Permafrost, a major component of the cryosphere, is undergoing rapid degradation due to climate change, human activities, and other external disturbances, profoundly impacting ecosystems, hydroclimate, engineering geological stability, and infrastructure. In Northeast China, the thermal dynamics of Xing'an permafrost (XAP) are particularly complex, complicating the accurate assessment of its spatial extent. Many earlier mapping efforts, despite significant progress, fall short in accounting for some key local geo-environmental factors. Thus, this study introduces a new approach that incorporates four key driving factors-biotic, climatic, physiographic, and anthropogenic-by integrating multisource datasets and in situ observations. Four machine learning (ML) models [random forest (RF), support vector machine (SVM), logistic regression (LR), and extreme gradient boosting (XGB)] are applied to simulate permafrost distribution and probability, as well as to evaluate their performance. The results indicate that models' accuracy, ranked from highest to lowest, is as follows: RF (area under the curve (AUC) =0.88 and accuracy =0.81), XGB (0.86 and 0.77), LR (0.81 and 0.73), and SVM (0.76 and 0.66), with RF emerging as the most effective model for permafrost mapping in Northeast China. Analysis of the relationships between predictors and permafrost occurrence probability (POP) indicates that vegetation and snow cover exert nonlinear effects on permafrost, while human activities significantly reduce POP. Additionally, finer soil textures and higher soil organic matter content are positively correlated with increased POP. The modeling results, combined with field survey data, also show that permafrost is more prevalent in lowlands than in uplands, confirming the symbiotic relationship between permafrost and wetlands in Northeast China. This spatial variation is influenced by local microclimates, runoff patterns, and soil thermal properties. The primary sources of model error are uncertainties in the accuracy of multisource datasets at different scales and the reliability of observational data. Overall, ML models demonstrate great potential for mapping permafrost in Northeast China.

Landslides are destructive natural disasters that cause catastrophic damage and loss of life worldwide. Accurately predicting landslide displacement enables effective early warning and risk management. However, the limited availability of on-site measurement data has been a substantial obstacle in developing data-driven models, such as state-of-the-art machine learning (ML) models. To address these challenges, this study proposes a data augmentation framework that uses generative adversarial networks (GANs), a recent advance in generative artificial intelligence (AI), to improve the accuracy of landslide displacement prediction. The framework provides effective data augmentation to enhance limited datasets. A recurrent GAN model, RGAN-LS, is proposed, specifically designed to generate realistic synthetic multivariate time series that mimics the characteristics of real landslide on-site measurement data. A customized moment-matching loss is incorporated in addition to the adversarial loss in GAN during the training of RGAN-LS to capture the temporal dynamics and correlations in real time series data. Then, the synthetic data generated by RGAN-LS is used to enhance the training of long short-term memory (LSTM) networks and particle swarm optimization-support vector machine (PSO-SVM) models for landslide displacement prediction tasks. Results on two landslides in the Three Gorges Reservoir (TGR) region show a significant improvement in LSTM model prediction performance when trained on augmented data. For instance, in the case of the Baishuihe landslide, the average root mean square error (RMSE) increases by 16.11%, and the mean absolute error (MAE) by 17.59%. More importantly, the model's responsiveness during mutational stages is enhanced for early warning purposes. However, the results have shown that the static PSO-SVM model only sees marginal gains compared to recurrent models such as LSTM. Further analysis indicates that an optimal synthetic-to-real data ratio (50% on the illustration cases) maximizes the improvements. This also demonstrates the robustness and effectiveness of supplementing training data for dynamic models to obtain better results. By using the powerful generative AI approach, RGAN-LS can generate high-fidelity synthetic landslide data. This is critical for improving the performance of advanced ML models in predicting landslide displacement, particularly when there are limited training data. Additionally, this approach has the potential to expand the use of generative AI in geohazard risk management and other research areas. (c) 2024 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

Stubble burning is a conventional technique of residue management that has affected the physio-chemical properties of the soils. In soil sciences, dielectric properties of soils using radio and microwave-based remote sensing have huge applications. Thus, presented paper has studied the burning effects of stubble on soil's physical, chemical and dielectric properties ($\varepsilon {{\prime}} $epsilon ' and $\varepsilon {{\prime \prime}}$epsilon ''). Moreover, the experimentally observed soil's dielectric data has been explored with various classical Machine Learning (ML) and Neural Network (NN) based regression models. The soil samples were taken from the fields of Punjab, India, in the October-November months following a multistage soil sampling method. Then, Dak-12 open-ended coaxial probe (DOCP) has been used in alliance with a two-port Vector Network Analyzer (VNA) E5071C, Agilent Technologies, to investigate the dielectric properties of soil samples. The obtained results indicate that physio-chemical and dielectric properties have been strongly affected by burning as well as because of the presence of high concentrations of ash residues.$ \varepsilon {{\prime}} $epsilon ' and $\varepsilon {{\prime \prime}}$epsilon '' variations with depth indicate that ash residues can seep up to depths of 10 cm in a single burning process. Moreover, the continuous burning of stubble can have permanent effects on soil's properties. Among considered regression models, the Deep NN-based regression model has given the most accurate predictions of the regressor variables $\varepsilon {{\prime}} $epsilon ' and $\varepsilon {{\prime \prime}}$epsilon '', with a root-mean-square-error (RMSE) of 0.06 and 0.07, respectively. Stubble burning has visible effects on physical, chemical as well as dielectric properties of soil. The burning of stubble damages natural ecosystem and essential nutrients which decrease fertility of soil. Also, the resultant residue ash becomes permanent part of soil profile and alters basic properties of soil. Moreover, exploration of ML-based regression models suggests the tremendous applications of data-centric models in soil and material sciences.