Data stratification toward advanced flood waste estimation: A case study in South Korea

Area of Science:

• Environmental Science
• Disaster Management
• Data Science

Background:

• Effective flood waste management is crucial for mitigating disaster impact and preventing secondary pollution.
• Accurate waste quantity estimation is vital for efficient recovery planning and policy development.
• Existing disaster datasets often contain disparities that hinder precise waste prediction.

Purpose of the Study:

• To develop an advanced flood waste estimation technique using data stratification.
• To assess the effectiveness of different stratification characteristics and sequences on prediction accuracy.
• To provide a transferable strategy for improving disaster data utilization in waste management.

Main Methods:

• A total of 90 flood cases in South Korea were analyzed.
• Data stratification was applied based on administrative region (AR), urbanization rate (UR), and disaster type and coastal accessibility (DC).
• The impact of single-stage and successive stratifications, along with different stratification orders, on waste prediction was evaluated.

Main Results:

• Data stratification significantly improved flood waste prediction accuracy.
• Both single-stage and successive stratifications demonstrated effectiveness in enhancing predictions.
• The optimal stratification sequence for improved waste prediction was identified as urbanization rate, administrative region, and disaster type and coastal accessibility (UR, AR, DC).
• This specific stratification order led to enhanced waste prediction in 74 out of 90 cases.

Conclusions:

• Data stratification is an effective strategy for identifying similar contexts within disaster data and reducing disparities.
• The proposed stratification method enhances the accuracy of flood waste estimation, crucial for disaster recovery.
• This approach offers a valuable and transferable solution for countries facing challenges with disaster data gaps.