The European Information Society

This paper looks at different datasets obtained from an airborne Light Detection And Ranging (LiDAR) system and compares the reliability of two contemporary analysis approaches. Estimates of different stand parameters, such as top tree height, were derived using regression analysis and a segmentation approach on data obtained from small-footprint laser scan were contrasted with the field measurements in 7 plots, specifically volume and basal area. Plots of 2,500m² containing plantations of Sitka spruce (Picea sitchensis Bong. Carr.) were scanned with two different point densities in years 2003 and 2004. These plots were divided into training and test regions of 625 m² each. Regression analysis was performed using percentiles corresponding to the canopy tree height at different vertical levels and a segmentation method was used to delineate individual tree crowns where tree metrics can be determined. The bias of the estimated values for the stand volume and basal area ranged from 1.21 to 6.49 m³ha^-1 (0.17 to 0.92 %) and - 2.69 to 1.23 m²ha^-1(- 3.9 to 1.7 %), respectively; and the bias calculated from the segmentation using 0.5 and 1m dataset ranged between - 349.77 to - 434.76 m³ha^-1 (- 49.7 to - 61.8 %) for the stand volume and - 33.36 to - 42.24 m²ha^-1 (- 48.5 to - 61.4 %) for the basal area. The results showed that the regression models estimated stand volume and basal more accurately compared with values calculated from the segmentation. Furthermore, it is shown that there was no significant difference in the estimates from the regression model when using different point densities.

An efficient forest∈dexforest management requires accurate and cost-effective measurements of forest inventory parameters. The cost of LiDAR surveys are directly dependent on the number and size of validation plots as well as the sampling density of points needed to adequately estimate forest inventory parameters. This study investigates (i) the spatial variability of the forest stand, i.e., the effect of the area chosen on the prediction of forest parameters and (ii) the relation between prediction accuracy and sampling point density for the estimation of top height, basal area and volume at plot level. Assessment of the stand’s spatial variability was accomplished by comparing the accuracy of the top height estimations, using the 99^th percentile of a normalised distribution of points, over areas of different size. Original sampling density was synthetically reduced to 10, 5, 4, 3, 2, 1, 0.50, 0.33, 0.25 and 0.20 returns per m². Forest parameters were subsequently estimated for each point density by means of 99^th percentile (top height) and linear regression models (basal area and volume). Predictions were validated using 11 stands, each containing one 50x50 m² plot. Results show that the optimum area for forest parameters prediction is 1600 m² with an average top height accuracy of 95.05% and a standard deviation of 3.41%. Larger sizes will merely increase the cost of field data collection without improving accuracy. Interestingly, top height predictions were slightly more accurate for lower point densities. Linear equations yielded RMSEs of 3.28-5.28 m²/ha and 29.41-36.04 m³/ha for basal area and volume respectively. There were therefore small differences in terms of accuracy of predicted parameters for different point densities, which indicates that once a good DTM is created, future LiDAR surveys can be accomplished over the same area at lower sampling densities, and thus reducing the costs but without disregarding estimation accuracy.

An approach for monitoring network optimization is presented which estimates the characteristics of a spatial phenomenon based on the measured values in order to perform the optimization. Based on a phenomenon model it is determined where additional information is needed. During this optimization process also the limitations and constraints of the geosensor network are taken into account. After deriving the model based approach from theoretical considerations, the presented approach is evaluated by means of a table top experiment.

The article presents a conceptual framework for computations with imprecise values. Typically, the treatment of imprecise values differs from the treatment of precise values. While precise computations use a single number to characterize a value, computations with imprecise values must deal with several numbers for each value. This results in significant changes in the program code because values are represented, e.g., by expectation and standard deviation and both values must be considered within the computations. It would be desirable to have a solution where only limited changes in very specific places of the code are necessary. The mathematical concept of lifting may lead to such a solution.

The SIM-DL theory has been developed to enable similarity measurement between concept specifications using description logics. It thus closes the gap between similarity theories from psychology and formal representation languages from the AI community, such as the Web Ontology Language (OWL). In this paper, we present the results of a human participants test which investigates the cognitive plausibility of SIM-DL, that is, how well the rankings computed by the similarity theory match human similarity judgments. For this purpose, a questionnaire on the similarity between geographic feature types from the hydrographic domain was handed out to a group of participants. We discuss the set up and the results of this test, as well as the development of the according hydrographic feature type ontology and user interface. Finally, we give an outlook on the future development of SIM-DL and further potential application areas.

The overarching objective of this paper is to introduce a novel Fast, Efficient, and Scalable k-means k-means (FES-k-means*) algorithm. This algorithm is designed to increase the overall performance of the standard k-means clustering technique. The FES-k-means* algorithm uses a hybrid approach that comprises the k-d tree data structure, the nearest neighbor query, the standard k-means algorithm, and Mashor’s adaptation rate. The algorithm is tested using two real datasets and two synthetic datasets and is employed twice on all four datasets. The first trial consisted of previously MIL-SOM* trained data, and the second was on raw, untrained data. The approach presented with this method enables unfounded knowledge discovery, otherwise unclaimed by conventional clustering methods. When used in conjunction with the MIL-SOM* training technique, theFES-k-means* algorithm reduces the computation time and produces quality clusters. In particular, the robust FES-k-means* method opens doors to (1) faster cluster production than conventional clustering methods, (2) scalability allowing application in other platforms, and its ability to handle small and large datasets, compact or scattered, and (3) efficient geospatial data analysis of large datasets. All of the above makes FES-k-means* live up to defending its well-deserved name—Fast, Efficient, and Scalable k-means (FES-k-means*). The findings of this study are vital to the relatively new and expanding subfield of geospatial data management.

This paper introduces DBSCAN-MO, an algorithm for density-based clustering of point objects on a planar surface with moving obstacles. This algorithm extends a well known spatial clustering method, named DBSCAN, which has been initially proposed to cluster point objects in a static space. DBSCAN-MO is able to form a set of spatio-temporal clusters and may be readily customized to complex dynamic environments. A prototype system, which implements the algorithm, developed in Java and tested through a series of synthetic datasets, is also presented.

Trajectory data is of crucial importance for a vast range of applications involving analysis of moving objects behavior. Unfortunately, the extraction of relevant knowledge from trajectory data is hindered by the lack of semantics and the presence of errors and uncertainty in the data. This paper proposes a new analytical method to reveal the behavioral characteristics of moving objects through the representative features of migration trajectory patterns. The method relies on a combination of Fuzzy c-means, Subtractive and Gaussian Mixture Model clustering techniques. Besides, this method enables splitting the analysis into sections in order to differentiate the whole migration into i) migration-to-destination, ii) reverse-migration. The method also identifies places where moving objects’ cumulate and increase in number during the moves (bottleneck points). It also computes the degree of importance for a given point or probability of existence of an object at a given coordinate within a certain confidence degree, which in turn determines certain zones having different degrees of importance for the move, i.e. critical zones of interest. As shown in this paper, other techniques are not capable to elaborate similar results. Finally, we present experimental results using a trajectory dataset of migrations of white storks (Ciconia ciconia).

In this paper we propose a methodology for mining very large spatio-temporal datasets. We propose a two-pass strategy for mining and manipulating spatio-temporal datasets at different levels of detail (i.e., granularities). The approach takes advantage of the multi-granular capability of the underlying spatio-temporal model to reduce the amount of data that can be accessed initially. The approach is implemented and applied to real-world spatio-temporal datasets. We show that the technique can deal easily with very large datasets without losing the accuracy of the extracted patterns, as demonstrated in the experimental results.

This paper describes the creation of a set of small-areas for the reporting of census data in the Republic of Ireland. The current areas used for reporting the results of the quinquennial population censuses are known as Electoral Divisions; they are large compared with similar reporting areas in Northern Ireland, they have widely varying populations and considerable internal social heterogeneity which makes them unsuitable for a wide variety of planning tasks. We describe an automated method of creating a suitable census geography which uses existing digital map and gazetteer data. We describe its structure and operation, validation and its application to nationwide. The areas have a prescribed minimum size, are designed to be consistently small, nest into the existing ED geography, cover the whole country, are constrained by natural boundaries, use streets as their unifying feature, and are reasonably homogenous.

Land-use modelling and spatial scenarios have gained increased attention as a means to meet the challenge of reducing uncertainty in the spatial planning and decision-making. Several organisations have developed software for land-use modelling. Many of the recent modelling efforts incorporate cellular automata (CA) to accomplish spatially explicit land-use change modelling. Spatial interaction between neighbour land-uses is an important component in urban cellular automata. Nevertheless, this component is calibrated through trial-and-error estimation. The aim of the current research project has been to quantify and analyse land-use neighbourhood characteristics and impart useful information for cell based land-use modelling. The results of our research is a major step forward, because we have estimated rules for neighbourhood interaction from really observed land-use changes at a yearly basis. This higher temporal granularity gives a more realistic foundation for estimating neighbourhood interaction rules to be applied in for example land-use cellular automata.

Based on principles of panorama maps we present an interactive visualization technique that generates multi-perspective views of complex spatial environments such as virtual 3D landscape and city models. Panorama maps seamlessly combine easily readable maps in the foreground with 3D views in the background – both within a single image. Such nonlinear, non-standard 3D projections enable novel focus & context views of complex virtual spatial environments. The presented technique relies on global space deformation to model multi-perspective views while using a standard linear projection for rendering which enables single-pass processing by graphics hardware. It automatically configures the deformation in a view-dependent way to maintain the multi-perspective view in an interactive environment. The technique supports different distortion schemata beyond classical panorama maps and can seamlessly combine different visualization styles of focus and context areas. We exemplify our approach in an interactive 3D tourist information system.

This paper describes an extended framework for scenario based spatial decision support for constructing new network infrastructure and its application in the domains of telecommunication, forestry and energy. There is an increasing need to provide new planning paradigms to support very expensive strategic investment decisions in new network infrastructure in these domains. The planning processes are still dominated by an expert approach based on empirical knowledge and manual implementation. With this conventional approach it is impossible to visualize and to consider different planning scenarios within a reasonable cost and time frame. We combined the powerful analytical and visualization capabilities of a Geographic Information System (GIS) with mathematical methods of graph theory and combinatorial optimization. This conceptual approach extends the basic spatial decision support model with a knowledge based module for scenario parameterization and graph generation, a module for geodata integration and processing, an operations research optimization module and a multi-level visualization module supporting the need of different communication channels within the decision making process.

The purpose of multi-modal route planners is to provide the user with the optimal route between trip start and destination, where the route may utilize several transportation modes including public transportation. The optimal route is defined over a set of evaluation criteria considered by the user during the route selection process. Especially in the case of multi-modal transportation, numerous evaluation criteria play a role in the traveler’s route choice. Thus the number of requested search parameters in the route planner may be large, and the user interface is overcrowded easily. Based on a set of pedestrian routes that are optimized for various criteria in multi-modal, inner-urban transportation networks of two European cities, an exploratory study based on Principal Components Analysis (PCA) identifies underlying factors that capture the correlations among route selection criteria. The results show how the variability of routes can be parsimoniously described with a smaller set of components, and how these findings can be used to simplify the user interface design of multi-modal route planners.

Local cultural heritage documents are characterized by contents strongly attached to a territory (i.e. Geographical references). Numerous corpora of such local documents become available and a challenging task is to process them automatically in order to retrieve and to make explicit the geographical information that they contain. The research reported in this paper aims at developing a toolset that teachers could use, first to retrieve travel stories from these corpora, and then to study the itineraries reported in these travel stories. To provide an adequate support to teachers, we propose two computational models from which we have built a Geographical Information Retrieval toolset in tune with travel stories characteristics. The paper demonstrates that these quite simple computational models are well fitted to process automatically (at discourse level) travel stories and to make explicit the geographical itineraries reported in such texts.

Building and maintaining Spatial Data Infrastructures (SDIs) in large networks of public bodies is a great challenge, in particular for European countries since the INSPIRE project has become a EU directive in the last months. However, SDIs can have different goals and thus different requirements depending on which type of processes they aim to serve.

In this paper we consider a highly coupled SDI where different actors join the SDI at different times, and each actor is responsible for providing geographical data during the joining phase and for maintaining them updated afterwards. The main result is the definition and the experimental validation of an SDI architecture where the well-known problems of conflation and semantic harmonization can be approached in a consistent way; moreover, two less-evident problems, called instance integration and ownership definition are precisely identified and integrated in the same architecture.

The number of multi-sourced heterogeneous spatial datasets continues to grow and the fragmentation of organizational arrangements has caused much technical and non-technical heterogeneity. Spatial Data Infrastructures aim to facilitate spatial data use and sharing, and can be an effective platform to aid in data integration. This paper discusses the technical and non-technical heterogeneities of multi-sourced spatial data within the holistic framework of Spatial Data Infrastructure. The paper capitalizes on research and case studies undertaken within Australia. The paper also introduces Geo-WebServices as a means of facilitating spatial data integration and interoperability. Geo-WebService can provide a platform to assess the level of Integrability and readiness of multi-sourced datasets. The results of this research aim to assist practitioners in developing the necessary technical tools including geo web-services and guidelines for effective data integration.

The concept for an operational service for natural disaster situations requires a scenario driven data access to different sensor information for all phases of a disaster management. This also includes the actual availability of image information of the earth surface concerning the specific requirements of each phase. From the temporal point of view, spaceborne data acquisition does not offer a sufficient data availability in order to support all different phases in specific crisis situations. Especially the event phase cannot be supported as required.

In this paper we describe a concept for on demand remote sensing image data acquisition and a rapid information flow within a crisis management system, which allows to support the decision making process for different crisis scenarios and user groups in charge. The investigation focuses on an airborne data acquisition platform as well as on the development of a multi platform geo-service framework to improve the risk management capacities in mountainous regions by realizing an integrated pre-operational service. The demonstrator includes the client applications for building up an overall crisis management system including mobile units, a mobile command centre, web based presentations and open interfaces to other systems. Additionally, it is described how Very High Resolution (VHR) remote sensing data could be used in this context. A landslide susceptibility mapping has been produced using airborne LIDAR data and QUICKBIRD satellite imagery.

This paper describes a framework of methods for the measurement and evaluation of users’ performance in interaction with a web-GIS. The framework involves testing of a system with real-world users, streaming of the user’s screen during the evaluation and the analysis of web-server log files after the evaluation. We have developed and tested this method within a project called RIV together with real-world users that are using a web-GIS. We found out that users have different strategies when interacting with a web-GIS that offers different manners of interaction. The findings that we present in this paper are useful for developers and designers of web-GIS and can help improving such systems.