Wildfire assessment using machine learning algorithms in different regions

被引:0
|
作者
Moghim, Sanaz [1 ]
Mehrabi, Majid [1 ]
机构
[1] Sharif Univ Technol, Dept Civil Engn, Azadi Ave, Tehran, Iran
来源
FIRE ECOLOGY | 2024年 / 20卷 / 01期
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 50 条
  • [31] Gully Erosion Susceptibility Assessment Using Different Machine Learning Algorithms: A Case Study of Shazand Watershed in Iran
    Majid Mohammady
    Aliakbar Davudirad
    Environmental Modeling & Assessment, 2024, 29 : 249 - 261
  • [32] Gully Erosion Susceptibility Assessment Using Different Machine Learning Algorithms: A Case Study of Shazand Watershed in Iran
    Mohammady, Majid
    Davudirad, Aliakbar
    ENVIRONMENTAL MODELING & ASSESSMENT, 2024, 29 (02) : 249 - 261
  • [33] Machine learning algorithms applied to wildfire data in California's central valley
    Hernandez, Kassandra
    Hoskins, Aaron B.
    TREES FORESTS AND PEOPLE, 2024, 15
  • [34] Landslide susceptibility assessment using locally weighted learning integrated with machine learning algorithms
    Hong, Haoyuan
    EXPERT SYSTEMS WITH APPLICATIONS, 2024, 237
  • [35] Integrating hydrological parameters in wildfire risk assessment: a machine learning approach for mapping wildfire probability
    Khodaee, Mahsa
    Easterday, Kelly
    Klausmeyer, Kirk
    ENVIRONMENTAL RESEARCH LETTERS, 2024, 19 (11):
  • [36] Comparison of Machine Learning Algorithms on Different Datasets
    Uysal, Elif
    Ozturk, Ali
    2018 26TH SIGNAL PROCESSING AND COMMUNICATIONS APPLICATIONS CONFERENCE (SIU), 2018,
  • [37] Do we need different machine learning algorithms for QSAR modeling? A comprehensive assessment of 16 machine learning algorithms on 14 QSAR data sets
    Wu, Zhenxing
    Zhu, Minfeng
    Kang, Yu
    Leung, Elaine Lai-Han
    Lei, Tailong
    Shen, Chao
    Jiang, Dejun
    Wang, Zhe
    Cao, Dongsheng
    Hou, Tingjun
    BRIEFINGS IN BIOINFORMATICS, 2021, 22 (04)
  • [38] Assessment of Machine Learning algorithms for automated monitoring
    Rotuna, Carmen-Ionela
    Dumitrache, Mihail
    Sandu, Ionut-Eugen
    ROMANIAN JOURNAL OF INFORMATION TECHNOLOGY AND AUTOMATIC CONTROL-REVISTA ROMANA DE INFORMATICA SI AUTOMATICA, 2022, 32 (03): : 73 - 84
  • [39] Landslide susceptibility assessment with machine learning algorithms
    Marjanovic, Milos
    Bajat, Branislav
    Kovacevic, Milos
    2009 INTERNATIONAL CONFERENCE ON INTELLIGENT NETWORKING AND COLLABORATIVE SYSTEMS (INCOS 2009), 2009, : 273 - +
  • [40] Wildfire Susceptibility Mapping Using Five Boosting Machine Learning Algorithms: The Case Study of the Mediterranean Region of Turkey
    Abujayyab, Sohaib K. M.
    Kassem, Moustafa Moufid
    Khan, Ashfak Ahmad
    Wazirali, Raniyah
    Ozturk, Ahmet
    Toprak, Ferhat
    ADVANCES IN CIVIL ENGINEERING, 2022, 2022