Background and State-of-the-Art

The computational forecast of states of the environment in distributed, dynamic systems is certainly compute intensive; if the application is, in addition, time critical and mission critical, since it is related to an emergency situation and must support real-time management decisions, the demands on computing resources both in terms of speed and reliability multiply. Since complex and detailed real-time models also require large volumes of real-time data as inputs, this also puts high demands on the data acquisition systems and data transfer. And finally, the large volumes of information generated have to be made available to human operators and decision makers that will be distributed and possibly mobile in the field, in a reliable, an easy to understand manner.

While most of the methods proposed are used, individually, in planning applications, their computational requirements, in particular if integrated, usually precludes application in a real-time management and decision support environment. The use of High-Performance Computing and Networking Technology, based on flexible and cost-efficient distributed client/server technology could open the possibility for a new class of applications.

Typical examples of this class of new HPCN decision-support applications include:

  • accidental release of toxic chemicals, both from transportation accidents as well as from process plants and storage facilities, nuclear accidents;

  • spills of toxic substances or oil in river systems or in the coastal marine environment;

  • urban air pollution (photochemical smog and alarm models) coupled to real-time traffic information and control systems;

  • regional flood forecasting and river basin management (operational control);

  • forest fire management.

In all these cases, the basic modeling technology of 2D/3D dynamic, physically based models is reasonably well understood; however, these models are most often used in a laboratory setting, which allows extensive pre- and post-processing of inputs and outputs, as well as the running of batch jobs over possibly many hours of compute time.

© Copyright 1995-2018 by:   ESS   Environmental Software and Services GmbH AUSTRIA | print page