Wildfire assessment using machine learning algorithms in different regions

被引:7
作者
Moghim, Sanaz [1 ]
Mehrabi, Majid [1 ]
机构
[1] Sharif Univ Technol, Dept Civil Engn, Azadi Ave, Tehran, Iran
基金
美国海洋和大气管理局;
关键词
Wildfire; Fire regime; Machine learning; Random forest; Logistic regression; Fire Susceptibility map; Conservation ecology; SUPPORT VECTOR MACHINE; FOREST-FIRE DANGER; LOGISTIC-REGRESSION; SPATIAL-PATTERN; CLIMATE-CHANGE; BOREAL FOREST; SUSCEPTIBILITY; CANADA; PREDICTION; PROVINCE;
D O I
10.1186/s42408-024-00335-2
中图分类号
Q14 [生态学(生物生态学)];
学科分类号
071012 ; 0713 ;
摘要
BackgroundClimate change and human activities are two main forces that affect the intensity, duration, and frequency of wildfires, which can lead to risks and hazards to the ecosystems. This study uses machine learning (ML) as an effective tool for predicting wildfires using historical data and influential variables. The performance of two machine learning algorithms, including logistic regression (LR) and random forest (RF), to construct wildfire susceptibility maps is evaluated in regions with different physical features (Okanogan region in the US and Jam & eacute;sie region in Canada). The models' inputs are eleven physically related variables to output wildfire probabilities.ResultsResults indicate that the most important variables in both areas are land cover, temperature, wind, elevation, precipitation, and normalized vegetation difference index. In addition, results reveal that both models have temporal and spatial generalization capability to predict annual wildfire probability at different times and locations. Generally, the RF outperforms the LR model in almost all cases. The outputs of the models provide wildfire susceptibility maps with different levels of severity (from very high to very low). Results highlight the areas that are more vulnerable to fire. The developed models and analysis are valuable for emergency planners and decision-makers in identifying critical regions and implementing preventive action for ecological conservation. AntecedentesEl cambio clim & aacute;tico y las actividades humanas son dos de las fuerzas principales que afectan la intensidad, duraci & oacute;n, y frecuencia de los incendios, lo que puede conducir a riesgos e incertidumbres en los ecosistemas. Este estudio us & oacute; la t & eacute;cnica de aprendizaje autom & aacute;tico (machine learning) como una herramienta efectiva para predecir incendios de vegetaci & oacute;n usando datos hist & oacute;ricos y variables influyentes. La performance de dos algoritmos del aprendizaje autom & aacute;tico, incluyendo regresiones log & iacute;sticas (LR) y bosques al azar (Random Forest, RF), para construir mapas de susceptibilidad a los incendios, fue evaluado en regiones con diferentes caracter & iacute;sticas f & iacute;sicas (la regi & oacute;n de Okanogan en los EEUU y la de Jam & eacute;sie en Canad & aacute;). Los inputs del modelo fueron once variables f & iacute;sicamente relacionadas y cuyos resultados fueron las probabilidades de incendios.ResultadosLos resultados indican que las variables m & aacute;s importantes en esas dos & aacute;reas son la cobertura vegetal, la temperatura, el viento, la elevaci & oacute;n, la precipitaci & oacute;n, y el NDVI (Indice Normalizado de Vegetaci & oacute;n). Adicionalmente, los resultados revelan que ambos modelos tienen la capacidad de generar espacial y temporalmente la predicci & oacute;n de la probabilidad anual de la ocurrencia de incendios en tiempos y ubicaciones diferentes. Generalmente, el RF excede al modelo LR casi todos los casos. Como resultado, el modelo provee de mapas de susceptibilidad con diferentes niveles de severidad (desde muy altos a muy bajos). Los resultados tambi & eacute;n resaltan las & aacute;reas que son m & aacute;s vulnerables al fuego. Los modelos desarrollados y el an & aacute;lisis son muy valiosos para los que planifican las emergencias y, para los decisores, para identificar regiones cr & iacute;ticas e implementar acciones preventivas para la conservaci & oacute;n ecol & oacute;gica.
引用
收藏
页数:20
相关论文
共 104 条
[1]   Impact of anthropogenic climate change on wildfire across western US forests [J].
Abatzoglou, John T. ;
Williams, A. Park .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2016, 113 (42) :11770-11775
[2]   Assessing fire hazard potential and its main drivers in Mazandaran province, Iran: a data-driven approach [J].
Adab, Hamed ;
Atabati, Azadeh ;
Oliveira, Sandra ;
Gheshlagh, Ahmad Moghaddam .
ENVIRONMENTAL MONITORING AND ASSESSMENT, 2018, 190 (11)
[3]   Landfire hazard assessment in the Caspian Hyrcanian forest ecoregion with the long-term MODIS active fire data [J].
Adab, Hamed .
NATURAL HAZARDS, 2017, 87 (03) :1807-1825
[4]   Historical range of variability in eastern Cascades forests, Washington, USA [J].
Agee, JK .
LANDSCAPE ECOLOGY, 2003, 18 (08) :725-740
[5]   Evaluation of flood susceptibility mapping using logistic regression and GIS conditioning factors [J].
Al-Juaidi, Ahmed E. M. ;
Nassar, Ayman M. ;
Al-Juaidi, Omar E. M. .
ARABIAN JOURNAL OF GEOSCIENCES, 2018, 11 (24)
[6]   Spatial Modelling of Gully Erosion Using GIS and R Programing: A Comparison among Three Data Mining Algorithms [J].
Arabameri, Alireza ;
Pradhan, Biswajeet ;
Pourghasemi, Hamid Reza ;
Rezaei, Khalil ;
Kerle, Norman .
APPLIED SCIENCES-BASEL, 2018, 8 (08)
[7]   Mapping Forest Fuels through Vegetation Phenology: The Role of Coarse-Resolution Satellite Time-Series [J].
Bajocco, Sofia ;
Dragoz, Eleni ;
Gitas, Ioannis ;
Smiraglia, Daniela ;
Salvati, Luca ;
Ricotta, Carlo .
PLOS ONE, 2015, 10 (03)
[8]   Human-started wildfires expand the fire niche across the United States [J].
Balch, Jennifer K. ;
Bradley, Bethany A. ;
Abatzoglou, John T. ;
Nagy, R. Chelsea ;
Fusco, Emily J. ;
Mahood, Adam L. .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2017, 114 (11) :2946-2951
[9]   Contrasting the role of human- and lightning-caused wildfires on future fire regimes on a Central Oregon landscape [J].
Barros, Ana M. G. ;
Day, Michelle A. ;
Preisler, Haiganoush K. ;
Abatzoglou, John T. ;
Krawchuk, Meg A. ;
Houtman, Rachel ;
Ager, Alan A. .
ENVIRONMENTAL RESEARCH LETTERS, 2021, 16 (06)
[10]   Performance Evaluation of Machine Learning Methods for Forest Fire Modeling and Prediction [J].
Binh Thai Pham ;
Jaafari, Abolfazl ;
Avand, Mohammadtaghi ;
Al-Ansari, Nadhir ;
Tran Dinh Du ;
Hoang Phan Hai Yen ;
Tran Van Phong ;
Duy Huu Nguyen ;
Hiep Van Le ;
Mafi-Gholami, Davood ;
Prakash, Indra ;
Hoang Thi Thuy ;
Tran Thi Tuyen .
SYMMETRY-BASEL, 2020, 12 (06)