulnerable populations is critical to be able to accurately assess susceptibility to flood hazards. In many countries, variables relevant to vulnerability are only offered at a national level while some have a selection of relevant variables available at various administrative levels. Here we provide a methodology to be used globally to extract building level vulnerability data by integrating remote sensing products. First, we focus on residential urban areas using remote sensing to specify specific locations to assess. Second, we identify building characteristics and observable indicators of vulnerability using a crowdsourcing approach from satellite electro-optical remote sensing and ground-based images along roads from Google Street View. This process is performed through a Mechanical Turk interface using a stratified sampling approach by area at the district level. High resolution vulnerability data can serve to better prioritize areas in disaster preparedness, response, recovery, and mitigation efforts.

Urban Flood Vulnerability at Street Level (127KB .xlsx file).

Here we provide a methodology to be used globally to extract building-level vulnerability data by integrating remote sensing products. This study presents a crowdsourcing methodology used by a team of expert GIS users to identify structural characteristics of buildings (e.g., building materials, roof type, number of floors) and of immediately surrounding infrastructure (e.g., street drains, street material, street slope) relevant to socioeconomic indicators of flood vulnerability. An open-source local computer version of Mechanical Turk (MTurk) was used as a platform to collect variables. An interface using the MTurk environment was implemented with two primary inputs: 1) a survey document that includes our variables of interest for collection, and 2) geographic coordinates associated with remote sensing imagery and Google Street View (GSV). The data collected using this methodology obeyed the following sections. a) Survey Design: it involves the consolidation of a survey instrument. We selected variables that are critical in flood risk assessment, particularly those occurring in urban settings. To ensure consistency in data collection during the turking process, we were faced with the need to standardize variables by the consolidation of a codebook. Finally, we developed a simple HTML script to collect these variables through the mechanical turk process. b) Selection of Sample Sites: Creation of random points by stratified sampling. Then, GIS users ensure that the point is near to any building location and has GSV coverage. c) MTurk Setup: Require the input of survey template and geographic coordinated sample points by the installation of a local version of MTurk by Danvk (2018). We ran the local MTurk module with the template HTML task script that pulls from the list of sample sites and includes both the AWS link to the footprint image and a link to GSV for the turkers to review the area further. d) Data Collection and Analysis: collected data at each geographic coordinate marked sample site point by examining the corresponding GSV image and recording variable characteristics using our survey. We evaluated the data collected for each variable through exploratory analysis, mapping, and statistical methods. Finally, we performed inter-rater agreement and intraclass correlation to evaluate consistency for a sample. A crowdsourcing methodology and high-resolution vulnerability data can serve to better prioritize areas in disaster preparedness, response, recovery, and mitigation efforts.

Data set:

Velez R., C. Aime, D. Calderon, L. Carey, N. Bermudez, Y. Gorokhovich, C. Hultquist, G. Yetman, R. S. Chen, and A. Kruczkiewicz. 2024. Urban Flood Vulnerability at Street Level, Preliminary Release . Palisades, New York: NASA Socioeconomic Data and Applications Center (SEDAC). https://doi.org/10.7927/jmdt-fv53. Accessed DAY MONTH YEAR.

Peer-reviewed publication:

In preparation.

This is a preliminary open data release, pending peer review of the data and associated journal articles. Following the peer review process, data curation will be completed by the NASA Socioeconomic Data and Applications Center (SEDAC) and the data will be disseminated through the SEDAC catalog.