Monitoring of forest conditions during fire danger

Abstract

A variety of information on forest ecosystem can be obtained by using remote sensing techniques nowadays. The use of space data of forest monitoring is economically beneficial, because it allows to quickly obtain objective information needed by forest users to solve practical problems. Satellite data provides wide coverage of forest lands, high accuracy of results, and high frequency of received data. Space images of the territory of Ovruch district of Zhytomyr region of Ukraine were selected for the study in the summer of 2020. The definition of the rock composition is carried out by methods of the guided classification, namely Bayesian classifier. It is found that 70% of forests are pine, the aspen, hornbeam, birch, alder and ash breeds of tree are found in smaller numbers. In Ukraine according to statistical data for 2000-2020, 51,4 thousand hectares of forest plantations were damaged and destroyed by forest fires. Therefore, objective and timely information on the fire effects is necessary for the solving a wide class of applied problems of forestry. Determining the area of damaged forests is an important task in assessing the ecological and the economic damage caused to forest management by forest fires. In the work technologies for defining the forest area, in which the fire occurred, are reviewed using the space images of Landsat 8 sattelite. For detecting burned areas and levels of damage a normalized NBR pre and post fire index and DNBR index are used. For forecasting forest fires a mathematical model based on Bayesian theorem and a quarterly thematic map with classes of fire danger are created. To check the accuracy of the results of the projected model, a combination of the thematic map with a fire layer of the defined areas is carried out. This software product is flexible and versatile enough, it can be easily adapted to use not only for the determination of burned forest lands, but also for other territories.