Using machine learning models, remote sensing, and GIS to investigate the effects of changing climates and land uses on flood probability

Journal of Hydrology, Journal Year: 2020, Volume and Issue: 595, P. 125663 - 125663

Published: Oct. 27, 2020

Language: Английский

---

Landslide Susceptibility Mapping Using Machine Learning Algorithms and Remote Sensing Data in a Tropical Environment

International Journal of Environmental Research and Public Health, Journal Year: 2020, Volume and Issue: 17(14), P. 4933 - 4933

Published: July 8, 2020

We used AdaBoost (AB), alternating decision tree (ADTree), and their combination as an ensemble model (AB-ADTree) to spatially predict landslides in the Cameron Highlands, Malaysia. The models were trained with a database of 152 compiled using Synthetic Aperture Radar Interferometry, Google Earth images, field surveys, 17 conditioning factors (slope, aspect, elevation, distance road, river, proximity fault, road density, river normalized difference vegetation index, rainfall, land cover, lithology, soil types, curvature, profile stream power topographic wetness index). carried out validation process area under receiver operating characteristic curve (AUC) several parametric non-parametric performance metrics, including positive predictive value, negative sensitivity, specificity, accuracy, root mean square error, Friedman Wilcoxon sign rank tests. AB (AUC = 0.96) performed better than AB-ADTree 0.94) successfully outperformed ADTree 0.59) predicting landslide susceptibility. Our findings provide insights into development more efficient accurate that can be by makers land-use managers mitigate hazards.

Language: Английский

---

Deep learning neural networks for spatially explicit prediction of flash flood probability

Geoscience Frontiers, Journal Year: 2020, Volume and Issue: 12(3), P. 101076 - 101076

Published: Dec. 17, 2020

Flood probability maps are essential for a range of applications, including land use planning and developing mitigation strategies early warning systems. This study describes the potential application two architectures deep learning neural networks, namely convolutional networks (CNN) recurrent (RNN), spatially explicit prediction mapping flash flood probability. To develop validate predictive models, geospatial database that contained records historical events geo-environmental characteristics Golestan Province in northern Iran was constructed. The step-wise weight assessment ratio analysis (SWARA) employed to investigate spatial interplay between floods different influencing factors. CNN RNN models were trained using SWARA weights validated receiver operating technique. results showed model (AUC = 0.832, RMSE 0.144) performed slightly better than 0.814, 0.181) predicting future floods. Further, these demonstrated an improved compared previous studies used same area. network successful capturing heterogeneity patterns Province, resulting can be development plans response general policy implication our suggests design, implementation, verification systems should directed approximately 40% area characterized by high very susceptibility flooding.

Language: Английский

---

Modeling groundwater potential using novel GIS-based machine-learning ensemble techniques

Journal of Hydrology Regional Studies, Journal Year: 2021, Volume and Issue: 36, P. 100848 - 100848

Published: June 26, 2021

The present study has been carried out in the Tabriz River basin (5397 km2) north-western Iran. Elevations vary from 1274 to 3678 m above sea level, and slope angles range 0 150.9 %. average annual minimum maximum temperatures are 2 °C 12 °C, respectively. rainfall ranges 243 641 mm, northern southern parts of receive highest amounts. In this study, we mapped groundwater potential (GWP) with a new hybrid model combining random subspace (RS) multilayer perception (MLP), naïve Bayes tree (NBTree), classification regression (CART) algorithms. A total 205 spring locations were collected by integrating field surveys data Iran Water Resources Management, divided into 70:30 for training validation. Fourteen conditioning factors (GWCFs) used as independent inputs. Statistics such receiver operating characteristic (ROC) five others evaluate performance models. results show that all models performed well GWP mapping (AUC > 0.8). MLP-RS achieved high validation scores = 0.935). relative importance GWCFs was revealed slope, elevation, TRI HAND most important predictors presence. This demonstrates ensemble can support sustainable management resources.

Language: Английский

---

Improved flood susceptibility mapping using a best first decision tree integrated with ensemble learning techniques

Geoscience Frontiers, Journal Year: 2020, Volume and Issue: 12(3), P. 101105 - 101105

Published: Dec. 13, 2020

Improving the accuracy of flood prediction and mapping is crucial for reducing damage resulting from events. In this study, we proposed validated three ensemble models based on Best First Decision Tree (BFT) Bagging (Bagging-BFT), Decorate Random Subspace (RSS-BFT) learning techniques an improved susceptibility in a spatially-explicit manner. A total number 126 historical events Nghe An Province (Vietnam) were connected to set 10 influencing factors (slope, elevation, aspect, curvature, river density, distance rivers, flow direction, geology, soil, land use) generating training validation datasets. The via several performance metrics that demonstrated capability all elucidating underlying pattern occurrences within research area predicting probability future Based Area Under receiver operating characteristic Curve (AUC), Decorate-BFT model achieved AUC value 0.989 was identified as superior over RSS-BFT (AUC = 0.982) Bagging-BFT 0.967) models. comparison between previously reported literature confirmed our provided reliable estimate susceptibilities their maps are trustful early warning systems well development mitigation plans.

Language: Английский

---

Characterization of groundwater potential zones in water-scarce hardrock regions using data driven model

Environmental Earth Sciences, Journal Year: 2021, Volume and Issue: 80(24)

Published: Nov. 27, 2021

Language: Английский

---

GIS-based spatial modeling of snow avalanches using four novel ensemble models

The Science of The Total Environment, Journal Year: 2020, Volume and Issue: 745, P. 141008 - 141008

Published: July 20, 2020

Language: Английский

---

Application of Advanced Machine Learning Algorithms to Assess Groundwater Potential Using Remote Sensing-Derived Data

Remote Sensing, Journal Year: 2020, Volume and Issue: 12(17), P. 2742 - 2742

Published: Aug. 24, 2020

Groundwater (GW) is being uncontrollably exploited in various parts of the world resulting from huge needs for water supply as an outcome population growth and industrialization. Bearing mind importance GW potential assessment reaching sustainability, this study seeks to use remote sensing (RS)-derived driving factors input advanced machine learning algorithms (MLAs), comprising deep boosting logistic model trees evaluate their efficiency. To do so, results are compared with three benchmark MLAs such boosted regression trees, k-nearest neighbors, random forest. For purpose, we firstly assembled different topographical, hydrological, RS-based, lithological altitude, slope degree, aspect, length, plan curvature, profile relative position, distance rivers, river density, topographic wetness index, land use/land cover (LULC), normalized difference vegetation index (NDVI), lineament, lineament lithology. The spring indicator was divided into two classes training (434 springs) validation (186 a proportion 70:30. dataset springs accompanied by were incorporated outputs validated indices accuracy, kappa, receiver operating characteristics (ROC) curve, specificity, sensitivity. Based upon area under ROC tree (87.813%) generated similar performance (87.807%), followed (87.397%), forest (86.466%), neighbors (76.708%) MLAs. findings confirm great modelling potential. Thus, application can be suggested other areas obtain insight about GW-related barriers toward sustainability. Further, based on algorithm depicts high impact RS-based factor, NDVI 100 influence, well influence river, RSP variables 46.07, 43.47, 37.20 respectively,

Language: Английский

---

Preprocessing approaches in machine-learning-based groundwater potential mapping: an application to the Koulikoro and Bamako regions, Mali

Hydrology and earth system sciences, Journal Year: 2022, Volume and Issue: 26(2), P. 221 - 243

Published: Jan. 18, 2022

Abstract. Groundwater is crucial for domestic supplies in the Sahel, where strategic importance of aquifers will increase coming years due to climate change. potential mapping a valuable tool underpin water management region and, hence, improve drinking access. This paper presents machine learning method map groundwater potential. illustrated through its application two administrative regions Mali. A set explanatory variables presence developed first. Scaling methods (standardization, normalization, maximum absolute value and max–min scaling) are used avoid pitfalls associated with reclassification. Noisy, collinear counterproductive identified excluded from input dataset. total 20 classifiers then trained tested on large borehole database (n=3345) order find meaningful correlations between or absence variables. Maximum standardization proved most efficient scaling techniques, while tree-based algorithms (accuracy >0.85) consistently outperformed other classifiers. The flow rate data were calibrate results beyond standard metrics, thereby adding robustness predictions. southern part study area better prospect, which consistent geological climatic setting. Outcomes lead three major conclusions: (1) picking best performers out number recommended as good methodological practice, (2) metrics should be complemented additional hydrogeological indicators whenever possible (3) variable contributes minimize expert bias.

Language: Английский

---

Groundwater Potential Mapping in Hubei Region of China Using Machine Learning, Ensemble Learning, Deep Learning and AutoML Methods

Natural Resources Research, Journal Year: 2022, Volume and Issue: 31(5), P. 2549 - 2569

Published: July 9, 2022

Language: Английский

---