Hybrid Models Incorporating Bivariate Statistics and Machine Learning Methods for Flash Flood Susceptibility Assessment Based on Remote Sensing Datasets

Remote Sensing, Journal Year: 2021, Volume and Issue: 13(23), P. 4945 - 4945

Published: Dec. 5, 2021

Flash floods are considered to be one of the most destructive natural hazards, and they difficult accurately model predict. In this study, three hybrid models were proposed, evaluated, used for flood susceptibility prediction in Dadu River Basin. These integrate a bivariate statistical method fuzzy membership value (FMV) machine learning methods support vector (SVM), classification regression trees (CART), convolutional neural network (CNN). Firstly, geospatial database was prepared comprising nine conditioning factors, 485 locations, non-flood locations. Then, train test models. Subsequently, receiver operating characteristic (ROC) curve, seed cell area index (SCAI), accuracy evaluate performances The results reveal following: (1) ROC curve highlights fact that CNN-FMV had best fitting performance, under (AUC) values success rate 0.935 0.912, respectively. (2) Based on performance evaluation methods, all better capabilities than their respective single Compared with models, AUC SVM-FMV, CART-FMV, 0.032, 0.005, 0.055 higher; SCAI 0.05, 0.03, 0.02 lower; accuracies 4.48%, 1.38%, 5.86% higher, (3) indices, between 13.21% 22.03% study characterized by high very susceptibilities. proposed especially CNN-FMV, have potential application assessment specific areas future studies.

Language: Английский

---

Wildfire Risk Assessment in Liangshan Prefecture, China Based on An Integration Machine Learning Algorithm

Remote Sensing, Journal Year: 2022, Volume and Issue: 14(18), P. 4592 - 4592

Published: Sept. 14, 2022

Previous wildfire risk assessments have problems such as subjectivity of weight allocation and the linearization statistical models, resulting in generally low robustness generalization ability fire assessment models. Therefore, this paper, we explored potential integration machine learning algorithms to build Based on analyzing data’s spatial temporal distribution, selected 10 triggering factors topography, meteorology, vegetation, human activities, using frequency ratio (FR) provide uniform data representation factors. Next, used Bayesian optimization (BO) algorithm perform hyperparametric solutions for various models: support vector (SVM), random forest (RF), extreme gradient boosting (XGBoost). Finally, constructed an acquire a grading map importance evaluation corresponding each factor. For validation purposes, Liangshan Prefecture Sichuan Province specific study area obtained MCD64A1 burned product extract extent areas from 2011 2020. The accuracy, kappa coefficient, under curve (AUC) were then applied assess predictive power consistency classification maps. experimental analysis showed that among three FR-BO-XGBoost had best performance region (AUC = 0.887), followed by FR-BO-RF 0.876) FR-BO-SVM 0.820). feature result indicated area’s most significant effects wildfires precipitation, NDVI, land cover, maximum temperature. proposed method avoided subjective weighting model problems. Compared with previous methods, it automatically acquired wildfire, which certain advantages assessment, was worthy further promotion.

Language: Английский

---

Modelling flood susceptibility based on deep learning coupling with ensemble learning models

Journal of Environmental Management, Journal Year: 2022, Volume and Issue: 325, P. 116450 - 116450

Published: Oct. 11, 2022

Language: Английский

---

Assessment of flood susceptibility mapping using support vector machine, logistic regression and their ensemble techniques in the Belt and Road region

Geocarto International, Journal Year: 2022, Volume and Issue: 37(25), P. 9817 - 9846

Published: Jan. 6, 2022

Floods have occurred frequently all over the world. During 2000–2020, nearly half (44.9%) of global floods in Belt and Road region because its complex geology, topography, climate. Therefore, providing an insight into spatial distribution characteristics flood susceptibility this is essential. Here, a database was established with 11 conditioning factors, 1500 flooded points, non-flooded points selected by improved method. Subsequently, rare combination logistic regression support vector machine, integrated heterogeneous framework, applied to generate ensemble map. Based on it, concept ecological vulnerability synthesis index field introduced study, comprehensive (FSCI) proposed quantify degree each country sub-region. At results, model has excellent accuracy, highest AUC value 0.9342. The high zones are mainly located southeastern part Eastern Asia, most Southeast Asia South account for 12.22% 9.57% total study area, respectively. From regional perspective, it can be found that had FSCI 4.69, while East Central Europe showed significant characteristics. national 66 countries region, 20 level (FSCIn > 0.8), which face greatest threat flooding. These results able facilitate reasonable mitigation measures develop at critical locations lays theoretical basis quantifying or scale.

Language: Английский

---

Exploring the additional value of class imbalance distributions on interpretable flash flood susceptibility prediction in the Black Warrior River basin, Alabama, United States

Journal of Hydrology, Journal Year: 2022, Volume and Issue: 610, P. 127877 - 127877

Published: April 28, 2022

Language: Английский

---

Assessment Analysis of Flood Susceptibility in Tropical Desert Area: A Case Study of Yemen

Remote Sensing, Journal Year: 2022, Volume and Issue: 14(16), P. 4050 - 4050

Published: Aug. 19, 2022

Flooding is one of the catastrophic natural hazards worldwide that can easily cause devastating effects on human life and property. Remote sensing devices are becoming increasingly important in monitoring assessing disaster susceptibility hazards. The proposed research work pursues an assessment analysis flood a tropical desert environment: case study Yemen. base data for this were collected organized from meteorological, satellite images, remote data, essential geographic various sources used as input into four machine learning (ML) algorithms. In study, RS (Sentinel-1 images) to detect flooded areas area. We also Sentinel application platform (SNAP 7.0) Sentinel-1 image detecting zones locations. Flood spots discovered verified using Google Earth Landsat press create inventory map Four ML algorithms flash (FFS) Tarim city (Yemen): K-nearest neighbor (KNN), Naïve Bayes (NB), random forests (RF), eXtreme gradient boosting (XGBoost). Twelve conditioning factors prepared, assessed multicollinearity, with inventories parameters run each model. A total 600 non-flood points chosen, where 75% 25% training validation datasets. confusion matrix area under receiver operating characteristic curve (AUROC) validate maps. results obtained reveal all models had high capacity predict floods (AUC > 0.90). Further, terms performance, tree-based ensemble (RF, XGBoost) outperform other algorithms, RF algorithm provides robust performance = 0.982) flood-prone only few adjustments required prior value lies fact being tested first time Yemen assess susceptibility, which be assess, example, earthquakes, landslides, disasters. Furthermore, makes significant contributions effort reduce risk disasters, particularly This will, therefore, help enhance environmental sustainability.

Language: Английский

---

Flood Hazard Index Application in Arid Catchments: Case of the Taguenit Wadi Watershed, Lakhssas, Morocco

Land, Journal Year: 2022, Volume and Issue: 11(8), P. 1178 - 1178

Published: July 28, 2022

During the last decade, climate change has generated extreme rainfall events triggering flash floods in short periods worldwide. The delimitation of flood zones by detailed mapping generally makes it possible to avoid human and economic losses, especially regions at high risk flooding. Taguenit basin, located southern Morocco, is a particular case. this basin method Flood Hazard Index (FHI) GIS geographic information systems environment was based on multi-criteria analysis, taking into consideration seven parameters influencing these phenomena, namely rainfall, slope, flow accumulation, drainage network density, distance from rivers, permeability, land use. Average annual data for 37 years (1980 2016) used study floodplain mapping. A weight calculated each parameter using Analytical Hierarchy Process (AHP) method. combination maps different made draw up final map classified five intervals: very high, moderate, lower presenting, respectively, 8.04%, 20.63%, 31.47%, 15.36%, 24.50% area basin. reliability tested susceptibility analysis. results model are similar those previous historical events. Realistic applicable solutions have been proposed minimize impact as much possible.

Language: Английский

---

Flood susceptibility mapping in an arid region of Pakistan through ensemble machine learning model

Stochastic Environmental Research and Risk Assessment, Journal Year: 2022, Volume and Issue: 36(10), P. 3041 - 3061

Published: Feb. 11, 2022

Language: Английский

---

Flood susceptibility mapping using hybrid models optimized with Artificial Bee Colony

Journal of Hydrology, Journal Year: 2023, Volume and Issue: 624, P. 129961 - 129961

Published: July 19, 2023

Language: Английский

---

Artificial neural network for flood susceptibility mapping in Bangladesh

Heliyon, Journal Year: 2023, Volume and Issue: 9(6), P. e16459 - e16459

Published: May 23, 2023

The objective of the research is to investigate flood susceptibility in Sylhet division Bangladesh. Eight influential factors (i.e., elevation, slope, aspect, curvature, TWI, SPI, roughness, and LULC) were applied as inputs model. In this work, 1280 samples taken at different locations based on non-flood characteristics; these, 75% inventory dataset was used for training 25% testing. An artificial neural network develop a model, results plotted map using ArcGIS. According finding, 40.98% 499433.50 hectors) study area found within very high-susceptibility zone, 37.43% 456168.76 are highly susceptible zone. Only 6.52% 15% low medium zones, respectively. model validation show that overall prediction rate around 89% success 98%. study's findings assist policymakers concerned authorities making risk management decisions order mitigate negative impacts.

Language: Английский

---