Research Topic

Dynamic data-driven modeling and simulation of atmospheric dispersion processes using adaptive observations

Air pollution, especially air pollution due to accidental releases of chemical and radioactive substances, can have a critical impact on human health and well-being. Therefore, monitoring of the atmospheric dispersion of these substances is essential for disaster response. The goal is to obtain repeatedly updated estimates of the current process state and of characteristic process parameters during its evolution. Based on these estimates, forecasts can be computed helping to understand and assess possible future impacts of the process.

In order to meet this objective, the atmospheric dispersion process is modeled and simulated taking into account a number of ongoing observations taken by a distributed sensor network to minimize effects of uncertainty. For even more accurate results, the concept of steerable adaptive observations is used: The sensors are mounted on vehicles which deploy adaptively so that the best possible results can possibly be obtained. This leads to an optimal control problem linked to an inverse problem, which is in many approaches solved on a central supercomputer, together with the data-driven simulation of the process.

This research project focuses on distributed and hierarchical approaches which are better suited to meet the online and real-time requirements of a mobile sensor network. Data is processed locally, a certain amount of information is exchanged between neighboring sensor units, decisions are made based on locally available information and the global problem is solved cooperatively. For this research, concepts from the fields of distributed data assimilation, optimal control and reduced order modeling play an important role.

Key Research Area

K2) Simulation based optimization
R2) Optimal control
D1) Cyber-physical systems


Tobias Ritter


Dolivostra├če 15

D-D-64293 Darmstadt



+49 6151 16 - 24373


+49 6151 16 - 24404




ritter (at) gsc.tu...

 Print |  Impressum |  Sitemap |  Search |  Contact |  Privacy Policy
zum Seitenanfangzum Seitenanfang