Spatial variability of soil water erosion: Comparing empirical and intelligent techniques DOI Creative Commons
Ali Golkarian, Khabat Khosravi, Mahdi Panahi

et al.

Geoscience Frontiers, Journal Year: 2022, Volume and Issue: 14(1), P. 101456 - 101456

Published: Aug. 22, 2022

Soil water erosion (SWE) is an important global hazard that affects food availability through soil degradation, a reduction in crop yield, and agricultural land abandonment. A map of susceptibility first vital step management conservation. Several machine learning (ML) algorithms optimized using the Grey Wolf Optimizer (GWO) metaheuristic algorithm can be used to accurately SWE susceptibility. These include Convolutional Neural Networks (CNN CNN-GWO), Support Vector Machine (SVM SVM-GWO), Group Method Data Handling (GMDH GMDH-GWO). Results obtained these compared with well-known Revised Universal Loss Equation (RUSLE) empirical model Extreme Gradient Boosting (XGBoost) ML tree-based models. We apply methods together frequency ratio (FR) Information Gain Ratio (IGR) determine relationship between historical data controlling geo-environmental factors at 116 sites Noor-Rood watershed northern Iran. Fourteen are classified topographical, hydro-climatic, cover, geological groups. next divided into two datasets, one for training (70% samples = 81 locations) other validation (30% 35 locations). Finally model-generated maps were evaluated Area under Receiver Operating Characteristic (AU-ROC) curve. Our results show elevation rainfall erosivity have greatest influence on SWE, while texture hydrology less important. The CNN-GWO (AU-ROC 0.85) outperformed models, specifically, order, SVR-GWO GMDH-GWO (AUC 0.82), CNN GMDH 0.81), SVR XGBoost 0.80), RULSE. Based RUSLE model, loss ranges from 0 2644 t ha–1yr−1.

Language: Английский

Hybrid Machine Learning Approach for Gully Erosion Mapping Susceptibility at a Watershed Scale DOI Creative Commons
Sliman Hitouri, Antonietta Varasano, Meriame Mohajane

et al.

ISPRS International Journal of Geo-Information, Journal Year: 2022, Volume and Issue: 11(7), P. 401 - 401

Published: July 14, 2022

Gully erosion is a serious threat to the state of ecosystems all around world. As result, safeguarding soil for our own benefit and from actions must guaranteeing long-term viability variety ecosystem services. developing gully susceptibility maps (GESM) both suggested necessary. In this study, we compared effectiveness three hybrid machine learning (ML) algorithms with bivariate statistical index frequency ratio (FR), named random forest-frequency (RF-FR), support vector machine-frequency (SVM-FR), naïve Bayes-frequency (NB-FR), in mapping GHISS watershed northern part Morocco. The models were implemented based on inventory total number 178 points randomly divided into 2 groups (70% used training 30% validation process), 12 conditioning variables (i.e., elevation, slope, aspect, plane curvature, topographic moisture (TWI), stream power (SPI), precipitation, distance road, stream, drainage density, land use, lithology). Using equal interval reclassification method, spatial distribution was categorized five different classes, including very high, moderate, low, low. Our results showed that high classes derived using RF-FR, SVM-FR, NB-FR covered 25.98%, 22.62%, 27.10% area, respectively. area under receiver (AUC) operating characteristic curve, precision, accuracy employed evaluate performance these models. Based (ROC), RF-FR achieved best (AUC = 0.91), followed by SVM-FR 0.87), then 0.82), contribution, line Sustainable Development Goals (SDGs), plays crucial role understanding identifying issue “where why” occurs, hence it can serve as first pathway reducing particular area.

Language: Английский

Citations

37
Assessment of soil erosion risk using an integrated approach of GIS and Analytic Hierarchy Process (AHP) in Erzurum, Turkiye DOI
Derya Mumcu Küçüker, Daniela CEDANO GIRALDO

Ecological Informatics, Journal Year: 2022, Volume and Issue: 71, P. 101788 - 101788

Published: Aug. 27, 2022

Language: Английский

Citations

36
A comparison of machine learning models for suspended sediment load classification DOI Creative Commons
Nouar AlDahoul, Ali Najah Ahmed, Mohammed Falah Allawi

et al.

Engineering Applications of Computational Fluid Mechanics, Journal Year: 2022, Volume and Issue: 16(1), P. 1211 - 1232

Published: May 24, 2022

The suspended sediment load (SSL) is one of the major hydrological processes affecting sustainability river planning and management. Moreover, sediments have a significant impact on dam operation reservoir capacity. To this end, reliable applicable models are required to compute classify SSL in rivers. application machine learning has become common solve complex problems such as modeling. present research investigated ability several data. This investigation aims explore new version classifiers for classification at Johor River, Malaysia. Extreme gradient boosting, random forest, support vector machine, multi-layer perceptron k-nearest neighbors been used values divided into multiple discrete ranges, where each range can be considered category or class. study illustrates two different scenarios related number categories, which five 10 with time scales, daily weekly. performance proposed was evaluated by statistical indicators. Overall, achieved excellent data under various scenarios.

Language: Английский

Citations

32
Mapping of Water-Induced Soil Erosion Using Machine Learning Models: A Case Study of Oum Er Rbia Basin (Morocco) DOI
Ahmed Barakat,

Mouadh Rafai,

Hassan Mosaid

et al.

Earth Systems and Environment, Journal Year: 2022, Volume and Issue: 7(1), P. 151 - 170

Published: June 12, 2022

Language: Английский

Citations

31
Spatial variability of soil water erosion: Comparing empirical and intelligent techniques DOI Creative Commons
Ali Golkarian, Khabat Khosravi, Mahdi Panahi

et al.

Geoscience Frontiers, Journal Year: 2022, Volume and Issue: 14(1), P. 101456 - 101456

Published: Aug. 22, 2022

Soil water erosion (SWE) is an important global hazard that affects food availability through soil degradation, a reduction in crop yield, and agricultural land abandonment. A map of susceptibility first vital step management conservation. Several machine learning (ML) algorithms optimized using the Grey Wolf Optimizer (GWO) metaheuristic algorithm can be used to accurately SWE susceptibility. These include Convolutional Neural Networks (CNN CNN-GWO), Support Vector Machine (SVM SVM-GWO), Group Method Data Handling (GMDH GMDH-GWO). Results obtained these compared with well-known Revised Universal Loss Equation (RUSLE) empirical model Extreme Gradient Boosting (XGBoost) ML tree-based models. We apply methods together frequency ratio (FR) Information Gain Ratio (IGR) determine relationship between historical data controlling geo-environmental factors at 116 sites Noor-Rood watershed northern Iran. Fourteen are classified topographical, hydro-climatic, cover, geological groups. next divided into two datasets, one for training (70% samples = 81 locations) other validation (30% 35 locations). Finally model-generated maps were evaluated Area under Receiver Operating Characteristic (AU-ROC) curve. Our results show elevation rainfall erosivity have greatest influence on SWE, while texture hydrology less important. The CNN-GWO (AU-ROC 0.85) outperformed models, specifically, order, SVR-GWO GMDH-GWO (AUC 0.82), CNN GMDH 0.81), SVR XGBoost 0.80), RULSE. Based RUSLE model, loss ranges from 0 2644 t ha–1yr−1.

Language: Английский

Citations

30