Scientific journal
Scientific Review. Geographical Sciences

Personal portfolio

OPTIMIZATION OF THE CONTEXTUAL ALGORITHM PARAMETERS ON FOREST FIRES DETECTION USING EARTH REMOTE SENSING DATA

Пономарчук А.И.
The contextual algorithm MOD14 (use of spectroradiometer MODIS data) for active fires detection uses calculation parameters adapted to conditions of a planet as a whole — so that to minimize a number of false alarms. In the northern boreal forests with tall vegetation and dense forest canopy method sensitivity sharply decreases, leading to numerous droppings of rather large fires in a wood zone. Work is devoted to optimizing parameters of the contextual algorithm to increase the detection reliability (number of confirmed fires to total number of detected thermal anomalies). Under existing conditions, the number of confirmed fires have tripled, with an overall accuracy of detection remained at the level of «standard» parameters. The basic gain of the confirmed fires has fallen to fires of a technogenic origin. This result is gained for threshold values of brightness temperature in the 21st channel of MODIS (range 4 mm) 295K/290K (day/night) while standard algorithm uses the values 310K/305K.