Where FR is the numerical index of fire risk, Vi the vegetation variable (with 110 classes), Hj indicates the proximity to human habitation (with 1-4 classes), S, indicates slope factor (with 1-4 classes) and Rk is road/fire line factor (with 15 classes). The subscripts i, j, k, 1 indicate sub-classes based on importance determining the fire risk.
After obtaining the fire risk map (Fig. 8) in Motichur range (part of Rajaji National Park) attempt was made to suggest response routes for extinguishing forest fires. The forest type maps obtained by using Remote Sensing data have been used to assign non-directional costs under different vegetation category and digital elevation model was used for giving directional costs using GDIRGRAD programme compatible to be used in ILWIS. Finally the GROUTES programme was used to trace final response route plan from the source i.e., forest range head quarter to high fire risk areas (Porwal, 1998). A final map showing response routes planned and dropped out fire risk map obtained in the beginning was developed.
I u>j fu HUrLlfl Hltf
1 LHI HIM CM
I u>j fu HUrLlfl Hltf
1 LHI HIM CM
Figure 8: Fire Risk Map (part of Rajaji National Park)
Forest fires cause significant damage to the forest ecosystem. In Central Himalayan region forest fires occur between April to June annually when the weather is hot and dry. Usually the south facing slopes are prone to fire due to direct sun insulation and inflammable litters of pine and dry deciduous trees at the forest floor. The presence of habitation, roads, footpaths etc., and their distance from such sites indicate an additional yardstick for the occurrence of forest fires. Extensive forest area during summer of 1995 in the Western and Central Himalaya drew wide attention of the forest managers and environmentalists. This study attempts to provide estimates about forest fire damaged areas using digital satellite remote sensing data in the Tehri district of Garhwal Himalaya (Pant et al., 1996).
The study area is characterized by hilly and mountainous terrain supporting varied forest types and composition controlled by altitude, variety of land use/ land cover types along with perpetual snow cover on the mountain peaks. Pine is the dominant forest type and is most susceptible to fire almost every year particularly near habitation.
The Indian Remote Sensing Satellite-1B, LISS-II (IRS-B, LISS-II) data of pre-fire and post fire period (1993-95) were studied and analysed digitally in the IBM RS/6000, EASI/PACE computer system. The supervised per pixel classification and digital enhancement approaches have been used to identify the forest fire affected areas along with other cover types. Prior to this digital geometric correction of satellite images have been done using 1:250,000 scale Survey of India topographical sheets and both the images were masked with respect to district boundary. Digital enhancement techniques facilitated to choose correct training sets for supervised classification technique using maximum likelihood classifier. The training sets were assigned based on the ground truth information collected from fire burnt areas and surrounding cover types. Out of the various enhancement techniques the best result has been obtained by making the colour composite image of IR, NDVI and intensity under equal stretching.
The total area affected under forest fire has been estimated as 910.01 km2 or 20.58% of total geographical area of 4421.26 km2. This includes forest burnt area as 168.88 km2 or 3.38% of total geographical area, partially burnt forest area (area under active fire) as 473.69 km2 or 10.71% of total geographical area and the partially burnt fallow land/grassland/scrub land as 267.44 km2 or 6.05%. The forested area identified under smoke plumes has been estimated as 130.96 km2 or 2.96% of total forests area. The overall accuracy of classification has been assessed as 88%.
Was this article helpful?