Road hierarchy with integration of attributes using fuzzy-AHP

Road objects in a network data model are categorised into a hierarchical structure in accordance with their functions and capacities. In this study; five road attributes derived from semantic, geometric and topological properties of network data set (i.e. road class, road length and centralities of degree, closeness and betweenness) are utilised for the creation of road network hierarchy. The relationships with each attributes except road class and their effects on the determination of road importance are analysed by using a distribution graph and the equation of Pearson correlation coefficient. For creating road network hierarchy, integration process is achieved through the application of fuzzy analytic hierarchy process assuming attributes as the fuzzy criteria. The integration process is followed by the calculation of new priority attribute that indicates the importance of road objects. At the end of the process, road class, which is the most important attribute, is also used for the validation of proposed methodology. The results show that the new priority value of a road is superior to its each attribute value in hierarchical organisation.     

Using Neural Nets to Model the Spatial Distribution of Seasonal Homes

Seasonal and second homes are important aspects of recreational tourism. Owning a summer cottage, time-share condominium, hunting cabin, or a part-time residence in a location away from home affects development patterns in significant ways. This paper presents the results of using an artificial neural network and a geographic information systems-based approach to identify and quantify the principal predictors of seasonal home distribution within the Lake States region of Minnesota, Wisconsin, and Michigan. Representative variables, such as proximity and spatial configuration of lakes, proximity to the Great Lakes shore, surrounding forest acreage, and public land access have been quantified using geographic information systems at the minor civil division (MCD) scale for the three states. The GIS data have then been fed into artificial neural networks to enable these pattern recognition tools to identify the principal predictors of seasonal home distribution in the Upper Great Lakes States.     

On the importance of training methods and ensemble aggregation for runoff prediction by means of artificial neural networks

ABSTRACT

Artificial neural networks (ANNs) become widely used for runoff forecasting in numerous studies. Usually classical gradient-based methods are applied in ANN training and a single ANN model is used. To improve the modelling performance, in some papers ensemble aggregation approaches are used whilst in others, novel training methods are proposed. In this study, the usefulness of both concepts is analysed. First, the applicability of a large number of population-based metaheuristics to ANN training for runoff forecasting is tested on data collected from four catchments, namely upper Annapolis (Nova Scotia, Canada), Biala Tarnowska (Poland), upper Allier (France) and Axe Creek (Victoria, Australia). Then, the importance of the search for novel training methods is compared with the importance of the use of a very simple ANN ensemble aggregation approach. It is shown that although some metaheuristics may slightly outperform the classical gradient-based Levenberg-Marquardt algorithm for a specific catchment, none performs better for the majority of the tested ones. One may also point out a few metaheuristics that do not suit ANN training at all. On the other hand, application of even the simplest ensemble aggregation approach clearly improves the results when the ensemble members are trained by any suitable algorithms.

EDITOR D. Koutsoyiannis; ASSOCIATE EDITOR E. Toth     

Revisiting gateway cities: connecting hubs in global networks to their hinterlands

Gateway cities have received much attention from urban geographers. In spite of outstanding contributions being made, we think that the concept needs to be revisited with regard to regional development implications. Bringing together research on global production networks (GPNs) and world cities, this article shows that gateway cities are critical for development in networks, generating impulses for peripheral locations by engaging them in processes of “strategic coupling.” Yet, gateway cities also concentrate segments of GPNs to the detriment of their hinterlands. We conceptualize gateway cities with the aid of five features: logistics and transport, industrial processing, corporate control, service provision and knowledge generation. Our concept allows for an understanding of cities in global and regional economic processes beyond corporate headquarters, corporate services and governance – that is, beyond the boundaries of existing research. It unsettles traditional understandings of strategic coupling and world cities, filling a lacuna on city–hinterland connections.     

Hierarchical polygonization for generating and updating lane-based road network information for navigation from road markings

Lane-based road network information, such as lane geometry, destination, lane changing, and turning information, is important in vehicle navigation, driving assistance system, and autonomous driving. Such information, when available, is mainly input manually. However, manual methods for creating and updating data are not only costly but also time-consuming, labor-intensive, and prone to long delays. This paper proposes a hierarchical polygonization method for automatic generation and updating of lane-level road network data for navigation from a road marking database that is managed by government transport department created by digitizing or extraction from aerial images. The proposed method extends the hierarchy of a road structure from ‘road–carriageway–lane’ to ‘road–carriageway–lane–basic lane’. Basic lane polygons are constructed from longitudinal road markings, and their associated navigational attributes, such as turning information and speed limit, are obtained from transverse road markings by a feature-in-polygon overlay approach. A hierarchical road network model and detailed algorithms are also illustrated in this paper. The proposed method can accelerate the process of generating and updating lane-level navigation information and can be an important component of a road marking information system for road management.     

Self-Organized Cognitive Maps

This paper argues cognitive mapping is a learning process that can be simulated by a self-organizing neural network. The learning of city locations was considered in two studies. One study focused on the learning of city locations on four continents. Results indicated the neural network aligned the cities producing systematic errors similar to those in human cognitive maps. A second study had a neural network learn a biased sample of city locations in the United States. Results indicated a non-linear relationship between cognitive and physical distances. Self-organized cognitive maps naturally produce this non-linear relationship when information from more than one scale is mapped into one space.     

遗传算法优化BP网络在滑坡灾害预测中的应用研究

在陕西省宝鸡市附近长寿沟地区滑坡详细调查和遥感解译的基础上,完成了1∶10000滑坡编目图。通过使用GIS的水文分析功能,运用正反DEM技术,将长寿沟地区划分为216个自然斜坡单元,其中包括123个滑坡单元和93个未发生滑坡单元,分析滑坡发生与坡高、坡度、坡向、坡形、人类工程活动和水文地质条件影响因子之间的统计规律。利用经遗传算法优化后的BP神经网络对80个滑坡样本和40个未滑坡样本进行训练学习,然后再利用训练好的网络对预测样本进行评价分析。结果表明:43个已滑坡单元中只有3个被误判为无滑坡,正确率为9302%,53个未滑坡单元中有10个被预测为滑坡,正确率为8113%,总体正确率为8646%。通过对被预测为滑坡的10个斜坡单元进行分析,发现这些单元在坡形、坡高等影响因素的组合上已经具备了发生滑坡的条件,虽然目前没有发生滑坡,但作为潜在的滑坡危险区,可以为滑坡灾害预测预报和防灾减灾工作提供参考。     

考虑块段效应的RBF神经网络在土壤空间插值中的应用

定义了土壤变异中的块段效应,提出了考虑块段效应的RBF神经网络空间插值模型(RBFBE法),以期提高土壤水分和养分的插值精度;并以江苏省扬州市区北部某试区为例,通过A、B两种训练方案,将RBFBE法对土壤含水率与有机质含量的插值结果与常规的RBF神经网络(RBFANN法)空间插值结果及普通克里金法(OK法)插值结果进行了对比分析。结果表明:与OK法相比,RBFBE法能使土壤特性的均方误差M_se缩小19.0%~62.2%,预测吻合度G提高8.9%~28.8%;与输入信息相同的RBFANN法相比,RBFBE法亦能使土壤特性的均方误差M_se缩小10.0%~48.8%,预测吻合度G提高3.4%~22.0%;此外,研究也表明RBFBE法具有较强的泛化能力。考虑块段效应的RBF神经网络方法有利于拓展人工神经网络方法在土壤特性空间插值中的应用范围,具有一定的应用前景。     

Robust algorithms for polyhedral modelling of fractured rock mass structure

This paper presents robust algorithms for application in polyhedral modelling of rock mass structure based on discrete fracture networks. The algorithms improve the robustness of current polyhedral modellers to challenging or degenerate geometries and geometrical errors. These issues are inevitably encountered when the fracture network realisations incorporate large numbers of fractures, or regions of densely packed fractures are modelled, or when the number of fracture network realisations interrogated is large. Several stages in the polyhedral modelling process which are sensitive to geometrical errors are reviewed and algorithms which are robust and capable of handling topological errors in the input fracture network geometry are presented.