Evacuation zone modeling under climate change: A data-driven method

Kun Xie, Kaan Ozbay, Yuan Zhu, Hong Yang

Research output: Contribution to journalArticle

Abstract

Predetermined evacuation zones can be used to estimate the demand of evacuees, which is helpful in assessing the resilience of transportation systems in the presence of natural disasters. Evacuation zones defined based on current road networks and environmental and demo-economic characteristics of a region cannot remain the same in the future because long-term climate change such as the rise of sea level would have major impacts on hurricane-related risks. Traditional methods for the prediction of future evacuation zones rely heavily on the storm surge models and could be time-consuming and costly to use. This study develops a novel grid cell-based data-driven method that can predict future evacuation zones under climate change without running the expensive storm surge models. The map of Manhattan, which is the central area of New York City, was uniformly split into 45 × 45 m2 grid cells as the basic geographical units of analysis. A decision tree and a random forest were used to capture the relationship between grid cell-specific features, such as geographical features, evacuation mobility, and demo-economic features, and current zone categories that could reflect the risk levels during hurricanes. Tenfold cross validation was used to evaluate model performance and it was found that the random forest outperformed the decision tree in term of the accuracy and kappa statistic. The random forest was used to predict the delineation of evacuation zones in the 2050s and 2090s based on the predicted sea-level rises and changes of demo-economic features. Compared with the current zoning, the areas with a need for evacuation are expected to expand in the future. The proposed method can be used to promptly estimate the future evacuation zones under different sea-level rise scenarios and can provide the convenience to assess transportation system resilience in the context of climate change.

Original languageEnglish (US)
Pages (from-to)1-9
Number of pages9
JournalJournal of Infrastructure Systems
Volume23
Issue number4
DOIs
StatePublished - Dec 1 2017

Fingerprint

Sea level
Climate change
Hurricanes
Decision trees
Economics
Zoning
Disasters
Statistics

Keywords

  • Climate change
  • Emergency management
  • Evacuation zone
  • Hurricane
  • Random forest
  • Resilience

ASJC Scopus subject areas

  • Civil and Structural Engineering

Cite this

Evacuation zone modeling under climate change : A data-driven method. / Xie, Kun; Ozbay, Kaan; Zhu, Yuan; Yang, Hong.

In: Journal of Infrastructure Systems, Vol. 23, No. 4, 01.12.2017, p. 1-9.

Research output: Contribution to journalArticle

Xie, Kun ; Ozbay, Kaan ; Zhu, Yuan ; Yang, Hong. / Evacuation zone modeling under climate change : A data-driven method. In: Journal of Infrastructure Systems. 2017 ; Vol. 23, No. 4. pp. 1-9.
@article{1b1399bf02764e07a2c5aaad4c2a726b,
title = "Evacuation zone modeling under climate change: A data-driven method",
abstract = "Predetermined evacuation zones can be used to estimate the demand of evacuees, which is helpful in assessing the resilience of transportation systems in the presence of natural disasters. Evacuation zones defined based on current road networks and environmental and demo-economic characteristics of a region cannot remain the same in the future because long-term climate change such as the rise of sea level would have major impacts on hurricane-related risks. Traditional methods for the prediction of future evacuation zones rely heavily on the storm surge models and could be time-consuming and costly to use. This study develops a novel grid cell-based data-driven method that can predict future evacuation zones under climate change without running the expensive storm surge models. The map of Manhattan, which is the central area of New York City, was uniformly split into 45 × 45 m2 grid cells as the basic geographical units of analysis. A decision tree and a random forest were used to capture the relationship between grid cell-specific features, such as geographical features, evacuation mobility, and demo-economic features, and current zone categories that could reflect the risk levels during hurricanes. Tenfold cross validation was used to evaluate model performance and it was found that the random forest outperformed the decision tree in term of the accuracy and kappa statistic. The random forest was used to predict the delineation of evacuation zones in the 2050s and 2090s based on the predicted sea-level rises and changes of demo-economic features. Compared with the current zoning, the areas with a need for evacuation are expected to expand in the future. The proposed method can be used to promptly estimate the future evacuation zones under different sea-level rise scenarios and can provide the convenience to assess transportation system resilience in the context of climate change.",
keywords = "Climate change, Emergency management, Evacuation zone, Hurricane, Random forest, Resilience",
author = "Kun Xie and Kaan Ozbay and Yuan Zhu and Hong Yang",
year = "2017",
month = "12",
day = "1",
doi = "10.1061/(ASCE)IS.1943-555X.0000369",
language = "English (US)",
volume = "23",
pages = "1--9",
journal = "Journal of Infrastructure Systems",
issn = "1076-0342",
publisher = "American Society of Civil Engineers (ASCE)",
number = "4",

}

TY - JOUR

T1 - Evacuation zone modeling under climate change

T2 - A data-driven method

AU - Xie, Kun

AU - Ozbay, Kaan

AU - Zhu, Yuan

AU - Yang, Hong

PY - 2017/12/1

Y1 - 2017/12/1

N2 - Predetermined evacuation zones can be used to estimate the demand of evacuees, which is helpful in assessing the resilience of transportation systems in the presence of natural disasters. Evacuation zones defined based on current road networks and environmental and demo-economic characteristics of a region cannot remain the same in the future because long-term climate change such as the rise of sea level would have major impacts on hurricane-related risks. Traditional methods for the prediction of future evacuation zones rely heavily on the storm surge models and could be time-consuming and costly to use. This study develops a novel grid cell-based data-driven method that can predict future evacuation zones under climate change without running the expensive storm surge models. The map of Manhattan, which is the central area of New York City, was uniformly split into 45 × 45 m2 grid cells as the basic geographical units of analysis. A decision tree and a random forest were used to capture the relationship between grid cell-specific features, such as geographical features, evacuation mobility, and demo-economic features, and current zone categories that could reflect the risk levels during hurricanes. Tenfold cross validation was used to evaluate model performance and it was found that the random forest outperformed the decision tree in term of the accuracy and kappa statistic. The random forest was used to predict the delineation of evacuation zones in the 2050s and 2090s based on the predicted sea-level rises and changes of demo-economic features. Compared with the current zoning, the areas with a need for evacuation are expected to expand in the future. The proposed method can be used to promptly estimate the future evacuation zones under different sea-level rise scenarios and can provide the convenience to assess transportation system resilience in the context of climate change.

AB - Predetermined evacuation zones can be used to estimate the demand of evacuees, which is helpful in assessing the resilience of transportation systems in the presence of natural disasters. Evacuation zones defined based on current road networks and environmental and demo-economic characteristics of a region cannot remain the same in the future because long-term climate change such as the rise of sea level would have major impacts on hurricane-related risks. Traditional methods for the prediction of future evacuation zones rely heavily on the storm surge models and could be time-consuming and costly to use. This study develops a novel grid cell-based data-driven method that can predict future evacuation zones under climate change without running the expensive storm surge models. The map of Manhattan, which is the central area of New York City, was uniformly split into 45 × 45 m2 grid cells as the basic geographical units of analysis. A decision tree and a random forest were used to capture the relationship between grid cell-specific features, such as geographical features, evacuation mobility, and demo-economic features, and current zone categories that could reflect the risk levels during hurricanes. Tenfold cross validation was used to evaluate model performance and it was found that the random forest outperformed the decision tree in term of the accuracy and kappa statistic. The random forest was used to predict the delineation of evacuation zones in the 2050s and 2090s based on the predicted sea-level rises and changes of demo-economic features. Compared with the current zoning, the areas with a need for evacuation are expected to expand in the future. The proposed method can be used to promptly estimate the future evacuation zones under different sea-level rise scenarios and can provide the convenience to assess transportation system resilience in the context of climate change.

KW - Climate change

KW - Emergency management

KW - Evacuation zone

KW - Hurricane

KW - Random forest

KW - Resilience

UR - http://www.scopus.com/inward/record.url?scp=85018741518&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85018741518&partnerID=8YFLogxK

U2 - 10.1061/(ASCE)IS.1943-555X.0000369

DO - 10.1061/(ASCE)IS.1943-555X.0000369

M3 - Article

AN - SCOPUS:85018741518

VL - 23

SP - 1

EP - 9

JO - Journal of Infrastructure Systems

JF - Journal of Infrastructure Systems

SN - 1076-0342

IS - 4

ER -