Exploring forest fire susceptibility and management strategies in Western Himalaya: Integrating ensemble machine learning and explainable AI for accurate prediction and comprehensive analysis

Environmental Technology & Innovation, Journal Year: 2024, Volume and Issue: 35, P. 103655 - 103655

Published: May 5, 2024

Forest fires pose a significant threat to ecosystems and socio-economic activities, necessitating the development of accurate predictive models for effective management mitigation. In this study, we present novel machine learning approach combined with Explainable Artificial Intelligence (XAI) techniques predict forest fire susceptibility in Nainital district. Our innovative methodology integrates several robust — AdaBoost, Gradient Boosting Machine (GBM), XGBoost Random Deep Neural Network (DNN) as meta-model stacking framework. This not only utilises individual strengths these models, but also improves overall prediction performance reliability. By using XAI techniques, particular SHAP (SHapley Additive exPlanations) LIME (Local Interpretable Model-agnostic Explanations), improve interpretability provide insights into decision-making processes. results show effectiveness ensemble model categorising different zones: very low, moderate, high high. particular, identified extensive areas susceptibility, precision, recall F1 values underpinning their effectiveness. These achieved ROC AUC above 0.90, performing exceptionally well an 0.94. The are remarkably inclusion confidence intervals most important metrics all emphasises robustness reliability supports practical use management. Through summary plots, analyze global variable importance, revealing annual rainfall Evapotranspiration (ET) key factors influencing susceptibility. Local analysis consistently highlights importance rainfall, ET, distance from roads across models. study fills research gap by providing comprehensive interpretable modelling that our ability effectively manage risk is consistent environmental protection sustainable goals.

Language: Английский

---

Eigenvector Spatial Filtering Enhancing Natural Hazards Vulnerability Assessment in a Susceptible Urban Environment: A Case Study of Izmir Earthquake in Turkey

Environmental Technology & Innovation, Journal Year: 2024, Volume and Issue: 35, P. 103666 - 103666

Published: May 10, 2024

The increasing risk of earthquakes in urban areas has made it crucial to develop accurate vulnerability models for city infrastructure and systems. We aimed assess compare the effectiveness different analysis techniques predicting earthquake specific context Izmir, Turkey. One central hypothesis this research determine whether integrating Eigenvector Spatial Filtering (ESF) into both regression machine learning algorithms would yield a comparable enhancement model performance. performed modeling (EVM) by considering (ⅰ) only seismic-related variables (SRV) (ⅱ) ESF using Moran's eigenvector maps (MEMs). For each approach, we evaluated predictive performance two simple regression-based models; generalized linear (GLM) additive (GAM), complex ones; boosting (GBM), random forest (RF). study utilized five primary indicators encompassing geotechnical, physical, structural, social, facilities data. was assessed evaluation metrics including Root Mean Square Error (RMSE), Absolute (MAE), adjusted R2. results indicated that optimal candidate consisted key variables: altitude, building height, distance safety gathering places, Peak Ground Acceleration (PGA), population density. found decision-tree-based methods better than schemes. RF exhibited highest training data (RMSE = 0.59, R2 0.71), while GBM outperformed other test 0.79, 0.78). However, incorporating EVM revealed methods, particularly GLM, obtained improvement accuracy 0.94 vs 0.76 0.56 0.71 SRV + MEMs approach). Significant differences were observed between GLM-GBM GLM-RF comparisons, as well GAM-GBM GAM-RF comparisons. findings are expected be helpful informed decision-making, targeted reduction, development effective policies strategies enhance preparedness resilience face seismic events highly susceptible

Language: Английский

---

Megafires and koala occurrence: a comparative analysis of field data and satellite imagery

Australian Mammalogy, Journal Year: 2024, Volume and Issue: 46(2)

Published: March 25, 2024

Megafires can have a devastating effect on koala populations. With climate change increasing habitat vulnerability to wildfires, understanding how efficiently measure the impact of these events koalas is essential. We analysed relationship between 2019-2020 megafires and probability occurrence in Mid North Coast NSW. found that two on-field one satellite-derived variables measuring fire severity equally explained occurrence. The decreased with severity. This supports use remote sensing imagery monitor future populations region.

Language: Английский

---

Forest Fire Severity and Koala Habitat Recovery Assessment Using Pre- and Post-Burn Multitemporal Sentinel-2 Msi Data

Forests, Journal Year: 2024, Volume and Issue: 15(11), P. 1991 - 1991

Published: Nov. 11, 2024

Habitat loss due to wildfire is an increasing problem internationally for threatened animal species, particularly tree-dependent and arboreal animals. The koala (Phascolartos cinereus) endangered in most of its range, large areas forest were burnt by widespread wildfires Australia 2019/2020, mostly dominated eucalypts, which provide habitats. We studied the impact fire three subsequent years recovery on a property South-East Queensland, Australia. A classified Differenced Normalised Burn Ratio (dNBR) calculated from pre- post-burn Sentinel-2 scenes encompassing local study area was used assess regional koala-habitat types. geometrically structured composite burn index (GeoCBI), field-based assessment, classify severity impact. To detect lower levels recovery, manual classification multitemporal dNBR used, enabling direct comparison images between years. In our area, suitable habitat occupied only about 2%, 10% that wildfire. From five types studied, one upland type more severely extensively than others but recovered vigorously after first year, reaching same extent as other two alluvial showed negligible impact, likely their sheltered locations. second all impacted further, almost equal, recovery. third year there no detectable change therefore notable vegetative growth. Our field data revealed can probably measure general vegetation present not tree via epicormic shooting coppicing. Eucalypt foliage growth critical resource koala, so verification seems necessary unless more-accurate remote sensing methods such hyperspectral imagery be implemented.

Language: Английский

---