AIO - L. Alfonso

Maximising information content for optimal performance of water catchments

Reason of the research

Development, population growth and climate change are factors that are affecting the way how water resources are being managed. The rapid change of land use and the evident changes of rainfall patterns and intensities all over the world during the last years are sufficient reasons to think that not only extreme events, but also normal hydrological events are going to be more and more difficult to handle in the future. A vital part of the modern water management is the measurement of the different processes of the hydrologic cycle. For this purpose, monitoring networks provide data that is used by models to generate useful information content which is used by managers of water systems and decision-makers to keep a good water system performance. Nevertheless, monitoring networks are expensive to maintain and update, because they use devices that are exposed to the weather conditions and vandalism, which reduce the quality of the collected data. Therefore, the number of equipped points should be minimal, but still must be able to get data that accurately describe the system. This situation, together with the fact that the collected data has a discrete nature, implies that better methods for interpolating are required, as well as new methods for collecting data, especially during extreme events.

Project plan

The proposed research looks for maximising information content from monitoring networks and models for optimal performance of water systems. The possibility to reduce the monitoring networks without affecting the accuracy of the predictions, the use of hydrological and hydrodynamic models to interpolate adequately the information to non-measured points and the use of wireless technologies for the participation of the community in extreme-events monitoring are the main ideas considered to develop.
The first step is the data acquisition of ground-based monitoring as well as radar and remote sensing information, as well as available hydrological and hydrodynamic models. These data and models will be analyzed in terms of reliability and confidence, and will be used to evaluate different techniques for interpolating information, based on information theory approaches. For this purpose, a pilot area will be tested. Towards the maximisation of the information content, three stages are identified in the proposed methodology: 1) “before modelling”, which includes the manipulation of the monitoring networks information to produce the best input data for modelling; 2) “during modelling”, which looks for the best way of using the input data by looking for different mathematical modelling conception and for the best places where data should be collected. ; 3) “after modelling”, which uses complementary modelling for managing errors.

Project results

As a first result in this research, an information theory-based method for localising water-level monitoring stations has been developed. The placement a set of monitoring points such that reduces the uncertainty, understood from the Shannon's information theory perspective, related to water-level measurements in a water system, has been achieved. The multiobjective problem is posed as looking for the set of stations that are mutually independent from the information viewpoint, in which each station have the maximum possible amount of information content. The first condition is evaluated with the informational concept of total correlation, whereas the second is assessed by using the marginal entropy approach. Solutions are obtained by a pragmatic approach (PA) and by the evolutionary optimisation algorithm NSGAII.

The methodology is applied in a Dutch water system, where entropy and transinformation maps are generated. Interesting results, besides the effective location of stations, include the close relationship between hydraulic control structures, the definition of information flow and the characterisation of "areas of information" within the hydrodynamic system.

Promoter

The promotor is prof. Roland K. Price, PhD, MSc and the supervisor dr.ir. Arnold Lobbrecht.