The methods and results of seismic hazard zoning are investigated for the Russian Federation territory and abroad. The input data used in the zoning procedure, aimed at revealing the boundaries of areas with stable seismic intensity attenuation parameters have been analyzed. The zoning procedure has been developed for determining the boundaries of territories, within which the macroseismic field parameters (i.e., the coefficients in N.V. Shebalin’s equation; the orientation of the elliptic isoseist axes, as well as the ratio of their semi-axes) show close values in each point. Examples are given in distinguishing zones characterized by quasi-stable parameters for separate regions, as well as in calibrating the computer model of the macroseismic field. The case studies are considered in the presence of a complete set of input data and under conditions of their insufficiency. The efficiency of the macroseismic field calibrated models application in the Extremum system in order to increase the reliability of near real time earthquake loss estimations is shown.

Abstract. Impact database development and application for risk analysis and management promote the usage of self-learning computer systems with elements of artificial intelligence. Such system learning could be successful when the databases store the complete information about each event, parameters of the simulation models, the range of its application, and residual errors. Each new description included in the database could increase the reliability of the results obtained with application of simulation models. The calibration of mathematical models is the first step to self-learning of automated systems. The article describes the events' database structure and examples of calibrated computer models as applied to the impact of expected emergencies and risk indicator assessment. Examples of database statistics usage in order to rank the subjects of the Russian Federation by the frequency of emergencies of different character as well as risk indicators are given.

Abstract. Impact databases development and application for risk analysis and management promotes the usage of self-learning computer systems with elements of artificial intelligence. Such systems learning could be successful when the databases store the complete information about each event, parameters of the simulation models, the range of its application and residual errors. Each new description included in the database could increase the reliability of the results obtained with application of simulation models. The calibration of mathematical models is the first step to self-learning of automated systems. The article describes the events' database structure, and examples of calibrated computer models as applied to the impact of expected emergencies and risk indicators assessment. Examples of database statistics usage in order to rank the subjects of the Russian Federation by the frequency of emergencies of different character, as well as risk indicators are given.

No abstract

This paper is devoted to applications of the “Extremum” loss simulation system to two damaging earthquakes which occurred in Croatia in 2020. We provide a calibration procedure of mathematical models used for shaking intensity simulation. The regional macroseismic field parameters, such as the coefficients in the macroseismic field equation; the ratio between the longer (b) and the shorter (a) axes of the higher elliptical isoseismals (the flattening ratio k); the angle that specifies the orientation of the macroseismic field, in particular, the azimuth of the longer axis in the isoseismal ellipse, were all based on extensive macroseismic data acquired for the Balkan region and on the data for an analogous area with similar seismotectonic parameters in the Caucasus. We obtained a fairly good consistency between the results of simulation applied to the impact of the 2020 Croatia earthquakes and observations, confirming that the calibration of the macroseismic model by the Extremum system was both reasonable and effective for enhancing reliability for real time loss estimation.