CBRN modelling: Application to water contamination: Fid-Bau Portal

CBRN modelling: Application to water contamination

Griffiths, I.H.

This paper has provided a very brief overview of two areas of CBRN modelling that have been very actively pursued in recent years: optimal sensor placement and sensor data fusion (including network fusion and source term estimation). Significant progress has been made in both of these areas in CBRN defence, many lessons have been learned and adaptations and extensions to theory and technique have been developed. The work and techniques maybe of relevance to water contamination monitoring. All of the approaches discussed rely on a reliable and validated forward model. For CBRN modelling this is used to simulate very large numbers of possible releases and must execute extremely rapidly and, for practical application, offer automated operation. For water contamination and quality monitoring an equivalent forward model of flow within the water supply system with similar capability would be necessary. Even with the types of sampling reduction techniques developed for CBRN modelling, it would still need to be capable of running tens of thousands of simulations in an hour and include optimisation in memory and data handling. However, this many simulations are likely to be required to obtain reliable and statistically robust results, as it has been for CBRN modelling. The CBRN modelling applications have also shown the importance of sensors outputting to the data fusion algorithms their actual measured values for parameters and not simply an alarm status (as may be obtained from thresholding the output time series or some similar signal processing technique). Without this more complete information, the performance of the algorithms is significantly degraded. It would be expected that this will also be the case for data fusion within water contamination monitoring to determine whether an event has happened and where the source is. Inevitably this does place an increased burden on the network but techniques are being developed to alleviate this, such as distributing the fusion algorithms. The results of the CBRN modelling research have shown a great deal of progress and efforts are still continuing. There is also research into other areas such as methods for automatically processing and manipulating input data such as buildings for optimal use by models, handling uncertainty and uncertainty reduction in model predictions, more advanced decision aids and consequence management tools, the presentation of complex risk information, and information management on fragile and limited bandwidth networks. Some of this work may result in techniques that could have relevance to water contamination. The underlying commonalities between some of the applications of CBRN modelling and water contamination monitoring, which this paper has attempted to emphasise, suggest that collaboration and leverage of effort between the two areas could be highly beneficial to both research communities and result in improved capabilities for end users.