Monitoring fatigue damage is complex because ships are designed to crack after 20 years, and typically cracks will only occur after a long time. A novel fatigue damage sensor has been invented to indicate damage in corresponding aluminium welded joints. Fatigue damage is accelerated by introducing a stress concentration in the sensor w.r.t. the stress in the structure. New in this fatigue damage sensor compared to sensors invented in the past is the fact that this sensor is both in geometry and fatigue damage mechanisms similar to aluminium welded joints. Based on the time it takes for the sensor to crack, a prediction can be made of the expected lifetime of corresponding aluminium welded joints. This prediction is based on the in the VOMAS joint industry project developed Total Stress Concept in which fatigue resistance of all aluminum welded joints is combined in a single SN curve. It was found that with the total stress concept, resistance of welded joints is significantly higher than used in current fatigue design methods. Since the fatigue resistance increases with the total stress concept it is also investigated what the effects of implementing the total stress fatigue resistance curves at Damen would be on the fatigue design of high speed craft.

Gaining insights into anticipated future expenses is essential for both the planning and construction phases of a project. A significant factor in future costs, particularly over extended periods, is the variability of material and labor prices. the change in costs over time, known as ’cost escalation,’ has been the focus of numerous research efforts. These studies primarily aim to forecast the Construction Cost Index (CCI), a composite index representing a standardized array of materials and labor typical in construction projects. This research presents a novel approach to forecast cost escalations, tailored to individual construction projects, addressing the shortcomings of predictions of a generic Construction Cost Index (CCI). Traditional CCI predictions, while providing some foresight, are limited by their generic nature and often overlook the specific material and labor variations within different projects. Additionally, current research generally fails to account for the uncertainty in the forecasts of the change in the material price and the estimate of the construction cost. In response to these challenges, this research pivots around the following research question: How can cost escalation for various types of construction projects be predicted, accounting for uncertainties in final construction costs and the forecast? Addressing this question involves selecting distinct indices for various project resources, such as steel, concrete, and labor, and combining these predicted indices in line with each resource’s cost. To achieve this, the study evaluates several time-series forecasting models, namely the vector error correction model (VECM), the vector autoregression (VAR) model, and the Holt-Winters model. For each model an automatic forecasting process is created, which automatically checks the suitability of the model, select the appropriate variables (in case of a multivariate model) and selects parameters. These models are evaluated for their accuracy across different time-frames and forecasting horizons. The Holt-Winters model, in particular, showed promise in providing reliable confidence intervals and point forecasts. The study’s testing phase revealed varying degrees of success, as volatile indices like copper and steel proved to be challenging, whereas forecasts for less volatile indices achieved higher accuracy. This outcome suggests room for improvement in refining these forecasting methods. The project-specific forecasts, including cost uncertainties, are developed by inputting cost estimates associated to the material price at current day value. The cost estimates input includes material and labor costs and their uncertainties. The uncertainties are represented through a three v point estimate (optimistic, pessimistic, and most likely values). This input is then transformed into a PERT distribution which is a transformation of the Beta distribution. The final stage involves combining the PDFs of each project activities cost (considering the cost of each specific resource) with the monthly forecast PDFs in a Monte Carlo simulation, providing a detailed cost distribution histogram of total cost and the individual resources cost. The tool’s functionality was demonstrated through a case study on a highway construction project. In this demonstration, specific project data, including material and labor cost estimates, were inputted into a hybrid web and Excel interface. This setup facilitates visualization and manipulation of project information. The tool processes these inputs via the AFP and Monte Carlo simulation, yielding comprehensive outputs such as histograms and statistical properties of the output. This is visible for total project costs and resource-specific escalations. This demonstration effectively showcased the tool’s capability to offer detailed insights into cost escalation, addressing the variability and uncertainty in construction projects. It underscored the tool’s alignment with the study’s objective of providing nuanced, project-specific cost escalation forecasts, moving beyond traditional CCI predictions. In conclusion, this study introduces a tool that caters to specific project resources, timelines, and uncertainties in cost escalations. While current limitations prevent its immediate practical application, this proof-of-concept lays the groundwork for future improvements. Focus of future research should be on refining accuracy and comparing the tool’s forecasts with escalation of historic projects to establish more robust insight into the actual escalation of projects.

The design of offshore and coastal infrastructures, sand nourishment and other ’soft’ coastal interventions require the analysis of environmental variables (e.g. wind, waves, rainfall) that can potentially cause the failure of such structures. Processes such as overtopping, beach erosion, and coastal flooding can result from a combined action of two or more physical processes. Traditional infrastructure design practices assume the highest load previously experienced as the design load, regardless of possible interactions between variables (or processes). This may lead to a misrepresentation of critical design loads. This thesis presents a methodology for defining infrastructure design loads accounting for their interdependence. The methodology is general and is based on regular vines. Vines are graphical tools for defining high dimensional distribution functions through pair-copula construction. With this premise in mind, the main effort was concentrated in formulating a series of steps to integrate several stages of the design: from the processing of raw data up to the choice of design loads for any specific design purpose. The vine-based methodology was applied to the design of a breakwater in Galveston Bay, Texas. This application showed that accounting for the interdependence between design variables provides a more comprehensive description of the physical system acting on the infrastructure. However, the vine-based method is computationally demanding. Hence, the applicability of this methodology should be evaluated on a case by case basis. In parallel, the possibility to define goodness of fit test for vine-copula based on the concept of tree-equivalent classes is explored. The focus is on model selection strategies based on graphical and statistical properties of the vines. The main motivation to investigate model selection strategies for vines is the considerably large computational time needed to fit all regular vines in more than 6 nodes to the data. In this thesis, a novel algorithm is developed to facilitate the implementation of vines in higher dimensions (vines with more than 6 nodes). This algorithm significantly reduces the computational effort to select a regular vine by allowing the user to test only a subgroup of vines in n-nodes constructed on specific characteristics of the vines in (n-1)-nodes.

The rapid development in statistics, information technology, and computational power have en- abled numerous innovative methods to emerge in Structural Health Monitoring (SHM) for structural damage detection, both in practical application and research. The intention of such methods is to use the data obtained from the monitoring system to extract sufficient information to identify damage types that may appear in the structure. However, the following type of questions are mostly unanswered for realistic structural types and monitoring systems: Which responses of the structure should be monitored? Which sensor locations and sensor combinations carry the most information? Among the currently available computational algorithms, which one is the most suit- able one in this context? Hence, this study aims to contribute on this front and provide answers for practical applicability. A comprehensive study of a realistic, prestressed concrete bridge built by the cantilever method - the Lezíria Bridge in Portugal is undertaken to provide insight into these questions. The engineering challenge is studied in a probabilistic framework where the uncertainty sources are model uncertainty and measurement uncertainty. The Bayesian paradigm is used to handle the uncertainty component and a validated Finite Element (FE) model is applied to capture the mechanical behavior. The influence of the potential damage scenario, severity of damage, sensor type, the combination of sensors, prior knowledge of the structure, and the extent of uncertainty of the FE model and measurements are analyzed in this thesis. The informativeness of the Damage Identification (DI) process is reflected by the information content of its resulting posterior distributions. The information content is quantified using measures based on information entropy and the area of credible regions. It is demonstrated that selecting different responses to monitor may lead to a significant change in the informativeness of the result. It is also shown that for all analyzed cases using the most informative sensor type provides adequate information that reflects the damage state, while the rest types could only complement very limited information. In addition, the hybrid Markov Chain Monte Carlo(MCMC) seems more efficient and effective to conduct Bayesian inference. The findings provide valuable, quantitative insight into the design of new monitoring systems.

Drought is a recurrent extreme climate event that strongly affects every spectre of the natural environment and human lives (Madadgar andMoradkhani, 2014). There were numerous drought episodes recorded in MidwesternUSA, particularly in California. It has affected many aspects, including water shortages, groundwater overdraft, critically low stream flow (Diffenbaugh et al., 2015) and economical loss. In contrast, California as the third most extensive area in the U.S., is occupied most of the land use for agricultural purposes. In order to capture the trend of hydrological drought in California, understanding towards the risk of hydrological drought in comprehensive way should be conducted. In a broader sense, the quantification of hazard of hydrological drought will be a preliminary step mitigating the drought event, as an integral part of risk management. This research is aimed (i) to understand the characteristic of hydrological drought using statistical tool; (ii) to build a model to estimate the low stream-flow using the relationship between hydro-meteorological and hydrological drought; and (iii) to quantify the risk ratio of the hydrological drought duration using a statistical method. In general, the thesis will be divided into two big parts, hydrological drought severity and hydrological drought duration. The former part will be analysed using Bayesian Network to estimate the mean monthly discharge in California. The latter part is evaluated using Generalised Linear Model to study the risk of having the expected number of days without discharge in California under different cases of scenarios. Besides gaining information from modelling hydrological drought, this thesis will also try to answer the formulated research questions. The research questions focus upon the statistical perspective towards hydrological drought, by explaining the input data for the model and the advantage(s) and disadvantage(s) of such models.

Extreme weather events seem to happen more often nowadays. One of these extreme events was the California Drought between 2012 and 2016. The socioeconomic and environmental impacts of this drought were enormous: thousands square kilometers of agricultural land fallowed, thousands of lost jobs, salinization problems and forest fires. As a result of the drought, multiple research is conducted related to the hydrological and environmental impacts. However, the role of the root-zone storage capacity during the drought period is not investigated well. This root-zone storage capacity is the water available for plants to transpire and to grow. It influences the partitioning between the transpiration and the run-off rates, which controls the fundamental processes in ecosystem functioning (such as floods, droughts and groundwater recharge). Scientific evidence is growing that this root-zone storage behaves dynamically. The underlying assumption is that plants adapt their root system to climatic and environmental changes. One of these changes related to drought is the availability of snow, which is an important source of the rivers in California. It is important to know how the rootzone is changing due to the snow cover changes during the California Drought to understand the hydrological behaviour of the Californian rivers. For that, the Merced River basin in the Sierra Nevada is analyzed on the relationship between snow cover and root-zone storage capacity. To analyze these relationship, this research consists of two parts: trend analysis and hydrological modeling. The purpose of the first is to see whether there is a change in snow cover, temperature and precipitation and to learn about the dynamics between precipitation, temperature and snow cover. This is done by conducting a trend analysis with the Mann-Kendall test for precipitation, temperature and snow-free days. These snowfree days are calculated based on snow cover data from a MODIS satellite product (MOD10A1) and a method of regional snow-line elevation. Additional to this, a multivariate regression analysis was executed to related snow-free days, precipitation and temperature to each other. Furthermore, scenarios for snow-free days were conducted to analyze the effect of the CaliforniaDrought. The purpose of the second part is to relate the change in snow cover to the root-zone storage capacity. This is done by building a hydrological model calibrated on the snow cover data. By doing this, the maximum soil moisture deficit over a 15 year interval could be calculated, which is an estimation for the root-zone storage capacity. The trend analysis showed a clear downward trend in winter precipitation, but not a significant trend in temperature. The snow-free days are increasing at all elevations, but only significant up to 3300 m.a.s.l. The maximum trend related to the observed snow-free days was 7 days/year at 2300 m.a.sl. and the average for the whole catchment was 4 days/year. The multivariate regression analysis showed that the temperature influence ismore important than the precipitation input, however the maximum R2 was around 0.7. At higher elevations it was not so clear whether precipitation or temperature described snow-free days well enough. The scenarios re veiled the enormous effects of the California Drought on the total amount of snow-free days and confirmed the importance of temperature during this extreme event. At 2300 m.a.s.l. it differs 40 snow-free days with a situation with normal winter temperatures. The effect of temperature is also visualized in the hydrological model, where the maximum soil moisture deficit was not found in the drought period but slightly before it. This is caused by the combination of small precipitation amounts and a low temperature (no melt). In the drought period the soil moisture deficit was limited by the melt from the snow. This is also confirmed by the analysis of the snow storage in the hydrological model. However, the maximum of the soil moisture deficit is gradually increasing of the past 30 years which hints to the adaptive behaviour of the root-zone. The trend analysis and the hydrological model showed that the amount of snow in the Merced River basin is decreased due to the California Drought. At 2300 m.a.s.l., the snow-free days are changed from 170 to over 250. The maximumdifference in snow storage before and after 2010 is 100mmaveraged over the whole basin. The decrease of snow during this period had a limiting effect on the soil moisture deficit. Due to the melting water, no maximum values of soil moisture deficit were found. However, slightly before the California Drought a maximumwas found. This maximum is related to a cold but dry period, where not a lot of melting water was generated. Overall, the maximum of the soil moisture deficit is increasing over the past 30 years. This hints to the adaptive behaviour of plants. In summary, the California Drought caused a decrease of the amount of snow and resulted in limiting of the soil moisture deficit. Simultaneously the soil moisture deficit is increasing. The limiting effect of snow melt on the soil moisture deficit gives plants the time to cope with the changes.

Before the Spanish conquest of Mexico, around the year 1519, the Valley of Mexico was a closed basin. As a result, at the bottom of the valley, an extensive system of shallow lakes, lagoons, and swamps was formed due to precipitation and permanent river’s discharge from the Sierra Nevada mountains. This lacustrine system occupied around 1000km2 of the total surface of the valley. Lakes Zumpango, Xaltocan, Chalco, Xochimilco, Texcoco, and Mexico were distinguished. The capital of the Aztec empire, Tenochtitlan, was founded and built on an island in the middle of Lake Mexico. The Aztecs were known for their impressive constructions and hydraulic structures. At the time of the Spanish conquest, they had a complex system of approximately 95 hydraulic structures (Palerm, 1973), of which the most impressive one was the Nezahualcoyotl dike. This structure was roughly sixteen kilometers long, eight meters’ height and three and a half meters’ width. Its principal function was to protect the city of Tenochtitlan from high water levels in Lake Texcoco. Nowadays, there are no remains of the dike and most of the lakes were drained. The purpose of this thesis is to characterize the lacustrine system and the Nezahualcoyotl at the time of the Spanish conquest of Mexico City dike by using historical documentation and present-day climate and terrain data. This in order to assess the reliability of the dike as a flood defense mechanism and to compare it to modern safety levels. The dike was tested for one failure mechanism: Overflow. A Markov chain and Copula models are proposed in order to create a synthetic time series of precipitation and evaporation. Through a hydrological balance, the water elevation at Lake Texcoco was obtained. In this way, it was possible to provide an estimation of the water level fluctuation in the lake each year during the wet season. In total, a thousand years of synthetic data were generated. To the author’s knowledge, this is the first time that an attempt is made to compare the Aztec design criteria with present time standards. This research illustrates, from an engineering point of view, the possible design criteria of the Nezahualcoyotl dike and the uncertainties surrounding it. This work can be used as a guideline to assess the reliability of other ancient structures or present-day constructions all over the world whose design is largely based on informal criteria where information for the reliability assessment is scarce.

This study aims to explore the possibility of employing remote sensing images to build a probabilistic flood extent forecasting model. This model is constructed and tested in two study areas: New Orleans and Miami. Images that recorded flooding events are first performed with segmentation method Seed Region Growing, and segmented images are classified by Maximum Likelihood classifier. Area detected as water subtracting the permanent water area is the detected flood extent. In total there are nineteen images being processed. The flood detection result is validated by flooded locations from NOAA flood reports and the news, and the accuracy is at 70.4%. The detection result, with flood conditioning factors which include precipitation, sea level, elevation, drainage capacity and distance to the water area, is the input to the probabilistic forecasting model. All inputs are standardised to a common grid system and every cell in that grid system contains a set of data. Two kinds of model structures are proposed and both models are trained with logistic regression and probit regression, both of which are the members of the Generalised Linear Model(GLM). The first kind of model structure is only tested in the New Orleans study area and the second kind of model structure is examined at both study areas. The precision of the first model structure is at 20% with a kappa value at zero. For the second model structure, over-sampling method SMOTE is used to increase the number of data points of the class 'flooded'. The highest precision of the second model structure at New Orleans is 12.6% and at Miami 23.6%, and the highest cohen's kappa values are 0.127 and 0.131 for New Orleans and Miami respectively. The first model structure actually failed in building a success model at a large portion of the study area due to limited records. For the second kind of model structure, most variables are linked to the flooding by the model correctly. The precipitation has a positive relation with flooding, especially when time effect is considered. Elevation reduces the probability of flooding. At the Miami study area where no sea dike exists the sea level has a strong positive relation with flooding. Drainage capacity used in the New Orleans study area does not show an influence on flooding, which requires modelling the intricate drainage system more accurately. In Miami study area, when the study area is confined to the seaside and Miami Beach area, the accuracy of prediction is improved, which informs that land-use type is crucial to be considered in the input. This study innovative collected information from several remote sensing images of different flood events and applied the information to build a probabilistic model, which shows that the information provided by mages could link flooding conditioning factors with flooding. It is recommended to incorporate the remote sensing technique in the flood extent forecast model in the future.

Sewer assets face a great challenge for being underground assets and for experiencing constant deterioration and aging. Reactive decisions are no longer technically nor economically viable, and hence asset managers are migrating to the implementation of proactive strategies. Despite this, managers are making decisions that are poorly justified, based mostly in intuition. This is why it is important to improve the means to implement a more objective and justified proactive approach. Risk-based decisions are a proactive strategy that may help the prioritisation and optimisation Sewer assets face a great challenge for being underground assets and for experiencing constant deterioration and aging. Reactive decisions are no longer technically nor economically viable, and hence asset managers are migrating to the implementation of proactive strategies. Despite this, managers are making decisions that are poorly justified, based mostly in intuition. This is why it is important to improve the means to implement a more objective and justified proactive approach. Risk-based decisions are a proactive strategy that may help the prioritisation and optimisation of maintenance and inspection strategies. It requires for its implementation the estimation of probabilities of failure and their consequences. In this thesis, the analysis is oriented on the failure probabilities on sewer systems, especially on understanding the influences of different characteristics of the pipe (explanatory variables) on these probabilities. The outcome of the thesis answers the following research question: “What is the influence of different characteristics on the probability of failure of the sewer pipes?” Two sub-questions were stated to guide the research and help answering the main research question: 1.What defects can lead to sewer structural failure? 2.What physical characteristics make sewer pipes more prone to fail/deteriorate? For the analysis the Dutch municipalities of Breda and Almere facilitated the results of several years of CCTV inspections of their sewers. Due to the lack of availability of collapse records, this study considered a proxy of collapse, which is loss of watertightness on the pipes. Furthermore, defects that potentially cause loss of watertightness are used as a proxy of the loss of watertightness on a pipe. Defects are obtained from the CCTV reports. The first sub-question was answered based on the proxy failure under study. It was found in the literature review ten defects that potentially cause loss of watertightness, which were classified according to the coding of the standard NEN 3399. These defects are: cracks (BAB), break (BAC), defective connection (BAH), intruding sealing material (BAI), displaced joint (BAJ), porous pipe (BAN), soil visible through defect (BAO), void visible through defect (BAP), infiltration (BBF) and exfiltration (BBG). The second sub-question was answered based on some explanatory variables that were available in the dataset of the inspections, which included: sewer system type, materials, shape, diameter, length, and pipe’s above ground material. To answer this question, a descriptive statistical analysis and two survival methods were implemented: a non-parametric and a semi-parametric model. The non-parametric model consists of an extended version of the Nelson-Aalen estimator of the cumulative hazard (ENE) and its derivative the Extended survival estimator (ESE). Implementing ESE, each characteristic was analysed using the aggregate information of the defects that potentially cause loss of watertightness, and each defect individually. This was done to identify the influence in the failure probabilities (the probabilities of occurrence of defects that potentially cause loss of watertightness). The semi-parametric model that was used is the Cox proportional hazard function, used to estimate the risk ratio associated with one unit increase in one of the characteristics under study. Results of the ESE model showed that: The defects displaced joint (BAJ), infiltration (BBF) and defective connection (BAH), are the ones that have more incidence on the loss of watertightness for the two municipalities. Based on the information of aggregate defects it was observed for both municipalities that the median survival probability is past 14 years. Also, the analysis showed that stormwater sewers have a lower survival probability than foulwater sewers, that PVC pipes have a higher survival probability than concrete pipes, and that shorter pipes have a higher survival probability than longer pipes. In the case of the diameters, for the aggregate defects and defects displaced joint (BAJ) and defective connection (BAH), smaller diameters have a higher survival probability than larger ones. But this tendency is the other way around for defects like cracks (BAB), break (BAC) and porous pipes (BAN), where the diameter has a higher survival probability when are larger than when they are smaller. Characteristics shape and above ground material were only analysed for Breda. It was observed that egg-shaped pipes have a lower survival probability than circular shapes, and that pipes with green fields and floor tiles above them have a lower survival probability than pipes that have above asphalt and pavement. For both municipalities, the results of ESE showed that material is that characteristic that influences the most the probabilities of failure. Results of the Cox proportional model showed that: Almere’s results met the proportionality assumption and showed that the characteristics sewer system type and length are the one that influences the most the failure probabilities. Breda’s data is no appropriate to be used with this model, as it does not meet the proportionality assumption. Recommendations of analysing Breda with an extended Cox are given, as this version of the model allows to use time-dependent variables. of maintenance and inspection strategies. It requires for its implementation the estimation of probabilities of failure and their consequences. In this thesis, the analysis is oriented on the failure probabilities on sewer systems, especially on understanding the influences of different characteristics of the pipe (explanatory variables) on these probabilities. The outcome of the thesis answers the following research question: “What is the influence of different characteristics on the probability of failure of the sewer pipes?” Two sub-questions were stated to guide the research and help answering the main research question: 1.What defects can lead to sewer structural failure? 2.What physical characteristics make sewer pipes more prone to fail/deteriorate? For the analysis the Dutch municipalities of Breda and Almere facilitated the results of several years of CCTV inspections of their sewers. Due to the lack of availability of collapse records, this study considered a proxy of collapse, which is loss of watertightness on the pipes. Furthermore, defects that potentially cause loss of watertightness are used as a proxy of the loss of watertightness on a pipe. Defects are obtained from the CCTV reports. The first sub-question was answered based on the proxy failure under study. It was found in the literature review ten defects that potentially cause loss of watertightness, which were classified according to the coding of the standard NEN 3399. These defects are: cracks (BAB), break (BAC), defective connection (BAH), intruding sealing material (BAI), displaced joint (BAJ), porous pipe (BAN), soil visible through defect (BAO), void visible through defect (BAP), infiltration (BBF) and exfiltration (BBG). The second sub-question was answered based on some explanatory variables that were available in the dataset of the inspections, which included: sewer system type, materials, shape, diameter, length, and pipe’s above ground material. To answer this question, a descriptive statistical analysis and two survival methods were implemented: a non-parametric and a semi-parametric model. The non-parametric model consists of an extended version of the Nelson-Aalen estimator of the cumulative hazard (ENE) and its derivative the Extended survival estimator (ESE). Implementing ESE, each characteristic was analysed using the aggregate information of the defects that potentially cause loss of watertightness, and each defect individually. This was done to identify the influence in the failure probabilities (the probabilities of occurrence of defects that potentially cause loss of watertightness). The semi-parametric model that was used is the Cox proportional hazard function, used to estimate the risk ratio associated with one unit increase in one of the characteristics under study. Results of the ESE model showed that: The defects displaced joint (BAJ), infiltration (BBF) and defective connection (BAH), are the ones that have more incidence on the loss of watertightness for the two municipalities. Based on the information of aggregate defects it was observed for both municipalities that the median survival probability is past 14 years. Also, the analysis showed that stormwater sewers have a lower survival probability than foulwater sewers, that PVC pipes have a higher survival probability than concrete pipes, and that shorter pipes have a higher survival probability than longer pipes. In the case of the diameters, for the aggregate defects and defects displaced joint (BAJ) and defective connection (BAH), smaller diameters have a higher survival probability than larger ones. But this tendency is the other way around for defects like cracks (BAB), break (BAC) and porous pipes (BAN), where the diameter has a higher survival probability when are larger than when they are smaller. Characteristics shape and above ground material were only analysed for Breda. It was observed that egg-shaped pipes have a lower survival probability than circular shapes, and that pipes with green fields and floor tiles above them have a lower survival probability than pipes that have above asphalt and pavement. For both municipalities, the results of ESE showed that material is that characteristic that influences the most the probabilities of failure. Results of the Cox proportional model showed that: Almere’s results met the proportionality assumption and showed that the characteristics sewer system type and length are the one that influences the most the failure probabilities. Breda’s data is no appropriate to be used with this model, as it does not meet the proportionality assumption. Recommendations of analysing Breda with an extended Cox are given, as this version of the model allows to use time-dependent variables.

Extreme value analyses (EVA) are often used to determine the frequency of extreme events. The length of the available observations is an important aspect when performing EVA. It is generally known that more available data results in better estimates with less uncertainties. The main objective of this research report was to assess what the influence of the length of the observations is when inferring rare events. This was done by first analyzing the sensitivity of inferred return levels from synthetic data from three known distributions. Also, three case studies were analyzed to observe the sensitivity of inferred return levels and return periods from the observations. The results of the analyses were that a larger sample size generally leads to a higher confidence in the estimates of inferred return levels from synthetic data. However, there will always remain some uncertainty associated with the estimates. The confidence in the inferred return levels from observations also generally increase for an increasing sample size. However, this can not always be observed and other aspects can be more dominant than an increasing sample size. No correlation could be observed between the sample size and the inferred return periods. The conclusion of the research was that, while there is a positive correlation between the sample size and the confidence in the estimates, there will always remain some uncertainties. It is therefore important to always communicate the uncertainties associated with estimates.