Mapping of the wind erodible fraction of soil by bidirectional gated recurrent unit (BiGRU) and bidirectional recurrent neural network (BiRNN) deep learning models

Article Computer Science, Information Systems

Arabic aspect based sentiment analysis using bidirectional GRU based models

Mohammed M. Abdelgwad et al.

Summary: Aspect-based Sentiment Analysis (ABSA) achieves a fine-grained analysis of the aspects and sentiments in a document or sentence, with proposed deep learning models showing improved performance in Arabic hotel review datasets.

JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES (2022)

添加到收藏夹

Article Environmental Sciences

Predicting wind erosion rate using portable wind tunnel combined with machine learning algorithms in calcareous soils, southern Iran

Monireh Mina et al.

Summary: The study successfully predicted wind erosion rates in different land uses and soil types in Fars Province, southern Iran, by combining easily measurable soil properties with data mining approaches. Results showed significant effects of land use and soil type on wind erosion rates, with bare lands and Entisols soil type exhibiting the highest rates. The non-linear regression models using easily measurable soil properties outperformed linear regression models, and support vector regression models coupled with a genetic algorithm showed the highest performance in predicting wind erosion rates.

JOURNAL OF ENVIRONMENTAL MANAGEMENT (2022)

添加到收藏夹

Article Environmental Sciences

Determination of the threshold velocity of soil wind erosion using a wind tunnel and its prediction for calcareous soils of Iran

Mahrooz Rezaei et al.

Summary: Determining the threshold velocity (TV) is crucial for assessing wind erosion. This study used pedotransfer functions (PTFs) and easily measurable soil properties to predict the TV in calcareous soils of Fars Province, southern Iran. The results showed that soil surface shear strength (SS) and aggregate mean weight diameter (MWD) were the most influential parameters affecting the TV. The predictive models developed using SS and penetration resistance (PR), as well as MWD and PR, had the highest performance for predicting the TV. PTF 5, which only required easily measurable soil properties, was selected as the final model for TV prediction.

LAND DEGRADATION & DEVELOPMENT (2022)

添加到收藏夹

Article Multidisciplinary Sciences

Novel deep learning hybrid models (CNN-GRU and DLDL-RF) for the susceptibility classification of dust sources in the Middle East: a global source

Hamid Gholami et al.

Summary: Dust storms have negative consequences on ecosystems, climate, and weather conditions. This study develops deep learning hybrid models to classify the susceptibility of dust sources in the Middle East and mitigate its negative effects.

SCIENTIFIC REPORTS (2022)

添加到收藏夹

Article Multidisciplinary Sciences

Assessment of the uncertainty and interpretability of deep learning models for mapping soil salinity using DeepQuantreg and game theory

Aliakbar Mohammadifar et al.

Summary: This research introduces a new combined modelling approach for mapping soil salinity in the Minab plain in southern Iran. Deep learning models were used and their uncertainty and interpretability were assessed. The results showed that the 1DCNN-LSTM hybrid model performed slightly better than the DBM model. Four controls (evaporation, sand content, precipitation, and vertical distance to channel) were identified as the most important factors for soil salinity.

SCIENTIFIC REPORTS (2022)

添加到收藏夹

Article Soil Science

Vis-NIR spectroscopy predicts threshold velocity of wind erosion in calcareous soils

Monireh Mina et al.

Summary: This study demonstrated that using reflectance spectroscopy combined with machine learning algorithms is a promising technique for large-scale assessment of wind erosion.

GEODERMA (2021)

添加到收藏夹

Article Environmental Sciences

Spatiotemporal modeling of land subsidence using a geographically weighted deep learning method based on PS-InSAR

Huijun Li et al.

Summary: The increasing demand for water resources during urbanization has led to continuous groundwater exploitation, resulting in severe land subsidence and posing a threat to sustainable development. Developing land subsidence models can help understand the phenomenon and enhance prevention strategies to address the issue.

SCIENCE OF THE TOTAL ENVIRONMENT (2021)

添加到收藏夹

Article Geography, Physical

Using the Boruta algorithm and deep learning models for mapping land susceptibility to atmospheric dust emissions in Iran

Hamid Gholami et al.

Summary: This study utilized the Boruta algorithm and two DL models to map land susceptibility to wind erosion in parts of southeastern Iran, identifying precipitation, digital elevation model, and soil organic carbon as the most important factors. Predictive models constructed using test and training datasets exhibited satisfactory performance, with the simple RNN model slightly outperforming the RBM model. The study concluded that these new methods in aeolian geomorphology may provide accurate spatial maps of dust sources for targeting mitigation of detrimental dust effects on climate, ecosystems, and human health.

AEOLIAN RESEARCH (2021)

添加到收藏夹

Article Geosciences, Multidisciplinary

Assessment of the interpretability of data mining for the spatial modelling of water erosion using game theory

Aliakbar Mohammadifar et al.

Summary: This study applied 15 data mining models to predict land susceptibility to water erosion hazard in the Kahorestan catchment, southern Iran, with BGAM and PLS showing the best and worst performance in predicting water erosion hazard. Game theory analysis identified factors extracted from DEM as the most important controls on predicted severity of soil erosion by water. Game theory was found to be a valuable technique for assessing the interpretability of predictive models.

CATENA (2021)

添加到收藏夹

Article Environmental Sciences

Evaluation of different boosting ensemble machine learning models and novel deep learning and boosting framework for head-cut gully erosion susceptibility

Wei Chen et al.

Summary: This study aimed to assess gully head-cut erosion susceptibility and identify erosion prone areas using boosting ensemble machine learning algorithms. Results showed altitude, land use, distance from roads, and soil characteristics as key factors affecting head-cut gully erosion. The DB method demonstrated higher predictive accuracy than other approaches, with potential applications in soil conservation and preventing threats to human activities.

JOURNAL OF ENVIRONMENTAL MANAGEMENT (2021)

添加到收藏夹

Article Computer Science, Information Systems

The Road Map to Classify the Potential Risk of Wind Erosion

Hana Stredova et al.

Summary: This paper aims to develop a methodological road map to identify key soil-climatic conditions that make soil vulnerable to wind erosion and demonstrate the road map in a case study using a relevant data source. By combining all relevant soil-climatic data and evaluating different mechanisms of wind erosion for sand and clay simultaneously, it is possible to set threshold values based on our method that trigger landscape adjustments or indicate the need for specific management.

ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION (2021)

添加到收藏夹

Article Environmental Sciences

Investigation of the 2018 Shiraz dust event: Potential sources of metals, rare earth elements, and radionuclides; health assessment

Sajjad Abbasi et al.

Summary: In May 2018, an unprecedented dust storm in Shiraz was found to be mainly influenced by air masses from Saudi Arabian deserts. The dust settled in Shiraz showed similarities with shale, sandstone, and Saudi Arabian soils, indicating their sources. High concentrations of toxic elements and rare earth elements in the dust were attributed to emissions from industrial activities and exhausts. The risk assessment revealed higher health risks for children compared to adults, with different pathways showing varying levels of risk.

CHEMOSPHERE (2021)

添加到收藏夹

Article Environmental Sciences

Optimization modelling to establish false measures implemented with ex-situ plant species to control gully erosion in a monsoon-dominated region with novel in-situ measurements

Asish Saha et al.

Summary: This study conducted gully erosion susceptibility mapping and land use planning optimization in rugged badland topography of eastern India using machine learning algorithms and an ecological geography based optimization model. The validation results showed that the Ecogeography based optimization (EBO) model was the most accurate model for predicting gully erosion susceptibility.

JOURNAL OF ENVIRONMENTAL MANAGEMENT (2021)

添加到收藏夹

Article Environmental Sciences

Integrated modelling for mapping spatial sources of dust in central Asia-An important dust source in the global atmospheric system

Hamid Gholami et al.

Summary: Spatial mapping of dust sources in Central Asia using integrated modelling approach including deep learning, game theory, and Gaussian copula-based multivariate model identified key factors for controlling dust emissions, assessed model performance, and determined hybrid copula-gcForest model as the most accurate for predicting susceptibility classes. Bulk density, precipitation, and coarse fragment content were highlighted as the most important factors contributing to the predictive model output.

ATMOSPHERIC POLLUTION RESEARCH (2021)

添加到收藏夹

Article Geosciences, Multidisciplinary