There are currently a significant number of two-dimensional (2-D) and three-dimensional (3-D) inversion codes available for magnetotelluric (MT) data. Through various 2-D inversion algorithms suggested so far, the classical Occam's inversion, the data space Occam's inversion, the nonlinear conjugate gradient (NLCG) method, and the Gauss–Newton (GN) method are fundamental driving methods to find optimum earth models, and OCCAM, DASOCC, NLCG, and MT2DInvMatlab are possible candidates one can find in the public domain that implement these algorithms for 2-D MT inversions, respectively. In this study, we investigate the pros and cons (strength and weakness) of these codes to help one use them efficiently in practical works and, as an introductory guide, further develop (sophisticate or extend) them, especially for the 3-D case. To achieve this goal, we applied each one of the four aforementioned codes on a profile of real MT field dataset. Then, further investigations have been done by performing several inversion tests to see how each code can find the appropriate model to reconstruct the subsurface resistivity structure. Numerical experiments show that the two parameters, regularization and target misfit, in addition to the main criteria of inversion (such as the forward and the sensitivities calculation method, and the type of inversion algorithm), are very important to produce the expected model in inversion. The regularization parameter that acts to trade off between model norm and data misfit can affect the inversion process in terms of both the computational efficiency and the accuracy of the obtained model. Also, lack of insufficient precision to choose the target misfit can lead the inversion to produce and reach an incorrect model. 相似文献
Petrographical and geochemical studies of Silurian Niur sandstones, Derenjal Mountains, Central Iran, were carried out to infer their provenance and tectonic setting. Modal analysis data of 37 medium sand size and well-sorted samples revealed that most quartz is composed of monocrystalline grains with straight to slightly undulos extinction and about 3 % polycrystalline quartz has inclusions, such as rutile needles. The sandstones are classified as quartzarenite, sublitharenite, and subarkose types based on framework composition and geochemistry. Petrographic studies reveal that these sandstones contain quartz, feldspars, and fragments of sedimentary rocks. The detrital modes of these sandstones indicate that they were derived from recycled orogen and stable cratonic source. Major and trace element contents of them are generally depleted (except SiO2) relative to upper continental crust which is mainly due to the presence of quartz and absence of Al-bearing minerals. Modal composition (e.g., quartz, feldspar, and lithic fragments) and discrimination diagrams based on major elements, trace elements (Ti, La, Th, Sc, and Zr), and also such ratios as La/Sc, Th/Sc, La/Co, and Th/Co, in sandstones suggest a felsic igneous source rock and quartzose polycyclic sedimentary provenance in a passive continental margin setting. Furthermore, high Zr/Sc values in these sandstones are considered as a sign of recycling. We indicated paleo-weathering conditions by modal compositions, the CIA index and Al2O3?+?K2O?+?Na2O% vs. SiO2% bivariate for these sandstones. Based on these results, although recycling is important to increase the maturity of the Niur sandstones, humid climate conditions in the source area have played a decisive role. 相似文献
Propagation of ion acoustic waves in plasmas containing electrons, positrons and high relativistic ions is investigated. It
is shown that the Korteweg-de Vries (KdV) equation describes the nonlinear waves in this media. The amplitude and energy of
the KdV solitary waves are derived and the effects of relativistic ions on these quantities are discussed. 相似文献
In this paper, the ion-acoustic solitons in a weakly relativistic electron-positron-ion plasma have been investigated. Relativistic
ions, Maxwell-Boltzmann distributed positrons and nonthermal electrons are considered in collisionless warm plasma. Using
a reductive perturbation theory, a Korteweg-de Vries (KdV) equation is derived, and the relativistic effect on the solitons
is studied. It is found that the amplitude of solitary waves of the KdV equation diverges at the critical values of plasma
parameters. Finally, in this situation, the solitons of a modified KdV (mKdV) equation with finite amplitude is derived. 相似文献
Ion acoustic shock waves (IASWs) are studied in a plasma consisting of electrons, positrons and ions. Boltzmann distributed
positrons and superthermal electrons are considered in the plasma. The dissipation is taken into account the kinematic viscosity
among the plasma constituents. The Korteweg–de Vries–Burgers (KdV–Burgers) equation is derived by reductive perturbation method.
Shock waves are solutions of KdV–Burgers equation. It is observed that an increasing positron concentration decreases the
amplitude of the waves. Furthermore, in the existence of the kinematic viscosity among the plasma, the shock wave structure
appears. The effects of ion kinematic viscosity (η0) and the superthermal parameter (k) on the ion acoustic waves are found. 相似文献
The time-varying Sun as the main source of space weather affects the Earth??s magnetosphere by emitting hot magnetized plasma in the form of solar wind into interplanetary space. Solar and geomagnetic activity indices and their chaotic characteristics vary abruptly during solar and geomagnetic storms. This variation depicts the difficulties in modeling and long-term prediction of solar and geomagnetic storms. On the other hand, the combination of neurofuzzy models and spectral analysis has been a subject of interest due to their many practical applications in modeling and predicting complex phenomena. However, these approaches should be trained by algorithms that need to be carried out by an offline data set, which influences their performance in online modeling and prediction of time-varying phenomena. This paper proposes an adaptive approach for multi-step ahead prediction of space weather indices by extending the regular singular spectrum analysis and locally linear neurofuzzy models to adaptive approaches. The combination of these recursive approaches fulfills requirements of long-term prediction of solar and geomagnetic activity indices. The results demonstrate the power of the proposed method in online prediction of space weather indices. 相似文献
Reservoir simulators model the highly nonlinear partial differential equations that represent flows in heterogeneous porous media. The system is made up of conservation equations for each thermodynamic species, flash equilibrium equations and some constraints. With advances in Field Development Planning (FDP) strategies, clients need to model highly complex Improved Oil Recovery processes such as gas re-injection and CO2 injection, which requires multi-component simulation models. The operating range of these simulation models is usually around the mixture critical point and this can be very difficult to simulate due to phase mislabeling and poor nonlinear convergence. We present a Machine Learning (ML) based approach that significantly accelerates such simulation models. One of the most important physical parameters required in order to simulate complex fluids in the subsurface is the critical temperature (Tcrit). There are advanced iterative methods to compute the critical point such as the algorithm proposed by Heidemann and Khalil (AIChE J 26,769–799, 1980) but, because these methods are too expensive, they are usually replaced by cheaper and less accurate methods such as the Li-correlation (Reid and Sherwood 1966). In this work we use a ML workflow that is based on two interacting fully connected neural networks, one a classifier and the other a regressor, that are used to replace physical algorithms for single phase labelling and improve the convergence of the simulator. We generate real time compositional training data using a linear mixing rule between the injected and the in-situ fluid compositions that can exhibit temporal evolution. In many complicated scenarios, a physical critical temperature does not exist and the iterative sequence fails to converge. We train the classifier to identify, a-priori, if a sequence of iterations will diverge. The regressor is then trained to predict an accurate value of Tcrit. A framework is developed inside the simulator based on TensorFlow that aids real time machine learning applications. The training data is generated within the simulator at the beginning of the simulation run and the ML models are trained on this data while the simulator is running. All the run-times presented in this paper include the time taken to generate the training data and train the models. Applying this ML workflow to real field gas re-injection cases suffering from severe convergence issues has resulted in a 10-fold reduction of the nonlinear iterations in the examples shown in this paper, with the overall run time reduced 2- to 10-fold, thus making complex FDP workflows several times faster. Such models are usually run many times in history matching and optimization workflows, which results in compounded computational savings. The workflow also results in more accurate prediction of the oil in place due to better single phase labelling.
