Manfred Kleidorfer

Comparison of different uncertainty techniques in urban stormwater quantity and quality modelling

Urban drainage models are important tools used by both practitioners and scientists in the field of stormwater management. These models are often conceptual and usually require calibration using local datasets. The quantification of the uncertainty associated with the models is a must, although it is rarely practiced. The International Working Group on Data and Models, which works under the IWA/IAHR Joint Committee on Urban Drainage, has been working on the development of a framework for defining and assessing uncertainties in the field of urban drainage modelling. A part of that work is the assessment and comparison of different techniques generally used in the uncertainty assessment of the parameters of water models. This paper compares a number of these techniques: the Generalized Likelihood Uncertainty Estimation (GLUE), the Shuffled Complex Evolution Metropolis algorithm (SCEM-UA), an approach based on a multi-objective auto-calibration (a multialgorithm, genetically adaptive multi-objective method, AMALGAM) and a Bayesian approach based on a simplified Markov Chain Monte Carlo method (implemented in the software MICA). To allow a meaningful comparison among the different uncertainty techniques, common criteria have been set for the likelihood formulation, defining the number of simulations, and the measure of uncertainty bounds. Moreover, all the uncertainty techniques were implemented for the same case study, in which the same stormwater quantity and quality model was used alongside the same dataset. The comparison results for a well-posed rainfall/runoff model showed that the four methods provide similar probability distributions of model parameters, and model prediction intervals. For ill-posed water quality model the differences between the results were much wider; and the paper provides the specific advantages and disadvantages of each method. In relation to computational efficiency (i.e. number of iterations required to generate the probability distribution of parameters), it was found that SCEM-UA and AMALGAM produce results quicker than GLUE in terms of required number of simulations. However, GLUE requires the lowest modelling skills and is easy to implement. All non-Bayesian methods have problems with the way they accept behavioural parameter sets, e.g. GLUE, SCEM-UA and AMALGAM have subjective acceptance thresholds, while MICA has usually problem with its hypothesis on normality of residuals. It is concluded that modellers should select the method which is most suitable for the system they are modelling (e.g. complexity of the models structure including the number of parameters), their skill/knowledge level, the available information, and the purpose of their study.

Performance and sensitivity analysis of stormwater models using a Bayesian approach and long-term high resolution data

Stormwater models are important tools in the design and management of urban drainage systems. Understanding the sources of uncertainty in these models and their consequences on the model outputs is essential so that subsequent decisions are based on reliable information. Model calibration and sensitivity analysis of such models are critical to evaluate model performance. The aim of this paper is to present the performance and parameter sensitivity of stormwater models with different levels of complexities, using the formal Bayesian approach. The rather complex MUSIC and simple KAREN models were compared in terms of predicting catchment runoff, while an empirical regression model was compared to a process-based build-up/wash-off model for stormwater pollutant prediction. A large dataset was collected at five catchments of different land-uses in Melbourne, Australia. In general, results suggested that, once calibrated, the rainfall/runoff models performed similarly and were both able to reproduce the measured data. It was found that the effective impervious fraction is the most important parameter in both models while both were insensitive to dry weather related parameters. The tested water quality models poorly represented the observed data, and both resulted in high levels of parameter uncertainty.

Impact of input data uncertainties on urban stormwater model parameters.

The use of urban drainage models requires careful calibration, where model parameters are selected in order to minimize the difference between measured and simulated results. It has been recognized that often more than one set of calibration parameters can achieve similar model accuracy. A probability distribution of model parameters should therefore be constructed to examine the models sensitivity to its parameters. With increasing complexity of models, it also becomes important to analyze the model parameter sensitivity while taking into account uncertainties in input and calibration data. In this study a Bayesian approach was used to develop a framework for quantification of impacts of uncertainties in the model inputs on the parameters of a simple integrated stormwater model for calculating runoff, total suspended solids and total nitrogen loads. The framework was applied to two catchments in Australia. It was found that only systematic rainfall errors have a significant impact on flow model parameters. The most sensitive flow parameter was the effective impervious area, which can be calibrated to completely compensate for the input data uncertainties. The pollution model parameters were influenced by both systematic and random rainfall errors. Additionally an impact of circumstances (e.g. catchment type, data availability) has been recognized.

Stormwater quality models: performance and sensitivity analysis.

The complex nature of pollutant accumulation and washoff, along with high temporal and spatial variations, pose challenges for the development and establishment of accurate and reliable models of the pollution generation process in urban environments. Therefore, the search for reliable stormwater quality models remains an important area of research. Model calibration and sensitivity analysis of such models are essential in order to evaluate model performance; it is very unlikely that non-calibrated models will lead to reasonable results. This paper reports on the testing of three models which aim to represent pollutant generation from urban catchments. Assessment of the models was undertaken using a simplified Monte Carlo Markov Chain (MCMC) method. Results are presented in terms of performance, sensitivity to the parameters and correlation between these parameters. In general, it was suggested that the tested models poorly represent reality and result in a high level of uncertainty. The conclusions provide useful information for the improvement of existing models and insights for the development of new model formulations.

A case independent approach on the impact of climate change effects on combined sewer system performance

Design and construction of urban drainage systems has to be done in a predictive way, as the average lifespan of such investments is several decades. The design engineer has to predict many influencing factors and scenarios for future development of a system (e.g. change in land use, population, water consumption and infiltration measures). Furthermore, climate change can cause increased rain intensities which leads to an additional impact on drainage systems. In this paper we compare the behaviour of different performance indicators of combined sewer systems when taking into account long-term environmental change effects (change in rainfall characteristics, change in impervious area and change in dry weather flow). By using 250 virtual case studies this approach is--in principle--a Monte Carlo Simulation in which not only parameter values are varied but the entire system structure and layout is changed in each run. Hence, results are more general and case-independent. For example the consideration of an increase of rainfall intensities by 20% has the same effect as an increase of impervious area of +40%. Such an increase of rainfall intensities could be compensated by infiltration measures in current systems which lead to a reduction of impervious area by 30%.

Parallel flow routing in SWMM 5

The hydrodynamic rainfall-runoff and urban drainage simulation model SWMM (Storm Water Management Model) is a state of the art software tool applied likewise in research and practice. In order to reduce the computational burden of long simulation runs and to use the extra power of modern multi-core computers, a parallel version of SWMM is presented herein. The challenge has been to modify the software in such minimal way that the resulting code enhancement may find its way into the commercial and non-commercial software tools that depend on SWMM for its calculation engine. A pragmatic approach to identify and enhance only the critical parts of the software in terms of run-time was chosen in order to keep the code changes as low as possible. The enhanced software was first tested for coherence against the original code and then benchmarked on four different input scenarios ranging from a very small village to a medium sized urban area. For the investigated sewer systems a speedup of six to ten times on a twelve core system was realized, thus decreasing the execution time to an acceptable level even for tedious system analysis. A parallel version of SWMM for multi-core processors is herein presented.The enhanced software was first tested for coherence and then benchmarked.Changes were kept minimal in order to encourage adoption.A speedup of six to ten times on a twelve-core system was realized.

GIS-based applications of sensitivity analysis for sewer models

Sensitivity analysis (SA) evaluates the impact of changes in model parameters on model predictions. Such an analysis is commonly used when developing or applying environmental models to improve the understanding of underlying system behaviours and the impact and interactions of model parameters. The novelty of this paper is a geo-referenced visualization of sensitivity indices for model parameters in a combined sewer model using geographic information system (GIS) software. The result is a collection of maps for each analysis, where sensitivity indices (calculated for model parameters of interest) are illustrated according to a predefined symbology. In this paper, four types of maps (an uncertainty map, calibration map, vulnerability map, and design map) are created for an example case study. This article highlights the advantages and limitations of GIS-based SA of sewer models. The conclusion shows that for all analyzed applications, GIS-based SA is useful for analyzing, discussing and interpreting the model parameter sensitivity and its spatial dimension. The method can lead to a comprehensive view of the sewer system.

Optimization of measurement campaigns for calibration of a conceptual sewer model

To simulate hydrological models of combined sewer systems an accurate calibration is indispensable. In addition to all sources of uncertainties in data collection due to the measurement methods itself, it is a key question which data has to be collected to calibrate a hydrological model, how long measurement campaigns should last and where that data has to be collected in a spatial distributed system as it is neither possible nor sensible to measure the complete system characteristics. In this paper we address this question by means of stochastic modelling. Using Monte Carlo Simulation different calibration strategies (selection of measurement sites, selection of rainfall-events) and different calibration parameters (overflow volume, number of overflows) are tested, in order to evaluate the influence on predicting the total overflow volume of the entire system. This methodology is applied in a case study with the aim to calculate the combined sewer overflow (CSO) efficiency. It can be shown that a distributed hydrological model can be calibrated sufficiently when calibration is done on 30% of all existing CSOs based on long-term observation. Event based calibration is limited possible to a limited extend when calibration events are selected carefully as wrong selection of calibration events can result in a complete failure of the calibration exercise.

Identifying weak points of urban drainage systems by means of VulNetUD

This article presents the development and application of the software tool VulNetUD. VulNetUD is a tool for GIS-based identification of vulnerable sites of urban drainage systems (UDS) using hydrodynamic simulations undertaken using EPA SWMM. The benefit of the tool is the output of different vulnerability maps rating sewer surcharging, sewer flooding, combined sewer overflow (CSO) efficiency and CSO emissions. For this, seven predefined performance indicators are used to evaluate urban drainage systems under abnormal, critical and future conditions. The application on a case study highlights the capability of the tool to identify weak points of the urban drainage systems. Thereby it is possible to identify urban drainage system components which cause the highest performance decrease across the entire system. The application of the method on a real world case study shows for instance that a reduction of catchment areas which are located upstream of CSOs with relatively less capacity in the downstream sewers achieves the highest increases efficiency of the system. Finally, the application of VulNetUD is seen as a valuable tool for managers and operators of waste water utilities to improve the efficiency of their systems. Additionally vulnerability maps generated by VulNetUD support risk management e.g. decision making in urban development planning or the development of rehabilitation strategies.

Integrated rehabilitation planning of urban infrastructure systems using a street section priority model

Abstract Due to the changes in water management, from the construction of new systems to the rehabilitation of existing networks, integrated strategic rehabilitation management has become more and more important. Consequently, this paper shows a method that considers sewer and water distribution networks against environmental factors (e.g. street network and connected buildings) in a priority model. This approach aids in not only identifying areas where rehabilitation is technically necessary but also economically recommendable (e.g. completely depreciated). This is achieved by implementing factors of structural resiliency, vulnerability of the network, capital value and other network components (e.g. manholes, house connections). This model is applied to a medium sized case study in which the rehabilitation areas identified by the developed model are compared to the actual rehabilitation plans of the water and sewage companies. Results also show the application of the estimated ranking for rehabilitation planning with a fixed budget.