Application of forest fire detection index in forest fires: a case study of Aberdare forest

Abstract

Forest fire detection is an important step in wildfire pre-suppression process. Many organizations involved in forest monitoring and surveillance strive to achieve efficient and faster methods to detect forest fires. This is due to colossal amount of money spent in forest restoration and rehabilitation after fire occurrences. Timely detection and efficient interpretation of the satellite and other terrestrial images allow fire personnel to contain the fire at its initial stage and this will reduce damages caused to both flora and fauna besides the suppression cost. It has been difficult to predict, detect and control forest fires in Kenya and in many cases different agencies are caught unawares in the instances of fires witnessed in the country. Forest fires spread faster due to the presence of dry vegetation covers especially during dry season and this requires a faster way of detection to avoid escalated damages as witnessed in the previous fire occurrences in our country; the latest being the one in Aberdare forest which formed the basis for this project. The motivation behind this project was to have a smooth means for forest fire detection through the use of a forest fire index that will help in detecting flames and smokes in the near-real time and to enable concerned parties to react quickly in case of forest fire. The method borrowed from the vegetation classification to detect distinct reflectance from both flame and smokes. Smoke has been difficult to detect because of the weak infrared radiations emitted at its production and this makes sensors not to detect it. In this project smokes was detected adaptively in the region of interest with the help of a variable factor. The project was carried out in phases; MODIS terra and aqua image acquisition from https://ladsweb.modaps.eosdis.nasa.gov/search/ and terrestrial image from drone on 30th March 2018, extraction of colour components, normalization of the colour components, using the forest fire detection index for smoke and flame detection and lastly and representing the output from colour analysis for smokes and flames in image forms. The method can be used in near-real time forest fire detection resulting in a more cost-effective outcome than the traditional systems involving water tanks and watch towers. This method is efficient in forest fire detection at the early stage when both satellite and drone images are made readily available.