Final Dissemination Event, Amsterdam International Water Week
On Thursday 2nd November 2017 QUICS will host a practitioner focused event to share new knowledge and tools related to uncertainty in integrated catchment studies.
To register for this FREE event, visit the side-events page on the Amsterdam International Water Week website, deadline extended to 22nd October 2017. Note: be sure to select the registration category "I will only visit a side event" to get free registration! Registration allows also free entrance to the exhibition of the water week after attending the workshop.
The deadlines for compliance with the WFD are getting closer and closer. As most of the no regret and relatively cheap measures have already been taken, huge investments may be necessary for compliance. This requires an integrated approach to be able to derive the optimal set of measures in integrated catchment studies. However, given the complexity of these model based studies, associated uncertainties may be extremely high. This is due to the abstraction from reality to model representation with its simplifications and idealisations of the real systems comes with the unavoidable occurrence of uncertainties. The definition, recognition and consideration of these uncertainties is, therefore, of the utmost importance for applying such models and for the interpretation of model results, in real world problems.
In this field the EU funded FP7 Marie Curie Initial Training Network (ITN) Quantifying Uncertainty in Integrated Catchment Studies (QUICS) performs high quality research for developing and implementing uncertainty analysis tools for Integrated Catchment Modelling. The main aim of the QUICS ITN is to educate and train researchers capable of operating at academic research institutions, water utilities or other public bodies so as to provide them with a comprehensive understanding of water quality processes, uncertainty issues and knowledge of appropriate decision making strategies for integrated catchment management. So topics include:
- quantification and the propagation of uncertainty at significant temporal and spatial scales in catchments,
- approaches for minimising uncertainties in integrated models,
- techniques for model reduction of computationally expensive models,
- real world case studies on integrated catchment modelling,
- tools, which can be deployed by end users considering all aspects of modelling uncertainty and hence they are able to be used in the context of the decision-making process.
|09:00 – 09:30||Registration and Reception|
|09:30 – 11:00||3 Keynotes about Uncertainties in Integrated Catchment Studies given by international experts|
|11:00 – 11:40||Podium discussion with questions from the audience|
|11:40 – 12:00||Tea and coffee break|
|12:00 – 12:30||Poster presentations of Research from the project|
|12:30 – 13:45||Lunch and opportunity to view and discuss posters|
|13:45 – 15:15||Workshops about developed tools in 6 groups|
|15:15 – 18:00||Open discussion in small groups with drinks and snacks|
Keynote 1: Introduction to QUICS network - Bridging the gap between practitioners and academics - Alma Schellart and Francois Clemens
Dr. Alma Schellart is a Senior Lecturer of Water Engineering at the Department of Civil and Structural Engineering, University of Sheffield. Her research areas are sewer sediment transport, urban rainfall and energy in urban drainage systems, with a specific interest in quantifying uncertainty in modelling studies. Alma is the coordinator of QUICS, a Marie Skłodowska-Curie Initial Training Network, looking into quantifying uncertainty in integrated catchment studies. This training network involves seven other universities and several commercial partners in their exploration of uncertainty in different models and data collection methods, used for simulating water flow and quality throughout the whole catchment. The aim is to improve industry guidelines and train researchers to be both specialised in their own area, as well as have an eye for the bigger picture and the practical application of their research.
Prof. Dr. Francois Clemens has extensive experience of working on the interface between the water industry and academic research. He is part time professor on Urban Drainage at Delft University of Technology since 2000, and he was Head of the Urban Drainage Department at Witteveen+Bos from 1996 – 2013, and since 2013 he is Head of the Industrial Hydrodynamics Department at Deltares. Francois’ work encompasses a wide range of aspects related to urban drainage research, from uncertainty in urban drainage modelling and risk analysis, to the development of novel monitoring systems for sewer processes.
Keynote 2: The multiple scales of uncertainty: an engineer's perspective - Evangelina Belia
Dr. Evangelina Belia is a process engineer and the co-founder of Primodal Inc. She has over 20 years of experience in wastewater treatment. She specializes in combining mathematical modelling with her experience in operations to solve optimization problems around the globe. More recently she has been implementing in the field real time data quality methods and dynamic modelling for real time plant control. For the past nine years, she has been working in the field of model-based uncertainty evaluation as the Chair of a joint IWA/WEF Task Group on Design and Operational Uncertainty (DOUT). Her focus is the integration of dynamic simulation, explicit uncertainty evaluations and static design guidelines into a cohesive approach for the explicit and size-appropriate design of water resource recovery facilities. The group's Scientific and Technical Report is under way with publication expected in 2018.
Keynote 3: Coping with uncertainties in the design of urban drainage systems: Why ignorance might not be the best option - Manfred Kleidorfer
Dr. Manfred Kleidorfer, has a Master in Civil Engineering (2005) and PhD in Technical Sciences (2010). Since 2015 he is Associate Professor at the University of Innsbruck. His research encompasses water management, with a special focus on modelling of drainage and supply systems including systems analysis, uncertainties and model calibration. Dr. Kleidorfer has published more than 100 papers in journals and conference proceeding and lead several national and international research projects in water management with respect to climate change adaptation and sustainable development. Dr. Kleidorfer is an active member in different international working groups. He is secretary of the IWA/IAHR specialist groups "International Working Group on Data and Models" and member of the management team of the IWA working group "Modelling Integrated Urban Water Systems". Current research activities are consequences of upcoming challenges as climate change or aging infrastructure and to find solutions for adaptation, mitigation and rehabilitation.
Podium Discussion with Questions from the audience
Alma Schellart (University of Sheffield)
Francois Clemens (Delft University of Technology)
Evangelina Belia (Primodal Inc.)
Manfred Kleidorfer (University of Innsbruck)
Elliot Gill (CH2M)
Franz Tscheikner-Gratl (Delft University of Technology)
Pico presentation of posters
Each pico presentation is 3 minutes long, introducing the poster of an aspect of each QUICS fellow's work. The posters will be displayed throughout the whole day and can be discussed during the breaks for lunch, coffee and snacks.
Summary of Posters:
Francesca Cecinati (University of Bristol): Optimal temporal resolution of merged radar – gauge rainfall for urban applications
Carla Camargos (Justus-Liebig-University of Giessen): How uncertainty of simulating water resources is affected by different input data information content
Mahmood Mahmoodian (LIST): Data-driven surrogate modelling for simplification of an urban drainage simulator
Omar Wani (EAWAG): Parameter estimation of hydrologic models using censored and binary observations
Antonio Moreno Ródenas (Delft University of Technology): Uncertainty analysis of water quality predictions at integrated urban water systems
Franz Tscheikner-Gratl (Delft University of Technology): Dealing with the propagation of uncertainties in Integrated Catchment Studies
Manoranjan Muthusamy (University of Sheffield): Geostatistical upscaling of rain gauge data to support uncertainty analysis of lumped urban hydrological models
Nazmul Beg (University of Coimbra): Flow investigation in a Gully-Manhole drainage system
Arturo Torres-Matallana (LIST): TUnc: a R package for temporal uncertainty propagation in urban drainage modelling accounting for input uncertainty quantification
Ambuj Sriwastava (University of Sheffield): Impact of modelling uncertainty on decisions to manage water quality failure caused by Combined Sewer Overflow spills
Alexandre Wadoux (Wageningen University): Sampling design optimisation for rainfall prediction using a non-stationary geostatistical model
Each workshop will be 30 minutes long and be repeated 3 times, so participants have the opportunity to participate in three different workshops.
Workshop 1: Improve rainfall estimations using Kriging with External Drift with non-stationary variance.
Presenters: Francesca Cecinati (University of Bristol) and Alexandre Wadoux (Wageningen University)
Rainfall data are one of the main inputs for many different models, but often too little attention is put in the correct estimation of rainfall and its uncertainty. This workshop will show the large differences that one can expect to observe using only rain gauges, using only radar data, or merging the two. Additionally, an advanced technique (Kriging with external drift with non-stationary variance – KED-NSV) to estimate the uncertainty in merged radar-rain gauge data will be presented and the improvement in terms of uncertainty reduction will be shown.
Workshop 2: Surrogate Modelling for Simplification of urban Drainage Simulators
Presenters: Mahmood Mahmoodian (LIST) and Omar Wani (EAWAG)
Currently, many detailed or complex urban drainage simulators (e.g InfoWorks) are popular among technicians, consultancy engineers, and researchers. These models help to solve various problems related to urban drainage systems’ management and control. However, such simulators are sometimes computationally expensive when it comes to tasks like calibration, sensitivity analysis, uncertainty quantification, and real-time control (RTC). Surrogate modelling is a promising technology to address this challenge and achieve faster run times that are tailored to one’s specific needs. In this workshop, we are going to present a data-driven Gaussian Process Emulator as a surrogate model for InfoWorks ICM simulator. We will illustrate and explain some selected results from this emulation. Also, the performance of simulator and emulator will be compared. Finally, we will briefly discuss uncertainty quantification in the context of these surrogate models.
Workshop 3: CEUPUB (Chaos Expansion Uncertainty Propagation in Ungauged Basins)
Presenters: Antonio Moreno Ródenas (Delft University of Technology) and Vasilis Bellos (CH2M)
This tool allows propagating uncertainties from input time-series and model parameters in Unit Hydrograph based flow estimations. CEUPUB generates a polynomial chaos expansion emulator of the response of a computationally expensive hydrodynamic model (from which a unit hydrograph response is derived). From this emulated structure, the forward propagation of uncertainties can be performed at a fraction of the original computational cost. The tool also includes the possibility to fit known synthetic unit hydrograph structures.
Workshop 4: Software tools for uncertain propagation of input variables in urban drainage modelling
Presenters: Arturo Torres-Matallana (LIST) and Ambuj Sriwastava (University of Sheffield)
Uncertainty propagation techniques for long term simulations in urban drainage modelling are often not addressed or are not feasible because the restriction of the computational time required to properly apply these techniques. To overcome this issue, we analysed the feasibility of performing a full procedure for uncertainty propagation of input variables for water quantity and quality by using a simplified model implemented in R as a case study. We present the software tools developed for uncertainty propagation, analysis of results and visualization, and illustrate them using a case study for quantifying uncertainties. Moreover, we extend the case study with a multi objective optimisation for selection of decision alternatives by minimising the risk of water quality failure caused by the pollutant load released to the receiving water body in the combined sewer overflow spill.
Workshop 5: Modelling urban sediment transport: From surfaces to sewer network
Presenters: Manoranjan Muthusamy (University of Sheffield) and Nazmul Beg (University of Coimbra)
Exponential wash-off models is the most widely used method to predict sediment wash-off from urban surfaces. Although a number of studies proposed various modification to the original equation, these studies mostly looked into one parameter in isolation which is often rainfall intensity thereby ignoring the interactions of other parameters correspond to catchment and sediment characteristics. In the first part of this workshop we present an improved version of the wash-off exponential model where the effect of multiple parameters such as rainfall intensity, surface slope and initial load are all taken in to account. Once this washed off sediment from the urban surfaces enters into drainage structures such as manholes and gullies, some part of it get deposited inside the structures while the rest are transported through them. The fate of sediment depends on the hydraulics inside the structure. In the second part of the workshop, we present the hydraulics and the sediment transport inside different street gully and manholes using a multiphase hydraulic model.
Workshop 6: Calibration and uncertainty analysis of a SWAT model using SPOTPY
Presenter: Carla Camargos (Justus-Liebig-University of Giessen)
SPOTPY is a Python tool that enables the use of computational optimization techniques for calibration, uncertainty and sensitivity analysis. The tool is an open-source package that runs on Windows, Linux and Mac. Linking a model and complex algorithms can be challenging. SPOTPY aims to simplify this task, making the implementation as easy as possible, while providing complex analysis techniques. Once the user connects the model with SPOTPY, testing the effect of different analysis strategies and analyzing the model performance is straightforward. The package allows the choice among twelve algorithms for calibration, nine pre-built parameter distribution functions and sixteen objective functions or likelihoods. This workshop will present the code, clarifying how to make the link between SPOTPY and SWAT and enabling the participants to apply the method for different models.