Advancing forest fire prediction: A multi-layer stacking ensemble model approach

Earth Science Informatics, Journal Year: 2025, Volume and Issue: 18(3)

Published: Feb. 19, 2025

Language: Английский

---

Explainable Sinkhole Susceptibility Mapping Using Machine-Learning-Based SHAP: Quantifying and Comparing the Effects of Contributing Factors in Konya, Türkiye

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(6), P. 3139 - 3139

Published: March 13, 2025

Sinkholes, naturally occurring formations in karst regions, represent a significant environmental hazard, threatening infrastructure, agricultural lands, and human safety. In recent years, machine learning (ML) techniques have been extensively employed for sinkhole susceptibility mapping (SSM). However, the lack of explainability inherent these methods remains critical issue decision-makers. this study, Konya Closed Basin was mapped using an interpretable model based on SHapley Additive exPlanations (SHAP). The Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Machine (LightGBM) algorithms were employed, interpretability results enhanced through SHAP analysis. Among compared models, RF demonstrated highest performance, achieving accuracy 95.5% AUC score 98.8%, consequently selected development final map. analyses revealed that factors such as proximity to fault lines, mean annual precipitation, bicarbonate concentration difference are most variables influencing formation. Additionally, specific threshold values quantified, effects contributing analyzed detail. This study underscores importance employing eXplainable Artificial Intelligence (XAI) natural hazard modeling, SSM example, thereby providing decision-makers with more reliable comparable risk assessment.

Language: Английский

---

Post-Fire Burned Area Detection Using Machine Learning and Burn Severity Classification with Spectral Indices in İzmir: A SHAP-Driven XAI Approach

Fire, Journal Year: 2025, Volume and Issue: 8(4), P. 121 - 121

Published: March 21, 2025

This study was conducted to precisely map burned areas in fire-prone forest regions of İzmir and analyze the spatial distribution wildfires. Using Sentinel-2 satellite imagery, burn severity first classified using dNBR dNDVI indices. Subsequently, machine learning (ML) algorithms—RF, XGBoost, LightGBM, AdaBoost—were employed classify unburned areas. To enhance model performance, hyperparameter optimization applied, results were evaluated multiple accuracy metrics. found that RF achieved highest with an overall 98.0% a Kappa coefficient 0.960. In comparison, classification based solely on spectral indices resulted accuracies 86.6% (dNBR) 81.7% (dNDVI). A key contribution this is integration Explainable Artificial Intelligence (XAI) through SHapley Additive exPlanations (SHAP) analysis, which used interpret influence environmental variables area classification. SHAP analysis made decision processes transparent identified dNBR, dNDVI, SWIR/NIR bands as most influential variables. Furthermore, analyses confirmed variations reflectance across fire-affected are critical for accurate delineation, particularly heterogeneous landscapes. provides scientific framework post-fire ecosystem restoration, fire management, disaster strategies, offering decision-makers data-driven effective intervention strategies.

Language: Английский

---

Modeling the Spatial Distribution of Wildfire Risk in Chile Under Current and Future Climate Scenarios

Fire, Journal Year: 2025, Volume and Issue: 8(3), P. 113 - 113

Published: March 15, 2025

Wildfires pose severe threats to terrestrial ecosystems by causing loss of biodiversity, altering landscapes, compromising ecosystem services, and endangering human lives infrastructure. Chile, with its diverse geography climate, faces escalating wildfire frequency intensity due climate change. This study employs a spatial machine learning approach using Random Forest algorithm predict risk in Central Southern Chile under current future climatic scenarios. The model was trained on time series dataset incorporating climatic, land use, physiographic variables, burned-area scars as the response variable. By applying this three projected scenarios, forecasts distribution probabilities for multiple periods. model’s performance high, achieving an Area Under Curve (AUC) 0.91 testing 0.87 validation. accuracy, True Positive Rate (TPR), Negative (TNR) values were 0.80, 0.87, 0.73, respectively. Currently, prediction Mediterranean-type areas central Araucanía are most at risk, particularly agricultural zones rural–urban interfaces. However, projections indicate southward expansion overall increase scenarios become more pessimistic. These findings offer framework policymakers, facilitating evidence-based strategies adaptive management effective mitigation risk.

Language: Английский

---

Explainable machine learning for predictive modeling of blowing snow detection and meteorological feature assessment using XGBoost-SHAP

PLoS ONE, Journal Year: 2025, Volume and Issue: 20(3), P. e0318835 - e0318835

Published: March 28, 2025

Accurate forecasting of blowing snow events is vital for improving numerical models processes, yet traditional predictive methods often lack interpretability. This study leverages eXtreme Gradient Boosting (XGBoost) to detect using meteorological and flux monitoring data from three weather stations in the Alps. Through 5-fold cross-validation, model achieved impressive performance metrics, with precision rates exceeding 0.94 non-blowing 0.77-0.80 events. The SHAP framework was employed analyze relative importance factors, revealing that maximum wind speed (WS-MAX), average (WS-AVG), air temperature (AT), humidity (AH) are most influential factors. Additionally, Partial dependence plots (PDP) demonstrated a linear correlation between increased WS-MAX probability snow, while WS-AVG showed diminishing returns beyond 10 m/s. Notably, AT below -3°C strongly correlates occurrence, whereas above exhibits negative relationship. Relative plays significant role, values 60% stabilizing peaking near 100%. research contributes drifting event dynamics by integrating explainable artificial intelligence techniques (XAI), thereby interpretability supporting data-driven decision-making applications.

Language: Английский

---

Analysing Fire Propagation Models: A Case Study on FARSITE for Prolonged Wildfires

Fire, Journal Year: 2025, Volume and Issue: 8(5), P. 166 - 166

Published: April 23, 2025

With increasing wildfire severity and duration driven by climate change, accurately predicting fire behavior over extended time frames is critical for effective management mitigation of such wildfires. Fire propagation models play a pivotal role in these efforts, providing simulations that can be used to strategize respond active fires. This study examines the area simulator (FARSITE) model’s performance simulating recent events persisted 24 h with limited firefighting intervention mostly remote access areas across diverse ecosystems. Our findings reveal key insights into prolonged scenarios potentially informing improvements operational long-term predictive accuracy, as comparisons indexes showed reasonable results between detected fires from information resource systems (FIRMSs) first following days. A case Madeira Island highlights integration real-time weather predictions post-event data analysis. analysis underscores potential combining accurate forecasts retrospective validation improve capabilities dynamic environments, which guided development software platform designed analyse ongoing real-time, leveraging image satellite predictions.

Language: Английский

---

Interpretable machine learning for coastal wind prediction: Integrating SHAP analysis and seasonal trends

Journal of Coastal Conservation, Journal Year: 2025, Volume and Issue: 29(3)

Published: May 6, 2025

Language: Английский

---

Evaluation of Three Algorithms and Forest Fire Risk Prediction in Zhejiang Province of China

Forests, Journal Year: 2024, Volume and Issue: 15(12), P. 2146 - 2146

Published: Dec. 5, 2024

Forest fires represent a paramount natural disaster of global concern. Zhejiang Province has the highest forest coverage rate in China, and are one main disasters impacting management region. In this study, we comprehensively analyzed spatiotemporal distribution based on MODIS data from 2013 to 2023. The results showed that annual incidence shown an overall downward trend 2023, with occurring more frequently winter spring. By utilizing eight contributing factors fire occurrence as variables, three models were constructed: Logistic Regression (LR), Random (RF), eXtreme Gradient Boosting (XGBoost). RF XGBoost demonstrated high predictive ability, achieving accuracy rates 0.85 0.92, f1-score 0.84 AUC values 0.892 0.919, respectively. Further analysis using revealed elevation precipitation had most significant effects fires. Additionally, predictions risk generated by indicated is southern part Province, particularly Wenzhou Lishui areas, well southwest Hangzhou area north Quzhou area. future, can be predicted site models, providing scientific reference for aiding prevention mitigation impacts

Language: Английский

---