· · · Institutional Repository

Urban flood risk analyses suffer from a lack of quantitative historical data on flooding incidents. Data collection takes place on an ad hoc basis and is usually restricted to severe events. The resulting data deficiency renders quantitative assessment of urban flood risks uncertain. The study reported in this thesis reviews existing approaches to quantitative flood risk analysis and evaluation of urban flooding guidelines. It proceeds to explore historical data on flooding incidents from municipal call centres in two cities in the Netherlands with the final aim to quantitatively assess urban flood risk. The data from municipal call centres consist of texts describing citizens’ observations of urban drainage problems. The texts provide information about causes, locations and consequences of flooding incidents. Call information on flooding causes is used to identify causes of urban flooding through application of probabilistic fault tree analysis. Urban flooding probabilities are quantified as well as contributions of a range of causes to the overall flood probability. Call information on flooding consequences is used to draw risk curves for a range of consequence classes: separate risk curves are drawn for consequences associated with human health, damage to private property and damage related to traffic disturbance. The curves depict a combination of flood consequences of increasing severity and associated probabilities of occurrence. Health risk associated with urban flooding is evaluated additionally in a screening-level quantitative microbial risk assessment. The assessment is based on analyses of samples from flooding incidents and from combined sewers. Risk values from call data analysis are translated into monetary values and into numbers of people affected by flooding in order to obtain risk outcomes that can be weighed against investments to reduce flood risk. It is discussed how outcomes in monetary terms differ from those based on numbers of affected people affected. The effectiveness of urban flood reduction strategies is assessed based on a comparison of flood risk values associated with three main failure mechanisms causing urban flooding. The effectiveness of existing strategies for flood risk control is discussed and potential improvements are indicated. Finally the acceptability of flood risk is discussed in view of the quantitative flood risk outcomes of this thesis. It is shown how quantitative risk values based on call data provide a starting point for the development of risk-based standards for urban flooding.

This study presents a first attempt to quantify tangible and intangible flood damage according to two different damage metrics: monetary values and number of people affected by flooding. Tangible damage includes material damage to buildings and infrastructure; intangible damage includes damages that are difficult to quantify exactly, such as stress and inconvenience. The data used are representative of lowland flooding incidents with return periods up to 10 years. The results show that monetarisation of damage prioritises damage to buildings in comparison with roads, cycle paths and footpaths. When, on the other hand, damage is expressed in terms of numbers of people affected by a flood, road flooding is the main contributor to total flood damage. The results also show that the cumulative damage of 10 years of successive flood events is almost equal to the damage of a singular event with a T = 125 years return period. Differentiation between urban functions and the use of different kinds of damage metrics to quantify flood risk provide the opportunity to weigh tangible and intangible damages from an economic and societal perspective.

Flooding in urban areas can be caused by heavy rainfall, improper planning or component failures. Quantification of these various causes to urban flood probability supports prioritisation of flood risk reduction measures. In many cases, a lack of data on flooding incidents impedes quantification of probability and risk. In the proposed paper we use municipal call data that describe citizens’ observations of urban flooding incidents to quantify flood probabilities in a quantitative fault tree analysis. Given the unstructured nature of call information, calls are first assigned to classes that correspond with causes of flooding as represented in the fault tree. Since manual classification of calls is very timeconsuming, pattern recognition routines are used to automatically classify the call data. The aim of this study is to assess whether by automatic classification of citizen’s calls, sufficient accuracy can be obtained to allow for use of the results in quantitative risk analysis. This is illustrated by application of automatic classification results in a quantitative fault tree analysis for urban flooding, for two cases with datasets of approximately 6000 calls. The results show that straightforward automatic classification routines can reach error rates below 20%. Largest classification errors occur for small classes, where few data are available to train the classifiers. Automatic classification errors lead to small deviations in the outcomes of quantitative fault tree analysis. Still, conclusions about the ranking of contributions to urban flooding that are to be drawn from fault tree analysis, remain intact.

Hydrological response times are often used to characterise runoff processes. They provide information about temporal resolution of catchments responses, thus of the required measurement resolutions of in-situ sensors as well as spatial sensors like rainfall radars. The objective of this study was to characterise response times for urban catchments in lowland areas equipped with looped, combined sewer systems. Subcatchments size vary between 0.3 and 7.8 ha. The results show that variation in catchment area size can explain only a part of variation in lag times between the subcatchments. Pipe length per subcatchment explains variations in lag times to a similar, low degree. Comparison of lag time values for looped networks and constructed branched versions of the same networks shows that lag time values are influenced by flow between subcatchments in about half of the looped networks, especially during low flow conditions. This means that in looped systems subcatchment area per exit point varies during a storm event and lag time cannot be related to a single catchment characteristic.

In the past, there has been an emphasis on the use of hydrodynamic models as a tool for urban drainage analysis. Limited availability of monitoring data and the perceived more limited resource requirements of models led to a preference for this approach. The last decade has seen a gradual development of water quantity and quality monitoring systems through the development of reliable and increasingly cost-efficient water level sensors, continuous water quality sensors and data communication and storage. Employment of monitoring systems has become feasible in a growing number of research and practical applications. Thus, a datadriven approach to system analysis, purely based on monitoring data, has presented itself as a viable alternative to model simulations. The advantage of monitoring data is that they provide direct information on parameters of interest, such as flooded locations, flood depths and combined sewer overflow concentrations, at a level of accuracy that simulation models are unable to achieve. This paper compares urban drainage analysis, for two complexity levels, based on model simulations and on monitoring data. The results show that, given the current state of technology and model instruments, monitoring systems are the most promising way to obtain reliable estimates of crucial urban drainage parameters.

Insurance databases form a promising data source that can be used to improve pluvial flood damage estimations. This paper describes the key characteristics of an insurance database on water related damages to private buildings and content in the Netherlands that has been made available for research. The paper presents preliminary results of a case study where insurance data are explored to find relationships between rainfall characteristics and pluvial flood damage. The results show that variations in damage are partly related to rainfall characteristics. More research on rainfall characteristics and other explanatory variables of flood damage is needed to capture the processes causing damage.

Data from call centres at two municipalities were analysed in order to quantify flooding frequencies and associated flood risks for three main failure mechanisms causing urban flooding. The aim was to find out whether current operational strategies are efficient for flood prevention and if directions for improvement could be found. The results show that quantified flood risk for the two cases is well above the standard which is defined in sewer management plans. The analysis pointed out that gully pot blockages are the main cause of flooding and handling gully pot blockages should therefore be a priority for sewer operators. Reactive handling of calls, as is currently applied, is inefficient if all calls are reacted upon since a small portion of all calls report serious consequences like flooding in buildings or wastewater flooding. Preventive cleaning of sewer pipes proves to be an efficient strategy to reduce flooding due to sewer blockages as flood risk associated with sewer blockages is lower in case of higher cleaning sewer frequencies. Sewer blockages often have serious consequences, thus preventive handling is to be preferred to reactive cleaning. According to the results of this analysis, reduction of flooding sewer overloading is not of primary concern, because serious consequences for this failure mechanism are rare compared to other failure mechanisms.

Urban pluvial flood management requires detailed spatial and temporal information on flood characteristics and damaging consequences. There is lack of quantitative field data on pluvial flooding resulting in large uncertainties in urban flood model calculations and ensuing decisions for investments in flood protection. In this paper four different data sources are discussed, based on literature and expert consultation, that are believed to be of value for the acquisition of quantitative data on pluvial flooding. Data assembled by insurance agencies on flood damage, call databases held by water authorities and emergency services and remote sensing images cover years of observational data that can be mined to obtain data on flood characteristics and occurrence. Flood monitoring using sensor technology can be effective to collect additional pluvial flood data, that is not captured by existing data sources.

Nieuwe KNMI-klimaatscenario’s: grotere rioleringen? Veel artikelen wijzen, na de openbaarmaking van de nieuwste inzichten van het KNMI over de ontwikkeling van het klimaat in Nederland, in die richting. Maar is dat niet te kort door de bocht?

Public health risks of urban pluvial flooding have so far received little attention in technical discussions. In this paper, the results of pathogen measurements in the sewer system of Utrecht and an urban flooding experiment are presented and used in an application of Quantitative Microbial Risk Assessment, an existing risk analysis method for the quantification of infection probabilities. This method uses ingested doses of pathogenic organisms for the calculation of infection probabilities. Ingested dose estimations are based on pathogen measurements. These samples have been analysed for concentrations of Campylobacter, Cryptosporidium and Giardia. Dose-response relations from literature are used to calculate infection probabilities for flood events. The results show that mean probabilities of obtaining a Campylobacter or Giardia infection as a result of contact with urban flood water are 2.8% and 0.6% per event respectively for adults and at least 5.7% and 1.0% per event for children, respectively. Infection probabilities for Cryptosporidium are about 1000 times lower than for Giardia. The infection probabilities found indicate that the health risk of urban flooding is higher than that of swimming in recreational freshwater environments, based on a comparison to the values for ‘acceptable risk’ as defined by the WHO for bathing water.

In this paper, a database of water-related insurance damage claims related to private properties and content was analysed. The aim was to investigate whether the probability of occurrence of rainfall-related damage was associated with the intensity of rainfall. Rainfall data were used for the period of 2003–2009 in the Netherlands based on a network of 33 automatic rain gauges operated by the Royal Netherlands Meteorological Institute. Insurance damage data were aggregated to areas within 10-km range of the rain gauges. Through a logistic regression model, high claim numbers were linked to maximum rainfall intensities, with rainfall intensity based on 10-min to 4-h time windows. Rainfall intensity proved to be a significant damage predictor; however, the explained variance, approximated by a pseudo-R2 statistic, was at most 34% for property damage and at most 30% for content damage. When directly comparing predicted and observed values, the model was able to predict 5–17% more cases correctly compared to a random prediction. No important differences were found between relations with property and content damage data. A considerable fraction of the variance is left unexplained, which emphasizes the need to study damage generating mechanisms and additional explanatory variables.

The objective of this paper is to establish relationships between rainfall extremes and damage data from Dutch insurance industry. Rainfall data are based on a network of 33 automatic rain gauges held by the Royal Netherlands Meteorological Institute. Rainfall characteristics, such as peak rainfall intensity and rainfall volume, are correlated with damage statistics of claims in the vicinity of the rain gauges. The results show that rainfallrelated damage mainly occurs during summer seasons. There is a weak relationship between property damage and rainfall intensities and between property damage and rainfall volumes for summer events. More data is needed to confirm these relationships. In a subsequent study this will be investigated by using weather radar data to obtain a higher spatial rainfall resolution and thus be able to include more insurance data.

In this paper, a database of water-related insurance damage claims related to private properties and content was analysed. The aim was to investigate whether high numbers of damage claims were associated with high rainfall intensities. Rainfall data were used for the period of 2003–2010 in the Netherlands based on a network of 33 automatic rain gauges operated by the Royal Netherlands Meteorological Institute. Insurance damage data were aggregated to areas within 10-km range of the rain gauges. Through a logistic regression model, high claim numbers were linked to maximum rainfall intensities, with rainfall intensity based on 10-min to 4-h time windows. Rainfall intensity proved to be a significant damage predictor; however, the explained variance, approximated by a pseudo-R2 statistic, was at most 34% for property damage and at most 30% for content damage. When directly comparing predicted and observed values, the model was able to predict 5–17% more cases correctly compared to a random prediction. No important differences were found between relations with property and content damage data. A considerable fraction of the variance is left unexplained, which emphasizes the need to study damage generating mechanisms and additional explanatory variables.

Traditional methods to evaluate flood risk mostly focus on storm events as the main cause of flooding. Fault tree analysis is a technique that is able to model all potential causes of flooding and to quantify both the overall probability of flooding and the contributions of all causes of flooding to the overall flood probability. This paper gives the results of a fault tree analysis for urban flooding for the case of Haarlem, a city of 105.000 inhabitants. Data from a complaint register, rainfall data and hydrodynamic model calculations are used to quantify the probabilities of the basic events in the fault tree. The flood probability that is calculated for Haarlem is 0.78/week. Gully pot blockages make the main contribution to flood probability: 79%, storm events contribute only 5%. This implies that in this case an increased efficiency of gully pot cleaning is a more effective strategy to reduce flood probability than to increase the drainage system capacity. Whether this is also the most cost-effective measure can only be decided if the risk calculation is completed with a quantification of the consequences of both types of events. To do this will be the next step in this study.

Van 3 tot en met 7 september 2012 is de negende Urban Drainage Modelling conferentie gehouden in Belgrado, de hoofdstad van Seme. Hieronder volgt een verslag van de indrukken die zijn opgedaan in Servië en de belangrijkste, wetenschappelijke ontwikkelingen die aan de orde zijn gekomen. Het congres had een vol wetenschappelijk programma, mede doordat de organisatie ook graag de stad en het land wilde promoten.