Investigating the Use of Citizen-Science Data as a Proxy for Flood Risk Assessment in New York City

Abstract

New York City has an extensive waterfront and numerous low-lying neighborhoods that make it highly exposed to the impacts of extreme weather events. With global warming increasing the frequency and intensity of these events, the City government has created the Stormwater Flood Maps to prepare for future floods, manage the associated risk, and inform residents and government agencies on flood hotspots. However, these maps rely on complex Hydrological and Hydraulic (H&H) models, which require expensive computational resources and may not always accurately reflect the built environment, e.g., urban surface, the stormwater system, etc. This study investigates the potential of citizen science data, specifically 311 service requests, as a tool for understanding and quantifying flood risks. 311 service requests provide novel and unique information for urban flood risk characterization as they are community-generated, updated daily, and geo-referenced, allowing for near-real-time insights into the changing flood risk landscape across the city. By analyzing data from Hurricane Ida in 2021, we found a moderate correlation (Pearson coefficient ? = 0.56) between the number of 311 requests related to flooding and the level of risk, estimated using the Stormwater Flood Map. Further, we characterized performance variations of 311 data as a proxy for risk across the city, revealing that some high-risk census tracts had only a limited number of 311 service requests, while other low-risk areas recorded a high volume of requests. This result highlights potential inconsistencies between the traditional H&H models and actual flood conditions, and underscores the value of citizen-generated data in characterizing flood risk. Despite its advantages, 311 data also has some shortcomings: its reliability is influenced by the tendency of communities to utilize the service. To address these limitations, we outline future studies to improve the accuracy of this methodology.