Natural emergency modeling and forecasting

Abstract

The purpose of this study is to develop forecast and analytical models (FAM) for predicting the most catastrophic natural emergencies caused by floods, earthquakes and forest fires. The article discusses predictive and analytical solutions for natural hazards for urbanized areas based on the Bayesian classifiers. The result is a formalized description of models for predicting forest fires, consequences of earthquakes and floods. The novelty of the models is due to the application of a unified scientific approach - the statistical processing method based on Bayes’ theorem. In contrast to the frequency probability determined by the relative frequency of occurrence of random events over sufficiently long observations, the Bayesian probability is the main forecasting method for constructing and training neural networks. Scientific forecasting of crises and incidents based on the Bayesian method and Bayesian networks requires a large amount of up-to-date data to model natural disasters, which is typical for frequently recurring negative events. Due to the lack of statistical data, the Bayesian method is not applicable to predicting catastrophic natural disasters that occur rarely but cause significant damage.