Establishment and assessment of urban meteorological disaster emergency response capability based on modeling methods. (September 2022)

• Record Type: Journal Article
• Authors: Zhou, Si-Yu; Huang, An-Chi; Wu, Jie; Wang, Ying; Wang, Long-Shuai; Zhai, Juan; Xing, Zhi-Xiang; Jiang, Jun-Cheng; Huang, Chung-Fu
• Is Part Of: International journal of disaster risk reduction. Volume 79(2022)
• Publication Date: 2022-09
• Subjects: Quantitative assessment standard -- Multiple linear regression -- Artificial neural network -- Support vector machine -- Analytic hierarchy process; Emergency management -- Periodicals; Risk management -- Periodicals; Disaster relief -- Periodicals; Hazard mitigation -- Periodicals; 363.34
• DOI: 10.1016/j.ijdrr.2022.103180
• Languages: English
• ISSNs: 2212-4209

Abstract

Abstract: Emergency response capability assessments can be performed to determine the state of emergency management and identify weaknesses, thus increasing preparedness and strengthening emergency rescue capabilities. Multiple linear regressions (MLRs), support vector machines (SVMs), and artificial neural networks (ANNs) are widely used for assessment and prediction. However, few studies have used them to evaluate urban meteorological disaster emergency response capabilities. This study used these mathematical models to evaluate urban meteorological disaster emergency response capabilities, and the model verification parameters and deviations were compared. The analytic hierarchy process formulated a comprehensive weighting system and a quantitative assessment standard. The meteorological disaster cases from 2010 to 2020 were used as the data source, and then 50 sets of sample data were obtained to enhance the operability and scientificity. MLR, SVM, and ANN were used to establish an urban meteorological disaster emergency response capability assessment and prediction model. The results indicate that all three models are effective. SVMs (mean squared error = 0.1074) provide excellent prediction, MLRs (mean squared error = 0.1184) provides satisfactory prediction, and ANNs (mean squared error = 0.3211) provides poor prediction. The proposed estimation methods provide an effective predictive evaluation model that reflects the actual conditions and can enable the government and other groups to perform inspections conveniently. It can not only check one by one according to the quantitative evaluation standards but also continuously improve the emergency management ability according to the forecast results and then enhance the ability to resist meteorological disasters in time.