David Thomas McCarthy

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.

Sewage pollution in urban stormwater runoff as evident from the widespread presence of multiple microbial and chemical source tracking markers.

The concurrence of human sewage contamination in urban stormwater runoff (n=23) from six urban catchments across Australia was assessed by using both microbial source tracking (MST) and chemical source tracking (CST) markers. Out of 23 stormwater samples human adenovirus (HAv), human polyomavirus (HPv) and the sewage-associated markers; Methanobrevibacter smithii nifH and Bacteroides HF183 were detected in 91%, 56%, 43% and 96% of samples, respectively. Similarly, CST markers paracetamol (87%), salicylic acid (78%) acesulfame (96%) and caffeine (91%) were frequently detected. Twenty one samples (91%) were positive for six to eight sewage related MST and CST markers and remaining two samples were positive for five and four markers, respectively. A very good consensus (>91%) observed between the concurrence of the HF183, HAv, acesulfame and caffeine suggests good predictability of the presence of HAv in samples positive for one of the three markers. High prevalence of HAv (91%) also suggests that other enteric viruses may also be present in the stormwater samples which may pose significant health risks. This study underscores the benefits of employing a set of MST and CST markers which could include monitoring for HF183, adenovirus, caffeine and paracetamol to accurately detect human sewage contamination along with credible information on the presence of human enteric viruses, which could be used for more reliable public health risk assessments. Based on the results obtained in this study, it is recommended that some degree of treatment of captured stormwater would be required if it were to be used for non-potable purposes.

Uncertainties in stormwater E. coli levels

Although water-quality monitoring programs have been widely used to identify and understand the level of pollution in urban stormwater systems, these data are often used without due consideration of the inherent uncertainties contained within these measurements. This study focuses on the uncertainties associated with the monitored levels of Escherichia coli, a common microbial indicator, in urban stormwater. Four sites located in Melbourne, Australia, were used to assess the uncertainty of six stormwater flow and E. coli variables: (1) discrete E. coli concentration, (2) stormwater flow rate, (3) stormwater event volume, (4) event mean concentration (EMC) of E. coli (i.e. a flow-weighted average of an events E. coli concentrations), (5) E. coli load for each measured event, and (6) site mean E. coli concentration (SMC) (i.e. a volume-weighted average of the E. coli EMCs). Uncertainties of discrete E. coli samples were greater than 30%, while the uncertainty in stormwater flow measurements averaged greater than 97%, mainly due to the high uncertainties in measurements of very low flows. Propagation of these uncertainties, through their respective formulas, found that E. coli EMC uncertainties varied between 10% and 52% and that uncertainties relating to SMC estimates ranged from 35% to 55%. These results show the importance of considering uncertainty when using monitored data sets for any application, including those relating to stormwater management decisions. Suggestions are made about how to increase the accuracies of E. coli monitoring in urban stormwater and how to balance the different sources of uncertainties so that the overall combined uncertainties are minimised while keeping costs at a minimum.

Intra-event variability of Escherichia coli and total suspended solids in urban stormwater runoff

Sediment levels are important for environmental health risk assessments of surface water bodies, while faecal pollution can introduce significant public health risks for users of these systems. Urban stormwater is one of the largest sources of contaminants to surface waters, yet the fate and transport of these contaminants (especially those microbiological) have received little attention in the literature. Stormwater runoff from five urbanized catchments were monitored for pathogen indicator bacteria and total suspended solids in two developed countries. Multiple discrete samples were collected during each storm event, allowing an analysis of intra-event characteristics such as initial concentration, peak concentration, maximum rate of change, and relative confidence interval. The data suggest that a catchments area influences pollutant characteristics, as larger catchments have more complex stormwater infrastructure and more variable pollutant sources. The variability of total suspended solids for many characteristics was similar to Escherichia coli, indicating that the variability of E. coli may not be substantially higher than that of other pollutants as initially speculated. Further, variations in E. coli appeared to be more commonly correlated to antecedent climate, while total suspended solids were more highly correlated to rainfall/runoff characteristics. This emphasizes the importance of climate on microbial persistence and die off in urban systems. Discrete intra-event concentrations of total suspended solids and, to a lesser extent E. coli, were correlated to flow, velocity, and rainfall intensity (adjusted by time of concentrations). Concentration changes were found to be best described by adjusted rainfall intensity, as shown by other researchers. This study has resulted in an increased understanding of the magnitude of intra-event variations of total suspended solids and E. coli and what physical and climatic parameters influence these variations.

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.

Urban stormwater harvesting - sensitivity of a storage behaviour model

The harvesting of urban stormwater to supply non-potable water demands is emerging as a viable option, amongst others, as a means to augment increasingly stressed urban water supply systems. This paper investigates the sensitivity of an urban stormwater harvesting systems capacity-yield-reliability relationship to variations in the behaviour modelling method used, focusing on the storage and demand components of a single reservoir system. The aim is to enhance our understanding of the appropriate computational method for assessing such volumetric reliability/storage capacity relationships. Four reference scenarios were developed, based on two different climates and two different water demand patterns. A sensitivity analysis was conducted, which considered the following computational, storage and demand parameters: yield-spillage order, modelling time-step, length of rainfall record, initial storage volume, open/closed storage surface, dead storage volume, diurnal and weekly pattern of water demand, and inter-annual variability of seasonal water demand. It was found that several parameters had an insignificant impact on the estimation of volumetric reliability for the scenarios tested, whilst the three most significant parameters were: length of rainfall record, inter-annual variability of seasonal demand, and storage surface type. Recommendations about the minimum length of rainfall record used and the inclusion of both the inter-annual variability of seasonal demand and net evaporative losses in the case of an open store are made.

Redefining the stormwater first flush phenomenon

The first flush in urban runoff has been an important, yet disputed phenomenon amongst many researchers. The vast differences in the evidence could be solely due to limitations of the first flush current definition and the approach used for its assessment. There is a need for revisiting the first flush theory in the light of its practical applications to urban drainage management practices. We propose that a catchments first flush behaviour is to be quantified by the runoff volume required to reduce a catchments stormwater pollutant concentrations to background levels. The proposed method for assessment of this runoff volume starts by finding the average catchment pollutant concentrations for a given increment of discharged volume using a number of event pollutographs. Non-parametric statistics are then used to establish the characteristic pollutograph by pooling statistically indifferent runoff increments (known as slices) together. This allows the identification of the catchments initial and background pollutant concentrations and for quantification of the first flush volume and its strength. The novel technique was used on seven catchments around Melbourne, Australia, with promising results. Sensitivity to the chosen increment of runoff (for which mean concentrations are calculated) indicated that when dealing with discrete flow-weighted water quality data, a suitable slice size should closely match the flow-weighting of samples. The overall sensitivity to runoff increment and level of significance was found to be negligible. Further research is needed to fully develop this method.

Biofilters for Stormwater Harvesting: Understanding the Treatment Performance of Key Metals That Pose a Risk for Water Use

A large-scale stormwater biofilter column study was conducted to evaluate the impact of design configurations and operating conditions on metal removal for stormwater harvesting and protection of aquatic ecosystems. The following factors were tested over 8 months of operation: vegetation selection (plant species), filter media type, filter media depth, inflow volume (loading rate), and inflow pollutant concentrations. Operational time was also integrated to evaluate treatment performance over time. Vegetation and filter type were found to be significant factors for treatment of metals. A larger filter media depth resulted in increased outflow concentrations of iron, aluminum, chromium, zinc, and lead, likely due to leaching and mobilization of metals within the media. Treatment of all metals except aluminum and iron was generally satisfactory with respect to drinking water quality standards, while all metals met standards for irrigation. However, it was shown that biofilters could be optimized for removal of iron to meet the required drinking water standards. Biofilters were generally shown to be resilient to variations in operating conditions and demonstrated satisfactory removal of metals for stormwater-harvesting purposes.

Toxicity characterization of urban stormwater with bioanalytical tools.

Stormwater harvesting has become an attractive alternative strategy to address the rising demand for urban water supply due to limited water sources and population growth. Nevertheless, urban stormwater is also a major source of surface water pollution. Runoff from different urban catchments with source contributions from anthropogenic activities and various land uses causes variable contaminant profiles, thus posing a challenging task for environmental monitoring and risk assessment. A thorough understanding of raw stormwater quality is essential to develop appropriate treatment facilities for potential indirect potable reuse of stormwater. While some of the key chemical components have previously been characterized, only scarce data are available on stormwater toxicity. We benchmarked stormwater samples from urban, residential and industrial sites across various Australian capital cities against samples from the entire water cycle, from sewage to drinking water. Six biological endpoints, targeting groups of chemicals with modes of toxic action of particular relevance for human and environmental health, were investigated: non-specific toxicity (Microtox and combined algae test), the specific modes of action of phytotoxicity (combined algae test), dioxin-like activity (AhR-CAFLUX), and estrogenicity (E-SCREEN), as well as reactive toxicity encompassing genotoxicity (umuC) and oxidative stress (AREc32). Non-specific toxicity was highly variable across sites. The baseline toxicity equivalent concentrations of the most polluted samples were similar to secondary treated effluent from wastewater treatment plants. Phytotoxicity results correlated well with the measured herbicide concentrations at all sites. High estrogenicity was found in two sampling events and could be related to sewage overflow. Genotoxicity, dioxin-like activity, and oxidative stress response were evident in only three of the samples where the stormwater drain was beside a heavy traffic road, confirming that road runoff is the potential source of contaminants, while the bioanalytical equivalent concentrations (BEQ) of these samples were similar to those of raw sewage. This study demonstrates the benefit of bioanalytical tools for screening-level stormwater quality assessment, forming the basis for the evaluation of future stormwater treatment and reuse schemes.

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.