COMPARATIVE ESTIMATION OF THE QUALITY OF FIRE DANGER INDEXES IN FORESTS
R. V. Kotelnikov, A. N. Chugaev
The Center of Forest Pyrology, Development of Forest Ecosystem Conservation, Forest Protection and Regeneration Technologies - Branch of the All-Russian Scientific-Research Institute of Silviculture and Mechanization of Forestry, Krasnoyarsk, Russian Federation
Keywords: forest fire frequency, correlation, interactive panel
Abstract
Weather conditions are one of the important factors, which affects forest fire danger. Taking into consideration different methodical approaches (in particular, to accounting for precipitation) and differences between forestry and natural climatic conditions, also different density the weather stations, the values calculated according by these methods differ significantly in different forest areas. In order to correctly estimate the methods quality and correctly interpret the results which were obtained, it was developed special toolkit allowing to do the necessary preliminary processing and data visualization in the form of an interactive panel. Existing approaches were based on linear Pearson’s correlation coefficient. At the same time this criterion can be applicable in case when the source data are distributed according to normal distribution law. ISDM-Rosleskhoz formed Big Data, which show it is not true. The new approach differs in that the source data are previously transform logarithm method, which increases accuracy of the obtained estimations. The developed toolkit allowed to conduct comparative analysis of the main methods, which are used in Russia. Method of humidity indicator taking into account hygroscopicity (PVG) shows the best results - (30 %), method of humidity indicator 2 (PV-2) is in the second place (26 %), method of humidity indicator 1 (PV-1) is located in the third place (20 %), Nesterov’s method is in the last place (13 %). The number of forest fires doesn’t allow to obtain reliable result in the other area (11 %). The prepared interactive map by the authors in combination with a dynamic graph allows to visualize comparative results to the forest areas level, inside subjects of the Russian Federation thereby simplifying interpretation the obtain results. The total results can be used for improvement forest fire danger scales depending on weather conditions.
|
