ISPRS Journal of Photogrammetry and Remote Sensing
28
Items for publication in this section are requested to be mailed to the Book Review Editor 0. Bethel, 508 Park Ridge Drive, West Lafayette, IN 47906, USA.
Advanced Geographic Data Modelling: Spatial Modelling and. Query Languages from 2D and 3D API plications. M. Molenaar and S. de Hoop (Editors). Geodetic Commission,
Delft, Netherlands, Pubiications in Geodesy (New Series), NO. 40, 1994, 251 pp., ISSN 90-6132-249-9. One of the most enduring questions for those with an interest in the research and development of geographic information systems (GE) concerns the data model by which real-world data is represented within such systems. Most users are familiar with the vector and raster data models which form the basis for almost all systems presently available. But are the existing data models sufficiently flexible for handling the diversity of applications to which GIS may reasonably be applied and to what extent does the use of one particular data model constrain us from capturing, storing, querying and visualising our data in its most appropriate form? These questions have been taken up enthusiastically by researchers in a range of disciplines including computer science, geography, geoinformatics, photogrammetry and surveying. In September 1994 two conferences were held which provided an opportunity for the early results of this research to be discussed - the international symposium on spatial data handling (SDH 6) in Edinburgh and the symposium of ISPRS Commission Ill in Munich. Both conferences had sessions concerning the modelling of geographic data; unfortunately, both conferences were in the week of 5-9 September. As a means to resolve this particular overlap problem, the Subcommission on GIS of the Netherlands Geodetic Commission, with help from members of the ISPRS working group on “Conceptual aspects of GE” arranged a workshop with the Department of Surveying and Remote Sensing at the Wageningen Agricultural University in the Netherlands, The workshop was restricted to under 40 participants and entitled “Advanced Geographic Data Modelling”. The speakers from the workshop have updated or rewritten their papers for the published proceedings, There are 19 papers, reflecting the focus of the original call for papers on conceptual modelling for GIS. The themes selected for the proceedings are: semantic data LWf?SJournal of Photogrammetry and Remote Sensing, 50(6): 28-29 8 19% Elsevier Science B.V. All rights reserved.
models; spatial querying; three-dimensional modelllng; dynamic data structures, fuzzy data models, multLscale models, query languages and cognitive aspects. of GIS. The majority of papers are from researchers in universities and particularly from departments of computer science. Some of the papers, such as those dealing with relational algebra and formal semantics use notations which may not be familiar to a general audience, but all papers are well written with helpful supporting text and black and white illustrations. Some particularly interesting papers for me were the keynote address by Jan van Roessel, in which the author traces his experience over 20 years with structuring data for vector overlay processing, showing how the notion of attribute propagation continues to be at the heart of new developments. Mark Gahegan provides formal semantics for the integration of image data into GIS. He works toward an automated means to extract features of interest from images in a goal-directed manner. The paper also illustrates how computer science can support the idea of users seeing different features of interest in the same image. Several papers address the modelling of truly threedimensional data, including a contribution by van Oosterom et al. which proposes an integrated design for a 3D GIS. The architecture is integrated in that 3D geometric data is stored with other geometric and nonspatial (thematic) data in a Postgres database. A range of applications are illustrated using a linked visualisation system. Two papers which directly consider the question of whether another kind of data model may be more appropriate for certain kinds of spatial data, are the papers by Gold and by Molenaar. The first demonstrates how the voronoi tessellation can offer distinct advantages over vector and raster tessellations, particularly when adjacency relationships are important; the second shows how the voronoi tessellation can form the basis of 30 (and higher-dimensional) representations based on tetrahedra. Common criticisms of the initial attempts by computer scientists to devise models for geographic data were that these models were incapable of fully capturing the nature of geographic objects, which are often
mal syntax for re
Richardson, who generalises satellite
i
papers and the of the arguments she presented in a 1968 paper to the Association of American Geogra duality of the “location” and “object” the world. For many readers, this paper is perhaps the place to start; it is a helpful, reflective synthesis on where research on data modelling has it is currently going. This reinforces gained from reading this proceedings, of an active GIS community producing high-quality research. This and similar output at the SDH and GISRUK conferences suggest a growing maturity to GIS research and a shift away from short-term problem-solvin S researchers eciding if the are taking a necessary step back an data models they have been living with really are the data models they need for the future. Neil Stuart (Edinburgh, UK)
from Satellite
e-
of enegre (Editor). Pub the International Cartographic Association by Elsevier, 1994, 303x213 mm, xxvi + 269 pp., $105.00, ISBN 0 08 042351 5. In 1988, the ICA Commission on Thematic Mapping from Satellite Imagery proposed that a guidebook on the topic be produced and this handsome volume is the result of that decision. Its production has involved a major international cooperative effort of specialists from many nations and draws on practical application experience to formulate general rules for cartographic
and a discussion
These are repeated
and page. There are colour illustrations, both of satellite images a
soil and agricultural studies, ocean and coastal-zone range of 16 mapping. These applic ions are from countries and involve ata from CDS OS, kandsat MSS, Landsat T , SPOT and NOAA/AVHRR. This is a well produced volume, with clear text and good-quality images. As a summary of what can and has been achieved around the world using remotely sensed data, this guidebook certainly meets its objectives and the editor must be congratulated for organising the contributions from such a long and diverse collection of authors into such a coherent whole. For anyone wishing an easy guide to the current contribution of satellite remote sensing to thematic mapping, this guide can be strongly recommended. D.A. Tait (Glasgow, UK)