Object-relational features for modeling and analysis of spatio-temporal data

At Technische Universität München extensive official, forestal, climatological, touristical geodata and metadata on the unique ecosystem Bavarian Forest National Park were gathered since 1996 to set up a GIS-platform used for research projects as well as for education. To improve data availability of this precious datapool the GIS-platform has been Web-enabled by migrating all its data content to a new integrated and highly scalable 3-tier GIS-environment, using ArcInfo, ArcSDE and Oracle(Spatial). Currently seven applications of different university chairs and institutions access this central geodata server for reading and writing, guided by metadata. The objective is to build an interdisciplinary sustainable GIS-platform with unique data on analyzing the ecosystem's short and long-term natural processes like its recovering behaviour and forest development after bark beetle outbreak. In this manner, the datapool is updated and extended automatically through its distributed write-access concept in an interdisciplinary network. The main objectives of this paper are new aspects of using object-relational features of the underlying database management system Oracle to improve the extensibility and flexibility of data models, enhance interoperability and analyzing potential as well as to ensure consistency by defining standards based on abstract data types. The potential is demonstrated considering two examples: First the scenario of deadwood spread due to bark beetle outbreak during 1994 and 2002 as well as spatio-temporal analysis of forest rejuvenation, that is all hope of getting back to a green, more beetle-resistant forest in future. The object-relational extension of data models can be combined with data mining techniques to analyse e.g. spreading patterns of bark beetle outbreak. The vision is to merge data that represents factors which influence bark beetle activity and to derive conclusions about the correlation of these factors. The second example describes the application of real time deer tracking based on object-relational features in Bavarian Forest National Park.