With the growing need for efficient mitigation strategies in liquefaction-prone regions, ensuring both seismic resilience and sustainability of infrastructure has become increasingly significant. This paper presents a datadriven probabilistic seismic demand model (PSDM) prediction and sustainability optimization framework to mitigate liquefaction-induced lateral deformation in regional mildly sloping ground improved with stone columns. The framework integrates finite element (FE) simulations with machine learning (ML) models, generating 1,200 ground FE models based on the key site attributes, such as ground inclination, soil properties, and stone column configurations. The performance of the selected ML models is evaluated through hyperparameter tuning by k-fold cross-validation, with the artificial neural network (ANN) outperforming other models in accurately predicting the PSDM. Subsequently, this framework is applied to a set of representative mildly sloping ground sites, enabling rapid PSDM prediction for each site with varying site attributes. Moreover, by incorporating cost and sustainability metrics, multi-objective optimization is performed using the developed ANN predictive model to maximize seismic performance while minimizing total carbon emissions and costs associated with ground improvement. Overall, the framework allows for rapid and accurate PSDM prediction and regional optimization, facilitating the identification of the optimal stone column configurations for efficient and sustainable liquefaction mitigation.

Evaluating the mechanical properties of deep soil mixing (DSM) requires destructive borehole coring because they are mainly situated underground. Although surface wave method offers potential for quality assurance, its complexity arises from multi-mode phenomena and the need to re-evaluate inversion results, often necessitating manual interpretation. This paper presents a data-driven surface wave framework to retrieve field DSM profile Vs over time, incorporating a mode-free forward operator and the Monte Carlo Tree Search (MCTS) inversion. Validations using synthetic data affirmed the framework's accuracy and efficiency in tracking subsurface shear wave velocity. In a real-world DSM site, the proposed method successfully captures the mechanical properties evolution across two DSM layers over the curing period, aligning well with site investigations and borehole coring. This pioneering monitoring framework integrates geotechnical engineering with geophysics expertise, underscoring the value of non-destructive seismic methods for measuring subsurface property evolution.

Data-driven constitutive models are increasingly addressing non-elastic and three-dimensional scenarios. However, their robustness can be significantly impacted by the inadequate integration of physical information. Accordingly, this study introduces a tensor-based physics-encoded neural network to characterize the constitutive behavior of soil, exemplified by isotropic hypoplasticity with dependency on pressure and porosity. The framework ensures strict adherence to fundamental physical laws for small strain, rate-independent isotropic constitutive behavior. The network utilizes stress tensor invariants and soil state parameters (porosity) as inputs, and outputs crucial coefficients for the tensorial constitutive relations. The model has been calibrated using only symmetric triaxial test data (both drained and undrained). The effectiveness of the neural network-based constitutive model has been validated through simulations of drained and undrained triaxial tests under various initial conditions, as well as boundary value problems with complex loading. The results demonstrate that the proposed model offers the following distinguishing advantages: 1) Applicability to both two- and threedimensional non-elastic cases, even when trained with two-dimensional data; 2) Strict adherence to fundamental physical laws, avoiding soft constraints; 3) An incremental, tensor-based architecture, suitable for integration in numerical software for boundary value problems.

Applications of Bayesian updating commonly treat soil parameters as random variables. A significant issue with this is that soil parameters are highly subjective. Therefore, using traditional parameter-based models, Bayesian analysis starts from a subjective prior and it is unclear how this may influence the overall results of a study. In this paper, Bayesian updating is combined with a data-driven method, known as CRACA (i.e., CReep And Consolidation Analysis), for predicting the settlement of embankments on soft soil. Importantly, the method directly ingests measured oedometer data and, therefore, avoids the subjectivity involved in parameter selection. Because parameters are not used, scaling factors are introduced that account uncertainty associated with the laboratory measurements and the automated interpretation process. These factors have an initial value of unity (returning the prior) and are updated in a Bayesian framework as settlement monitoring data are revealed over time to improve future forecasts. The model was applied to an embankment case history and was shown to result in a rapid improvement in the accuracy and a narrowing of the 95% confidence interval as settlement monitoring data are revealed to the model.

Large amounts of frozen carbon stored in the permafrost of the Tibetan Plateau (TP) are gradually thawing due to climate warming. The thaw of frozen carbon allows more active soil organic carbon (SOC) on the TP to participate in the global carbon cycle, which has usually been neglected in previous studies of permafrost carbon. This paper assesses the thawed SOC stock and its historical dynamics in TP permafrost areas based on a data driven scheme. The results show that the current permafrost area of the TP is 1.14 x 106 km2, and the SOC stock in the top 10 m of permafrost areas is 47.36 Pg. The active layer of the TP permafrost has a regional average thickness of 2.37 m during the baseline period (2000-2020), storing 18.19 Pg of SOC, accounting for about 38.4 % of the total SOC stock in the top 10 m, with the remaining more than 60 % perennially frozen in permafrost. The dynamics of thawed SOC are calculated based on baseline period SOC pools. Despite significant fluctuations, there is a general trend of 0.074 Pg SOC thaw per decade over 1901-2020 as the active layer has deepened. Significantly, the thaw trend of frozen SOC after 1976 is more prominent, reaching 0.420 Pg decade-1. Moreover, this paper reconstructs the SOC pool at the beginning of the last century by incorporating historical deep carbon emissions (1 10 m) in the current SOC pool. It recalculates the thaw trend of frozen carbon based on this and finds that the general thaw trend of SOC from 1901-2020 is more significant, with a rate of 0.092 Pg decade-1. Additionally, the dynamics of SOC pools in surface soils (0 1 m) of TP permafrost areas are also discussed. Our study highlights the importance of permafrost carbon thaw and provides valuable information for TP carbon studies.