Document Type
Presentation
Publication Date
8-1-2014
Abstract
Many of our natural resource management issues cannot be adequately informed by a single discipline or sub-discipline, and require an integration of information from multiple natural and human systems. As we are unable to observe and monitor more than a few important indicators there is a strong reliance on supplementing observed information with modelled information. Following a period of record drought in the 1990’s, the Australian government recognised the need for better quality, more integrated, and nationally consistent water information. The Australian Water Resources Assessment system (AWRA) is an integrated hydrological modelling system developed by CSIRO and Australian Bureau of Meteorology (the Bureau) as part of the Water Information Research and Development Alliance (WIRADA) to support the development of two new water information products produced by the Bureau. This paper outlines the informatics, systems implementation and integration challenges in the development and deployment of the proto-operational AWRA system. A key challenge of model integration is how you access and repurpose data, how you reconcile semantic differences between both models and disparate input data sources, how you translate terms when passing between often conceptually different modelling components and how you ensure consistent identity between real world objects. The rapid development of AWRA and simultaneous transfer to an operational environment also raised many additional challenges, such as supporting multiple technologies and differing development rates of each model component, while still maintaining a working system. Additionally the continentally sized model extent, combined with techniques relatively new to the hydrologic domain, such as data assimilation and continental calibration, have introduced significant computational overheads. While an in-house fit for purpose operational build of AWRA is currently under development within the Bureau, the research challenges undertaken early in AWRA’s development still hold many valuable lessons. We have found that the use of file standards such as NetCDF, services-based modelling, and scientific workflow technologies such as ‘The WorkBench’ combined with strong model governance has mostly reduced the burden of system development and deployment and exposes some important lessons for future integrated modelling and systems integration efforts.
Comments
Session R02, Semantic Integration in Water Resources