CORDIS - EU research results

Quantifying Uncertainty in Integrated Catchment Studies

Final Report Summary - QUICS (Quantifying Uncertainty in Integrated Catchment Studies)

The Water Framework Directive (WFD) is the most significant EU directive concerning the protection and management of surface water bodies. The directive requires all member states to take action to ensure that all surface water bodies within a river basin catchment achieve a good ecological status by a defined target date. Currently many surface waters currently do not have good ecological status, and planned interventions are required in order to improve these surface waters. In order to design effective interventions, a good understanding of surface water quality processes is required and it is important to predict the impact of any such interventions in order to justify investment. Many natural and human processes control surface water quality, these processes are highly complex with a range of pollutant sources, and pollutant transport and transformation processes that act at different spatial and temporal scales. For example, agricultural runoff processes and pollution may act over 100s km2 for months or longer, whereas pollution generated from urban drainage systems may act over 100s of meters and have timescales of tens of minutes.
Cost estimates by EU governments indicate that billions of euros will need to be spent over several decades to implement the WFD and achieve good ecological status in the EU’s surface water bodies. There is an increasing level of concern on the implementation cost of achieving good ecological status in natural water bodies in the EU. It is therefore imperative that any interventions designed to improve surface water quality are cost effective and work as anticipated. The impacts of any interventions designed to improve surface water quality are usually predicted by using computer models that simulate hydraulic and water quality related processes across the whole catchment (e.g. rainfall runoff, flow of water and nutrients through urban drainage systems and through rivers). Integrated water quality models are specifically designed to predict the quality of water in a catchment and are seen as tools to optimise the costs of meeting water quality objectives. Catchment scale water quality predictions are used to both design interventions and justify the investment costs.
However, existing sub-models in integrated catchment water quality models are known to contain significant predictive uncertainty. Methods have been developed by academic researchers to quantify this uncertainty. Once this uncertainty is quantified information can be gained on both the risk of failure of any interventions designed to improve surface water quality, as well as on actions that can be taken to reduce uncertainty. These academic uncertainty analysis methods are generally not utilised by consultants and end-users who design solutions and interventions to improve water quality. The Quantifying Uncertainty in Catchment Studies (QUICS) network has studied several academic frameworks and methods for analysis of uncertainty in integrated catchment modelling studies. The network has identified several challenges hindering uptake of uncertainty analysis among consultants and end-users: (1) a lack of people educated in both the use of uncertainty analysis techniques as well as experience in working with end-users, (2) a lack of knowledge on how uncertainty interacts between different spatial and temporal scales found in a catchment, (3) long run times for calculating multiple model simulation runs needed for uncertainty analysis when using commercial software preferred by end-users, and (4) the ways in which the WFD is interpreted by local and national environmental regulators who prefer certain approaches which can hinder the implementation of uncertainty analysis.
To promote the use of uncertainty analysis techniques to enable better informed decision making and hence to enable more cost effective implementation of the WFD, the overall aim of QUICS was to provide high levels of training to twelve Early Stage Researchers and four Experienced Researchers on water quality processes, statistical and computational uncertainty methods, current industry modelling techniques, policy and regulation. The QUICS project ran from June 2014 to May 2018 and the network comprised nine partners and eleven associate partners, including universities, research institutes, water authorities and engineering consultants. QUICS therefore contains leading water quality scientists, uncertainty experts and private sector water management practitioners and modellers. The full and associate partners provided complementary technical skills across the spatial and temporal scales of typical catchments and the skills and networks to facilitate implementation of uncertainty analysis by end-users and also hosted the researchers during secondments.
The QUICS team has developed various open source software tools in R and Python, to enable uncertainty analysis to be carried out with various existing software packages used by end users, and to make water quality uncertainty analysis more efficient.
The better scientific understanding developed during QUICS, as well as the developed statistically based uncertainty tools have the potential to provide more informed investment decision making and to deliver effective WFD solutions at lower cost. To ensure that this does occur, the QUICS team of researchers and academics have collaborated with Aquafin, Witteveen+Bos, RTC4Water, and Jacobs, to create practical case studies that generate an evidence base on the usefulness of incorporating different forms of uncertainty analysis in modelling studies. The new knowledge can be used to further develop risk based decision making techniques so as to deliver better informed decisions for WFD implementation. Approaches have been tested and developed on case study catchments throughout Europe. QUICS had two network catchments Eindhoven-De Dommel in the Netherlands and the Haute-Sûre catchment in Luxembourg / Belgium, data for which have been made available to all researchers by TU Delft and LIST respectively. Some testing used other case study catchments. The case studies have highlighted both barriers restricting future implementation of uncertainty analysis in practice, as well as creating an evidence base of the benefits of implementation of different forms of uncertainty analysis. For example, in some practical case studies it was not possible to incorporate a full uncertainty analysis as this would not satisfy the current requirements of the local environmental regulator. In other case studies, advantages were shown of employing methods for optimising data sampling, or of calculating ‘risk of failure’ of different designs rather than employing pass/fail criteria, or of utilising uncertainty quantification to highlight sources of uncertainty which can then subsequently be addressed. Case studies have also shown the potential overall uncertainty in water quality predictions, is not always as large as anticipated due to the impacts of different sources of uncertainty at different spatial and temporal scales. This suggests that the overall cost of implementing the WFD may not be as large as currently feared.
Further information and links to the developed models, code and publications about methods and case studies can be found on the project website Project coordinator, Dr Alma Schellart: