UPDATING FOREST FIRE ZONING IN THE ARCTIC REGION OF SIBERIA IN CONTEMPORARY FIRE REGIMES
P. D. Tret’yakov1,2, E. I. Ponomarev1,2
1Krasnoyarsk Science Centre of the Siberian Branch of Russian Academy of Science, Krasnoyarsk, Russian Federation
2Krasnoyarsk Science Centre of the Siberian Branch of Russian Academy of Science, V. N. Sukachev Institute of Forest, Russian Academy of Sciences, Siberian Branch, Krasnoyarsk, Russian Federation
Keywords: vegetation fires, zoning, thermally active pixels, logistic regression, GIS analysis, probability of fire events, Arctic region of Siberia
Abstract
The article discusses approaches to updating the forest fire zoning of the Arctic region of Siberia based on geoinformation analysis and machine learning methods. The analysis of vegetation categories, spectral indices (NDVI), meteorological data series (air temperature, insolation, precipitation, thunderstorm activity) and relief categories is presented in relation to information on remotely recorded thermal anomalies («thermally active pixels») obtained from satellite monitoring data on vegetation fires for 2001-2023. Using the logistic regression method (nonlinear S-shaped coupling function), the weighting coefficients of predictors determining the probability of fire events were determined, among which the air temperature (β = 1.27) and the vegetation index NDVI (β = -0.53) were characterized by the greatest significance. Summarizing the influence of all the considered predictors, four zones are identified in the study area, characterizing the probability of fire events. Verification of the model based on independent data on thermally active pixels for 2024-2025 and the ROC analysis confirmed a high predictive ability (AUC = 84.3 %). It was found that 78.4 % of all thermal anomalies in the period 2001-2025 occurred in zones classified as highly and very likely to cause fires, occupying 24.4 % of the region’s area. A comparative analysis with the existing schemes of forest zoning states the need to shift the boundaries of the Siberian subarctic region (1A) to the north and expand the West Siberian zone (2A, 2B), which is determined by the current dynamics of fire regimes and, probably, economic development of the territories. The results can be used to optimize the system of monitoring and planning fire protection measures in the Arctic under the conditions of predicted climatic changes.
|
