Spatiotemporal Heterogeneity Analysis of Factors Affecting Fire Occurrence in Forest and Wildland-Urban Interface

Abstract

Forest fire is a major issue faced by many countries worldwide. It’s a kind of disaster that is sudden, destructive and difficult to handle and rescue, and can pose a major threat to ecosystems, society and humans. Identifying the dominant factors leading to forest fires and understanding the spatiotemporal distribution patterns of driving factors in a particular area are crucial for the effective prevention and management of such disasters. Research conducted across multiple temporal scales holds unique guiding significance for both short-term and long-term fire prevention efforts. However, current academic research pays little attention to the spatiotemporal relationship between forest fires and their driving factors, with an even more significant lack of systematic research at different temporal scales. The wildland-urban interface, as an extension of forest area, displays unique characteristics in terms of combustible material composition and distribution. While the driving factors of wildland-urban interface fires differ to some extent from those affecting forest fires solely within forests, there is no unified definition of the wildland-urban interface in China. Based on the above, this thesis investigates the spatiotemporal correlation between forest fire occurrence and driving factors at the inter-annual scale considering spatiotemporal heterogeneity and identifies the dominant factors in Anhui province. Considering the seasonal characteristics of forest fires, this thesis analyzes the spatiotemporal correlation between forest fire occurrences and its driving factors at the inter-monthly scale and determines the dominant factors separately for the fire season and non-fire season. Additionally, this thesis compares the similarities and differences between the dominant factors driving forest fire occurrences at the inter-annual and inter-monthly scale. Based on the study of driving factors of inter-monthly forest fire occurrences, this thesis further reveals the spatiotemporal heterogeneity of factors driving wildland-urban interface fire occurrences at the inter-monthly scale and identifies the dominant factors separately for the fire season and non-fire season. Moreover, this thesis compares the similarities and differences between the dominant factors driving forest fires and wildland-urban interface fires at the same temporal scale. The main research contents and findings of the study are as follows:

(1)This work proposed a spatiotemporal model of forest fire occurrences based on the spatiotemporal information of forest fires and their driving factors at the inter-annual scale, using Geographically Temporally Weighted Regression (GTWR) techniques. The results are visualized using ArcGIS, revealing the spatiotemporal correlations between forest fires and changes in vegetational, topographical, meteorological, and socioeconomic factors. Innovatively, nighttime light is proposed as a socioeconomic factor influencing forest fire occurrences. This work find that nighttime light is closely related to forest fire occurrences. Compared to GDP, which is traditionally used as a socioeconomic factor in previous studies, nighttime light, with its high spatiotemporal resolution, independence, and objectivity, improves the performance of the research model. This work determines that socioeconomic factors have the greatest impact on forest fire occurrences at the inter-annual scale, providing a reference for long-term forest fire risk zoning and the formulation of forest fire management policies.

(2)This work improves the inter-annual forest fire occurrence model to better reflect the seasonal characteristics of forest fires at the inter-monthly scale. It was found that meteorological factors have the strongest dominant influence during the fire season, while socioeconomic factors have the strongest dominant impact during the non-fire season. The dominant effect of individual factors at the inter-monthly forest fire occurrences is similar but not identical across different seasons. Socioeconomic factors are important in both the fire prevention and non-fire prevention seasons. During the fire season, the relatively important meteorological factors are land surface temperature and relative humidity, whereas during the non-fire season, land surface maximum temperature and sunshine hours are more significant.

(3)This work reveals the spatiotemporal heterogeneity of wildland-urban interface fires occurrence at the inter-monthly scale based on the inter-monthly forest fire occurrence model. The results show that socioeconomic and meteorological factors play more dominant effect on wild-land interface fires occurrence than topographical and vegetation factors regardless of the season. The dominant influence of individual factors on wild-land interface fires occurrence is similar but not identical across different seasons. During the fire season, socioeconomic factors have a stronger dominant influence than meteorological factors, while during the non-fire season, meteorological factors have a stronger dominant impact than socioeconomic factors. This suggests that the current division of the fire season may not be applicable to wild-land interface fires. The study highlights the similarities and differences in the dominant factors driving inter-monthly forest fires and inter-monthly interface fires, providing a scientific basis for developing targeted fire prevention and management policies for different regions.

Date of Award	28 Jul 2024
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Siu Ming LO (Supervisor) & Jiping Zhu (External Supervisor)