Over recent decades, climatic changes and significant variations in all key components of the hydrological cycle have been observed in many regions worldwide, profoundly altering water availability, river flow regimes, and the concentration of nutrients and pollutants. Ecosystem is a key component of the terrestrial hydrological cycle as it shapes the hy drological functioning of catchments by regulating the long-term average partitioning of water into drainage and evaporative fluxes (i.e. latent heat). In response to a changing environment, ecosystems continuously adapt to allow the most efficient use of available energy and resources. However, direct quantification of how ecosystems adapts to climatic variability over long time periods and the mechanistic drivers thereof at the catchment scale is missing so far. As a consequence, it remains unclear how climatic variability, such as precipitation regime or canopy water demand, influences the partitioning of water fluxes, the hydrological response, and hydrological processes and transport mechanisms at the catchment-scale. Therefore, the overarching objective of this thesis is to address the follow ing main research question: How does climatic variability affect the hydrological response and transport mechanisms in a temperate-humid basin over multiple decades? All analysis in this thesis is carried out in a large river basin, the Neckar basin, Germany. A unique long-term dataset is used for this basin, consisting of 70 years of hydrometeorological and tracer data. Hydrological and transport processes in the basin are quantified using a state-of-the-art semi-distributed hydrological model that (i) includes spa tial heterogeneity in topography, vegetation and precipitation, (ii) accounts for ecosystem adaptation to climate variability via a time-varying root zone water storage capacity, and (iii) uses StorAge Selection (SAS) functions to account for mixing of tracers and to estimate time-varying water age distributions at catchment scale. Multi-objective calibration of the hydrological model using the long-term hydrometeorological and tracer dataset pro vides the basis for investigating how climatic variability affects hydrological and physical transport processes in the Neckar basin. The first research question focuses on ecosystem adaptation to climate variability via changes in root zone storage capacity. The root zone storage capacity is a critical factor reg ulating latent heat fluxes and thus the moisture exchange between land and atmosphere as well as the hydrological response and biogeochemical processes in terrestrial hydrological systems. To be survive, root systems of vegetation and the associated vegetation-accessible water storage capacity respond to the ever-changing conditions of its environment. How ever, as these changes occur at landscape scale and are mostly reflected by changes in the composition of plant species present in a specific spatial domain, fluctuations in root zone storage capacity occur largely at time-scales that reflect the life-cycles of individual plants. However, it remains unclear whether root zone storage capacity adapts to climatic variability and evolves over time, thereby reflecting ecosystem adaptation to changing conditions. The thesis investigates this for the Neckar basin by quantifying long-term changes in root zone storage capacity using two different methods, i.e. hydrological model calibration and an independent water balance estimation method. The analysis provides quantitative mechanistic evidence that root zone storage capacity significantly changes over multiple decades reflecting ecosystem adaptation to climatic variability. However, the analysis also suggests that accounting for temporal evolution of root zone storage capacity with a time-variable formulation of that parameter in a hydrological model does not sig nificantly improve its ability to reproduce the hydrological response and may therefore be of minor importance to predict the effects of a changing climate on the hydrological response. The second research question investigates the use of different isotopic tracers to estimate water age distributions, i.e. age distributions of water fluxes (“transit time distri butions”, TTD) and water stored in catchments (“residence time distributions”, RTD) as fundamental descriptors of hydrological functioning and catchment storage. These distri butions provide a way to quantitatively describe the physical link between the hydrological response of catchments and physical transport processes of conservative solutes. However, water age distributions cannot be directly observed, and instead have to be estimated with tracer-aided models. Stable isotopes (𝛿18O) and tritium (3H) are frequently used as tracers in environmental sciences to estimate age distributions of water. It has previously been argued that seasonally variable tracers, such as 𝛿18O, generally and systematically fail to detect the tails of water age distributions and therefore substantially underestimate water ages as compared to radioactive tracers, such as 3H. Early approaches often relied on simple lumped sine-wave or lumped parameter convolution integral models under the assumption that water storage in catchments is at steady state. Here, these methods are compared with the more realistic StorAge Selection (SAS) functions embedded in the dynamic hydrological model used in this thesis. By comparing water age distributions inferred from 𝛿18O and 3H with several different transport model implementations, this thesis demonstrates that previously reported underestimations of water ages are most likely not a result of the use of 𝛿18O or other seasonally variable tracers. Instead, these underestimations can be largely attributed to choices of model approaches and complexity. It is therefore strongly advocated to avoid the use of steady-state model types in combination with seasonally variable tracers and to instead adopt SAS-based or other time-variant model formulations that allow for the representation of transient conditions. The third and final research question investigates the effects of temporal variability of the hydrological response on physical transport processes over a spectrum of time scales from daily to multiple decades. Due to limited availability of tracer records over longer durations in many catchments, most previous studies focused on daily time scales to analyse temporal variability of water ages as metric of physical transport and the underlying drivers. To improve understanding of long-term transport dynamics, this thesis quantifies the variability in water ages, identifies the associated dominant controls from daily to multi-decadal time scales, and analyses the associated temporal evolution of water ages of streamflow and evaporation. It is shown that there are no major long term dynamics in water ages, driven by either internal processes or external transport variability. Consequently, the physical transport dynamics in the upper Neckar basin, and potentially in other basins with similar water age characteristics, are inferred to exhibit near-stationarity over multiple decades. Concluding, this thesis provides sufficient evidence that long-term varying root zone storage capacity significantly reflects ecosystem adaptation to climatic variability. However, the temporal evolution of root zone storage capacity does not control variation in the partitioning of water fluxes and has no significant effects on fundamental hydrological response characteristics of the studied semi-humid river basin in the near future under a changing climatic condition. In addition, the thesis suggests physical transport processes can be assumed to be near-stationary and predictable across multiple decades under either internal (i.e., time-variant root zone storage capacity) or external transport variability (i.e., climatic variability), which contrasts with the frequently reported fractal pattern in stream water solute dynamics. This finding is crucial for management of subsurface water quality and the design of restoration interventions for groundwater affected by legacy contamination such as nitrate.

Hydrological ecosystem services (HESS), also referred to as water-related ecosystem services, highlight the intricate connection between ecosystem services and the hydrological cycle. This relationship underscores that the magnitude of these services is largely dependent on the availability and quality of water. The inclusion of HESS ensures the transition from traditional top-down and single objective systems into multi-criteria and human-centered approaches that can prioritize activities with a broader spectrum of benefits. In particular, HESS ensures the uptake of international guidance, such as Integrated Water Resources Management (IWRM), Nature-based Solutions (NbS) or Natural Capital Accounting (NCA), by achieving a win-win co-development of water with, among others, land and ecosystems, while fully delivering the benefits to humans and society.

Hydrological modelling is an essential aspect of water resources management and has many applications, particularly water quality monitoring and the pollutant transmission. Recent research found that hydrology also plays a crucial role in plastic transport from land to the ocean, making it vital to understand the impact of spatially distributed precipitation on the hydrological processes. This leads to the question of how a hydrological model can help waste management and prevent plastic waste in the ocean. However, developing a suitable hydrological model with limited data availability, complex topography, and human interaction is a challenge. This research uses the Rio Las Vacas catchment as a case study and focuses on testing the applicability of the wflow hydrological model and assessing the impact of heterogeneous precipitation on the model result.

The research analysed the ground data available in Rio Las Vacas to verify and validate its quality. This study uses precipitation gauges, ERA5, and CHIRPS to assess the most suitable input for modelling rainfall-runoff in the Rio Las Vacas catchment. The catchment's hydrology was analysed and a conceptual model of the catchment was proposed. Using the wflow model, the hydrological response was modelled using the three rainfall products and the effect of uncertainty and spatially distributed rainfall was assessed.

The result shows that the Rio Las Vacas catchment suffers from a shortage of hydrometeorological information which is also of spurious quality. Only three rain gauges were available, and their analysis revealed high error variance and independence between stations. Hydrological characterisation analysis revealed significant human-induced changes. A calibrated wflow model effectively simulated the catchment’s hydrological response, demonstrating that global parameterisation was sufficient as a starting point for the model in this region. The soil water parameters are one of the most significant parameters in generating runoff for the model. The impact of spatially distributed precipitation is prominent in the model. This is one of the reasons for ERA5 to perform poorly in simulating the streamflow in the catchment. Based on the quantitative validation, the best modelling results came from using rain gauge data interpolated with the IDW method, followed by CHIRPS. However, the qualitative assessment shows CHIRPS is more reliable among the datasets.

In conclusion, this study shows the difficulties of modelling a data-scarce basin in the mountainous region. However, extensive data verification and validation, in combination with hydrological characterisation, helps to improve the model result. It also shows the precipitation inputs were crucial for a distributed model. The use of precipitation input with finer resolution could improve the model significantly. The study shows that the spatio-temporal distribution of the precipitation did not only affect the streamflow at the measurement point, but also the spatio-temporal runoff generated in the basin at pixel level. This could propagate to the water quality or plastic transport study. Among the datasets tested, CHIRPS emerged as the most reliable and suitable for hydrological modelling in this basin, both quantitatively and qualitatively.

Urban areas of coastal cities are increasingly susceptible to the consequences of climate change. In particular, the threat of increasing floods due to sea level rise and heavier rainfall is growing. Cartagena de Indias in Colombia is such a coastal city, where flooding is a significant issue, with floods becoming more frequent in recent decades. Arcadis, together with the ConAgua consortium, is currently investigating and designing mitigation strategies for the city’s water-related issues. The thesis focuses on defining and quantifying the sources of urban flood problems and assessing the impact of potential mitigation measures for reducing these flooding problems.

The study consists of four parts:
1. Identifying the main causes of floods using a data analysis;
2. Quantifying these findings with a numerical D-Hydro model;
3. Investigating suitable kind of mitigation measures and quantifying the impact of the best feasible mitigation options;
4. Evaluating the investigated mitigation options based on several criteria with the help of a multicriteria analysis.
Parts 1 and 2 focus on understanding the system. Parts 3 and 4 of the study focus on finding the best option to reduce floods.

The data analysis allows the development of a system description identifying the main causes of flooding. The analysis reveals that the limited capacity of the drainage system, high topographic variations, and extreme rainfall are the main causes of flooding in Cartagena. Extreme rainfall flows quickly from steep upstream areas to milder slope lower-lying areas, particularly those at the edge of the Ciénaga de la Virgen. It is hypothesised that downstream water level plays a role in the accumulation of runoff water, thereby influencing the frequency and intensity of floods in the neighbourhoods at the southern edge of the Ciénaga de la Virgen, especially with a relative sea level increase of approximately half a meter factored in. This hypothesis is checked by quantifying the impact of different scenarios with a numerical model.

A numerical D-Hydro model is used to quantify the impact of different meteorological conditions and downstream water levels on floods. The model’s simulation results show that rainfall intensity is a more dominant factor in flood generation than downstream water levels. Even with a return period of two years, rainfall causes significant flooding of almost 25% in the study area. The model also indicates an increasing influence of tidal conditions due to relative sea level rise. Still, the increased influence is mainly visible in the borders of the Ciénaga de la Virgen with minimal residential impact. These conclusions highlight the need for mitigation efforts to focus primarily on improving the areas influenced by the rainfall, especially for short return period rainfall events.

Various flood mitigation measures are considered for the southern border of the Ciénaga de la Virgen to reduce floods caused by the dominant intense rainfall. Wetland creation and evaporation measures are ruled out due to the dominant influence of rainfall-induced floods and the inability to prevent floods caused by extreme rainfall events respectively. After investigating the scale required, considering feasibility, the three types of measures that are converted into mitigation options are:
• Mitigation Option 1: Increasing channel dimensions,
• Mitigation Option 2: Increasing infiltration,
• Mitigation Option 3: Constructing retention areas.
These mitigation options are assessed for their impact on reducing flooded areas and flood levels. The three mitigation options are evaluated to recommend the best-preferred mitigation design. This evaluation employs a Multi-Criteria Analysis (MCA)...

The political interest in fish habitat suitability and, consequently, of fish populations has increased. The fish habitat suitability is a key factor for successful ecological restoration, for example via dam removal and implementation of fish passage. Furthermore, the fish population composition determines the ecological quality of water bodies within the Water Framework Directive (WFD).

The aim of this thesis is to develop an ecological model for fish habitat suitability in the river Rotte basin in the Netherlands that can be used in policy development for ecological restoration. The model serves as a decision support system to explain differences in fish population and evaluate the impact of management actions to modify fish habitat suitability. The river Rotte suits perfectly for this case study, because recently a fish passage has been realised between the Rotte and Nieuwe Maas to facilitate fish migration and expand living conditions. Furthermore, the river Rotte is a designated WFD water body, but the current status of the river varies between "poor" and "moderate" due to an unbalance between plant-loving and benthivorous fish species.

The model developed in this thesis is a Bayesian Belief Network model that predicts habitat factors for food preference and preference for habitat structure. The model is based on machine learning with a set of cases from monitoring data and predicts the probability distribution for fish habitat suitability for plant-loving and benthivorous fish species. The model has been applied to assess the impact of local conditions on fish habitat differentiation and to evaluate the impact of management actions on fish habitat suitability. The research shows that the Bayesian Belief Network model is very useable for policy making and facilitates the participation of various stakeholders. However, the current version of the model shows inadequate prediction accuracy and relies heavily on sampling data. This can be improved by expanding the scope of the model to include other water bodies in the Netherlands and by using metamodels for specific model variables.

Overall, the Bayesian Belief Network model is functional and usable for policy making, but further improvements are needed to enhance its prediction accuracy. This can be achieved through expanding the model's scope, evaluating its performance, and including more habitat factors in the model structure.

This study aimed to generate accurate evapotranspiration (ETA) data for the Mountain Aquifer region using the SEBAL model in Google Earth Engine (GEE) with a resolution of 30 m. The analysis provided valuable insights into evapotranspiration's spatial and temporal dynamics and its relationship with land use characteristics. Significant spatial and temporal variations in ETa were observed, ranging up to 2400 mm/year with an average of 600-800 mm/year. The generated data is then compared to the results obtained utilizing FAO's WaPOR product for the ETA calculation. Comparisons with previous studies highlighted the variability of ETa values due to different methodologies and spatial resolutions. The ETa values obtained in this study aligned with the average range reported by earlier research. However, discrepancies were observed with the FAO's WaPOR product, primarily due to methodological limitations and variations in parameter values across land use types.

To obtain and fully use long series of high-precision observations of water balance components, it is necessary to quantify and reduce random and systematic data errors. This thesis applies and evaluates a previously developed data fusion methodology for bias-correcting and noise-filtering remote sensing observations of water balance variables that results in a consistent set of estimates that close the water balance. The method combines monthly water balance constraints and probabilistic data models for each water balance variable (precipitation, evaporation, river discharge and water storage), and uses Markov chain Monte Carlo sampling and iterative smoothing to estimate data errors and water balance variables without the need for any ground-truth. The methodology is evaluated here by application to three river basins in China located in different climate zones (humid to semi-arid). The evaluation assesses (i) how the method performs in the case study basins and to what extent data error assumptions in the probabilistic data models are satisfied, (ii) how sensitive the results are to changes in the datasets, and (iii) whether the use of dataset ensembles rather than dataset pairs (as in the original method) changes the results and further improves the overall data fit. The findings for these three research questions are as follows. First, the posterior water balance estimate for humid and arid basins with average standard error of 6-10 mm/month for precipitation, 4-6 mm/month for evaporation, 8-14 mm/month for water storage. Significant increase in both precipitation and evaporation uncertainty during wet summer period and the data error assumption is violated for precipitation in Wuding basin. Second, the results of replacing precipitation are more sensitive for Baihe humid basin, with both precipitation and evaporation increased, while for Wuding basin, replacing evaporation datasets is more sensitive.Third, significant increases in likelihood value by fusing all precipitation and evaporation datasets, with posterior uncertainty of water balance components increased. And the data error assumption is satisfied for Wuding basin.
Therefore, the use of dataset ensembles as opposed to dataset pairs is recommended in further applications of the data fusion methodology. Additional applications may focus on applying the methodology across a wider range of river basins, using a wider range of ensemble datasets (including different GRACE solutions), as well as comparison of different data error models.

The water balance model has been an indispensable tool for quantifying water supply and demand and regulating water resources for decades. By convention, hydrologists use in-situ measurements of river discharge for model calibration. Whereas hardly is it possible for areas suffering from data storage. Researchers have been working on methods dealing with this problem for many years. Satellite datasets are potential substitutes for the in-situ observation of hydrological variables. This research proposes a monthly strategy using satellite time series as model inputs. This research aims to apply this strategy in the absence of gauged data for discharge simulations and predictions. Instead of streamflow, the strategy uses variables like actual evapotranspiration (ETa) and/or water storage for calibration. The research challenges lie in the situation that there can be significant errors in data derived from satellite products. Also, hydrological models designed for discharge simulation cannot necessarily function well when calibrated on other terms. This boils down to the research questions as follow:
1. How large are water balance data errors and to what extent can they be reduced?
2. How large are water balance model errors and to what extent can they be reduced?
3. To what extent does quantifying and reducing data and model errors eliminate trade-offs in fitting multiple datasets?
The strategy is constructed based on an error estimation and water balance data fusion method and the original and the advanced version of the Water Partition and Balance model (Wapaba), then tested in the Smoky Hill River catchment. When using unprocessed data, the water balance is not closed for the basin. The discharge simulation has the fitting precision index the Box-Cox transformed root mean squared error (TRMSE) in the range of 0.70 - 1.43 for different datasets and in calibration and validation period. Indexes of discharge fitting σma$ exceed 1.66. After closing the water balance with the mean time series of all fluxes, TRMSE decreases to 0.55. Considering data uncertainties, TRMSE is further declined to 0.29 and σ drops to 0.46. After that, the model structure is also improved. When using the modified model to calibrate on only ETa and TWSA for calibration (TRMSE for discharge = 0.87), the performance is similar to that of using the original Wapaba on all fluxes (TRMSE for discharge = 0.86). The fitting precision index σ for TWSA also decreases.

The research demonstrates the effectiveness of the data fusion method in correcting satellite time series and sheds light on the potential of application of this strategy in the ungauged area through the comparison of different calibration cases. After modification, the strategy is able to reproduce the flow regime, without using in-situ data, to the same degree as all three hydrological components (discharge, actual evapotranspiration and water storage) are used for calibration.

Hydrological models are often used to evaluate future changes in streamflow. Despite the strong aware- ness of non-stationarity in hydrological system characteristics, model parameters are often assumed stationary and obtained through calibration on past conditions. The representation of system change in hydrological models is challenging, as a lot of uncertainty abounds on changes in future climate and ecosystems. However, it is shown that ecosystems co-evolve with the prevailing climate conditions. There is increasing evidence that vegetation adapts its root zone storage capacity - considered as a key parameter in any hydrological model - corresponding to moisture deficits in the root zone. This is the main assumption underlying the water balance method. In combination with long-term water budget estimates from the Budyko framework, this method has the potential to meaningfully describe future climate-vegetation interactions within the context of process-based hydrological models. Accordingly, this study provides an exploratory analysis for six catchments in the Austrian Alps to in- vestigate future changes in root zone storage capacity and their impact on modelled streamflow in the past and under two emission scenarios in the future. Our findings show that, although parameter ranges of the root zone storage capacity significantly narrow-down for climate-based estimates, modelling performance on past streamflow is similar when using the calibrated and climate-based parameter sets. Following climate projections from 14 cli- mate models, adaptive climate-based parameter estimates are predicted to increase by 10-100% in all catchments in the future. However, little to no dissimilarity in modelled future streamflow is found when adaptation in root zone storage capacity is included in the hydrological model. Modelled differences in annual mean, maximum, and minimum flows remain within 5%, with slight increases for monthly streamflow and runoff coefficients. Thereby, little to no evidence is found that time-dynamic representation of root zone storage capacity significantly alters modelled future streamflow and suggests limited necessity for its inclusion in hydrological models to obtain rea- sonable descriptions of future streamflow in the investigated Alpine catchments.

Hydraulic properties of aquifer systems are usually estimated by conducting field experiments, which are called aquifer tests. Different approaches have been applied to simulate the drawdown data, such as graphical type curves, analytical solutions, and grid-based models. Since the type of groundwater system varies greatly from location to location, only a few general solutions have been developed. Computation of hydraulic properties are limited by the size and time step of grid-based model. Determining the boundary conditions is also difficult in actual groundwater system. Thus, in order to better simulate various aquifer systems, semi-analytic approaches have been promoted and used in aquifer test analysis. In this report, three different softwares aiming at aquifer tests simulation are presented. Both MLU and TTim are based on semi-analytical solutions and need users to build their own aquifer models, while AQTESOLV provides a base model for adding a variety of data and a stack of choices containing both analytical and semi-analytic solutions. Some benchmark analyses have been performed to assess the performance of TTim with limited types of aquifer systems. This research focuses on the application of TTim to different aquifer systems and the investigation of TTim's performance compared to the other two softwares. In this study, fifteen aquifer tests have been simulated with TTim. Test 0 uses hypothetical data to verify TTim's capabilities of retrieving specified parameters and reporting accurate confidence interval. Test 1 through Test 10 are pumping tests which are taken from reported field experiments and are grouped as confined systems, leaky systems, and unconfined systems. Test 11 through Test 14 present four slug tests with different top boundaries and well construction. The values and confidence intervals of the calibrated parameters are compared to the results of AQTESOLV and MLU. Improvement of conceptual models is carried out by model structure adjustments and parameters set adjustments. Different models' performance are assessed by root-mean-squared-error and AKAIKE Information Criterion (AIC). Most of the pumping tests and slug tests can be conceptualized using either ModelMaq and Model3D within TTim. Model3D is recommended when conceptualizing unconfined systems. The top boundary needs to be specified as `confined', and an additional thin aquifer needs to be added to simulate the specific yield, which is calibrated separately. The performance of TTim is similar, in general, to AQTESOLV and MLU. Well construction parameters cannot be calibrated with AQTESOLV, and only one aquifer system is available. TTim is more flexible and accurate than AQTESOLV when the groundwater system has information of multi aquifers and well construction. Modifications of parameters to be calibrated and model structure have been carried out to improve TTim's performance. It is concluded that aquifers with multi subdivided layers perform better when the well is partially penetrating or the observation wells have different depths. Calibration of well construction parameters may also contributes to a better simulation, but they are usually sensitive to the initial values. It is important to note that adding parameters may give better results, but whether this is significant needs to be tested by the AIC criterion.

Calibration and model prediction is always affected by uncertainty in the forcing data, response data and structural error in the model. The storm-dependent parameters are believed to be able to capture these errors and improve model prediction. The goal of this research will be to further investigate and develop the storm-based approach and compare it to more traditional approaches, including the use of static (time- invariant) parameters and the use of GLUE to capture model errors. The hypothesis in this paper is the ran- dom variation of storm-dependent parameters can capture the model error and improve the prediction. The storm-based method will apply a sensitivity analysis to identify the storm-dependent parameters that are most likely to vary by storms. The variation of storm-dependent parameters will give large changes in model performance which is measured by Nash–Sutcliffe efficiency. In the storm-based method, parameters will be calibrated storm by storm in the calibration period. Then in the validation period, streamflow will still be predicted storm by storm through picking each parameter set from the calibrated parameter sets. Besides, the effect of dryness and the optimal threshold for identifying the storm epochs on the model performance will also be explored in the storm-based method. The results obtained by the storm-based method will be compared with the method using static parameters and GLUE method. Six case studies calibrating a con- ceptual rainfall-runoff model with for parameters with daily data illustrate the improvement of prediction obtained by the storm-based method for dry basins. The extent of variation of storm-dependent parameters is very random in each case which indicates there is error in the model. Moreover, the extent of variation of storm-dependent parameters has no relation with initial water storage, rainfall characteristic and basin characteristics. Although the parameters cannot be predicted deterministically, they can be predicted prob- abilistically with the histogram or fitted distribution for the calibrated parameter sets in the future work. By making storm-dependent parameters vary with storms and other parameters constant, the storm-based method performs better for drier basins while worse for wetter basins compared to GLUE method and tradi- tional method. The logscore value obtained in storm-based method(e.g. -0.68 for one of the dry basin A) is larger than those obtained in the traditional method (e.g. -1.23 for basin A) and GLUE method (e.g. -0.67 for basin A). Additionally, the RMSE values for total flow obtained in the storm-based method are all smaller than those obtained in the traditional method, and GLUE method for dry basins. This suggests the storm-based method is more applicable for dry basins and this method should be better developed for wet basins. What is more, the extent of variation of storm-dependent parameters has no relation with basin characteristics but the mean and variation of the storm-dependent parameters can be obtained. Hence the extent of variation of parameters can be described probabilistically.

An accurate representation of water retention curves is important for various reasons. Traditional models already exist for the representation of these curves, with one of them being the van Genuchten model. When soil parameters are available, the van Genuchten model can be used to plot water retention curves. However, when these soil parameters are not available, regression can be performed to estimate and predict the water retention curves. Commonly, the Non-Linear Least Squares regression method is used in combination with a certain water retention model. Problems arise for inhomogeneous soils as the traditional water retention models tend to break down. To improve the representation of water retention curves, Gaussian Process regression will be implemented. This method will be combined with the Non-Linear Least Squares method to obtain new representations of water retention curves. These new curves are better in terms of curve fit and uncertainty, when compared to the traditional method. These comparisons can be made visually, by observing the plots and their confidence intervals, as well as quantitatively by computing the log-likelihoods of the different methods. When comparing the results of the log-likelihood computations for both methods, it follows that the value of the log-likelihood is greater for water retention data with correlated residuals. In the case where the residuals are uncorrelated, the log-likelihoods are equal for both methods and no improvements are observed.

This report introduces the use of Genetic Programming (GP) into hydrology by describing the results of GP using conceptual hydrological models as physical representation. First the possibilities of GP are tested on synthetic data, which results in a shortlist of good working objective functions and insight in the most important GP settings. The test on real data in the Belgium Ardennes showed that GP using the objective functions KG10, MM and Shafii performed better. Nevertheless all three models performed not well on simulating the low flows and high peaks. Furthermore GP using KG10 and MM both results in simple serial models which perform well overall, but bad on quick response runoff. Shafii resulted in parallel models which show quick response flow, however GP it is not able to capture the fast responses correctly (yet). GP has the potential to improve the understanding in the behaviour of catchments, however it still needs the human mind to observe, compare and analyse the modelling results. The main consideration with GP is to look for a balance between: model search space, objective function, randomness and (computational) time. The challenge is how to lead GP in an efficient way without removing the possibility of finding unknown patterns.

Drinking water company Dunea N.V. produces drinking water by artificially infiltrating water into the dunes. During this dune passage arsenic is mobilized causing arsenic levels up to 4 µg/l in the influent of DWTP Katwijk. In this thesis the effectiveness of the MnO4 – Fe(III) dosing was evaluated on RSF influent of Dunea’s treatment plant Katwijk. The goal of the experiments was to reduce the residual arsenic concentration to <1µg/l while minimizing negative effects on further treatment steps. Experiments were performed in a pilot filter filled with filter material originating from the RSFs in Katwijk to ensure conditions were similar to RSFs in Katwijk. The performance of MnO4 – Fe(III) dosing was evaluated at different filtration velocities (4, 6, 10 m/h) and at different top layer materials (anthracite (1.6-2.5mm) and pumice (2.5-3.5mm)). In this research multiple dosages were evaluated and residual As <1µg/l was consistently achieved by dosing 0.5 mg NaMnO4 /l and 1.0 mg Fe(III) /l. It was found that the effectiveness of the MnO4 – Fe(III) dosing was not influenced by the filtration velocity applied in the RSFs. For all the filtration velocities checked the removal efficiency ranged from 60-70%, resulting in As levels <1µg/l. The pumice used appeared to show slightly higher removal rates than the anthracite that was used. However, for full scale application purposes in Katwijk the used anthracite was better suited. The pumice filter clogged faster leading to a pressure loss of 95 cmWc after 6 days while the anthracite filter reached a pressure loss of 73 cmWc. Due to the coarse top layer less particles were removed in the top layer and caused the sand underneath to be loaded more and to clog more quickly. The AOCF process did not negatively influence the filter performance. Mn-, Fe- removal and nitrification all still performed to standards and the turbidity of the RSF effluent decreased from 0.3 FTU to 0.03 FTU. Based on preliminary experiments the sedimentation properties of the backwash water changed and the sedimentation velocity increased. The MnO4-Fe(III) dosage can effectively reduce the residual arsenic concentration to <1µg/l and is ready to be applied at Katwijk. However, the negative effects on the runtime of the RSFs can’t be overlooked. The runtime of the current filters is expected to decrease by about 50% causing a higher stress on the backwash water treatment.

Polder areas as typical in the Netherlands require real time control on the water levels. Efficient pumping is a subject with increasing interest due to a required use of 20% green energy by 2020 set by the government. In 2010, the energy requirements already exceeded 175 GWh per year. For optimal control, water level predictions are important. However, they are uncertain due to errors in model structure, parameter estimation, initial states and observed forcing data. Especially the rainfall was found to contribute to water level prediction uncertainty. The uncorrected radar rainfall forcing data available in real time underestimate the corrected radar rainfall measurement over 100%, worsening with increasing rainfall intensities.

The aim of the research was to asses whether data assimilation can reduce the uncertainty on the water level predictions in a Delfland polder. The polder is modelled with a conceptual rainfall-runoff model in Sobek RR, while the data assimilation is implemented as an ensemble Kalman filter in OpenDA.
In order to manipulate the states in Sobek RR for data assimilation, the model was extended by a black-box wrapper that was newly created in Java for this research.

To asses the theoretical applicability of data assimilation in a polder area, four twin experiments were carried out, which differed only in the model components to which noise was applied. The four set-ups included the use of a rainfall multiplier, additive noise on the model states, additive noise on the groundwater state and a single multiplier for all states. The results were compared visually and objectively through the probabilistic Nash Sutcliffe, a model efficiency coefficient.

All set ups for the twin experiment generated good results for the assimilated water level, mimicking the observed water levels closely with probabilistic Nash Sutcliffe values ranging between 0.92 to 0.97. However, only the set up with the rainfall multiplier exhibited good score measures for the hidden states, with an average of 0.6, showing a significant improvement over the case without assimilation. The other twin experiments failed to represent the hidden states, with average score measures below zero.

The data assimilation set-up with the rainfall multiplier was then applied using water level observations from Tedingerbroek polder. The pump pattern and magnitude of water levels of the modelled and observed water level differed significantly. The influence of the different model components on the modelled water level was assessed, from which was concluded that these could not account for the difference in modelled and observed values. Possible explanations are due to a non-representative measurement point in the polder or additional unknown water fluxes coming into the polder.

Recommended further research should focus on creating a larger observation network; implementing multiple water level observation locations, collecting pump discharges and taking measurements of the storage in greenhouse basins. The last could give a method to validate the assimilation of the hidden states, whereas the pump discharges should be used as additional observation in the data assimilation.