Forest-fire detection by means of lidar.

Abstract

ABSTRACT: The fundamentals of smoke sensing based on a single-wavelength lidar technique are discussed and experimental results of forest-fire smoke detection using a Nd:YAG laser lidar reported. The investigations include tracing the smoke-plume evolution, estimating the alarm promptness, smoke-plume location and profiling via azimuth scanning, and detection of fires out of line-of-sight and in extremely unfavorable visibility conditions. The possibility of automated lidar-based fire surveillance is discussed on the basis of the results obtained. Finally, a neural-network technique for the automatic recognition of smoke-plume signatures in lidar signals is considered and several neural-network architectures and learning rules tested and compared. The best neural-network algorithm lead to 22 false alarms and 27 misdetections out of ~1.7×104 patterns containing in 231 complete lidar signals, of which 122 had smoke signatures. 1