Perform Uncertainty Analysis Research Articles

Stochastic error propagation with independent probability distributions in LCA does not preserve mass balances and leads to unusable product compositions—a first quantification

Abstract Purpose To mitigate the effects of the triple planetary crisis of climate change, pollution and biodiversity loss, a system-based approach to estimating environmental impacts—such as life cycle assessment (LCA)—is critical. International standards recommend using uncertainty analysis to improve the reliability of LCA, but there has been debate about how to do this for many years. In particular, in order to characterise uncertainty in the inputs and outputs of each unit process in an LCA, a prevalent approach is to represent each one by an independent probability distribution. Thus, any physical relationships between inputs and outputs are ignored, which causes two potential errors during Monte Carlo simulation (a popular method for propagating uncertainty through an LCA model). First, the sum of the inputs to a unit process may not equal the sum of the outputs (i.e. there may be a mass imbalance), and second, the proportions of each input and output may be unrealistic (e.g. too much cement in a concrete production unit process). However, while some literature has discussed the problem, it has not yet been quantified. Methods Therefore, this paper investigates the extent to which existing uncertainty characterisation approaches, where there is a lack of parameterisation or correlations in databases, lead to mass imbalances and unrealistic variations in unit process compositions when performing uncertainty analysis. The matrix-based structure of LCA and the standard uncertainty analysis procedure using Monte Carlo (MC) simulation to propagate uncertainty are described. We apply the procedure to a concrete production process. Two uncertainty characterisation approaches are also explored to assess the effect of data quality scoring on mass imbalances and the mass contribution of each exchange (i.e. production compositions). Results and discussion For median data quality scores and using a typical (basic + additional uncertainty) uncertainty characterisation approach, the 1000-iteration MC simulation leads to mass imbalances ranging from − 49 to + 30% of the original mass and found that the mass imbalance exceeded existing prescribed plausibility limits on 62.7% of MC runs. On average across all exchanges, the exchange mass exceeded the 5% plausible variation limit on 77.7% of MC runs. This means that the final concrete product compositions are unlikely to be realistic or functionally equivalent to one another. We discuss the appropriateness of using universal variances for the underlying normal distribution for data quality scores (“additional uncertainty”) when input exchange quantities are of different scales. Additionally, we discuss potential solutions to the mass imbalance problem and their suitability for implementation at a database scale. Conclusions We have quantified, for the first time, the significant impact that uncertainty characterisation via independent probability distributions has on maintaining mass balances and plausible product compositions in unit processes. To overcome these challenges, databases would need to be parameterised and have the ability to sum quantities to perform mass balance checks during uncertainty analysis.

Predicting ground vibration during rock blasting using relevance vector machine improved with dual kernels and metaheuristic algorithms

The ground vibration caused by rock blasting is an extremely hazardous outcome of the blasting operation. Blasting activity has detrimental effects on both the ecology and the human population living in proximity to the area. Evaluating the magnitude of blasting vibrations requires careful evaluation of the peak particle velocity (PPV) as a fundamental and essential parameter for quantifying vibration velocity. Therefore, this study employs models using the relevance vector machine (RVM) approach for predicting the PPV resulting from quarry blasting. This investigation utilized the conventional and optimized RVM models for the first time in ground vibration prediction. This work compares thirty-three RVM models to choose the most efficient performance model. The following conclusions have been mapped from the outcomes of the several analyses. The performance evaluation of each RVM model demonstrates each model achieved a performance of more than 0.85 during the testing phase, there was a strong correlation observed between the actual ground vibrations and the predicted ones. The analysis of performance metrics (RMSE = 21.2999 mm/s, 16.2272 mm/s, R = 0.9175, PI = 1.59, IOA = 0.8239, IOS = 0.2541), score analysis (= 93), REC curve (= 6.85E−03, close to the actual, i.e., 0), curve fitting (= 1.05 close to best fit, i.e., 1), AD test (= 11.607 close to the actual, i.e., 9.790), Wilcoxon test (= 95%), Uncertainty analysis (WCB = 0.0134), and computational cost (= 0.0180) demonstrate that PSO_DRVM model MD29 outperformed better than other RVM models in the testing phase. This study will help mining and civil engineers and blasting experts to select the best kernel function and its hyperparameters in estimating ground vibration during rock blasting project. In the context of the mining and civil industry, the application of this study offers significant potential for enhancing safety protocols and optimizing operational efficiency.

Accelerating regional-scale groundwater flow simulations with a hybrid deep neural network model incorporating mixed input types: A case study of the northeast Qatar aquifer

ABSTRACT This study presents the ‘Dual Path CNN-MLP’, a novel hybrid deep neural network (DNN) architecture that merges the strengths of convolutional neural networks (CNNs) and multilayer perceptrons (MLPs) for regional groundwater flow simulations. This model stands out from previous DNN approaches by managing mixed input types, including both imagery and numerical vectors. Such flexibility allows the diverse nature of groundwater data to be efficiently utilized without the need to convert it into a uniform format, which often leads to oversimplification or unnecessary expansion of the dataset. When applied to the northeast Qatar aquifer, the model demonstrates high accuracy in simulating transient groundwater flow fields, benchmarked against the well-established MODFLOW model. The model's efficacy is confirmed through k-fold cross-validation, showing an error margin of less than 12% across all examined locations. The study also examines the model's ability to perform uncertainty analysis using Monte Carlo simulations, finding that it achieves around 1% average absolute percentage error in estimating the mean hydraulic head. Errors are mostly found in areas with significant variations in the hydraulic head. Switching to this ML model from the conventional MODFLOW simulator boosts computational efficiency by about 99%, showcasing its advantage for tasks like uncertainty analysis in repetitive groundwater simulations.

From threat to opportunity: Hydrologic uncertainty regionalization across large domains

Study regionThe study was carried out in Northern Canada (90,000 km2) and Southern Canada (18,000 km2). These basins represent case studies with a larger application target of high latitude regions. Study focusAs model domains become large, ungauged basins are inevitably encountered, and basin heterogeneity increases. This study aims to develop the computationally frugal Model Agnostic Uncertainty Transfer (MAUT) method to regionalize an uncertainty analysis from a set of donor basins to a receiver. The goal is an evaluation of the MAUT method for regionalizing over sufficiently heterogeneous domains to be applicable to high latitude data sparse regions. New hydrological insights for the regionWe propose the Model Agnostic Uncertainty Transfer (MAUT) method, a surrogate method to perform uncertainty analysis in large-domain modelling with a focus on high-latitude regions where data scarcity is especially severe. With the MAUT method, antecedent precipitation and simulated flow are related. Deviation from this relationship is assumed to represent modelling uncertainty and is quantified by quantile regression lines. These lines act as transfer functions requiring only antecedent precipitation to generate uncertainty bounds. The MAUT method requires uncertainty donors and a target receiver model. Key results suggest the method is relatively insensitive to precipitation dataset differences. Simulation length was the most sensitive input, with 15–20 years being ideal for reasonable results. Overall, the MAUT method is viable for high latitude domains.

Discharge estimation in compound channels with converging and diverging floodplains using an optimised Gradient Boosting Algorithm

ABSTRACT River discharge estimation is vital for effective flood management and infrastructure planning. River systems consist of a main channel and floodplains, collectively forming a compound channel, posing challenges in discharge calculation, particularly when floodplains converge or diverge. In the present study, ML algorithms such as XGBoost, CatBoost, and LightGBM were developed to predict discharge in a compound channel. PSO algorithm is applied for optimization of hyperparameters of gradient boosting models, denoted as PSO-XGBoost, PSO-LightGBM, and PSO-CatBoost. ML model discharge predictions were validated with existing empirical models and feature importance was explored using SHAP and sensitivity analysis. Results show that all three gradient-boosting algorithms effectively predict discharge in compound channels and are further enhanced by application of PSO algorithm. The R2 values for XGBoost, PSO-XGBoost, CatBoost, and PSO-CatBoost exceed 0.95, whereas they are above 0.85 for LightBoost and PSO-LightBoost. PSO-CatBoost performance is better than other models based on findings of statistical performance parameters, uncertainty analysis, reliability index, and resilience index for prediction of discharge in a compound channel with converging and diverging flood plains. The findings of this study validate the suitability of the proposed models especially optimized with PSO is recommended for predicting discharge in a compound channel.

GeoPDNN 1.0: a semi-supervised deep learning neural network using pseudo-labels for three-dimensional shallow strata modelling and uncertainty analysis in urban areas from borehole data

Abstract. Borehole data are essential for conducting precise urban geological surveys and large-scale geological investigations. Traditionally, explicit modelling and implicit modelling have been the primary methods for visualizing borehole data and constructing 3D geological models. However, explicit modelling requires substantial manual labour, while implicit modelling faces problems related to uncertainty analysis. Recently, machine learning approaches have emerged as effective solutions for addressing these issues in 3D geological modelling. Nevertheless, the use of machine learning methods for constructing 3D geological models is often limited by insufficient training data. In this paper, we propose the semi-supervised deep learning using pseudo-labels (SDLP) algorithm to overcome the issue of insufficient training data. Specifically, we construct the pseudo-labels in the training dataset using the triangular irregular network (TIN) method. A 3D geological model is constructed using borehole data obtained from a real building engineering project in Shenyang, Liaoning Province, NE China. Then, we compare the results of the 3D geological model constructed based on SDLP with those constructed by a support vector machine (SVM) method and an implicit Hermite radial basis function (HRBF) modelling method. Compared to the 3D geological models constructed using the HRBF algorithm and the SVM algorithm, the 3D geological model constructed based on the SDLP algorithm better conforms to the sedimentation patterns of the region. The findings demonstrate that our proposed method effectively resolves the issues of insufficient training data when using machine learning methods and the inability to perform uncertainty analysis when using the implicit method. In conclusion, the semi-supervised deep learning method with pseudo-labelling proposed in this paper provides a solution for 3D geological modelling in engineering project areas with borehole data.

On the Uncertainty Analysis of the Data-Enabled Physics-Informed Neural Network for Solving Neutron Diffusion Eigenvalue Problem

We performed uncertainty analysis and further numerical studies on the data-enabled physics-informed neural network (DEPINN). The purpose of DEPINN is to accurately and efficiently use a small amount of prior data to solve the neutron diffusion eigenvalue equations based on the physics-informed neural network. However, in practical engineering experiments, these prior data are acquired through different kinds of sensors, which are inevitably polluted by noise. Numerical results of three typical benchmark problems show that the classical DEPINN is not so robust with respect to noise. To improve the noise robustness, we propose an interval loss function to deal with the noisy prior data term; the weight of the noisy prior data term is also set to be noise dependent. Numerical results show that the proposed framework effectively enhances the robustness of DEPINN and improves the efficiency of utilizing the noisy prior data and thus promotes the engineering application of DEPINN.

High-fidelity Modeling of Multilayer Building Process in Electron Beam Powder Bed Fusion: Build-quality Prediction and Formation-Mechanism Investigation

High-fidelity simulations of powder bed fusion (PBF) additive manufacturing have made significant progress over the past decade. In this study, an efficient two-dimensional frame was developed for simulating the electron beam PBF process with hundreds of tracks for the direct prediction of the build quality. The applicable parameter range of the developed model was determined by comparing the heat transfer with that in three-dimensional cases. Subsequently, powder deposition and selective melting were coupled for a continuous simulation of the multilayer process. Three powder deposition models were utilized to generate random powder particles, and their effects on the packing structure and the resultant simulated build quality were investigated. The predicted build quality was validated using experimental results from independent studies. By reproducing the building process, the defect development mechanism in a multilayer process was revealed for the coalescence behaviors of randomly distributed powder particles, which also confirmed the importance of simulation at the high-fidelity powder scale. The effects of key process parameters during multilayer and multi-track processes on the build quality were systematically investigated. In particular, the formation statuses of all tracks during the simulated building process were recorded and analyzed statistically, which provided crucial information on the printing process for understanding the building mechanism or performing uncertainty analysis.

Flexural Performance Sensitivity and Uncertainty Analysis of Embedded Concrete Sleepers in Preplaced Aggregate Concrete

Flexural Performance Sensitivity and Uncertainty Analysis of Embedded Concrete Sleepers in Preplaced Aggregate Concrete

Large-scale agent-based simulation model of pedestrian traffic flows

Mobility patterns of pedestrians at a very high spatial and temporal resolution support urban planning strategies and facilitate a better understanding of the transport system. In this study, we develop an agent-based model that simulates disaggregated pedestrian traffic flows at a regional scale. This model is designed to overcome limitations of existing approaches in pedestrian traffic modelling by incorporating several decision processes that are defined by probabilistic rules. These rules include activity type, mode, and route choices. The results of a case study in Salzburg (Austria) show traffic flows concentrated in the city centre and along the river. Performed complexity analysis justifies the structure and logic of the modelled system by showing improvements in results during the stepwise implementation of model concepts. We also performed uncertainty analysis to provide information on the accuracy of the modelled results, showing strong and moderate correlations against observational data.

Uncertainty Analysis and Design of Air Suspension Systems for City Buses Based on Neural Network Model and True Probability Density

The accuracy of uncertainty analysis in suspension systems is closely tied to the precision of the probability distribution of sprung mass. Consequently, traditional assumptions regarding the probability distribution fail to guarantee the accuracy of uncertainty analyses results. To achieve more precise uncertainty analysis outcomes, this paper proposes a data-driven approach for analyzing the uncertainties in bus air suspension systems. Firstly, a bus vehicle dynamics model is established to investigate the influence of sprung mass on suspension system performance. Subsequently, a deep neural network model is trained using road test data, for the accurate identification of the sprung mass. The historical mass of the bus is then computed using vehicle network data to obtain the true probability density of the sprung mass. Lastly, the real probability distribution of the sprung mass is utilized to perform uncertainty analysis on the bus suspension system, and the results are compared with those obtained by assuming a probability distribution. Comparative analysis reveals substantial disparities in uncertainty response, with a maximum relative error of 9% observed for wheel dynamic loads, thus emphasizing the significance of precise probability distribution information concerning the sprung mass.

Numerical Evaluation of Hydrodynamic Coefficients in Roll Motion

The determination of ship motions in a seaway heavily depends on hydrodynamic coefficients. However, the coefficients related to roll-motion exhibit nonlinear behavior due to viscous effects and relatively small wave damping, which pose significant challenges. To overcome these challenges, several studies have employed computational fluid dynamics (CFD) to analyze hydrodynamic performance, leveraging the advancements in computational performance. Nevertheless, limited research has investigated the error and uncertainty of these numerical solutions. In this study, we conducted a forced oscillation test on a ship-like section that induces lateral vortex shedding using the OpenFOAM with overset grid method. We validated the numerical results of the roll hydrodynamic coefficients with experiments and the results of potential theory. We used the procedure proposed by the ITTC to perform uncertainty analysis on the grid size and time step size, which presented the overall uncertainty of the numerical analysis results. Our findings indicate that the grid uncertainty of the added moment of inertia was significant in the low frequency region, while the time step uncertainty was significant in the high frequency region.

Uncertainty analysis and optimization of laser thermal pain treatment

Uncertainty in operating parameters during laser thermal pain treatment can yield unreliable results. To ensure reliability and effectiveness, we performed uncertainty analysis and optimization on these parameters. Firstly, we conducted univariate analysis to identify significant operational parameters. Next, an agent model using RBNN regression determined the relationship between these parameters, the constraint function, and the target function. Using interval uncertainty analysis, we obtained confidence distributions and established a nonlinear interval optimization model. Introducing RPDI transformed the model into a deterministic optimization approach. Solving this with a genetic algorithm yielded an optimal solution. The results demonstrate that this solution significantly enhances treatment efficacy while ensuring temperature control stability and reliability. Accounting for parameter uncertainties is crucial for achieving dependable and effective laser thermal pain treatment. These findings have important implications for advancing the clinical application of this treatment and enhancing patient outcomes.

Modeling for dependent competing failure processes of subsea pipelines considering parameter uncertainty based on dynamic Bayesian network

Leakages in subsea pipelines can have serious environmental and economic consequences, making it essential to develop reliable evaluation methods to prevent failures. Up to now, no comprehensive assessment methods focusing on parameter uncertainty of lifetime distribution functions in the dependent competing failure of subsea pipelines has been developed. This paper proposes a novel method for evaluating the reliability of subsea pipelines with dependent competing failure processes and parameter uncertainty, using dynamic Bayesian network (DBN). The independent variables and dependent variables of physical models are converted into parent nodes and child nodes of DBNs, respectively. Uncertain parameters of the degradation failure and sudden failure equations are denoted by root nodes and assigned to probability distributions in the DBN model. The interaction between degradation failure and sudden failure of subsea pipelines is studied, and two DBNs of competing failure based on variable degradation increment and variable degradation rate are established. Mutual information analysis finds out the important parameters affecting the reliability of subsea pipelines. The parameters in degradation failure model have little effects on the failure of subsea pipelines at first, but their impact is increasing over time. According to 3-sigma rule, three different situations are defined to perform uncertainty analysis of parameters in the developed DBN model. It shows that increasing the values of uncertain parameters, the reliability of the subsea pipelines would have obvious decrease earlier.

A copula model of extracting DEM-based cross-sections for estimating ecological flow regimes in data-limited deltaic-branched river systems

For operational flood control and estimating ecological flow regimes in deltaic branched-river systems with limited surveyed cross-sections, accurate river stage and discharge estimation using public domain Digital Elevation Model (DEM)-extracted cross-sections are challenging. To estimate the spatiotemporal variability of streamflow and river stage in a deltaic river system using a hydrodynamic model, this study demonstrates a novel copula-based framework to obtain reliable river cross-sections from SRTM (Shuttle Radar Topographic Mission) and ASTER (Advanced Spaceborne Thermal Emission and Reflection) DEMs. Firstly, the accuracy of the CSRTM and CASTER models was assessed against the surveyed river cross-sections. Thereafter, the sensitivity of the copula-based river cross-sections was evaluated by simulating river stage and discharge using MIKE11-HD in a complex deltaic branched-river system (7000 km2) of Eastern India having a network of 19 distributaries. For this, three MIKE11-HD models were developed based on surveyed cross-sections and synthetic cross-sections (CSRTM and CASTER models). The results indicated that the developed Copula-SRTM (CSRTM) and Copula-ASTER (CASTER) models significantly reduce biases (NSE>0.8; IOA>0.9) in the DEM-derived cross-sections and hence, are capable of satisfactorily reproducing observed streamflow regimes and water levels using MIKE11-HD. The performance evaluation metrics and uncertainty analysis indicated that the MIKE11-HD model based on the surveyed cross-sections simulates with higher accuracies (streamflow regimes: NSE>0.81; water levels: NSE>0.70). The MIKE11-HD model based on the CSRTM and CASTER cross-sections, reasonably simulates streamflow regimes (CSRTM: NSE>0.74; CASTER: NSE>0.61) and water levels (CSRTM: NSE>0.54; CASTER: NSE>0.51). Conclusively, the proposed framework is a useful tool for the hydrologic community to derive synthetic river cross-sections from public domain DEMs, and simulate streamflow regimes and water levels under data-scarce conditions. This modelling framework can be easily replicated in other river systems of the world under varying topographic and hydro-climatic conditions.

Diving deeper: Mesopelagic fish biomass estimates comparison using two different models

A growing population on a planet with limited resources demands finding new sources of protein. Hence, fisheries are turning their perspectives towards mesopelagic fish, which have, so far, remained relatively unexploited and poorly studied. Large uncertainties are associated with regards to their biomass, turn-over rates, susceptibility to environmental forcing and ecological and biogeochemical role. Models are useful to disentangle sources of uncertainties and to understand the impact of different processes on the biomass. In this study, we employed two food-web models – OSMOSE and the model by Anderson et al. (2019, or A2019) – coupled to a regional physical–biogeochemical model to simulate mesopelagic fish in the Eastern Tropical South Pacific ocean. The model by A2019 produced the largest biomass estimate, 26 to 130% higher than OSMOSE depending on the mortality parameters used. However, OSMOSE was calibrated to match observations in the coastal region off Peru and its temporal variability is affected by an explicit life cycle and food web. In contrast, the model by A2019 is more convenient to perform uncertainty analysis and it can be easily coupled to a biogeochemical model to estimate mesopelagic fish biomass. However, it is based on a flow analysis that had been previously applied to estimate global biomass of mesopelagic fish but has never been calibrated for the Eastern Tropical South Pacific. Furthermore, it assumes a steady-state in the energy transfer between primary production and mesopelagic fish, which may be an oversimplification for this highly dynamic system. OSMOSE is convenient to understand the interactions of the ecosystem and how including different life stages affects the model response. The combined strengths of both models allow us to study mesopelagic fish from a holistic perspective, taking into account energy fluxes and biomass uncertainties based on primary production, as well as complex ecological interactions.

Integrated Science Teaching in Atmospheric Ice Nucleation Research: Immersion Freezing Experiments.

This paper introduces hands-on curricular modules integrated with research in atmospheric ice nucleation, which is an important phenomenon potentially influencing global climate change. The primary goal of this work is to promote meaningful laboratory exercises to enhance the competence of students in the fields of science, technology, engineering, and math (STEM) by applying an appropriate methodology to laboratory ice nucleation measurements. To achieve this goal, three laboratory modules were developed with 18 STEM interns and tested by 28 students in a classroom setting. Students were trained to experimentally simulate atmospheric ice nucleation and cloud droplet freezing. For practical training, this work utilized a simple freezing assay device called the West Texas Cryogenic Refrigerator Applied to Freezing Test (WT-CRAFT) system. More specifically, students were provided with hands-on lessons to calibrate WT-CRAFT with deionized water and apply analytical techniques to understand the physicochemical properties of bulk water and droplet freezing. All procedures to implement the developed modules were typewritten during this process, and shareable read-ahead exploration materials were developed and compiled as a curricular product. Additionally, students conducted complementary analyses to identify possible catalysts of heterogeneous freezing in the water. The water analyses included: pH, conductivity, surface tension, and electron microscopy-energy-dispersive X-ray spectroscopy. During the data and image analysis process, students learned how to analyze droplet freezing spectra as a function of temperature, screen and interpret the data, perform uncertainty analyses, and estimate ice nucleation efficiency using computer programs. Based on the formal program assessment of learning outcomes and direct (yet deidentified) student feedback, we broadly achieved our goals to (1) improve their problem-solving skills by combining multidisciplinary science and math skills and (2) disseminate data and results with variability and uncertainty. The developed modules can be applied at any institute to advance undergraduate and graduate curricula in environmental science.

Geomechanical assessment of a large-scale CO2 storage and insights from uncertainty analysis

Geological storage of CO2 provides a feasible solution to curb CO2 emissions to the atmosphere and relieve climate change. However, as CO2 is injected into the reservoir, the increased pore pressure may result in fault reactivations or damaging geomechanical changes to the caprock, which could be detrimental to the containment of injected CO2. On the other hand, uncertainty is inevitable due to the limited information of the complex subsurface system. In this study, we develop a workflow to assess the geomechanical impact of CO2 geological storage and include management of sources of uncertainties associated with the subsurface system. We utilize the hypothetical large-scale carbon sequestration situated in the Southern San Joaquin Valley in California, USA as a case study. We firstly set up a three-dimensional (3D) geomechanical model using a non-linear quasi-static finite element method. The applied workflow has features including: (1) Coulomb friction behavior to model fault slippage, (2) CAD representation of land surface and faults, (3) non-uniform pressure increase based on dynamic reservoir simulations, and (4) variable rock properties for formation layers. We perform uncertainty analyses using a response surface methodology and a Box-Behnken design to assess the significance of subsurface uncertainties to geomechanical responses. We find that the intact caprock will not have a shear failure even considering the subsurface uncertainties. However, the faults can reactivate and give potential flow pathways in the caprock. The presented workflow integrates advanced geomechanical models and uncertainty analysis, which can be readily applied to other CO2 storage projects to mitigate geomechanical sources of risk.

Development and implementation of a metaphase DNA model for ionizing radiation induced DNA damage calculation

Objective. To develop a metaphase chromosome model representing the complete genome of a human lymphocyte cell to support microscopic Monte Carlo (MMC) simulation-based radiation-induced DNA damage studies. Approach. We first employed coarse-grained polymer physics simulation to obtain a rod-shaped chromatid segment of 730 nm in diameter and 460 nm in height to match Hi–C data. We then voxelized the segment with a voxel size of 11 nm per side and connected the chromatid with 30 types of pre-constructed nucleosomes and 6 types of linker DNAs in base pair (bp) resolutions. Afterward, we piled different numbers of voxelized chromatid segments to create 23 pairs of chromosomes of 1–5 μm long. Finally, we arranged the chromosomes at the cell metaphase plate of 5.5 μm in radius to create the complete set of metaphase chromosomes. We implemented the model in gMicroMC simulation by denoting the DNA structure in a four-level hierarchical tree: nucleotide pairs, nucleosomes and linker DNAs, chromatid segments, and chromosomes. We applied the model to compute DNA damage under different radiation conditions and compared the results to those obtained with G0/G1 model and experimental measurements. We also performed uncertainty analysis for relevant simulation parameters. Main results. The chromatid segment was successfully voxelized and connected in bps resolution, containing 26.8 mega bps (Mbps) of DNA. With 466 segments, we obtained the metaphase chromosome containing 12.5 Gbps of DNA. Applying it to compute the radiation-induced DNA damage, the obtained results were self-consistent and agreed with experimental measurements. Through the parameter uncertainty study, we found that the DNA damage ratio between metaphase and G0/G1 phase models was not sensitive to the chemical simulation time. The damage was also not sensitive to the specific parameter settings in the polymer physics simulation, as long as the produced metaphase model followed a similar contact map distribution. Significance. Experimental data reveal that ionizing radiation induced DNA damage is cell cycle dependent. Yet, DNA chromosome models, except for the G0/G1 phase, are not available in the state-of-the-art MMC simulation. For the first time, we successfully built a metaphase chromosome model and implemented it into MMC simulation for radiation-induced DNA damage computation.

Machine learning opportunities to conduct high-fidelity earthquake simulations in multi-scale heterogeneous geology

The 2019 Le Teil earthquake is an illustrative example of a moderate (MW4.9) yet damaging event, occurring at shallow depth (≈1 km) in a region with little to no geophysical data available. Therefore, using a high-fidelity wave propagation code, we performed numerical simulations of the Le Teil earthquake in a highly uncertain framework, investigating several seismic sources and geological set-ups. With respect to the former aspect, a point-source model and an extended kinematic fault model were compared. The latter aspect was investigated by comparing a 1D-layered to a 3D geological model. Those models were enhanced with random fluctuations, in order to obtain three alternative non-stationary random geological fields. The synthetic waveforms obtained from regional geophysical models were globally coherent with the recorded ones. The extended fault source model seemed more realistic than the point-source model. In addition, some geological random fields improved the synthetics’ agreement with the recordings. However, the three random field samplings led to a high variability in induced ground motion responses. Given the computational burden of high-fidelity simulations, we used two dimensionality reduction methods, namely the Principal Component Analysis (PCA) and a deep neural network (3D UNet), to investigate this variability. The methods were applied to a database of 40,000 3D geological random fields. Both the PCA and the 3D UNet condensed the variability of the 3D geological fields into a few components. These were sufficient to reconstruct the original fields with great accuracy. More importantly, the seismic response arising from the propagation throughout the reconstructed fields was in excellent agreement with the response of the original geological fields in more than 75% of the dataset. By building a structured ensemble of complex geological fields from their reduced representation, it may become possible to find a relationship between the reduced representation and the generated ground motion. Thus, our study proves the interest of dimensionality reduction to perform uncertainty analyses in complex geological media.