A GIS-Based Support Vector Machine Model for Flash Flood Vulnerability Assessment and Mapping in China

Junnan Xiong, Jin Li, Weiming Cheng, Nan Wang, Liang GuoISPRS Int. J. Geo Inf 2019 · DOI · OpenReview

TL;DR

This paper presents a novel GIS-based approach using Support Vector Machines (SVM) to map flash flood vulnerability across China. By integrating exposure and disaster reduction capabilities, the study identifies high-risk areas, offering crucial insights for urban planning and disaster management.

What Problem Does This Paper Address?

Flash floods are a significant natural disaster globally, causing substantial loss of life and property. China, in particular, is highly susceptible due to its diverse geography, extreme weather events, and intensive land use. While flash flood vulnerability assessments exist, comprehensive national-level analyses for China have been lacking. This gap hinders effective disaster prevention and mitigation strategies, including informed urban planning and resource allocation. This study aims to fill this gap by providing a detailed, spatially explicit vulnerability assessment for flash floods across China.

Methods

The authors developed a multi-dimensional index system to assess flash flood vulnerability, categorizing indicators into "exposure" and "disaster reduction capability." They employed the Analytic Hierarchy Process (AHP) to determine the relative importance of these indicators. The core of their methodology lies in using a Support Vector Machine (SVM) model, a powerful machine learning algorithm, to integrate these factors and predict vulnerability. The SVM model was chosen for its ability to handle complex, non-linear relationships and its robustness in dealing with incomplete information, which is common in disaster assessment. The study also utilized spatial autocorrelation analysis (Moran's I) to understand the spatial clustering of vulnerability.

Datasets

The study utilized a comprehensive set of datasets, primarily sourced from the National Mountain Flood Disaster Investigation Project (NMFDIP) and other scientific data platforms. Key data categories included:

  • Exposure Data: Information on roads, buildings, flood control projects, enterprises, institutions, monitoring and warning facilities, and river networks.
  • Socio-economic Data: Population density, GDP, and land use types.
  • Environmental Data: Topography (Digital Elevation Model - DEM), soil, and vegetation.
  • Geomorphological Data: Regionalization data.

These datasets were collected at various resolutions, with the NMFDIP providing extensive data from 2013-2016. The study used county-level administrative units as the basic assessment unit.

Key Insights and Ideas

  • Exposure and Disaster Reduction are Key: The study reinforces the understanding that vulnerability is a function of both how exposed a region is to a hazard and its capacity to reduce or cope with that hazard.
  • SVM's Power in Complex Systems: The application of SVM highlights its effectiveness in modeling complex, multi-factorial phenomena like disaster vulnerability, where traditional linear models might fall short.
  • Spatial Patterns Matter: The detailed mapping of vulnerability reveals distinct regional disparities, emphasizing that a one-size-fits-all approach to disaster management is insufficient.
  • Uncoordinated Development: A critical finding is the potential disconnect between socio-economic development and disaster reduction capabilities in some regions, suggesting that rapid development doesn't automatically translate to increased resilience.

Key Results

The study classified China's counties into five vulnerability grades: extremely low, low, moderate, high, and extremely high.

  • High Vulnerability Areas: Approximately 10.5% of the land area (around 100 million hectares) falls into the high and extremely high vulnerability categories. These areas are predominantly located in the south and southeast regions of China.
  • Moderate Vulnerability: 19.18% of the land area (around 180 million hectares) exhibits moderate vulnerability, concentrated in central, northern, and southwestern China.
  • Low Vulnerability: A significant portion, 39.43% of the land area (around 370 million hectares), shows low or extremely low vulnerability, primarily in the northwest regions.
  • Spatial Agglomeration: Local spatial autocorrelation analysis identified "High-High" clusters of vulnerability in provinces like Zhejiang, Fujian, Jiangxi, Hunan, Guangxi, Chongqing, and Beijing.

Limitations and Future Work

The authors acknowledge several limitations:

  • Data Availability: The disaster reduction capability indicators were limited to susceptibility and coping ability due to a lack of national data on resilience.
  • Spatial and Temporal Resolution: The study's spatial resolution and temporal scale might limit its fine-grained applicability.
  • Missing Factors: Factors like disaster intensity and density, while suggested as important, were not fully incorporated into this specific assessment.

For future work, the authors suggest:

  • Dynamic Models: Developing dynamic flood risk assessment models that build upon the established framework.
  • Additional Factors: Incorporating more comprehensive indicators such as population age structure, cultural level, existing medical conditions, risk density, and disaster intensity for more localized assessments.
  • Advanced Methods: Exploring more complex variables and methods to assess resilience and adaptability.
  • Policy Integration: Using the vulnerability maps to guide urban planning, building distribution, and overall flash flood management strategies.

Summary updated

Abstract

Flash floods are one of the natural disasters that threaten the lives of many people all over the world every year. Flash floods are significantly affected by the intensification of extreme climate events and interactions with exposed and vulnerable socio-economic systems impede regional development processes. Hence, it is important to estimate the loss due to flash floods before the disaster occurs. However, there are no comprehensive vulnerability assessment results for flash floods in China. Fortunately, the National Mountain Flood Disaster Investigation Project provided a foundation to develop this proposed assessment. In this study, an index system was established from the exposure and disaster reduction capability categories, and is based on analytic hierarchy process (AHP) methods. We evaluated flash flood vulnerability by adopting the support vector machine (SVM) model. Our results showed 439 counties with high and extremely high vulnerability (accounting for 10.5% of the land area and corresponding to approximately 100 million hectares (ha)), 571 counties with moderate vulnerability (accounting for 19.18% of the land area and corresponding to approximately 180 million ha), and 1128 counties with low and extremely low vulnerability (accounting for 39.43% of the land area and corresponding to approximately 370 million ha). The highly-vulnerable counties were mainly concentrated in the south and southeast regions of China, moderately-vulnerable counties were primarily concentrated in the central, northern, and southwestern regions of China, and low-vulnerability counties chiefly occurred in the northwest regions of China. Additionally, the results of the spatial autocorrelation suggested that the “High-High” values of spatial agglomeration areas mainly occurred in the Zhejiang, Fujian, Jiangxi, Hunan, Guangxi, Chongqing, and Beijing areas. On the basis of these results, our study can be used as a proposal for population and building distribution readjustments, and the management of flash floods in China.
BibTeX
@article{xiong2019gis,
  author = {Junnan Xiong and Jin Li and Weiming Cheng and Nan Wang and Liang Guo},
  title = {{A GIS-Based Support Vector Machine Model for Flash Flood Vulnerability Assessment and Mapping in China}},
  journal = {ISPRS Int. J. Geo Inf.},
  year = {2019},
  doi = {10.3390/ijgi8070297},
}
Figure