Computers & Geosciences最新文献

With fiber-optic seismic acquisition development, continuous dense seismic monitoring is becoming increasingly more accessible. Repurposing fiber cables in telecommunication conduits makes it possible to run seismic studies at low cost, even in locations where traditional seismometers are not easily installed, such as urban areas. However, due to the large volume of continuous streaming data, data collected in such a manner will go to waste unless we significantly automate the processing workflow. We train a convolutional neural network (CNN) for earthquake detection using 3000 events from a publicly available catalog and data acquired over three years by fiber cables in telecommunication conduits under the Stanford University campus. We performed a hyperparameter search both on the network architecture itself (e.g., number of layers, number of parameters) and on its training parameters, showing that CNNs with a small number of layers are sufficient for performing this detection task with high accuracy. We introduce a novel method for combining the deep learning results on fiber-optic and seismometer data to improve detection accuracy, dramatically reducing the false detection rate that is often a problem when processing large time-scale noisy continuous data. Consequently, we demonstrate that enhancing two sparse seismometer stations with an urban fiber system allows for the reliable detection of small earthquakes despite a low signal-to-noise ratio. We scale this processing method over three years of continuous data and show that this system reliably detects local small-amplitude earthquakes down to magnitudes as low as 0.5, leading to the discovery of previously uncataloged events.

The simulation of geological facies in an unobservable volume is essential in various geoscience applications. Given the complexity of the problem, deep generative learning is a promising approach to overcome the limitations of traditional geostatistical simulation models, in particular their lack of physical realism. This research aims to investigate the application of generative adversarial networks and deep variational inference for conditionally simulating channelized reservoir in underground volumes. In this paper, we review the generative deep learning approaches, in particular the adversarial ones and the stabilization techniques that aim to facilitate their training. We also study the problem of conditioning deep learning models to observations through a variational Bayes approach, comparing a conditional neural network model to a Gaussian mixture model. The proposed approach is tested on 2D and 3D simulations generated by the stochastic process-based model Flumy. Morphological metrics are utilized to compare our proposed method with earlier iterations of generative adversarial networks. The results indicate that by utilizing recent stabilization techniques, generative adversarial networks can efficiently sample complex target data distributions.

Oil source correlation can be used to identify the origin of crude oil by linking crude oil to source rocks; however, the manual methods, which are limited by the sample or parameter quantity or imbalanced datasets, are facing uncertainties. Although the existing multivariate statistical techniques can alleviate this problem, they are facing difficulties in processing imbalanced datasets and quantifying source beds. Therefore, a novel oil-source correlation analysis model called SVM-SelectKBest combining a support vector machine (SVM) with a feature selection algorithm to mitigate the common issue of dataset imbalance in oil-source correlations is proposed in this paper. The SVM-SelectKBest offers advantages over normal SVM by dynamically selecting the most relevant features and fine-tuning model parameters to achieve higher accuracy and better generalizability in complex datasets. SVM compensates for class imbalances by heavily penalizing the misclassification of the minority class, and SelectKBest streamlines the feature set to enhance SVM's effectiveness on critical variables. Furthermore, a shallow neural network (SensoryAttentionNet) is introduced into the proposed model to address the issue of quantifying the source bed proportions in crude oil. The result show that SVM-SelectKBest has better performance in identifying key geochemical parameters and discriminating oil source correlation, its accuracy in unbalanced datasets is improved by near 40% compared to SVM. The model obtains 25 key geochemical parameters such as C17 n-heptadecane, Pr pristane, and C18 n-octadecane, it achieves F1 scores of 1.0 on the training, validation, and test sets. SensoryAttentionNet also performs robustly, with a low variance of 0.05 between its predicted and actual values. All the results indicate the effectiveness of the proposed method in dealing with the imbalance problem in oil-source source correlation datasets and in determining the proportional contribution of source beds in crude oil.

Digital elevation models obtained from LiDAR surveys typically have a few meters or sub-meter resolution. DEM-derived products in such a fine resolution may not be desired for several circumstances, such as matching the resolution with other spatial datasets, preparing input data for hydrological models, and reducing the computational cost. This leads to DEM coarsening for further river network extraction. An alternative could be to derive the river flow paths in the original DEM resolution and use this information to obtain the coarser river networks (a procedure known as flow directions upscaling). This approach is the macroscale hydrology benchmark for deriving river networks with spatial resolution on the order of a few kilometers or even larger, based on the available DEM with tenths or hundreds of meters resolution. However, no study has applied this procedure for the change of scale involving fine-resolution LiDAR DEM. This research evaluated for the first time in literature a flow direction upscaling algorithm for deriving relatively coarse-resolution (30, 100, and 200m) river networks from very fine-resolution (1 m) flow paths obtained from LiDAR DEM. Two river basins of contrasting characteristics located in Northeast Brazil are studied. Results were evaluated through visual inspection, percentage within buffer (PWB) metrics, and river length comparison. It is shown that using an upscaling algorithm improves the ability of the coarse network to preserve river networks’ spatial patterns across multiple scale changes. Considering both basins, PWB ranged from 80% to 100% (average of 97%) for the upscaling procedure, while the DEM resampling resulted in PWB between 40% and 100% (average of 85%). A flow direction upscaling algorithm already used for macroscale hydrology proved helpful for the LiDAR-related shift in scale, outperforming the DEM resampling. Increasing the scale change augments the difference in performance between them, making the upscaling procedure more recommended. In addition, such an upscaling procedure provided drainage networks in the 100-m and 200-m resolutions with higher quality than the one obtained in the 30-m resolution directly from a globally available DEM.

In this paper, we present a new open-source software “Open-AMA” developed to investigate atmospheric circulation dynamics and environmental research. Open AMA presents an integral package to conduct several air mass analyses. It appears to be a powerful, versatile software package developed to meet the needs of researchers using python and C++ in order to facilitate and speed up working time. This software seamlessly integrates new models for source identification based on air trajectories and ambient air pollution concentration data and enhances certain existing ones. Beyond source identification, it offers a rich array of functionalities for making it automatic, quick and easy to get access many kinds data including gridded meteorological data, trajectory calculations, synoptic parameter extraction from back-trajectories. All this functionalities can be used through a user-friendly graphical interface. Open-AMA can be a significant leap forward in air quality research and analysis, empowering researchers with the tools they need to make informed decisions and address pressing environmental and public health challenges and enhance understanding of pollutant origins in the atmosphere.

Spectroscopic analytical techniques such as Fourier Transform Infrared Spectroscopy (FTIR), Raman or hyperspectral imaging are important in modern geosciences. In recent years there has been a shift from using exclusively single-spot analyses to 2-dimensional maps. Maps help to reveal patterns in a sample that would not be detected by single point measurements. Filtering and extracting signal information from multiple combined pixels can help improve the signal-to-noise ratio and thus the precision of the data. The combination of multi-layer numerical datasets obtained from different instruments or measurement settings opens up the possibility of exploring and investigating individual datasets in much greater detail. However, the amount of data and information in the dataset increases significantly when high-resolution spectroscopic and spatial data is used instead of spot analyses, thus making the data examination and data validation more challenging and time consuming. To investigate large datasets, we have developed SpecXY, a software solution for preparing, editing, extracting, and comparing spatially resolved spectral datasets. SpecXY aims to provide a user-friendly and open-source software solution for working with spectroscopic data by providing a simple user interface that is accessible to all users with basic computer skills. Advanced users with a basic understanding of MATLAB® programming can adapt, customise and extend SpecXY due to its modular and function-based program structure. SpecXY also provides innovative algorithms for analyzing spectral data, such as Monte Carlo deconvolution of peaks, and hybrid classification and filtering based on spectra in combination with user knowledge. Two examples illustrate possible applications of SpecXY: (1) multidimensional classification and correlation of spatially resolved spectroscopic data and quantified chemical element maps, and (2) classification, filtering, quantification of H₂O in minerals and profile extraction of a high-resolution spectroscopic data set measured by Fourier Transform Infrared (FTIR) Spectroscopy coupled to a Focal Plane Array (FPA) detector.

Nowadays, there is a growing trend that direct current (DC) field surveys are shifting towards challenging areas characterized by mountainous topography and electrical anisotropy. Given these complex geological settings, there is an urgent need for 3-D DC forward modeling software capable of effectively addressing large-scale problems and delivering accurate modeling results to interpret field data. However, most open-source software packages face certain limitations, such as the high numerical cost to handle complex surface topography, the lack of consideration for anisotropic conductivity, the absence of mesh refinement techniques to guarantee accuracy in forward modeling, and the lack of parallel computing techniques to solve large-scale problems. In this study, we develop an efficient and highly accurate 3-D DC anisotropic forward modeling software, namely DC3DPAFEM, using the adaptive finite element algorithm based on the unstructured tetrahedral mesh. Firstly, we construct a strong compatible boundary value problem (BVP) for 3-D anisotropic DC problems by adopting a specialized secondary potential approach to handle the surface topography efficiently. Then, we develop a goal-oriented adaptive mesh refinement (AMR) technique to ensure accurate forward modeling results, even with a coarse initial mesh. To ensure time and memory efficiency, we employ a robust conjugate gradient (CG) algorithm preconditioned by the algebraic multigrid (AMG) solver to solve the large-scale linear system of equations resulting from complex geological structures. We aim to investigate the performance of the AMG scheme in anisotropic DC cases. Furthermore, we incorporate the domain decomposition technique into the iterative solution scheme for further efficiency gains. This technique significantly improves computing efficiency for large-scale problems in parallel clusters. Finally, we conduct comprehensive performance tests for DC3DPAFEM using a two-layer anisotropic model and a 3-D complex model with undulating terrain. The results of both examples validate the accuracy of DC3DPAFEM, as they closely align with the analytical solutions and the solutions obtained from the existing 3-D DC forward modeling code. Compared to traditional direct solver MUMPS and ILU-preconditioned iterative solvers, DC3DPAFEM exhibits highly scalable performance for large-scale problems, offering significant advantages in terms of memory and time consumption. Overall, DC3DPAFEM demonstrates substantial advances in efficiency, accuracy, and practicality through a series of numerical examples. This open-source code provides an efficient and available tool for developing a 3-D DC inversion method that can deal with large-scale problems involving intricate topography and anisotropic media.

We present $ConvEQ$ as a tool for discriminating seismic phases, leveraging artificial intelligence technique (Convolutional Neural Network) for short-time Frequency Transform of the seismic signal. Timely detection of the vertical ($P$) wave from an earthquake can generate a warning several tens of precious seconds before the more destructive waves strike. We propose a train-for-each-station approach for an Internet-of-Things-based Smart Earthquake Early Warning System, where lightweight neural networks trained for the seismic data belonging to each station are implemented on edge devices directly interfaced with seismometers. The approach has the potential to get the most from the sparse seismic network for Pakistan and other third-world countries. We train networks for multi-station and single-station data and achieve 96% and 99% accuracy, respectively, proving that train-for-each-station maximizes accuracy. The total processing time (including preprocessing and inference) is about $30\mathrm{ms}$ for each event, thus suitable for real-time deployment. We further compare the performance of $ConvEQ$ on simulated real-time data with several state-of-the-art contemporary algorithms. Our proposed approach demonstrates a robust response on diverse metrics. The $ConvEQZ$ classifies the vertical seismic signal component with high accuracy and the $ConvEQX$ can classify any seismic data component, inculcating robustness against connectivity issues.

Traditional methods for fire point segmentation (FPS) in satellite remote sensing images (RSIs) overly rely on threshold judgment, which are greatly affected by factors such as regional time and show poor generalization. Besides, due to the difference between natural scene images (NSIs) and RSIs, directly apply NSIs-based deep learning methods to forest fire RSIs without any modification fails to achieve satisfactory results. To address these issues, first, we construct a Landsat8 RSI-FPS dataset covering different years, seasons and regions. Then, for the first time, we apply salient object detection (SOD) to FPS in forest fire monitoring and propose a novel network FPS-U²Net to improve the performance of FPS. FPS-U²Net is based on U²Netp (a lightweight U²Net), to make better use of the multi-level features from adjacent encoders, we propose multi-level aggregation module (MAM), which is placed between the encoder and decoder at the same stage to aggregate the adjacent multi-scale features and capture richer contextual information. To make up for the weakness of BCE loss, we employ the hybrid loss, BCE + IoU, for the training of the network, which can guide the network learn the salient information from pixel and map levels. Extensive experiments on three datasets demonstrate that our FPS-U²Net significantly outperforms the state-of-the-art semantic segmentation and SOD methods. FPS-U²Net can accurately segment fire regions and predict clear local details.

The simultaneous prediction of the subsurface distribution of facies and acoustic impedance ($I_P$) from fullstack seismic data requires solving an inverse problem and is fundamental in natural resources exploration, carbon capture and storage, and environmental risk management. In recent years, deep generative models (DGM), such as variational autoencoders (VAE) and generative adversarial networks (GAN), were proposed to reproduce complex facies patterns honoring prior geological information. Variational Bayesian inference using inverse autoregressive flows (IAF) can be performed to infer the solution to a geophysical inverse problem from the encoded latent space of such pre-trained DGM. Successful applications of such approach on crosshole ground-penetrating radar synthetic data inversion demonstrated that the technique's accuracy is comparable to that of Markov chain Monte Carlo (MCMC) inference methods, while significantly reducing the computational cost. Nonetheless, these application examples did not account for the spatial uncertainty affecting the facies-dependent continuous physical property, from which the geophysical data are calculated. This uncertainty can significantly affect the inversion accuracy and its applicability to real data. In this work, specific VAE and GAN architectures are proposed to simultaneously predict facies and co-located $I_P$, while accounting for their spatial uncertainties. The two types of generative networks are used in Bayesian inversion with IAF for the inversion of seismic data. The results are found to reproduce the statistics of the training images and solve the seismic inversion problem accurately, comparably to MCMC inversion. Furthermore, advantages and limitations of the two DGMs are evaluated by comparing the results obtained.