Exploration Geophysics - Volume 50, Issue 5, 2019

To improve the inversion results of roadway or tunnel transient electromagnetic detection data, particle swarm optimisation (PSO) and damped least squares (DLS) algorithms are combined. This overcomes the shortcomings of the PSO algorithm, namely that it has lower optimisation efficiency in the later stage, and the DLS algorithm, which requires manual assignment of initial values. With reference to actual stratum data, the mathematical model is established, and the inversion trial calculation is carried out using PSO, DLS and the proposed combined PSO–DLS algorithm. The results show that the combined PSO–DLS algorithm has higher optimisation efficiency without manual assignment of the starting model. The combined PSO–DLS algorithm is used to invert the measured transient electromagnetic data of the mine roadway, which realises separation of the mine roof and floor resistivity anomalies. Compared with PSO and DLS, the interpretation results for PSO–DLS are most consistent with the logging data, geological data and roadway exposure conditions.

 

Click to view

 

A three-dimensional (3D) magnetotelluric (MT) survey was carried out to delineate subsurface structures and possible fractures for geothermal development around Ulleung Island, Korea. Good-quality MT data were obtained throughout the survey region with respect to a remote reference located in Sawauchi, Japan, ∼ 900 km from the centre of the field site. 3D modelling and inversion were then performed, considering the sea effect on MT measurements near the coastline. The modelling results showed that seawater surrounding Ulleung Island affected MT data at frequencies below 10 Hz, depending on the distance from the shore. Based on this information, two types of 3D MT data inversion were performed: ordinary 3D inversion and 3D inversion using sea effects as a constraint. 3D inversion using the sea-effect constraint gave more reasonable and reliable results. In addition, the inversion that accounted for sea effects indicated a clear east–west conductive anomaly along the line connecting test boreholes GH-1 and GH-2; a high geothermal gradient of ∼ 100 °C/km was recorded. The geothermal gradients were lower, < 80 °C/km, at boreholes located in the southern (GH-3) and northern (GH-4) regions of the study area. These results suggest that deep circulation of fluid through deeply connected fractures, e.g. the anomaly observed in the east–west direction, carries heat from the bottom depths to near-surface layers. This would explain the large differences in gradients among sites in the study area.

Improving the accuracy of seismic wave propagation for imaging and inversion purposes often requires evaluating the validity of any underlying anisotropic assumption. Over the previous decades different models have been proposed to address the assumption of azimuthally anisotropic media; however, to our knowledge there is no published comparative analysis between these models that would allow practitioners to understand which provides more accurate theoretical predictions given specific field conditions. We evaluate two rock physics models for azimuthal anisotropy in widespread use (Mavko and Sayers) to determine which offers the better predictive power for benchmark laboratory data sets measured on three different kinds of dry rocks: Massillon Sandstone, Barre Granite and Ottawa Sand. We find that the Mavko model generally provides more accurate predictions, with a maximum 7% error for the consolidated Massillon Sandstone and Barre Granite rocks. Neither model provides very accurate approximations for Ottawa Sand due to the fact that this unconsolidated rock violates the underlaying assumption that the total rock compliance is affected only by the rock's matrix and crack compliances. We conclude that even though Mavko's approach provides more accurate predictions, both models are sufficiently accurate for simulating wave propagation in consolidated rocks with azimuthal anisotropy (e.g. well consolidated and cemented sandstones, and granites).

In this study, a data-driven linear filtering method is proposed to recover microseismic signals from noisy data/observations. The proposed method is based on the statistics of the background noise and the observation, which are directly extracted from the recorded data, obviating prior statistical knowledge of the microseismic source signal. The proposed method does not depend on any specific underlying noise statistics; therefore, it works for any type of noise, e.g. uncorrelated (random/white Gaussian), temporally correlated and spatially correlated noises. Consequently, the proposed method is suitable for microquake data sets that are recorded in contrastive noise environments. A mathematical analysis is presented to interpret the proposed method in two different ways. Furthermore, a number of practical concerns are discussed and their corresponding solutions are introduced. Finally, the proposed scheme is evaluated using both field and synthetic data sets and the experimental results show a reasonable and robust performance.

Fractures are important with regards to permeability within the basin subsurface; thus, understanding their generation within a given stress regime is crucial to the extraction of petroleum resources. A total of 517 naturally occurring fractures are identified on 12 resistivity image logs from the Exmouth and Beagle sub-basins, the Rankin Platform and the Exmouth Plateau in the Carnarvon Basin on the North West Shelf of Australia. The fractures have been identified on 12 image logs and can be divided into two sets: (1) electrically resistive and conductive fractures striking northeast–southwest; and (2) electrically resistive and conductive fractures striking east–west. There were 235 electrically resistive fractures identified that dominantly strike northeast–southwest, and 282 conductive fractures identified that dominantly strike east–west. The latter are considered to be open for fluid flow. The in situ stress field (orientations and magnitudes) is a major control on the ability of fractures to transmit fluid. This study identified 123 drilling-induced tensile fractures and 175 borehole breakouts present in 12 image logs, and a mean maximum horizontal stress orientation of 110°. Density logs and leak-off tests were used to calculate the in situ stress magnitudes with a vertical stress gradient () of 21.7 MPa km⁻¹, a minimum horizontal stress gradient of 16.8 MPa km⁻¹ and a maximum horizontal stress gradient of 23.4 MPa km⁻¹. This defines a strike-slip faulting stress regime for wells in this study in the Carnarvon Basin. The in situ stress and natural factures determined and identified in this study provide further clarity to the exploration and production processes occurring in the Carnarvon Basin.

Seismic inversion and reservoir identification are complicated problems with integrated data sets and need a velocity or density model to ensure the stability of the results. However, an elastic parameter model, such as velocity or density, is difficult to obtain with high precision by conventional migration or interpolation, especially during the early exploration stage, which contains few available logging data. The kriging method has proved to be an effective technique for mineral exploration and petroleum geophysics, and has led to a series of expansive techniques in earth science. In this paper, we proposed a new matching method, base value compensation (BVC), for integrating multiscale data sets into model elastic parameters. Considering the variety of original information, we used multiple constrain cokriging for three types of data set at different scales. These three data sets come from logging, seismic attributes and sedimentary facies information. The P-wave velocity model is stable and more representative of the subsurface condition, but we found that this method produces few points with anomalous values in the density model. Those abnormal points cause substantial loss of geological detail at sedimentary boundaries. In most modelling methods, this shortcoming is due to the fact that these three input parameters have significantly different properties. The proposed method can effectively solve this problem by matching the input data sets at a similar observational scale before modelling. We demonstrated this method using a case study in the South China Sea. The decrease in abnormal points in the final modelling results verifies the effectiveness of BVC.

We have studied the reflection dispersion signatures of a heterogeneously patchy saturated reservoir exhibiting attenuation and velocity dispersion in a seismic range of frequency. A modified patchy saturation model is employed to predict velocity dispersion and intrinsic attenuation behaviour associated with wave-induced pressure diffusion due to the presence of both mesoscopic and microscopic heterogeneities. Understanding the combined effects of microscopic squirt flow and wave-induced fluid flow related to mesoscopic heterogeneities on wave attenuation and associated dispersion would be important for determining the relative contribution of both of the interdependent energy loss mechanisms. The acoustic properties from the modified patchy saturation model were then employed in combination with an approximate expression for the frequency-dependent normal-incidence reflection coefficient of a dispersive reservoir rock, to give insights into the impacts of attenuation (and its associated velocity dispersion) and acoustic impedance contrast on the reflection dispersion characteristics. Analysis of the results indicates that the dispersion attribute of reflection coefficients and the associated cross-plot are more sensitive to fluid saturation and porosity than the magnitude of the reflection coefficient, and hence provide a promising approach to characterise fluid saturation and porosity in heterogeneous hydrocarbon reservoirs.

The wavelet transform is a well-known tool for the multiscale processing of geophysical data. Previous work has shown how it also could be used as a semi-automatic interpretation method for potential field data. This article demonstrates how to reduce the noise-sensitivity of the method when used in this manner, by using wavelets based on analytic signal amplitudes of different orders, and by using Hilbert transforms rather than horizontal derivatives. Although both the first horizontal derivative and the Hilbert transform of a data set change its phase by 90°, because Hilbert transforms do not boost the high-frequency component of a data set the method is less sensitive to noise. The method is applied to aeromagnetic data from the Cradle of Humankind, South Africa, with geologically reasonable results.

Compared with full elastic wave equations, pseudo-acoustic wave equations (PWEs) derived from the acoustic approximation can significantly reduce computational cost, thus they are widely used in seismic modelling, imaging and inversion in anisotropic media. However, these PWEs tend to suffer from inevitable SV-wave artefacts and numerical instabilities during wave propagation. The most effective solution to resolve the above issues is to adopt pure acoustic wave equations (PAWEs) which can produce pure P-wave responses without SV-wave residuals. Because the PAWE involves a complicated pseudo-differential operator (PDO), which is difficult to be solved using some common numerical algorithms. Here, we develop a new approach to implement pure acoustic wave propagation in 3D general anisotropic media. By minimising the difference between PDO and the new approximation, we compute optimal coefficients based on a least-squares method, and then derive the new time–space domain PAWEs in both 3D transversely isotropic and orthorhombic anisotropic media. Although the new PAWE contains a Poisson’s equation, we introduce a fast Poisson’s solver by using fast Fourier transform. To further improve computational efficiency, parameter gridding, coefficient reutilisation and graphic processing unit acceleration strategies are employed during wavefield extrapolation. Several numerical examples demonstrate that our proposed schemes can generate a pure and stable acoustic wavefield with high accuracy in general anisotropic media.