Emergency flood detection using multiple information sources: Integrated analysis of natural hazard monitoring and social media data

Shoyama, Kikuko; Cui, Qinglin; Hanashima, Makoto; Sano, Hiroaki; Usuda, Yuichiro

doi:10.1016/j.scitotenv.2020.144371

Cited by 48 publications

(16 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Recent papers focused on analysing in parallel several sources of information for deriving awareness on emergency events. In Shoyama et al 2021, the impact of extensive floods in Japan has been assessed using hydrogeological information and tweets for detecting event outbreaks and phases in the affected areas.…”

Section: Social Media As Early Warning and Alerting Systemsmentioning

confidence: 99%

TriggerCit: Early Flood Alerting using Twitter and Geolocation -- a comparison with alternative sources

Bono¹,

Pernici²,

Fernandez-Marquez³

et al. 2022

Preprint

View full text Add to dashboard Cite

Rapid impact assessment in the immediate aftermath of a natural disaster is essential to provide adequate information to international organisations, local authorities, and first responders. Social media can support emergency response with evidence-based content posted by citizens and organisations during ongoing events. In the paper, we propose TriggerCit: an early flood alerting tool with a multilanguage approach focused on timeliness and geolocation. The paper focuses on assessing the reliability of the approach as a triggering system, comparing it with alternative sources for alerts, and evaluating the quality and amount of complementary information gathered. Geolocated visual evidence extracted from Twitter by TriggerCit was analysed in two case studies on floods in Thailand and Nepal in 2021.

show abstract

Section: Social Media As Early Warning and Alerting Systemsmentioning

confidence: 99%

TriggerCit: Early Flood Alerting using Twitter and Geolocation -- a comparison with alternative sources

Bono¹,

Pernici²,

Fernandez-Marquez³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Detecting changes in a given sequence of data is a problem of critical importance in a variety of disparate fields and has been studied extensively in the statistics and econometrics literature for the past 40+ years. Recent applications include finding changes in terrorism-related online content (Theodosiadou et al, 2021), intrusion detection for cloud computing security (Aldribi et al, 2020), and monitoring emergency floods through the use of social media data (Shoyama et al, 2021), among many others. The general problem of change point detection may be considered from a variety of viewpoints; for instance, it may be considered in either "online" (sequential) and "offline" (retrospective) settings, under various types of distributional assumptions, or under specific assumptions on the type of change points themselves.…”

Section: Introductionmentioning

confidence: 99%

“…We first note the test statistics employed in Matteson and James (2014) are based on so-named degenerate U-statistics and therefore have a limit distribution that is rather complicated and was not given in their work. Note Matteson and James (2014) and other authors recently applying their method (e.g., Theodosiadou et al (2021), Shoyama et al (2021)) use the permutation method to obtain p-values. An infinite series expansion for the corresponding limiting statistic (with fixed d) was later given by Biau et al (2016) that depends on an an infinite sequence of eigenvalues λ 1 , λ 2 , .…”

Section: Introductionmentioning

confidence: 99%

Change point detection in high dimensional data with U-statistics

Boniece¹,

Horváth²,

Jacobs³

2022

Preprint

View full text Add to dashboard Cite

We consider the problem of detecting distributional changes in a sequence of high dimensional data. Our proposed methods are nonparametric, suitable for either continuous or discrete data, and are based on weighted cumulative sums of U-statistics stemming from Lp norms. We establish the asymptotic distribution of our proposed test statistics separately in cases of weakly dependent and strongly dependent coordinates as min{N, d} → ∞, where N denotes sample size and d is the dimension, and also provide sufficient conditions for consistency of the proposed test procedures under a general fixed alternative with one change point. We further assess finite sample performance of the test procedures through Monte Carlo studies, and conclude with two applications to Twitter data concerning the mentions of U.S. Governors and the frequency of tweets containing social justice keywords.

show abstract

“…Currently, few studies on the combination of social media and other data sources have been produced. It thus remains unclear how social media data can (i) be effectively integrated with hazard-monitoring data and (ii) provide emergency managers with appropriate information for better land-use planning and early warning support 37 .…”

mentioning

confidence: 99%

Integrating multiple information sources for landslide hazard assessment: the case of Italy

Franceschini

Rosi

Soldato

et al. 2022

Sci Rep

View full text Add to dashboard Cite

Landslides are the most frequent and diffuse natural hazards in Italy causing the greatest number of fatalities and damage to urban areas. The integration of natural hazard information and social media data could improve warning systems to enhance the awareness of disaster managers and citizens about emergency events. The news about landslide events in newspapers or crowdsourcing platforms allows fast observation, surveying and classification. Currently, few studies have been produced on the combination of social media data and traditional sensors. This gap indicates that it is unclear how their integration can effectively provide emergency managers with appropriate knowledge. In this work, rainfall, human lives, and earmarked fund data sources were correlated to “landslide news”. Analysis was applied to obtain information about temporal (2010–2019) and spatial (regional and warning hydrological zone scale) distribution. The temporal distribution of the data shows a continuous increase from 2015 until 2019 for both landslide and rainfall events. The number of people involved and the amount of earmarked funds do not exhibit any clear trend. The spatial distribution displays good correlation between “landslide news”, traditional sensors (e.g., pluviometers) and possible effects in term of fatalities. In addition, the cost of soil protection, in monetary terms, indicates the effects of events.

show abstract

Emergency flood detection using multiple information sources: Integrated analysis of natural hazard monitoring and social media data

Cited by 48 publications

References 27 publications

TriggerCit: Early Flood Alerting using Twitter and Geolocation -- a comparison with alternative sources

TriggerCit: Early Flood Alerting using Twitter and Geolocation -- a comparison with alternative sources

Change point detection in high dimensional data with U-statistics

Integrating multiple information sources for landslide hazard assessment: the case of Italy

Contact Info

Product

Resources

About