Deep learning methods for flood mapping: a review of existing applications and future research directions

Hydrology and earth system sciences, Journal Year: 2022, Volume and Issue: 26(16), P. 4345 - 4378

Published: Aug. 25, 2022

Abstract. Deep learning techniques have been increasingly used in flood management to overcome the limitations of accurate, yet slow, numerical models and improve results traditional methods for mapping. In this paper, we review 58 recent publications outline state art field, identify knowledge gaps, propose future research directions. The focuses on type deep various mapping applications, types considered, spatial scale studied events, data model development. show that based convolutional layers are usually more as they leverage inductive biases better process characteristics flooding events. Models fully connected layers, instead, provide accurate when coupled with other statistical models. showed increased accuracy compared approaches speed methods. While there exist several applications susceptibility, inundation, hazard mapping, work is needed understand how can assist real-time warning during an emergency it be employed estimate risk. A major challenge lies developing generalize unseen case studies. Furthermore, all reviewed their outputs deterministic, limited considerations uncertainties outcomes probabilistic predictions. authors argue these identified gaps addressed by exploiting fundamental advancements or taking inspiration from developments applied areas. graph neural networks operators arbitrarily structured thus should capable generalizing across different studies could account complex interactions natural built environment. Physics-based preserve underlying physical equations resulting reliable speed-up alternatives Similarly, resorting Gaussian processes Bayesian networks.

Language: Английский

---

Convolutional neural network (CNN) with metaheuristic optimization algorithms for landslide susceptibility mapping in Icheon, South Korea

Journal of Environmental Management, Journal Year: 2021, Volume and Issue: 305, P. 114367 - 114367

Published: Dec. 27, 2021

Language: Английский

---

Spatial flood susceptibility mapping using an explainable artificial intelligence (XAI) model

Geoscience Frontiers, Journal Year: 2023, Volume and Issue: 14(6), P. 101625 - 101625

Published: April 28, 2023

Floods are natural hazards that lead to devastating financial losses and large displacements of people. Flood susceptibility maps can improve mitigation measures according the specific conditions a study area. The design flood has been enhanced through use hybrid machine learning deep models. Although these models have achieved better accuracy than traditional models, they not widely used by stakeholders due their black-box nature. In this study, we propose application an explainable artificial intelligence (XAI) model incorporates Shapley additive explanation (SHAP) interpret outcomes convolutional neural network (CNN) analyze impact variables on mapping. This was conducted in Jinju Province, South Korea, which long history events. Model performance evaluated using area under receiver operating characteristic curve (AUROC), showed prediction 88.4%. SHAP plots land various soil attributes significantly affected light findings, recommend XAI-based future mapping studies interpretations outcomes, build trust among during flood-related decision-making process.

Language: Английский

---

Soil water erosion susceptibility assessment using deep learning algorithms

Journal of Hydrology, Journal Year: 2023, Volume and Issue: 618, P. 129229 - 129229

Published: Feb. 6, 2023

Accurate assessment of soil water erosion (SWE) susceptibility is critical for reducing land degradation and loss, mitigating the negative impacts on ecosystem services, quality, flooding infrastructure. Deep learning algorithms have been gaining attention in geoscience due to their high performance flexibility. However, an understanding potential these provide fast, cheap, accurate predictions lacking. This study provides first quantification this potential. Spatial are made using three deep – Convolutional Neural Network (CNN), Recurrent (RNN) Long-Short Term Memory (LSTM) Iranian catchment that has historically experienced severe erosion. Through a comparison predictive analysis driving geo-environmental factors, results reveal: (1) elevation was most effective variable SWE susceptibility; (2) all developed models had good prediction performance, with RNN being marginally superior; (3) maps revealed almost 40 % highly or very susceptible 20 moderately susceptible, indicating need control catchment. algorithms, catchments can potentially be predicted accurately ease readily available data. Thus, reveal great use data poor catchments, such as one studied here, especially developing nations where technical modeling skills processes occurring may

Language: Английский

---

RR-Former: Rainfall-runoff modeling based on Transformer

Journal of Hydrology, Journal Year: 2022, Volume and Issue: 609, P. 127781 - 127781

Published: April 1, 2022

Language: Английский

---

Enhancing flood susceptibility modeling using multi-temporal SAR images, CHIRPS data, and hybrid machine learning algorithms

The Science of The Total Environment, Journal Year: 2023, Volume and Issue: 871, P. 162066 - 162066

Published: Feb. 10, 2023

Language: Английский

---

A systematic review of trustworthy artificial intelligence applications in natural disasters

Computers & Electrical Engineering, Journal Year: 2024, Volume and Issue: 118, P. 109409 - 109409

Published: June 29, 2024

Artificial intelligence (AI) holds significant promise for advancing natural disaster management through the use of predictive models that analyze extensive datasets, identify patterns, and forecast potential disasters. These facilitate proactive measures such as early warning systems (EWSs), evacuation planning, resource allocation, addressing substantial challenges associated with This study offers a comprehensive exploration trustworthy AI applications in disasters, encompassing management, risk assessment, prediction. research is underpinned by an review reputable sources, including Science Direct (SD), Scopus, IEEE Xplore (IEEE), Web (WoS). Three queries were formulated to retrieve 981 papers from earliest documented scientific production until February 2024. After meticulous screening, deduplication, application inclusion exclusion criteria, 108 studies included quantitative synthesis. provides specific taxonomy disasters explores motivations, challenges, recommendations, limitations recent advancements. It also overview techniques developments using explainable artificial (XAI), data fusion, mining, machine learning (ML), deep (DL), fuzzy logic, multicriteria decision-making (MCDM). systematic contribution addresses seven open issues critical solutions essential insights, laying groundwork various future works trustworthiness AI-based management. Despite benefits, persist In these contexts, this identifies several unused used areas disaster-based theory, collects ML, DL techniques, valuable XAI approach unravel complex relationships dynamics involved utilization fusion processes related Finally, extensively analyzed ethical considerations, bias, consequences AI.

Language: Английский

---

Investigating the Role of the Key Conditioning Factors in Flood Susceptibility Mapping Through Machine Learning Approaches

Earth Systems and Environment, Journal Year: 2024, Volume and Issue: 8(1), P. 63 - 81

Published: Jan. 1, 2024

Abstract This study harnessed the formidable predictive capabilities of three state-of-the-art machine learning models—extreme gradient boosting (XGB), random forest (RF), and CatBoost (CB)—applying them to meticulously curated datasets topographical, geological, environmental parameters; goal was investigate intricacies flood susceptibility within arid riverbeds Wilayat As-Suwayq, which is situated in Sultanate Oman. The results underscored exceptional discrimination prowess XGB CB, boasting impressive area under curve (AUC) scores 0.98 0.91, respectively, during testing phase. RF, a stalwart contender, performed commendably with an AUC 0.90. Notably, investigation revealed that certain key variables, including curvature, elevation, slope, stream power index (SPI), topographic wetness (TWI), roughness (TRI), normalised difference vegetation (NDVI), were critical achieving accurate delineation flood-prone locales. In contrast, ancillary factors, such as annual precipitation, drainage density, proximity transportation networks, soil composition, geological attributes, though non-negligible, exerted relatively lesser influence on susceptibility. empirical validation further corroborated by robust consensus XGB, RF CB models. By amalgamating advanced deep techniques precision geographical information systems (GIS) rich troves remote-sensing data, can be seen pioneering endeavour realm analysis cartographic representation semiarid fluvial landscapes. findings advance our comprehension vulnerability dynamics provide indispensable insights for development proactive mitigation strategies regions are susceptible hydrological perils.

Language: Английский

---

Leveraging machine learning and open-source spatial datasets to enhance flood susceptibility mapping in transboundary river basin

International Journal of Digital Earth, Journal Year: 2024, Volume and Issue: 17(1)

Published: Feb. 9, 2024

Floods pose devastating effects on the resiliency of human and natural systems. flood risk management challenges are typically complicated in transboundary river basin due to conflicting objectives between multiple countries, lack systematic approaches data monitoring sharing, limited collaboration developing a unified system for hazard prediction communication. An open-source, low-cost modeling framework that integrates open-source models can help improve our understanding susceptibility inform design equitable strategies. This study datasets machine -learning techniques quantify across data-scare basin. The analysis focuses Gandak River Basin, spanning China, Nepal, India, where damaging recurring floods serious concern. is assessed using four widely used learning techniques: Long-Short-Term-Memory, Random Forest, Artificial Neural Network, Support Vector Machine. Our results exhibit improved performance Network Machine predicting maps, revealing higher vulnerability southern plains. demonstrates remote sensing prediction, mapping, environment.

Language: Английский

---

UAVs in Disaster Management: Application of Integrated Aerial Imagery and Convolutional Neural Network for Flood Detection

Sustainability, Journal Year: 2021, Volume and Issue: 13(14), P. 7547 - 7547

Published: July 6, 2021

Floods have been a major cause of destruction, instigating fatalities and massive damage to the infrastructure overall economy affected country. Flood-related devastation results in loss homes, buildings, critical infrastructure, leaving no means communication or travel for people stuck such disasters. Thus, it is essential develop systems that can detect floods region provide timely aid relief stranded people, save their livelihoods, protect key city infrastructure. Flood prediction warning implemented developed countries, but manufacturing cost too high developing countries. Remote sensing, satellite imagery, global positioning system, geographical information are currently used flood detection assess flood-related damages. These techniques use neural networks, machine learning, deep learning methods. However, unmanned aerial vehicles (UAVs) coupled with convolution networks not explored these contexts instigate swift disaster management response minimize Accordingly, this paper uses UAV-based imagery as method based on Convolutional Neural Network (CNN) extract features from images zone. This effective assessing local infrastructures zones. The study area flood-prone Indus River Pakistan, where both pre-and post-disaster collected through UAVs. For training phase, 2150 image patches created by resizing cropping source images. dataset train CNN model regions change has occurred. tested against validate it, which positive an accuracy 91%. Disaster organizations damages other assets worldwide proper responses help smart governance cities all emergent disasters addressed promptly.

Language: Английский

---