Affecting factors of wildfire occurrence in Dadu Terrace

• Main Authors: Po-Cheng Wang, 王柏程
• Other Authors: 林昭遠
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/cdx9jx

碩士 === 國立中興大學 === 水土保持學系所 === 104 === Wildfire occurred frequently in Dadu Terrace. According to 2011-2015 firefighting records of Fire Bureau, grass fire occupied 70% is the main fire types of the Terrace. This shows that the grassland is vulnerable to wildfire in the Terrace. In addition, once wildfire site nearby transportation facilities and building, life and property could be endangered if without prevention in advance. Therefore, to delineate the temporal and spatial distribution of wildfire risk is an important issue. An assessment model of wildfire was established in this study to explore the temporal and spatial risk of wildfire occurrence. Normalized Difference Vegetation Index, Normalized Difference Water Index, biomass of Guinea grass and meteorological data were used to assess the weights of the affecting factors of wildfire occurrence using factor analysis and then the temporal wildfire vulnerability in grassland could be obtained. Moreover, spatial distribution of wildfire risk is delineated based on the month which with the most vulnerability. The wildfire risk is calculated from the potential hazard and the vulnerability of causing fire. The potential burned area derived from Normalized Burn Ratio and the distance to roads and/or graves were considered as the potential hazard, and the Normalized Difference Water Index, Normalized Difference Vegetation Index and fuel index were applied to estimate the vulnerability of causing fire. The model of wildfire risk was also verified with fire-event records of Fire Bureau. Results show that the vulnerability values of October to the following year’s April are greater than 0.5 which could be regarded as high-risk grade comparing with the other months, and the trends of fire vulnerability versus fire records show a little of inconsistency. The values of March to November are classified as low-risk grade which are consistent with fire records. This reveals that the vulnerability index is unable to reflect the real fire occurrence especially at the higher value due to hazard ignored. Considering the potential hazard coupled with the highest vulnerability value of January, a model of wildfire risk was constructed in this study which could reasonably reflect the real world.