Landslide susceptibility mapping using machine learning algorithms and comparison of their performance at Abha Basin, Asir Region, Saudi Arabia

Geoscience Frontiers, Journal Year: 2020, Volume and Issue: 12(2), P. 639 - 655

Published: June 17, 2020

The current study aimed at evaluating the capabilities of seven advanced machine learning techniques (MLTs), including, Support Vector Machine (SVM), Random Forest (RF), Multivariate Adaptive Regression Spline (MARS), Artificial Neural Network (ANN), Quadratic Discriminant Analysis (QDA), Linear (LDA), and Naive Bayes (NB), for landslide susceptibility modeling comparison their performances. Coupling algorithms with spatial data types mapping is a vitally important issue. This was carried out using GIS R open source software Abha Basin, Asir Region, Saudi Arabia. First, total 243 locations were identified Basin to prepare inventory map different sources. All areas randomly separated into two groups ratio 70% training 30% validating purposes. Twelve landslide-variables generated modeling, which include altitude, lithology, distance faults, normalized difference vegetation index (NDVI), landuse/landcover (LULC), roads, slope angle, streams, profile curvature, plan length (LS), slope-aspect. area under curve (AUC-ROC) approach has been applied evaluate, validate, compare MLTs performance. results indicated that AUC values range from 89.0% QDA 95.1% RF. Our findings showed RF (AUC ​= ​95.1%) LDA ​941.7%) have produced best performances in other MLTs. outcome this maps would be useful environmental protection.

Language: Английский

---

COVID-19 and urban vulnerability in India

Habitat International, Journal Year: 2020, Volume and Issue: 103, P. 102230 - 102230

Published: Aug. 18, 2020

Language: Английский

---

Threats of climate and land use change on future flood susceptibility

Journal of Cleaner Production, Journal Year: 2020, Volume and Issue: 272, P. 122757 - 122757

Published: July 12, 2020

Language: Английский

---

Uncertainty pattern in landslide susceptibility prediction modelling: Effects of different landslide boundaries and spatial shape expressions

Geoscience Frontiers, Journal Year: 2021, Volume and Issue: 13(2), P. 101317 - 101317

Published: Oct. 22, 2021

In some studies on landslide susceptibility mapping (LSM), boundary and spatial shape characteristics have been expressed in the form of points or circles inventory instead accurate polygon form. Different expressions boundaries shapes may lead to substantial differences distribution predicted indexes (LSIs); moreover, presence irregular introduces uncertainties into LSM. To address this issue by accurately drawing polygonal based LSM, uncertainty patterns LSM modelling under two different shapes, such as circles, are compared. Within research area Ruijin City China, a total 370 landslides with information obtained, 10 environmental factors, slope lithology, selected. Then, correlation analyses between selected factors performed via frequency ratio (FR) method. Next, support vector machine (SVM) random forest (RF) points, polygons constructed point-, circle- polygon-based SVM RF models, respectively, Finally, prediction capabilities above models compared computing their statistical accuracy using receiver operating characteristic analysis, LSIs discussed. The results show that surfaces higher reliability express can provide markedly improved accuracy, those circles. Moreover, degree is present expression because there too few grid units acting model input variables. Additionally, errors measurement not most cases. addition, conditions lower mean values larger standard deviations point- circle-based models. overall superior SVM, similar affecting reflected

Language: Английский

---

Evaluation of multi-hazard map produced using MaxEnt machine learning technique

Scientific Reports, Journal Year: 2021, Volume and Issue: 11(1)

Published: March 22, 2021

Abstract Natural hazards are diverse and uneven in time space, therefore, understanding its complexity is key to save human lives conserve natural ecosystems. Reducing the outputs obtained after each modelling analysis present results for stakeholders, land managers policymakers. So, main goal of this survey was a method synthesize three one multi-hazard map evaluation hazard management use planning. To test methodology, we took as study area Gorganrood Watershed, located Golestan Province (Iran). First, an inventory different types including flood, landslides, gullies prepared using field surveys official reports. generate susceptibility maps, total 17 geo-environmental factors were selected predictors MaxEnt (Maximum Entropy) machine learning technique. The accuracy predictive models evaluated by drawing receiver operating characteristic-ROC curves calculating under ROC curve-AUCROC. model not only implemented superbly degree fitting, but also significant performance. Variables importance studied showed that river density, distance from streams, elevation most important respectively. Lithological units, elevation, annual mean rainfall relevant detecting landslides. On other hand, rainfall, lithological units used gully erosion mapping area. Finally, combining integrated created. demonstrated 60% subjected hazards, reaching proportion landslides up 21.2% whole territory. We conclude type may be useful tool local administrators identify areas susceptible at large scales research.

Language: Английский

---

Multi-hazard susceptibility mapping based on Convolutional Neural Networks

Geoscience Frontiers, Journal Year: 2022, Volume and Issue: 13(5), P. 101425 - 101425

Published: June 17, 2022

Multi-hazard susceptibility prediction is an important component of disasters risk management plan. An effective multi-hazard mitigation strategy includes assessing individual hazards as well their interactions. However, with the rapid development artificial intelligence technology, techniques based on machine learning has encountered a huge bottleneck. In order to effectively solve this problem, study proposes mapping framework using classical deep algorithm Convolutional Neural Networks (CNN). First, we use historical flash flood, debris flow and landslide locations Google Earth images, extensive field surveys, topography, hydrology, environmental data sets train validate proposed CNN method. Next, method assessed in comparison conventional logistic regression k-nearest neighbor methods several objective criteria, i.e., coefficient determination, overall accuracy, mean absolute error root square error. Experimental results show that outperforms algorithms predicting probability floods, flows landslides. Finally, maps three are combined create map. It can be observed from map 62.43% area prone hazards, while 37.57% harmless. hazard-prone areas, 16.14%, 4.94% 30.66% susceptible landslides, respectively. terms concurrent 0.28%, 7.11% 3.13% joint occurrence floods flow, respectively, whereas, 0.18% subject all hazards. The benefit engineers, disaster managers local government officials involved sustainable land mitigation.

Language: Английский

---

Assessment of forest fire severity and land surface temperature using Google Earth Engine: a case study of Gujarat State, India

Fire Ecology, Journal Year: 2024, Volume and Issue: 20(1)

Published: March 7, 2024

Abstract Forest fires are a recurring issue in many parts of the world, including India. These can have various causes, human activities (such as agricultural burning, campfires, or discarded cigarettes) and natural factors lightning). The present study presents comprehensive advanced methodology for assessing wildfire susceptibility by integrating diverse environmental variables leveraging cutting-edge machine learning techniques across Gujarat State, primary goal is to utilize Google Earth Engine compare locations Gujarat, India, before after forest fires. High-resolution satellite data were used assess amount types changes caused meticulously analyzes variables, i.e., slope orientation, elevation, normalized difference vegetation index (NDVI), drainage density, precipitation, temperature understand landscape characteristics susceptibility. In addition, sophisticated random regression model predict land surface based on set parameters. maps that result depict geographical distribution burn ratio forecasts, providing valuable insights into spatial patterns trends. findings this work show an automated temporal analysis utilizing may be successfully over wide range cover types, critical future monitoring such threats. impact severe, leading loss biodiversity, damage ecosystems, threats settlements.

Language: Английский

---

Spatial modeling, risk mapping, change detection, and outbreak trend analysis of coronavirus (COVID-19) in Iran (days between February 19 and June 14, 2020)

International Journal of Infectious Diseases, Journal Year: 2020, Volume and Issue: 98, P. 90 - 108

Published: June 20, 2020

ObjectivesCoronavirus disease 2019 (COVID-19) represents a major pandemic threat that has spread to more than 212 countries with 432,902 recorded deaths and 7,898,442 confirmed cases worldwide so far (on June 14, 2020). It is crucial investigate the spatial drivers prevent control epidemic of COVID-19.MethodsThis first comprehensive study COVID-19 in Iran; it carries out modeling, risk mapping, change detection, outbreak trend analysis spread. Four main steps were taken: comparison Iranian coronavirus data global trends, prediction mortality trends using regression detection random forest (RF) machine learning technique (MLT), validation modeled map.ResultsThe results show from February 19 2020, average growth rates (GR) total number Iran 1.08 1.10, respectively. Based on World Health Organisation (WHO) data, Iran's fatality rate (deaths/0.1 M pop) 10.53. Other countries' were, for comparison, Belgium – 83.32, UK 61.39, Spain 58.04, Italy 56.73, Sweden 48.28, France 45.04, USA 35.52, Canada 21.49, Brazil 20.10, Peru 19.70, Chile 16.20, Mexico– 12.80, Germany 10.58. The China 0.32 pop). Over time, heatmap infected areas identified two critical time intervals Iran. provinces classified terms death into large primary group three had outbreaks separate others. world shows distinguished other nine viral infection-related parameters. models showed an increasing but some evidence turning. A polynomial relationship was between infection province population density. Also, third-degree model recently, indicating subsequent measures taken cope have been insufficient ineffective. general similar world's, lower volatility. Change maps period March 11 18 provinces. worth noting LASSO MLT evaluate variables' importance, indicated most important variables distance bus stations, bakeries, hospitals, mosques, ATMs (automated teller machines), banks, minimum temperature coldest month.ConclusionsWe believe this study's are primary, fundamental step take managing controlling its

Language: Английский

---

Soil Science Challenges in a New Era: A Transdisciplinary Overview of Relevant Topics

Air Soil and Water Research, Journal Year: 2020, Volume and Issue: 13

Published: Jan. 1, 2020

Transdisciplinary approaches that provide holistic views are essential to properly understand soil processes and the importance of society will be crucial in future integrate distinct disciplines into studies. A myriad challenges faces science at beginning 2020s. The main aim this overview is assess past achievements current regarding threats such as erosion contamination related different United Nations sustainable development goals (SDGs) including (1) food production, (2) ensure healthy lives reduce environmental risks (SDG3), (3) water availability (SDG6), (4) enhanced carbon sequestration because climate change (SDG13). Twenty experts from sciences offer perspectives on important research directions. Special attention must paid some concerns effective conservation strategies; new computational technologies, models, situ measurements bring insights in-soil process spatiotemporal scales, their relationships, dynamics, thresholds; impacts human activities, wildfires, microorganisms thereby biogeochemical cycles relationships; microplastics a potential pollutant; (5) green technologies for rehabilitation; (6) reduction greenhouse gas emissions by simultaneous nitrous oxide emission. Manuscripts topics these particularly welcomed Air, Soil Water Research.

Language: Английский

---

A machine learning framework for multi-hazards modeling and mapping in a mountainous area

Scientific Reports, Journal Year: 2020, Volume and Issue: 10(1)

Published: July 22, 2020

Abstract This study sought to produce an accurate multi-hazard risk map for a mountainous region of Iran. The area is in southwestern has experienced numerous extreme natural events recent decades. models the probabilities snow avalanches, landslides, wildfires, land subsidence, and floods using machine learning that include support vector (SVM), boosted regression tree (BRT), generalized linear model (GLM). Climatic, topographic, geological, social, morphological factors were main input variables used. data obtained from several sources. accuracies GLM, SVM, functional discriminant analysis (FDA) indicate SVM most predicting flood hazards area. GLM best algorithm wildfire mapping, FDA avalanche risk. values AUC (area under curve) all five are greater than 0.8, demonstrating model’s predictive abilities acceptable. A approach can prove be very useful tool hazard management disaster mitigation, particularly modeling. maps valuable baselines area, providing evidence manage future human interaction with hazards.

Language: Английский

---