On the identification of ship coupled heave–pitch motions using neural networks

This paper outlines a procedure for the derivation of the differential equations describing the free response of a heaving and pitching ship from its stationary response to random waves. The coupled heave–pitch motion of a ship in random seas is modelled as a multi-dimensional Markov process. The partial differential equation describing the transition probability density function, known as the Fokker-Planck equation, for this process is derived. The Fokker-Planck equation is used to derive the random decrement equations for the coupled heave–pitch motion. The parameters in these equations are then identified using a neural network approach. The method is validated using numerical simulations and experimental results. The experimental data was obtained using an icebreaker ship model heaving and pitching in random waves. It is shown that the method produces good results when the system is lightly damped. An extension for using this method to identify couple heave–pitch motion in realistic seas is suggested.     

Crust and upper mantle heterogeneities in the southwest Pacific from surface wave phase velocity analysis

Direct earthquake-to-station Rayleigh and Love wave data observed on high gain broadband records are analyzed in order to improve the lateral resolution of the uppermost mantle in the southwest Pacific region. We used data of nine permanent Geoscope and Iris stations located in the southern hemisphere and nine other stations as part of two temporary networks, the first one installed in New Caledonia and Vanuatu (hereafter named Cavascope network) by ORSTOM and the EOST from Louis Pasteur University in Strasbourg (France) and the second one installed in the Fiji, Tonga and Niue islands (hereafter named Spase network) by Washington University in St. Louis (USA). In order to collect more significant details on the surficial structures, we included the analysis of short period waves down to 8 s. A multiple frequency filtering technique has been used to recover phase velocities of Rayleigh and Love waves for selected earthquakes with magnitude greater than 5.5 and with known centroid moment tensor (CMT). About 1100 well-distributed seismograms have been processed in the period range 8–100 s and corrections for topography and water depth have been applied to the observed phase velocities. The geographical distribution of phase velocity anomalies have then been computed using the tomographic method developed by Montagner [Montagner, J.P., 1986a. Regional three-dimensional structures using long-period surface waves. Ann. Geophys. 4 (B3), 283–294]. Due to a poor knowledge of dense, well-distributed, crustal thickness values and corresponding velocity models, we did not perform or speculate on the construction of an S-wave 3D velocity model; therefore, we limited this study to the interpretation of the phase velocity distribution. The location of phase velocity anomalies are well determined and the deviations are discussed within the framework of the geological context and compared with other tomographic models. At long periods, from 40 s to 100 s, our results agree well with most of previous studies: the tomographic imaging shows a large contrast between low and high phase velocities along the Solomon, New Hebrides and Fiji–Tonga trenches. The lowest phase velocity anomalies are distributed beneath northern and southern Fiji basins and the Lau basin (corresponding to the volume situated just above the dipping slabs), whereas the highest values are displayed beneath the Pacific plate and the eastern part of Indian plate downgoing under the North Fiji basin. At shorter periods, our results show that the phase velocity distributions are well correlated with the large structural crustal domains. The use of local temporary broadband stations in the central part of the studied area gives us the opportunity to observe surface waves showing well-dispersed trains, allowing extended velocity measurements down to 8 s although aliasing due to multipaths become important. The continental regions (Eastern Australia, New Guinea, Fiji islands and New Zealand) show low velocities which are likely due to thick continental crust, whereas the Tasmanian, D'Entrecasteaux, and the Northern and Southern Fiji basins are characterized by higher velocities suggesting thinner oceanic crust. Additional analysis including the anisotropic case and S-wave velocity inversion with depth is in progress.     

裂隙网络分维及其与渗透率的关系

在前人裂隙网络水实验室试验的基础上,依据分形理论对裂隙网络复杂系统进行简单的统计分析,获得定量描述这一复杂体系的特征值——分维,并查明它与裂隙网络渗透率之间的关系。从计算结果可以看出,裂隙网络(指有效裂隙网络)的分维值与渗透率之间存在明显的正相关关系,这有可能为裂隙网络渗透率的定量评价提供一个简单而又实用的方法。      

Improving neural network performance on the classification of complex geographic datasets

Neural Networks are now established computational tools used for search minimisation and data classification. They offer some highly desirable features for landuse classification problems since they are able to take in a variety of data types, recorded on different statistical scales, and combine them. As such, neural networks should offer advantages of increased accuracy. However, a barrier to their general acceptance and use by all but `experts' is the difficulty of configuring the network initially.  This paper describes the architectural problems of applying neural networks to landcover classification exercises in geography and details some of the latest developments from an ongoing research project aimed at overcoming these problems. A comprehensive strategy for the configuration of neural networks is presented, whereby the network is automatically constructed by a process involving initial analysis of the training data. By careful study of the functioning of each part of the network it is possible to select the architecture and initial weights on the node connections so the constructed network is `right first time'. Further adaptations are described to control network behaviour, to optimise functioning from the perspective of landcover classification. The entire configuration process is encapsulated by a single application which may be treated by the user as a `black box', allowing the network to the applied in much the same way as a maximum likelihood classifier, with no further effort being required of the user.     

Using computational intelligence techniques to model subglacial water systems

Measurements of water pressure beneath Trapridge Glacier, Yukon Territory, Canada show that the basal water system is highly heterogeneous. Three types of behaviour were recorded: pressure records which are strongly correlated, records which are strongly anticorrelated, and records which alternate between strong correlation and strong anticorrelation. We take the pressure in bore-holes that are connected to the evacuation route for basal water as the forcing, and the other pressures as the response to this forcing. Previous work (Murray and Clarke 1995) has shown that these relationships can be modelled using low-order nonlinear differential equations optimized by inversion. However, despite optimizing the model parameters we cannot be sure that the final model forms are themselves optimal. Computational intelligence techniques provide alternative methods for fitting models and are robust to missing or noisy data, applicable to non-smooth models, and attempt to derive optimal model forms as well as optimal model parameters. Four computational intelligence techniques have been used and the results compared with the more conventional mathematical model. These methods were genetic programming, artificial neural networks, fuzzy logic and self-organizing maps. We compare each technique and offer an evaluation of their suitability for modelling the pressure data. The evaluation criteria are threefold: (1) goodness of fit and an ability to predict subsequent data under different surface weather conditions; (2) interpretability, and the extent and significance of any new insights offered into the physics of the glacier; (3) computation time. The results suggest that the suitability of the computational intelligence techniques to model these data increases with the complexity of the system to be modelled.     

Monitoring qualitative spatiotemporal change for geosensor networks

Recent technological advances in geosensor networks demand new models of distributed computation with dynamic spatial information. This paper presents a computational model of spatial change in dynamic regions (such as may be derived from discretizations of continuous fields) founded on embeddings of graphs in orientable surfaces. Continuous change, connectedness and regularity of dynamic regions are defined and local transition rules are used to constrain region evolution and enable more efficient inference of a region's state. The model provides a framework for the detection of global high‐level events based on local low‐level ‘snapshot’ spatiotemporal data. The approach has particular relevance to environmental monitoring with geosensor networks, where technological constraints make the detection of global behaviour from local conditions highly advantageous.     

地震前兆台网观测数据质量分析——中国地震背景场探测项目实施

地震前兆台网的运行直接影响产出的观测数据质量。观测数据平均连续率和平均数据完整率可以反映地震前兆台网的运行质量。对中国地震背景场探测项目试行期间(2014年)及正式启动后2015年度的观测数据质量,从以上2个指标进行分析,并对比原有仪器运行质量,可知背景场项目试运行期间,各学科台网投入仪器运行稳定,观测数据符合相关要求,总体质量良好,观测仪器可以投入正式运行。     

基于开源数据和条件生成对抗网络的地形重建方法

如何使用少量的地形特征复原地形地貌一直为地学领域的难题。本文使用开源数据集提取地形特征要素,使用地形特征要素作为约束条件构建了用于生成DEM的条件生成对抗网络(Conditional Generative Adversarial Networks, CGAN),设计了基于开源DEM、开源DEM与遥感影像组合、以及5m高精度DEM提取地形特征要素生成DEM的对比实验,并对结果进行视觉效果、相关性分析以及地形因子的对比与评价。结果表明:(1)在视觉效果上,3种不同方式生成的DEM在视觉效果上均十分逼近原始5 m DEM,都远好于传统插值方法生成DEM,基于开源12.5m DEM提取要素和1m遥感影像的重建效果最接近于原始5 m DEM;(2)在相关性上,三种不同方式生成的DEM与原始5m DEM相关性均能达到0.75以上,组合开源数据提取要素重建DEM与原始5 m DEM相关性可达到0.85以上;(3)在地形因子方面,基于开源12.5 m DEM和1 m遥感影像提取要素重建DEM的坡度和坡向的分布趋势与原始5 m DEM最为一致。本文为高精度DEM建模提供了新的思路,在高精度DEM难以获取...