ScaDS.AI

As part of the Competence Centre ScaDS Dresden-Leipzig, the following methodological developments in particular were implemented and tested by the IOER under the application field of environmental and transport sciences:

• Processing of large amounts of data from real-time acquisition and efficient data access for the simulation of multi-model ensembles, coupled impact models and a decision support tool (topic area 1)
• Transformation, cleansing and database integration of data from different external and internal sources to expand and intensify the networking between data sources and models (topic area 2)
• Knowledge extraction, including through pattern recognition and data mining in (geo)databases as well as the use of new data warehouse components (topic area 3)
• New visualisation as well as efficient and intuitive interaction techniques for the representation of complexity and uncertainty ranges as well as for the extended usability of decision support tools (topic area 4)
• Flexible workflows and adaptive processes for recording, storing, extracting, transforming, analysing and interpreting data in model simulations and dynamic database applications (topic area 5)

To expand the possibilities for impact modelling and decision support, several developments from the methodological subject areas were implemented and tested. In particular, this included using innovative software architectures to realise efficiency gains in the processing and storage of very large volumes of geospatial and factual data, including from simulation calculations and data transfer (topic area 1).

In view of the heterogeneity of geodata and factual data sets, methods for semantic, syntactic and geometric transformation, cleansing and database integration of data were tested on-the-fly in order to be able to merge data from different sources and of comparable timeliness (topic area 2). This has created essential prerequisites for expanding and intensifying the networking between data sources and coupled models. A standardised data infrastructure is an important prerequisite for the inter-institutional use of data. Metadata retrieval and management played a central role in this.

In addition, new approaches to knowledge extraction were examined, for example by analysing textual and audiovisual documentation of participation processes with urban and regional development stakeholders, crowdsourcing and data mining in geodatabases (topic area 3). A new data warehouse component for the management of climate and impact modelling data as well as other geobasis and geospatial data was also developed and implemented. Conformity with the European INSPIRE spatial data infrastructure has been taken into account.

New methods were tested for the visualisation of results data, which illustrate the sensitivity of regional and local systems, including the uncertainty ranges and the performance of alternative courses of action, to urban and regional development stakeholders through multidimensional representations (topic area 4).

Last but not least, innovative methods were used to improve workflows in modelling and model coupling, but above all in the development of a web-based, spatial and addressee-oriented tool for decision support (topic area 5). This has significantly supported the transfer of the knowledge generated with the help of the impact models and their linkage with the climate models to the actors involved.

The research from phases 1 and 2 of ScaDS.AI will be continued in phase 3 as part of the DRESDEN-concept Research Group “Advanced Environmental Risk and Sustainability Modelling of Cities and Regions Using AI” (SITES.AI).