[Paper Review] Nowcasting Disaster Damage

Yury Kryvasheyeu, Haohui Chen+5 more|arXiv (Cornell University)|Apr 26, 2015

Public Relations and Crisis Communication45 references3 citations

TL;DR

This paper proposes using real-time Twitter activity to nowcast (predict in real time) disaster damage from large-scale events like Hurricane Sandy. By analyzing spatiotemporal patterns of hurricane-related tweets across 50 U.S. metropolitan areas, the authors demonstrate that per-capita Twitter activity strongly correlates with per-capita economic damage, enabling rapid, data-driven damage assessment during disasters.

ABSTRACT

Could social media data aid in disaster response and damage assessment? Countries face both an increasing frequency and intensity of natural disasters due to climate change. And during such events, citizens are turning to social media platforms for disaster-related communication and information. Social media improves situational awareness, facilitates dissemination of emergency information, enables early warning systems, and helps coordinate relief efforts. Additionally, spatiotemporal distribution of disaster-related messages helps with real-time monitoring and assessment of the disaster itself. Here we present a multiscale analysis of Twitter activity before, during, and after Hurricane Sandy. We examine the online response of 50 metropolitan areas of the United States and find a strong relationship between proximity to Sandy's path and hurricane-related social media activity. We show that real and perceived threats -- together with the physical disaster effects -- are directly observable through the intensity and composition of Twitter's message stream. We demonstrate that per-capita Twitter activity strongly correlates with the per-capita economic damage inflicted by the hurricane. Our findings suggest that massive online social networks can be used for rapid assessment ("nowcasting") of damage caused by a large-scale disaster.

Motivation & Objective

• To investigate whether social media activity can serve as a proxy for real-time disaster damage assessment.
• To examine how proximity to a hurricane's path influences social media engagement and message composition.
• To evaluate the predictive power of Twitter activity for economic damage at the metropolitan level.
• To develop a nowcasting framework using online social media for rapid disaster impact estimation.

Proposed method

• Collected and analyzed Twitter data from 50 U.S. metropolitan areas before, during, and after Hurricane Sandy.
• Mapped the spatiotemporal distribution of hurricane-related tweets to identify patterns of online activity.
• Calculated per-capita Twitter activity rates for each metropolitan area to normalize for population size.
• Correlated per-capita social media activity with official per-capita economic damage data from the National Hurricane Center.
• Used multiscale analysis to compare online behavior across regions with varying levels of physical impact.
• Employed statistical modeling to assess the strength and significance of the relationship between social media activity and actual damage.

Experimental results

Research questions

• RQ1 Can real-time social media activity serve as a reliable indicator of disaster damage?
• RQ2 How does the intensity and composition of social media messages relate to physical damage and perceived threat?
• RQ3 To what extent does proximity to a hurricane’s path predict social media engagement?
• RQ4 Is there a statistically significant correlation between per-capita Twitter activity and per-capita economic damage?
• RQ5 Can social media data be used to nowcast disaster impact with sufficient accuracy for emergency response?

Key findings

• There is a strong, statistically significant correlation between per-capita Twitter activity and per-capita economic damage caused by Hurricane Sandy.
• Twitter activity intensity increases in regions closer to the hurricane’s path, reflecting both real and perceived threats.
• The composition of disaster-related messages—such as mentions of evacuation, power outages, and damage—varies predictably with physical impact levels.
• The study demonstrates that social media data can be used to nowcast disaster damage in real time, offering a scalable alternative to traditional assessment methods.
• The predictive power of Twitter activity holds even after controlling for population size, indicating its utility as a normalized, real-time indicator.
• The findings suggest that online social networks can enhance situational awareness and support early warning and relief coordination during disasters.