Impact of sedimentary facies on machine learning of acoustic impedance from seismic data: lessons from a geologically realistic 3D model

2021 ◽  
pp. 1-52
Author(s):  
Hongliu Zeng ◽  
Yawen He ◽  
Leo Zeng
Keyword(s):  
Machine Learning ◽  
Seismic Data ◽  
Random Noise ◽  
Sedimentary Facies ◽  
Rock Properties ◽  
Prediction Errors ◽  
Case Scenario ◽  
Data Set ◽  
Single Well

We have developed a new machine-learning (ML) workflow that uses random forest (RF) regression to predict sedimentary-rock properties from stacked and migrated 3D seismic data. The training, validation, and testing are performed with 40 features extracted from a geologically realistic 46 × 66-trace model built in the Miocene Powderhorn Field in South Texas. We focus on the responses of the RF model to sedimentary facies and the strategies adopted to achieve better prediction with various data conditions. We apply explained variation (R2) and root-mean-square (rms) prediction errors to map the relationship between the quality of prediction and the sedimentary facies. In the single-well model, R2 and rms error maps highly resemble sand-percentage maps, or lithofacies maps, showing the facies control on the quality of the ML model. We observe that training with a small well data set (1–10 wells) leads to low and unstable test scores (R2 = 0.2–0.7). The R2 score increases and stabilizes with more (as many as 1000) training wells (R2 = 0.7–0.9), realizing the effective correction of facies bias. The stratigraphic and spatial features are useful and should be used. Weak to moderate random noise (−20 to −15 dB) slightly lowers the training score (R2 < 0.05) and should not be a major concern. Sparse well-supported models can outperform linear regression and model-based inversion and can be useful if caution is exercised. In the best-case scenario (500 wells), the predicted model largely duplicates the true model with a significant improvement in accuracy (R2 = 0.85) and stability. Such results can be applied in most, if not all, exploration and production practices.

scholarly journals Leveraging Road Characteristics and Contributor Behaviour for Assessing Road Type Quality in OSM

2021 ◽  
Vol 10 (7) ◽  
pp. 436
Author(s):  
Amerah Alghanim ◽  
Musfira Jilani ◽  
Michela Bertolotto ◽  
Gavin McArdle
Keyword(s):  
Machine Learning ◽  
Spatial Data ◽  
Classification Accuracy ◽  
Supervised Machine Learning ◽  
Machine Learning Techniques ◽  
Data Set ◽  
Semantic Inference ◽  
Road Type ◽  
The Impact

Volunteered Geographic Information (VGI) is often collected by non-expert users. This raises concerns about the quality and veracity of such data. There has been much effort to understand and quantify the quality of VGI. Extrinsic measures which compare VGI to authoritative data sources such as National Mapping Agencies are common but the cost and slow update frequency of such data hinder the task. On the other hand, intrinsic measures which compare the data to heuristics or models built from the VGI data are becoming increasingly popular. Supervised machine learning techniques are particularly suitable for intrinsic measures of quality where they can infer and predict the properties of spatial data. In this article we are interested in assessing the quality of semantic information, such as the road type, associated with data in OpenStreetMap (OSM). We have developed a machine learning approach which utilises new intrinsic input features collected from the VGI dataset. Specifically, using our proposed novel approach we obtained an average classification accuracy of 84.12%. This result outperforms existing techniques on the same semantic inference task. The trustworthiness of the data used for developing and training machine learning models is important. To address this issue we have also developed a new measure for this using direct and indirect characteristics of OSM data such as its edit history along with an assessment of the users who contributed the data. An evaluation of the impact of data determined to be trustworthy within the machine learning model shows that the trusted data collected with the new approach improves the prediction accuracy of our machine learning technique. Specifically, our results demonstrate that the classification accuracy of our developed model is 87.75% when applied to a trusted dataset and 57.98% when applied to an untrusted dataset. Consequently, such results can be used to assess the quality of OSM and suggest improvements to the data set.

A comparison of deep machine learning and Monte Carlo methods for facies classification from seismic data

Geophysics ◽  
2020 ◽  
Vol 85 (4) ◽  
pp. WA41-WA52 ◽  
Author(s):  
Dario Grana ◽  
Leonardo Azevedo ◽  
Mingliang Liu
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Monte Carlo ◽  
Monte Carlo Methods ◽  
Seismic Data ◽  
Transition Probabilities ◽  
Machine Learning Algorithms ◽  
Data Set ◽  
Facies Classification ◽  
The Neural Network

Among the large variety of mathematical and computational methods for estimating reservoir properties such as facies and petrophysical variables from geophysical data, deep machine-learning algorithms have gained significant popularity for their ability to obtain accurate solutions for geophysical inverse problems in which the physical models are partially unknown. Solutions of classification and inversion problems are generally not unique, and uncertainty quantification studies are required to quantify the uncertainty in the model predictions and determine the precision of the results. Probabilistic methods, such as Monte Carlo approaches, provide a reliable approach for capturing the variability of the set of possible models that match the measured data. Here, we focused on the classification of facies from seismic data and benchmarked the performance of three different algorithms: recurrent neural network, Monte Carlo acceptance/rejection sampling, and Markov chain Monte Carlo. We tested and validated these approaches at the well locations by comparing classification predictions to the reference facies profile. The accuracy of the classification results is defined as the mismatch between the predictions and the log facies profile. Our study found that when the training data set of the neural network is large enough and the prior information about the transition probabilities of the facies in the Monte Carlo approach is not informative, machine-learning methods lead to more accurate solutions; however, the uncertainty of the solution might be underestimated. When some prior knowledge of the facies model is available, for example, from nearby wells, Monte Carlo methods provide solutions with similar accuracy to the neural network and allow a more robust quantification of the uncertainty, of the solution.

Attenuation of random noise using denoising convolutional neural networks

2019 ◽  
Vol 7 (3) ◽  
pp. SE269-SE280
Author(s):  
Xu Si ◽  
Yijun Yuan ◽  
Tinghua Si ◽  
Shiwen Gao
Keyword(s):  
Neural Network ◽  
Data Processing ◽  
Seismic Data ◽  
Random Noise ◽  
Training Data ◽  
Prediction Errors ◽  
Network Learning ◽  
Deep Network ◽  
Residual Image ◽  
And Training

Random noise often contaminates seismic data and reduces its signal-to-noise ratio. Therefore, the removal of random noise has been an essential step in seismic data processing. The [Formula: see text]-[Formula: see text] predictive filtering method is one of the most widely used methods in suppressing random noise. However, when the subsurface structure becomes complex, this method suffers from higher prediction errors owing to the large number of different dip components that need to be predicted. Here, we used a denoising convolutional neural network (DnCNN) algorithm to attenuate random noise in seismic data. This method does not assume the linearity and stationarity of the signal in the conventional [Formula: see text]-[Formula: see text] domain prediction technique, and it involves creating a set of training data that are obtained by data processing, feeding the neural network with the training data obtained, and deep network learning and training. During deep network learning and training, the activation function and batch normalization are used to solve the gradient vanishing and gradient explosion problems, and the residual learning technique is used to improve the calculation precision, respectively. After finishing deep network learning and training, the network will have the ability to separate the residual image from the seismic data with noise. Then, clean images can be obtained by subtracting the residual image from the raw data with noise. Tests on the synthetic and real data demonstrate that the DnCNN algorithm is very effective for random noise attenuation in seismic data.

Reservoir Prediction Under Control of Sedimentary Facies

2017 ◽  
Vol 25 (03) ◽  
pp. 1750022
Author(s):  
Xiuwei Yang ◽  
Peimin Zhu
Keyword(s):  
Seismic Data ◽  
Low Frequency ◽  
Prediction Method ◽  
Rock Physics ◽  
Real Data ◽  
Sedimentary Facies ◽  
Seismic Inversion ◽  
Rock Properties ◽  
Reservoir Prediction ◽  
Frequency Model

Acoustic impedance (AI) from seismic inversion can indicate rock properties and can be used, when combined with rock physics, to predict reservoir parameters, such as porosity. Solutions to seismic inversion problem are almost nonunique due to the limited bandwidth of seismic data. Additional constraints from well log data and geology are needed to arrive at a reasonable solution. In this paper, sedimentary facies is used to reduce the uncertainty in inversion and rock physics modeling; the results not only agree with seismic data, but also conform to geology. A reservoir prediction method, which incorporates seismic data, well logs, rock physics and sedimentary facies, is proposed. AI was first derived by constrained sparse spike inversion (CSSI) using a sedimentary facies dependent low-frequency model, and then was transformed to reservoir parameters by sequential simulation, statistical rock physics and [Formula: see text]-model. Two numerical experiments using synthetic model and real data indicated that the sedimentary facies information may help to obtain a more reasonable prediction.

Stacking seismic data using local correlation

Geophysics ◽  
2009 ◽  
Vol 74 (3) ◽  
pp. V43-V48 ◽  
Author(s):  
Guochang Liu ◽  
Sergey Fomel ◽  
Long Jin ◽  
Xiaohong Chen
Keyword(s):  
Seismic Data ◽  
Field Data ◽  
Signal To Noise Ratio ◽  
Random Noise ◽  
Image Point ◽  
Local Correlation ◽  
Seismic Profiles ◽  
Signal To Noise ◽  
Prestack Migration

Stacking plays an important role in improving signal-to-noise ratio and imaging quality of seismic data. However, for low-fold-coverage seismic profiles, the result of conventional stacking is not always satisfactory. To address this problem, we have developed a method of stacking in which we use local correlation as a weight for stacking common-midpoint gathers after NMO processing or common-image-point gathers after prestack migration. Application of the method to synthetic and field data showed that stacking using local correlation can be more effective in suppressing random noise and artifacts than other stacking methods.

scholarly journals Protein structure prediction using AI and quantum computers

2021 ◽  
Author(s):  
Ben Geoffrey A S
Keyword(s):  
Machine Learning ◽  
Protein Structure ◽  
Protein Structure Prediction ◽  
Structure Prediction ◽  
Machine Learning Algorithms ◽  
Quantum Computers ◽  
Data Set ◽  
Backbone Atoms ◽  
Protein Contact Maps

This work seeks to combine the combined advantage of leveraging these emerging areas of Artificial Intelligence and quantum computing in applying it to solve the specific biological problem of protein structure prediction using Quantum Machine Learning algorithms. The CASP dataset from ProteinNet was downloaded which is a standardized data set for machine learning of protein structure. Its large and standardized dataset of PDB entries contains the coordinates of the backbone atoms, corresponding to the sequential chain of N, C_alpha, and C' atoms. This dataset was used to train a quantum-classical hybrid Keras deep neural network model to predict the structure of the proteins. To visually qualify the quality of the predicted versus the actual protein structure, protein contact maps were generated with the experimental and predicted protein structure data and qualified. Therefore this model is recommended for the use of protein structure prediction using AI leveraging the power of quantum computers. The code is provided in the following Github repository https://github.com/bengeof/Protein-structure-prediction-using-AI-and-quantum-computers.

Combining Supervised and Unsupervised Machine Learning Methods for Phenotypic Functional Genomics Screening

2020 ◽  
Vol 25 (6) ◽  
pp. 655-664
Author(s):  
Wienand A. Omta ◽  
Roy G. van Heesbeen ◽  
Ian Shen ◽  
Jacob de Nobel ◽  
Desmond Robers ◽  
...  
Keyword(s):  
Machine Learning ◽  
Training Set ◽  
Data Set ◽  
Genome Wide ◽  
Machine Learning Model ◽  
Exploratory Data ◽  
Interfering Rna ◽  
Insight Into ◽  
Training Sets

There has been an increase in the use of machine learning and artificial intelligence (AI) for the analysis of image-based cellular screens. The accuracy of these analyses, however, is greatly dependent on the quality of the training sets used for building the machine learning models. We propose that unsupervised exploratory methods should first be applied to the data set to gain a better insight into the quality of the data. This improves the selection and labeling of data for creating training sets before the application of machine learning. We demonstrate this using a high-content genome-wide small interfering RNA screen. We perform an unsupervised exploratory data analysis to facilitate the identification of four robust phenotypes, which we subsequently use as a training set for building a high-quality random forest machine learning model to differentiate four phenotypes with an accuracy of 91.1% and a kappa of 0.85. Our approach enhanced our ability to extract new knowledge from the screen when compared with the use of unsupervised methods alone.

Reservoir properties prediction integrating controlled-source electromagnetic, prestack seismic, and well-log data using a rock-physics framework: Case study in the Hoop Area, Barents Sea, Norway

2017 ◽  
Vol 5 (2) ◽  
pp. SE43-SE60 ◽  
Author(s):  
Pedro Alvarez ◽  
Amanda Alvarez ◽  
Lucy MacGregor ◽  
Francisco Bolivar ◽  
Robert Keirstead ◽  
...  
Keyword(s):  
Seismic Data ◽  
Barents Sea ◽  
Rock Physics ◽  
Positive Outcome ◽  
Rock Properties ◽  
Integrated Analysis ◽  
Reservoir Properties ◽  
Data Set ◽  
Controlled Source ◽  
Rotation Scheme

We have developed an example from the Hoop Area of the Barents Sea showing a sequential quantitative integration approach to integrate seismic and controlled-source electromagnetic (CSEM) attributes using a rock-physics framework. The example illustrates a workflow to address the challenges of multiphysics and multiscale data integration for reservoir characterization purposes. A data set consisting of 2D GeoStreamer seismic and towed streamer electromagnetic data that were acquired concurrently in 2015 by PGS provide the surface geophysical measurements that we used. Two wells in the area — Wisting Central (7324/8-1) and Wisting Alternative (7324/7-1S) — provide calibration for the rock-physics modeling and the quantitative integrated analysis. In the first stage of the analysis, we invert prestack seismic and CSEM data separately for impedance and anisotropic resistivity, respectively. We then apply the multi-attribute rotation scheme (MARS) to estimate rock properties from seismic data. This analysis verified that the seismic data alone cannot distinguish between commercial and noncommercial hydrocarbon saturation. Therefore, in the final stage of the analysis, we invert the seismic and CSEM-derived properties within a rock-physics framework. The inclusion of the CSEM-derived resistivity information within the inversion approach allows for the separation of these two possible scenarios. Results reveal excellent correlation with known well outcomes. The integration of seismic, CSEM, and well data predicts very high hydrocarbon saturations at Wisting Central and no significant saturation at Wisting Alternative, consistent with the findings of each well. Two further wells were drilled in the area and used as blind tests in this case: The slightly lower saturation predicted at Hanssen (7324/7-2) is related to 3D effects in the CSEM data, but the positive outcome of the well is correctly predicted. At Bjaaland (7324/8-2), although the seismic indications are good, the integrated interpretation result predicts correctly that this well was unsuccessful.

scholarly journals Prediction of Water Saturation from Well Log Data by Machine Learning Algorithms: Boosting and Super Learner

2021 ◽  
Vol 9 (6) ◽  
pp. 666
Author(s):  
Fahimeh Hadavimoghaddam ◽  
Mehdi Ostadhassan ◽  
Mohammad Ali Sadri ◽  
Tatiana Bondarenko ◽  
Igor Chebyshev ◽  
...  
Keyword(s):  
Machine Learning ◽  
Water Saturation ◽  
Machine Learning Algorithms ◽  
Rock Properties ◽  
Gradient Boosting ◽  
Data Set ◽  
Log Data ◽  
Gamma Density ◽  
Super Learner ◽  
Resistivity Log

Intelligent predictive methods have the power to reliably estimate water saturation (Sw) compared to conventional experimental methods commonly performed by petrphysicists. However, due to nonlinearity and uncertainty in the data set, the prediction might not be accurate. There exist new machine learning (ML) algorithms such as gradient boosting techniques that have shown significant success in other disciplines yet have not been examined for Sw prediction or other reservoir or rock properties in the petroleum industry. To bridge the literature gap, in this study, for the first time, a total of five ML code programs that belong to the family of Super Learner along with boosting algorithms: XGBoost, LightGBM, CatBoost, AdaBoost, are developed to predict water saturation without relying on the resistivity log data. This is important since conventional methods of water saturation prediction that rely on resistivity log can become problematic in particular formations such as shale or tight carbonates. Thus, to do so, two datasets were constructed by collecting several types of well logs (Gamma, density, neutron, sonic, PEF, and without PEF) to evaluate the robustness and accuracy of the models by comparing the results with laboratory-measured data. It was found that Super Learner and XGBoost produced the highest accurate output (R2: 0.999 and 0.993, respectively), and with considerable distance, Catboost and LightGBM were ranked third and fourth, respectively. Ultimately, both XGBoost and Super Learner produced negligible errors but the latest is considered as the best amongst all.

scholarly journals Design of a Fracture Detection System based on Deep Program in a Convolutional Neural Network

Webology ◽  
2021 ◽  
Vol 18 (2) ◽  
pp. 509-518
Author(s):  
Payman Hussein Hussan ◽  
Syefy Mohammed Mangj Al-Razoky ◽  
Hasanain Mohammed Manji Al-Rzoky
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Bone Fractures ◽  
Detection System ◽  
Training Data ◽  
Learning Models ◽  
Fracture Detection ◽  
Data Set ◽  
Final Fracture

This paper presents an efficient method for finding fractures in bones. For this purpose, the pre-processing set includes increasing the quality of images, removing additional objects, removing noise and rotating images. The input images then enter the machine learning phase to detect the final fracture. At this stage, a Convolutional Neural Networks is created by Genetic Programming (GP). In this way, learning models are implemented in the form of GP programs. And evolve during the evolution of this program. Then finally the best program for classifying incoming images is selected. The data set in this work is divided into training and test friends who have nothing in common. The ratio of training data to test is equal to 80 to 20. Finally, experimental results show good results for the proposed method for bone fractures.

Sign in / Sign up

Export Citation Format

Share Document