Estimation of Forest Fire Spread by Modeling in GIS Platform

摘要

Fire has been a source of disturbance for thousands of years. It has been found through research that forest fires mostly happen due to human interactions with the forests. The disturbances from humans mostly affect the forest areas that are nearer to the human habitations and roads. Statistical estimates infer that Tamil Nadu is one of the Indian states which is also frequently disturbed by forest fires. A spatial extent of about 4230.94 km2 of Tamil Nadu is vulnerable to human interventions from all the roads and settlements that can cause forest fires. This means that, no afforestation practice in future should be carried out within these areas. Instead, such practices in future could be encouraged in areas wherein there is ‘zero’ risk due to anthropogenic threats. This is essential to ensure the safer survival of any artificial plantations. If ‘Forest Fire Spread Model’ is prepared for these areas, the same will prove to be helpful in finding out the direction of fire in the event of a fire-spread from a spot. For any forest area, the factors that is responsible for the spread of a fire are, Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Aspect, Slope, and Land Surface Temperature (LST). Objective: To prepare a ‘forest fire spread model’ that could be used for predicting the direction of fire movement and estimating the area of fire spread after a stipulated time. Method: In this research, firstly a ‘Fire reach susceptibility index’ map has been prepared which could then be fed to a model that has be prepared using Python customizing facility in QGIS, to estimate the ‘forest fire spread’ in a course of time. The factors that have been considered in this study were Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Aspect, Slope, and Land Surface Temperature (LST). Findings: The ‘FRSI’ Map preparation phase considers all the necessary factors as suggested by the previously conducted researches. The hence prepared map has also been undergone through ‘accuracy assessment’ to ensure its reliability in its usage for the estimation of forest fire spread. This has been done by checking the correlation between the number of fire incidences and each district’s corresponding average FRSI value. This map and the number of fire incidences have a correlation coefficient of ‘0.94’, thereby making this map to achieve an accuracy of 97.43%. Thus, the average FRSI value and the number of fire incidences are relevant, which means that the prepared FRSI map is reliable. With the help of this map, ‘fire spread model’ has been prepared from which it has been estimated that, for a safer region in Tirunelveli district, the fire could spread to an extent of 38.56 km2 in 9 hours. Application: The ‘Forest Fire Spread Model’ prepared for these areas would prove to be helpful in finding out the direction of fire in the event of a fire-spread from a spot. Hence, this model could be useful to ‘Tamil Nadu Forest Department’ in predicting the most expected spot of arrival of fire in a time duration. This could ensure that plantations nearby are safeguarded from any possible adverse effects of human interventions.