Almost all the states of the world community have territories along which railroad tracks pass. In a number of states, these railway lines pass through forests. As known, railway facilities are a source of anthropogenic load on forested areas. This causes the occurrence of forest fires near railway infrastructure. For medium-term and long-term prediction of forest fire danger, it is often necessary to process data on the forest fire retrospective. In this paper, the processing of statistical data on forest fires has been carried out by using cluster analysis. The purpose of the study is cluster analysis and processing of statistical data on forest fires near the infrastructure facilities of Russian Railways in the context of assessing and predicting forest fire danger. Main objectives: 1) collecting data from forestries; 2) preparing samples; 3) classifying two sets of samples; 4) analysing clusters obtained. In the work, data processing has been carried out for a typical territory of the Republic of Buryatia (Russian Federation), along which the railway passes. Hierarchical clustering algorithms have been used for data processing. The initial information for data processing has included meteorological parameters, forest stands description, statistical data on forest fires and data on the location of railway infrastructure facilities. By using the QGIS software, digital maps have been built showing the spatial distribution of forest fire clusters near the railway infrastructure, forest conditions and typical meteorological conditions. Based on the results of the work, conclusions and proposals have been formulated on the use of cluster analysis in predicting forest fire danger. An analysis of the spatial distribution of forest fire clusters over the territory of the Zaigraevsky district of the Republic of Buryatia shows that a large number of forest fire clusters are located near the railway infrastructure. In fact, railway facilities are a serious source of anthropogenic load. In the future, it is necessary to define the so-called buffer zones near railway infrastructure facilities, whose forested areas require careful attention when predicting forest fire danger. The results of cluster analysis confirm the data of descriptive analysis for a number of positions. For example, forest fires occur near railway infrastructure, which is explained by the presence of attractive points in forest areas due to logging, hunting, picking berries and mushrooms, as well as recreational pressure. As a result of the descriptive analysis, reasonable assumptions have been made and explanations have been given for the presence of local maxima of forest fires for the period 2016-2020, as well as during the fire danger season. The article also presents a plan for further research in this direction.
Read full abstract