A statistical method for pre-estimating impacts from a disaster: A case study of floods in Kaduwela, Sri Lanka

Abstract

Quantification of impacts from a disaster is an important aspect in effectively managing the systemic risks arising from hazards. At present, damage assessments use complex modelling techniques which require time, capital, and a vast amount of data. This paper provides a scientific solution in finding an effective and rapid method to identify the extent of impact from a disaster event, by developing a statistical estimation model for disaster impacts using Multi-Criteria Analysis (MCA) and Pearson Correlation test. In the model developed, impacts of disasters are evaluated by considering the interconnected nature of systems and cascading nature of disasters, based on a Sri Lankan case study. Data from the case study, which denotes characteristics of floods, are used to correlate data categories containing details relevant to human activities in a disaster. Findings from the Pearson Correlation show strong correlations of 0.6–0.85 between modified variables. Also, the results of the MCA show that there are multiple correlations between the selected variables. These correlations can be used in deriving mathematical models to estimate human impacts. The models were tested for accuracy through a similarity test, by calculating the deviations of the model created, using the original figures. The model developed for calculating the affected population provided accurate outputs and a deviation of 0.98 when compared with the original data. The modelling process presented in this study could be further enhanced using additional data. Furthermore, it could be applied in analysing the rapid and approximate estimation of impacts on humans, due to events of disasters.