Aim: To describe the process for the conduct of a spatial analysis of hazards.

Introduction: Improvements to the rescue system require knowledge of incidents taking place in a given area, which need the intervention from rescue services. Such information ought to be scrupulously documented and inference methods should be plausible. Knowledge of issues is gained from past experience. This may be held on a database or derived from the experience of experts. Use of expert knowledge and experience is at times necessary, however, it is always weighed down with subjectivity. On the other hand, historical data is less subjective. The use of historical data requires an application of appropriate tools during the process of knowledge extraction and formulation of conclusions. It is possible to reveal a spatial dispersion of incidents from a focused statistical analysis of historical data, about fires and other local incidents, which occurred in a given area within a specified time frame. Prior research reveals that evolutionary change in the distribution density of critical incidents is very slow. This allows for an assumption that the density distribution of critical incidents, in a time frame of several years, is static. Therefore, the production of such an analysis can be very useful for planning purposes, to deal with issues concerning the distribution and capability levels of fire stations in context of applied standards for speed of response, to address critical incidents.

Methodology: Analysis and statistical forecasting.

Conclusions: Emergence of critical incidents in selected areas may be described with the aid of a stationery (in a particularly compartmentalized time frame) Poisson distribution. Forecasts based on this approach have significant credibility, subject to the proviso that the risk evaluation of a critical incident event will be made by aggregating forecasts from initial base areas. The smallest base area for such an analysis may be a square with an area of 1 km ² identified by topographic co-ordinates. The number of critical incidents should be counted for each square over a period of say, one year. The result will reveal a map with a historical density of critical incidents for an identified target area. Research reveals that by accepting an identified number of critical incidents in individual squares, as an anticipated number of critical incidents in a Poisson distribution, it is possible to forecast the number of critical incidents in the following year with a high level of confidence. Such an estimate, despite the simplistic approach, is a forecast with 90% credibility, a point, which this article sought to demonstrate.

Keywords: risk analysis, statistical analysis, statistical forecasting

Type of article: original scientific article