Integrating multi-hazard susceptibility and building exposure: A case study for Quang Nam province, Vietnam DOI Creative Commons
Chinh Luu, Giuseppe Forino,

Lynda Yorke

et al.

Published: Jan. 29, 2024

Abstract. Natural hazards have serious impacts worldwide on society, economy and environment. In Vietnam, throughout the years, natural caused a significant loss of lives as well severe devastation to houses, crops, transportation. This paper presents new model for multi-hazard (floods wildfires) exposure estimates using machine learning models, Google Earth Engine, spatial analysis tools typical Quang Nam province, Vietnam case study. By establishing context collected data climate impacts, geospatial database was built multiple hazard modelling, including an inventory climate-related wildfires), topography, geology, hydrology, features (temperature, wetness, wind), land use, building assessment. The susceptibility matrices were presented demonstrate profiling approach multi-hazards. results are explicitly illustrated floods wildfire buildings. Susceptibility models random forest provide accuracy AUC=0.882 0.884 wildfires, respectively. flood combined within semi-quantitative matrix assessing different combinations hazards. Digital risk maps wildfires aid identification areas prone potential can be used inform communities regulatory authorities how they develop implement long-term adaptation solutions.

Language: Английский

Exploring forest fire susceptibility and management strategies in Western Himalaya: Integrating ensemble machine learning and explainable AI for accurate prediction and comprehensive analysis DOI Creative Commons
Hoang Thi Hang, Javed Mallick, Saeed Alqadhi

et al.

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

Citations

18
Quantitative assessment of the GLOF risk along China-Nepal transboundary basins by integrating remote sensing, machine learning, and hydrodynamic model DOI
Manish Raj Gouli,

Kaiheng Hu,

Nitesh Khadka

et al.

International Journal of Disaster Risk Reduction, Journal Year: 2025, Volume and Issue: unknown, P. 105231 - 105231

Published: Jan. 1, 2025

Language: Английский

Citations

4
FFM: Flood Forecasting Model Using Federated Learning DOI Creative Commons
Muhammad Shoaib Farooq, Rabia Tehseen, Junaid Nasir Qureshi

et al.

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 24472 - 24483

Published: Jan. 1, 2023

Floods are one of the most common natural disasters that occur frequently causing massive damage to property, agriculture, economy and life. Flood prediction offers a huge challenge for researchers struggling predict floods since long time. In this article, flood forecasting model using federated learning technique has been proposed. Federated Learning is advanced machine (ML) guarantees data privacy, ensures availability, promises security, handles network latency trials inherent in by prohibiting be transferred over training. urges onsite training local models, focuses on transmission these models instead sending set towards central server aggregation global at server. proposed integrates locally trained eighteen clients, investigates which station flooding about happen generates alert specific client with five days lead A feed forward neural (FFNN) where expected. module FFNN predicts expected water level taking multiple regional parameters as input. The dataset different rivers barrages collected from 2015 2021 considering four aspects including snow melting, rainfall-runoff, flow routing hydrodynamics. successfully predicted previous happened selected zone during 2010 84 % accuracy.

Language: Английский

Citations

28
Machine learning-based predictions of current and future susceptibility to retrogressive thaw slumps across the Northern Hemisphere DOI Creative Commons

Jing Luo,

Guoan Yin, Fujun Niu

et al.

Advances in Climate Change Research, Journal Year: 2024, Volume and Issue: 15(2), P. 253 - 264

Published: March 7, 2024

Retrogressive thaw slumps (RTSs) caused by the thawing of ground ice on permafrost slopes have dramatically increased and become a common hazard across Northern Hemisphere during previous decades. However, gap remains in our comprehensive understanding spatial controlling factors, including climate terrain, that are conducive to these RTSs at global scale. Using machine learning methodologies, we mapped current future susceptibility distributions incorporating range environmental factors inventories. We identified thawing-degree days maximum summer rainfall as primary affecting susceptibility. The final ensemble map suggests regions with high very could constitute (11.6 ± 0.78)% Hemisphere's region. When juxtaposed (2000-2020) map, total area witness an increase ranging from (31.7 0.65)% (SSP585) (51.9± 0.73)% (SSP126) 2041-2060. insights gleaned this study not only offer valuable implications for engineering applications Hemisphere, but also provide long-term insight into potential change response change.

Language: Английский

Citations

6
Forest fire susceptibility assessment under small sample scenario: A semi-supervised learning approach using transductive support vector machine DOI
Tianwu Ma, Gang Wang, Rui Guo

et al.

Journal of Environmental Management, Journal Year: 2024, Volume and Issue: 359, P. 120966 - 120966

Published: April 26, 2024

Language: Английский

Citations

6
Accurate vegetation destruction detection using remote sensing imagery based on the three-band difference vegetation index (TBDVI) and dual-temporal detection method DOI Creative Commons
Chuanwu Zhao, Yaozhong Pan,

Shoujia Ren

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2024, Volume and Issue: 127, P. 103669 - 103669

Published: Jan. 25, 2024

Satellite remote sensing, as an important tool for Earth observation, has been widely used to monitor various vegetation destruction events (VDEs), such logging, wildfires and insect infestations. However, due the spectral diversity of VDE complexity background environments (BE), achieving accurate detection remains a challenge. To overcome this limitation, study developed novel index, called three-band difference index (TBDVI), which fully considered characteristics both BEs multiple VDEs, in complex scenarios. Three experiments were chosen prove performance TBDVI, including (1) possible changes; (2) (3) real events. The results showed that TBDVI was suitable change scenarios conditions, with F1 scores 0.906–0.979. Moreover, accurately identified extent caused by infestation, landslides, wildfires, floods, 0.922–0.965. Compared existing indices (VIs) (i.e., normalized (NDVI), moisture (NDMI) burn ratio (NBR)), obvious advantages reducing impact environment. In addition, exhibits cross-sensor applicability potential large-scale high-frequency monitoring. conclusion, is effective robust metric conservation management resources.

Language: Английский

Citations

5
Unveiling the thermal impact of land cover transformations in Khuzestan province through MODIS satellite remote sensing products DOI

Iraj Baronian,

Reza Borna,

Kamran Jafarpour Ghalehteimouri

et al.

Paddy and Water Environment, Journal Year: 2024, Volume and Issue: 22(4), P. 503 - 520

Published: June 5, 2024

Language: Английский

Citations

5
Quantifying soil erosion and influential factors in Guwahati's urban watershed using statistical analysis, machine and deep learning DOI
Ishita Afreen Ahmed, Swapan Talukdar, Mirza Razi Imam Baig

et al.

Remote Sensing Applications Society and Environment, Journal Year: 2023, Volume and Issue: 33, P. 101088 - 101088

Published: Nov. 10, 2023

Language: Английский

Citations

13
Integration of machine learning and hydrodynamic modeling to solve the extrapolation problem in flood depth estimation DOI Creative Commons
Huu Duy Nguyen, Dinh Kha Dang, Nhu Y Nguyen

et al.

Journal of Water and Climate Change, Journal Year: 2023, Volume and Issue: 15(1), P. 284 - 304

Published: Dec. 9, 2023

Abstract Flood prediction is an important task, which helps local decision-makers in taking effective measures to reduce damage the people and economy. Currently, most studies use machine learning predict flooding a given region; however, extrapolation problem considered major challenge when using these techniques rarely studied. Therefore, this study will focus on approach resolve flood depth by integrating (XGBoost, Extra-Trees (EXT), CatBoost (CB), light gradient boost machines (LightGBM)) hydraulic modeling under MIKE FLOOD. The results show that model worked well providing data needed build model. Among four proposed models, XGBoost was found be best at solving estimation of depth, followed EXT, CB, LightGBM. Quang Binh province hit floods with depths ranging from 0 3.2 m. Areas high are concentrated along downstream two rivers (Gianh Nhat Le – Kien Giang).

Language: Английский

Citations

13
Fire Risk Mapping Using Machine Learning Method and Remote Sensing in the Mediterranean Region DOI
Fatih Sivrikaya, Döndü Demirel

Advances in Space Research, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

Language: Английский

Citations

0