In full waveform inversion (FWI), Hessian information of the misfit function is of vital importance for accelerating the convergence of the inversion; however, it usually is not feasible to directly calculate the Hessian matrix and its inverse. Although the limited memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) or Hessian-free inexact Newton (HFN) methods are able to use approximate Hessian information, the information they collect is limited. The two methods can be interlaced because they are able to provide Hessian information for each other; however, the performance of the hybrid iterative method is dependent on the effective switch between the two methods. We have designed a new scheme to realize the dynamic switch between the two methods based on the decrease ratio (DR) of the misfit function (objective function), and we propose a modified hybrid iterative optimization method. In the new scheme, we compare the DR of the two methods for a given computational cost, and choose the method with a faster DR. Using these steps, the modified method always implements the most efficient method. The results of Marmousi and over thrust model testings indicate that the convergence with our modified method is significantly faster than that in the L-BFGS method with no loss of inversion quality. Moreover, our modified outperforms the enriched method by a little speedup of the convergence. It also exhibits better efficiency than the HFN method. 相似文献
Numerical implementation of the gradient of the cost function in a gradient‐based full‐ waveform inversion (FWI) is essentially a migration operator used in wave equation migration. In FWI, minimizing different data residual norms results in different weighting strategies of data residuals at receiver locations prior to back‐propagation into the medium. In this paper, we propose different scaling methods to the receiver wavefield and compare their performances. Using time‐domain reverse‐time migration (RTM), we show that compared to conventional algorithms, this type of scaling is able to significantly suppress non‐Gaussian noise, i.e., outliers. Our tests also show that scaling by its absolute norm produces better results than other approaches. 相似文献
Herein we investigate the coupled orbital and rotational dynamics of two rigid bodies modelled as polyhedra, under the influence of their mutual gravitational potential. The bodies may possess any arbitrary shape and mass distribution. A method of calculating the mutual potential’s derivatives with respect to relative position and attitude is derived. Relative equations of motion for the two body system are presented and an implementation of the equations of motion with the potential gradients approach is described. Results obtained with this dynamic simulation software package are presented for multiple cases to validate the approach and illustrate its utility. This simulation capability is useful both for addressing questions in dynamical astronomy and for enabling spacecraft missions to binary asteroid systems. 相似文献
The present paper presents the results of a technique based on Vertical Seismic Profiling (VSP) to quantify the offset of a fault in a context where seismic reflection profiles do not image any reflectors. The case study is located in Greece, in the Aigion area, on the south border of the Corinth Gulf. The Aigion fault, oriented East–West, and dipping at 60°N, has been intersected by the scientific well Aig-10 at 760 m in depth, but some uncertainty remained concerning its exact offset since the depth of the pre-rift sequence is open to debate due to the lack of subsurface data. The pre-rift consists of a Mesozoic low-porosity series (carbonates and radiolarite), while the syn-rift consists of poorly compacted conglomerates and turbidites.
Seismic diffractions on fault edges are expected to be present along the fault surface. We find that a few diffracted events are effectively recorded in P and S wave mode by the 4 components of the VSP survey, and critically refracted arrivals along the Aigion fault plane. These singular events have been used in order to refine the fault geometry and to determine its throw. Additionally, results from a 2D finite difference elastic seismic model of a single fault step geometry have been closely examined in order to illustrate the generation of the singular seismic events observed on the field data, such as P-wave and S-wave refracted and diffracted events, and support their interpretation with higher confidence. As a result, the seismic arrival patterns from full waveform seismic modelling confirm the characterisation of the main fault geometry (mainly its throw: about 200 m) in the well vicinity derived from the actual analysis of the 4 Component Aigion VSP data. 相似文献
Using infrared sensors to detect ice clouds in different atmospheric layers is still a challenge. The different scattering and absorption properties of longwave and shortwave infrared channels can be utilized to fulfill this purpose.In this study, the release of Suomi-NPP Cross-track Infrared Sounder(Cr IS) full spectrum resolution is used to select and pair channels from longwave(~ 15 μm) and shortwave(~4.3 μm) CO_2 absorption bands under stricter conditions, so as to better detect ice clouds. Besides, the differences of the weighting function peaks and cloud insensitive level altitudes of the paired channels are both within 50 h Pa so that the variances due to atmospheric conditions can be minimized. The training data of clear sky are determined by Visible Infrared Imaging Radiometer Suite(VIIRS) cloud mask product and used to find the linear relationship between the paired longwave and shortwave CO_2 absorption channels. From the linear relationship, the so-called cloud emission and scattering index(CESI) is derived to detect ice clouds. CESI clearly captures the center and the ice cloud features of the Super Typhoon Hato located above 415 h Pa. Moreover, the CESI distributions agree with cloud top pressure from the VIIRS in both daytime and nighttime in different atmospheric layers. 相似文献