This thesis investigates the efficacy of artificial intelligence (AI) models, particularly convolutional neural networks (CNNs) and U Net architectures, in reconstructing datasets with missing velocity data in river flow analysis. Optical flow and Particle Image Velocimetry (PIV) techniques have emerged as valuable tools for analyzing river flow patterns.
Through a comprehensive literature review, CNNs, and UNet models are identified as promising tools for this task due to their ability to capture intricate patterns in datasets. The study compares the performance of the U-Net model against a statistics-based hydrological benchmark model, revealing the superior performance of the U-Net model.
Furthermore, the analysis explores how the performance of AI models varies with differing quantities of missing data, by masking available data and comparing reconstructed values against the ground truth, highlighting the importance of data availability.
Additionally, the study investigates the influence of spatial patterns in training data on model performance, including patchy versus random missing data in the field of view, simulating more datasets more likely available in reality. This clarifies the challenges encountered in predicting grid points under different training dataset conditions.
Finally, the study identifies areas within the dataset that are particularly challenging to predict, shedding light on factors contributing to prediction errors. These findings underscore the potential of AI models in hydrological applications and provide valuable insights for future research in the field.
Our findings show that U net is capable of reconstructing velocity fields from a river flow better than an average benchmark that uses the average values, with varying accuracy depending on input data.
The average benchmark model had a relative error close to 0.2 in every instance, whereas the U-Net model showed relative errors ranging from 0.085 to 0.006. Errors from a patchy mask are ranging from 0.09 8 to 0.031.

The Eastern Nile Basin is a highly populated region dealing with inadequate energy systems and water scarcity issues. Moreover, its population is expected to increase in the following decades, leading to even higher electricity and water demands.
In order to sustainably meet this demands and guarantee access to electricity and water for all, new technologies and careful energy planning can play an important role.
In this context, floating solar power is a relatively new technology with promising advantages, such as the synergies between solar and hydropower resources, the exploitation of already existing infrastructures, and the reduction of evaporation rates and land use. These become even more relevant if seen in the context of the Eastern Nile Basin countries, where the need for efficient energy sources and solutions to the water scarcity issues are vital.

In this work, the role of floating solar power in the sustainable fulfillment of the increasing energy demand of the region is explored. The novelty of this study consist in the introduction of floating solar power in a long term regional energy system cost-optimization model (OSeMOSYS-TEMBA) at a single plant resolution. To do so, the single hydropower plants are also explicitly modelled, allowing both the spatial disaggregation of floating solar power plants and the connectivity between the countries via the Nile river. The regional approach is further enhanced by the presence of electricity trade links between countries, which connect the energy systems of the single countries directly.
Finally, the role of floating solar power on the energy system's footprints is evaluated in terms of CO2 emissions, land use and water savings. To this extent, a new methodology for land use accounting and pricing is proposed, and findings from previous studies are brought together to assess the evaporation reduction rates caused by the floating solar power plants.
This extended modelling framework is then used to analyse different scenarios, exploring hydrological regimes under different climate change projections and policy decisions such as the introduction of taxes for carbon emissions and land use change.

The results show that floating solar photovoltaics are a cost-optimal technology since early stages in the modelling horizon, and their full assumed potential is developed under every scenario. Their role in satisfying the energy demand of the whole region reaches 3\% of the generation mix in the reference scenario, but it increases to 4.3\% with the introduction of taxes on carbon emissions and land use. Moreover, the introduction of such policies cause an anticipation of floating solar power's capacity expansion. On the other hand, the tested climate change projections do not affect the results relevantly.
The sensitivity analyses, however, prove that the obtained results are very sensitive to the assumptions behind capacity constraints and costs of these technologies, which need more dedicated research.
As far as the energy system's footprints are concerned, the results show that the implementation of floating solar power can help reduce the total emissions and land use slightly, and cause evaporation reduction rates up to 376 million m\textsuperscript{3}/y (approximately 2\% of the total evaporation from hydropower reservoirs).
The optimal locations to invest in this technology are identified to be the largest hydropower plants in the system (Lake Nasser, the Grand Ethiopian Renaissance Dam and Merowe reservoir), but the reason of this choice relies in the very large size of these plants, which emerge for highest
FPV capacity deployment and water evaporation savings at the large scales considered.
Future research is still needed to reduce the uncertainty behind the key parameters (costs, capacity constraints), improve the representation of hydropower production, improve the evaporation assessments and investigate the effects of implementing floating solar power at smaller spatial and temporal scales.

Throughout sub-Saharan Africa, agricultural activities in urban and peri-urban areas are threatened by urban sprawl. These activities are important not just for the livelihoods of those who take part in it, but also for the food security of the city. Historically, in Kumasi 40% of the crops in the city are sourced from urban and peri-urban agriculture. However, Kumasi is also a city that is growing rapidly both in population and in area: since the 1980’s population has more than quadrupled and the area has increased more than tenfold. Because of this, there is a need to find ways to maintain urban and peri-urban agriculture under the threat of urban sprawl and the climate change in the future and also work well in the Kumasi context.
We used the agro-hydrological model AquaCrop informed by 4.5 years of local weather data, soil data and crop data to model the effectiveness of adaptations in maintaining food security for three neighbourhoods in Kumasi. Local farmer management practices that were determined by a survey of 150 Kumasi farmers made a distinction in management practices between adaptations and crops. This model gave yields and irrigation water uses for five crop groups that are commonly cited as crucial for food security. Combined with population growth predictions and land use and land cover analysis this allowed us to make a statement about how well adaptations can meet current and future demand of crops and how much space is needed to meet demand. This model was supplemented with a survey of vendors and farmers to investigate barriers against- and preferences for adaptations.
Backyard gardening and sack gardening turned out to be the best options to maintain peri-urban agriculture. These adaptations use the least space, for the highest yields. From the survey of farmers and vendors, no strong objections were found against the implementation of these adaptations. Over 80% of surveyed farmers and vendors felt that backyard gardening was a good adaptation to maintain peri-urban agriculture in Kumasi. Backyard gardening and sack gardening are optimally suited for growing vegetables and legumes, which are the easily perishable crops and thus benefit from a short supply line, which many vendors cite as solutions for spoilage.
It is possible to ensure self-sufficiency for these crops with 5-9% of total land in the Feyiase neighbourhood, 11-20% in Ejisu and with 14-22% in the Kwadaso neighbourhood. It is therefore recommended to target at least a majority self sufficiency, by reserving a >50% fraction of this land for backyards and sack gardening spaces. At the same time, any available marginal lands should be allocated for the growth of cereals and tubers, to allow for the production of these subsistence crops, until the production of the remaining demand is fully met by rural import. With these adaptations and recommendations, it is possible for peri-urban agriculture in Kumasi to maintain its important role throughout the coming decades.
In order to determine if there are any deviations from these conclusions, future research should focus on including empirical crop data tailored to AquaCrop or use a model that is better suited to represent the chaotic nature of (peri-) urban agriculture. While the model results do not differ significantly from the crop yields as found in literature, a model informed by local crop data can be an even better representation of the situation in Kumasi. Furthermore, a study into the long-term effects of adaptation on nutrition can reinforce our conclusions on food security.
Finally, there is an opportunity to develop more high-tech agricultural methods like greenhouses and aquaponics. There is also an opportunity to incorporate wastewater reuse schemes, following examples from other countries, such a urine reuse, or low-tech treatment with sludge harvest. For this, it is recommended to set up educational programs and pilots.

In order to evaluate measures to increase water availability in and around reservoirs it is necessary to have reliable reservoir water storage models. For that reason, it is necessary to assess the validity of these models. The aim of this research is to do this assessment of two models in data scarce semi- arid regions. The Nakamb ́e catchment is used as a case study. The study covers important aspects of hydrological modelling, such as model input (data) selection, hydrological model choice, calibration, model performance testing, parameter sensitivity, and reservoir water storage simulation.
For the selection of the optimal model forcing data diverse precipitation products are reviewed: CHIRPS, ERA5, and local measurements. Among the evaluated datasets, CHIRPS emerges as the superior choice validated against the local measurements. With respect to the potential evaporation, the combination of ERA5 and local measurements results in the most suitable potential evaporation data, leveraging the temporal and spatial aspects of ERA5 and the absolute values of the local measurements.
Comparing a lumped hydrological model (HBV) with a distributed model (SBM Wflow) in simulating river discharge reveals that the HBV model outperforms its counterpart in simulating discharge. This contrast in performance is attributed to potential overparameterization in the Wflow model, coupled with the complexities of parameter estimation in data-scarce areas. The HBV model, while bearing simplifications, benefits from a more comprehensive calibration process. The model performance is strongly influenced by the calibration efficiency, where the significantly shorter simulation time of the HBV model facilitates an extensive Monte Carlo sampling-based calibration, in contrast to Wflow’s time consuming manual parameter adjustment.
Additionally, the sensitivity analysis showed that in the HBV model, the parameters affecting actual evaporation are the most sensitive one. This emphasizes the importance of accurately simulating this component for the proper model performance. The Wflow model exhibits strong equifinality due to the many parameters within the model. The complexity of this model made it impossible to test all parameters and therefore only some parameters are tested.
Both reservoir water storage models studied, the HBV Reservoir Water Storage Model (HBV RWSM) and the Wflow reservoir module, can effectively simulate reservoir water storage fluctuations, although they differ in how the components are calculated. Due to data limitations, it is impossible to determine which, if any, of the models is correct. However, based on the downstream discharge the HBV RWSM displays a more promising performance.
In conclusion, the HBV model outperformed the SBM Wflow model in simulating discharge due to its simplicity and ease of calibration. Sensitivity analyses highlighted the significance of accurately representing actual evaporation. Both water balance models, the HBV RWSM and the Wflow reservoir module, performed similarly concerning the NSE values. The fluxes contributing to the water balance in the two reservoir water storage models differ significantly. The lack of data on these fluxes makes it impossible to determine which models performs best. Data limitations remain a significant hurdle in model evaluation, emphasizing the need for additional data collection, particularly upstream and downstream of the reservoir, to enhance reliability and reduce uncertainties.

Om goed te kunnen voorspellen waar, wanneer en hoeveel het gaat regenen is het belangrijk om satellietbeelden, weermodellen en grondmetingen te combineren. Het in 2007 opgezette project 'The Trans-African Hydro-Meteorological Obsevatory' (TAHMO) heeft de afgelopen jaren 500 goedkope weerstations geplaatst, verspreid over 21 landen in Sub-Sahara Afrika en wil uitbreiden naar 20.000 weerstations. De grondmetingen die met de al geplaatste weerstations zijn gedaan, kunnen gebruikt worden om weermodellen te ontwikkelen en te verbeteren.

Over de te verwachten regenval in Afrika is nog weinig bekend. Vaak wordt regen gemodelleerd met een Poissonproces. Daarom is vanuit ‘Delft Global Initiative’ de vraag gekomen of het Poissonproces een goed model kan zijn voor regenval in Afrika.

In dit onderzoek hebben we regendata van één weerstation gebruikt dat tijdens de moesson gegevens heeft verzameld op Mafia eiland, een eiland zo'n 20 kilometer voor de kust van Tanzania.

Een Poissonproces kan homogeen of inhomogeen zijn en we hebben de data op beide processen getoetst. Voor het homogene Poissonproces zijn veel verschillende toetsen te vinden in de literatuur. De Monte Carlo simulaties zijn gebaseerd is op de geïntegreerde verdelingsfunctie en gebruikt om de p-waarde te bepalen. Voor alle regen samen in de gehele observatieperiode was de p-waarde 0 en daarom kan de regen niet met een homogeen Poissonproces worden beschreven.

Voor de inhomogeen Poissonproces benadering hebben we ervoor gekozen om de intensiteit van de regen te benaderen met een stuksgewijs constante intensiteit. Met taut string werd de regen opgedeeld in 474 deelperiodes met een constante intensiteit die per deelperiode verschilt. Voor iedere deelperiode is getoetst of de regen met een homogeen Poissonproces kan worden beschreven. De p-waarde is voor iedere deelperiode bepaald met Monte Carlo simulaties en we werken met een significantieniveau van 5%. Voor 448 van de 474 deelperiodes kan de regen met een homogeen Poissonproces worden beschreven.

Een inhomogeen Poissonproces met stuksgewijs constante intensiteit lijkt daarom een redelijk bruikbaar model om de regen op Mafia eiland mee te beschrijven. Of dit model ook bruikbaar is voor alle andere weerstations van TAHMO zullen toekomstige studies moeten uitwijzen.

Rain-fed agriculture is the main source of food in Ghana therefore improving quantitative rainfall estimates is essential for local farmers to predict crop growth using vegetation models. Rainfall dynamics in the tropics is an ongoing topic of research due to their complexity and sub-grid precipitation variability. At the same time, tropical areas such as Ghana are the most affected by a lack of proper rain gauge network coverage. Traditional methods rely on empirical assumptions and statistical theories that require continuous calibration and still struggle to accurately represent local variability. The aim of this paper is to demonstrate the potential of a Deep Learning (DL) approach using bi-spectral information of water vapor imagery (WV) and thermal infrared (TIR) as a starting point to develop an effective alternative to the Cold Cloud Duration (CCD)
method which is a widely applied statistical technique by satellite rainfall products like Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) and Tropical Applications of
Meteorology using SATellite data (TAMSAT) that are specifically designed for Africa.
WV inhibition of low-level features assures the correct depiction of strong convective motions usually related to heavy rainfall which is crucial in tropical areas where convective rainfall is dominant. The addition of WV 7.3m is particularly beneficial in North Ghana as tropical systems
are advecting dry air from the nearby Sahara desert creating discontinuities in precipitation events which translates into dry intrusions and dry slots seen in the images of the WV channel.
The developed Deep learning model showed strong performances in binary classification where it outperformed IMERG-Final false alarms count resulting in lower rainfall overestimation (FBias < 2.0), although further research is needed to overcome the very poor relation between GEO-IR images and actual rainfall estimates at the surface.

This article uses Sentinel-1 satellite images to identify rice fields in Rwanda from 2017 to 2021 in an attempt to derive the current status of rice in Muvumba catchment, in the northern part of Rwanda. The timing of rice cultivation in each season in that area was identified as not homogeneous, but generally aligned with the local rainy season. The results identified after late 2019 show a large change, with flooding from extreme rainfall as a possible cause. The paddy fields may have been completely flooded and the infrastructure destroyed. Policy changes in Rwanda’s agriculture may also be a contributing factor. Among them, the implementation of land consolidation policies can influence some of the farmers’ options, for example, by withdrawing from the rice planting program. The absence of official data and field data makes this project not being able to provide definite reasons for the results. The open discussion of this project makes it exploratory and offers the possibility for potential follow-up studies. The uncertainty of the results also suggests that more attention needs to be focused on these topics. However, what is certain is that the use of remote sensing images to monitor rice has the potential to be cost-effective.

Recent research suggested that the digital elevation models can be considered, as an alternative to the altimetry data. However, the water prohibits the ability of monitoring the construction the waterbed, due to the loss of the returning signal. The elevation models have therefore a flat surface, which prohibits the extraction of the water level and volume variability.

The goal of this thesis was to develop a tool, that uses the elevations where no flatting had occurred, and create a linear representation of the shape of the lake or reservoir. The following research question was formulated:

What is the potential of applying elevation models to monitor the volume levels of lakes and reservoirs, when replacing the flat elevations with extrapolated depths?

The case study comprised of two parts. First, the validation of the water level extraction was conducted for pre-selected lakes and reservoirs. The results showed an average RMSE of 4.10 [m]. When analysing the distribution of lakes and reservoirs data, the RMSE was centred at 1.27 [m] (NED and ALOS) and 3.70 [m] (SRTM). The comparison between the three considered elevation models, showed a minor improvement regarding the relative water level. At the same time, it demonstrated that the number of waterbodies that can be monitored increases when applying the developed tool.

The second part of the case study aimed at providing an insight in the potential of applying the DEM for estimating the volume variability. An analysis of the USGS in-situ data compared to the results, showed that the overall RMSE decreased for each considered model. From the same analysis, the relative time-series could be determined for each model i.e., in case the minimum amount of surface area was available and in case the water level variation was significantly noticeable.

The results from this work indicated the potential of the global elevation models to monitor the volume levels of lakes and reservoirs. The developed tool demonstrated that elevations can be extrapolated, which would fit a more realistic shape of the reservoir at areas where the bathymetry was flattened or not found.

Accurate information for extracting the water level and assessing the volume variability of lakes and reservoirs currently depends on extensive hydrographic surveys. This work provides guidelines for an alternative method: the Linear Bathymetry for Digital Elevation Models tool.

This report describes the development of an early warning model for thunderstorm occurrence around Lake Victoria. It is the first model to use the TAHMO lightning sensor data to predict thunderstorms. This study hopes to contribute to the TWIGA focus area of increasing disaster resilience through forecasting and early warnings and could be used to reduce the impact of thunderstorms for the communities around Lake Victoria, and to aid the flights of Zipline’s medicine-delivery drones.

In this report, the design process of a high precision positioning system for Ghana is described. The system consists of two devices, a base station and a user operated rover. These devices provide precision positioning, using dual frequency receivers and the principle of differential positioning. The high precision positioning system can be utilized for multiple purposes. The system can open up business opportunities in surveying, can be used to make height maps for flood prediction and to gather meteorological data. The gathering of weather data is the main purpose of TWIGA, under which flag this project runs.
During this project, the main focus is to create a working product, which can be tested in Ghana as soon a possible. Therefore, the highest priority is to make a functional device in terms of accurate positioning, data connection and power supply. This device should be suitable for the local environmental conditions, such as tropical rain and a high humidity.

In this research, the flood damage caused by heavy rain and plastic blockage in the neighbourhood of Atonsu, Kumasi is quantified. Rapid urbanisation and poor waste management result in floods with an increasing impact in Kumasi, the fastest growing city of Ghana. This rapid growth of the city results in uncontrolled urbanisation of the floodplains and induces damage to homes during a flood event. Furthermore, due to a poor waste management system around thirty percent of the waste produced in Kumasi gets dumped in the streets and forms blockage of the water flow in the river. Moreover, the river is used as a dump not only by citizens but also by factories, which results in contaminated river water. As this research aims to give a first insight into the actual flood damage in Kumasi, simplified methods are applied, as they do not require a lot of data. To quantify the tangible damage, expressed in monetary values, depth-damage curves are constructed for different types of land use in the study area, to show the relation between the inundation depth or inundated area and the monetary damage. Besides that the intangible damage is quantified in the form of the number of lives lost and people exposed to waterborne diseases, both depending on the total number of people affected. The damage is quantified for floods induced by four different rain events, all with different peak precipitation, duration and frequency. The results indicate that the larger the rain event, the more damage it causes. It is furthermore estimated that plastic blockage of the flow of the river results in an increase of monetary damage of up to 5.7 percent and the number of people that are affected by the flood increases with up to 17.4 percent. Finally, the results of this research show the magnitude of flood damage in Kumasi and could be used to show the urge of measures that need to be taken to reduce the flood damage.

Plastic pollution in aquatic ecosystems is a global problem. Rivers transport large quantities of litter from land to the oceans. Plastics of different sizes and properties are widely present at various locations in river systems. Concerning macroplastics, previous studies mainly focussed on floating plastics, however, substantial parts of the litter can be transported underneath the water surface. Currently, this suspended litter load remains understudied. Submerged litter is presently monitored with the use of nets. The use of nets has several disadvantages. In large rivers, nets can usually, not be deployed over the full depth and width of the river. Moreover, using nets is labour intensive and requires fixed structures to be deployed. To overcome this problem, this thesis aimed to develop and test a new method for monitoring submerged riverine litter. In accordance with the fish detection abilities of sonar, it seems to have a high potential for identifying suspended underwater objects.

In this study, the litter detection ability of a low-cost single beam echosounder (Deeper CHIRP+) were investigated. Three different experiments were executed, in specific, controlled tests in an artificial environment, semi-controlled tests in a natural environment and litter monitoring in a naturally flowing river. The controlled tests, to get an insight into the scanning technique and detection abilities of the echosounder, were performed in the Kerkpolderbad in Delft. During these tests, the influence of actual object size, object depth and flow velocity on the sonar signal was investigated. The semi-controlled tests were carried out in the Rio de San Pedro, in Andalusia, Southern Spain. During these tests, several plastic targets were used and repeatedly released in the river, passing the sensor. For this, objects of different material properties and sizes were used. Lastly, plastic was monitored in the Guadalquivir and Guadalete rivers in Andalusia. In the Guadalquivir river, the sensor was operated together with nets for validation purposes. In the Guadalete river, monitoring took place for 18 hours from a pedestrian bridge, at different locations over the cross-section of the river and under varying tidal conditions.

The performed tests showed a significant relationship between the dimensions of the reflection signal, derived from the sonar observations and the actual object size. However, object orientation and deformation play a role and lead to deviations in the signal dimension results. A second relation, regarding flow velocity and signal dimensions, was observed. The larger the flow velocity, the smaller the sonar signal. Additionally, signal intensities can, for four out of the eight objects tested, be related to material properties but differences in signal intensities are relatively small. Regarding the river monitoring activities, suspended litter items can be counted, river tide is influencing litter transport and litter is present over the full river depth.

The following main conclusions are drawn based on this research:
- Echo sounding can be used for detecting suspended riverine macroplastics. Litter items can be counted, and fish can be discarded from the sonar readings by their specific displayed shapes.
- Litter size can be estimated when looking at the sonar readings, however, several factors, such as flow velocity, object orientation and deformation have to be taken into account when estimating litter size.
- In the Guadalete river, significantly more suspended litter is transported when river water flows into the sea compared to river water flowing inland. The counted litter items were approximately uniformly distributed over the river depth.

In general, using echo sounding for suspended litter monitoring is potentially useful to gain a better understanding of the suspended litter transport, from which prevention and mitigation strategies could be optimised. For further research, it is recommended to use an echosounder for which the raw sonar data can be exported as a standard digital file. Moreover, the set of test objects should be extended, including more variation in object size. To separate signal size and signal intensity, objects of different size but same material properties and objects of the same size but different material properties should be used for testing. Finally, other types of sonar such as side scan or multibeam sonars may potentially lead to more accurate sonar readings regarding litter size and material estimations.

This research assesses the impact of plastic waste accumulation on flood events with citizen observations in Kumasi, Ghana. In Kumasi, the fastest growing city in Ghana, flooding events have become more intense due to rapid urbanisation and land use intensification. Additionally, rapid population growth, urbanisation and lack of organised waste collection, brings along tonnes of solid (plastic) waste on the streets, riverbeds and -ways. Plastic waste creates blockages in drains and rivers, which causes flooding of open areas, streets and houses. To what degree plastic waste accumulation aggrevates flooding in Kumasi remains poorly understood. Therefore, this study investigates to what extent plastic accumulation impacts flood severity, in response to rainfall events in the city of Kumasi, Ghana. This is done by modelling different rainfall scenarios in a 1D-2D model developed with the software HEC-RAS. This model was forced with static and dynamic data available from open source databases and collected during fieldwork. Data collection took place during six consecutive weeks in May-June 2019 and consists of bathymetry and discharge measurements, a flood- and plastic waste survey performed together with citizens. The impact can be determined using different model scenarios varying in rainfall magnitude and in obstruction of riverways, serving as plastic accumulation. Flood severity was analysed in terms of extent, depth, velocity, local hydraulic impact and flow distribution. The model has been validated with two events, one during which velocity and stage was measured, which results in the same order of magnitude, modelled slightly lower than measured in the field. The second event was a historic rainfall event, which could be validated with community reported flood depths. From this validation it became clear the flood extent of the event was underestimated and depths close to the streams were overestimated. The results show that the selected rainfall events without blockage cover the studied area between 11-25% with water depths higher than 10 cm. The inundated areas are adjacent to the river streams and inundate flat and low lying areas. Comparing this to the scenarios where bridges where obstructed to 33% and 67% individually, the total inundated area remains more or less the same, varying between 0-3% in inundated area compared to the scenario without blockage. However, in terms of flood distribution, areas around the blocked bridge are impacted more, thus inundation areas shift within the observed study area. Comparing this to the rest of the area, upstream or downstream less flooded areas arise, making the change in total flood extent only change slightly. Furthermore, the two bridges that are obstructed have different impact on the flow in the study area, as the scenarios of obstructing the second bridge with 33% produce a numerical instability and therefore incorrect results. From the plastic waste survey it followed that plastic waste is very much abundant in the study area, thus the occurrence of plastic accumulation at the bottleneck locations seem inevitable. The results of this research have shown the main contribution to flooding is flow from upstream areas exceeding river capacity, while local blockages mainly aggravate flooding around the blocked locations.

Urban agriculture lies at the core of the Water Energy Food nexus and seems to provide a partial answer to confront modern trends such as population growth, climate change and resource depletion by increasing food security in cities and enhancing sustainability in an urban realm. The assembly of a WEF nexus framework taught, however, that most work that has been published on the nexus is very hypothetical and that the acquisition of quantitative data poses the biggest challenge in WEF nexus research. The mere absence of data collected at a local level impedes informed decision making on nexus sector integration and feasibility of sustainable solutions. This study attempts to bridge the existing knowledge gap and aimed to contribute to the quantification of the nexus regarding urban agriculture. It investigates the water, energy and nutrient demand of urban farms along with the presence of those resources in urban waters at three case study sites. Demands for water and nutrients (nitrogen & phosphorus) at a greenhouse in Amsterdam and a community farm and a container farm in East-Boston could be met by resources present in urban waters (rainwater and wastewater) in the direct vicinity. Whether enough energy is available to run each of these farms is related to the type of agriculture which is applied.

Urban pluvial flooding occurs when the run-off converted rainfall exceeds the capacity of the sewer or stormwater system (Houston et al., 2011). This can result in damage to ecology, infrastructure, disruption to human activities, injuries and in the worst-case scenario, loss of life (Biniyam et al., 2017). With the expected increase in the world’s population living in urban areas to 68% by 2050 (UN, 2018) and a more frequent occurrence of extreme weather events with a longer duration and higher intensity in the future (IPCC, 2012), there is a need to improve early warning system and disaster management (Restrepo-Estrada et al., 2017). Lopez et al. (2005) mentions that one of the key factors in hydrological models to determine accurate flood estimates is to have accurate rainfall input. However, many urban areas might lack this information because sensors are not available, or the number of sensors is too few to cover the entire region with an acceptable resolution (Restrepo-Estrada et al., 2017). Kidd et al. (2017) mentions that the density of rain gauges varies per region, with Europe and Eastern-Asia (including Japan) having a decent coverage by rain gauges, while the rest of the world has a sparse coverage of rain gauges. This study assesses the potential of citizen observatories for improving the spatial measurements of rainfall in comparison to a single professional station. Two types of citizen observatories are used, semi-professional stations (TU Delft) and citizen weather stations (Netatmo), that are located near the city of Rotterdam. In total there are 9 TU Delft stations and 73 Netatmo stations in a region of 256.7 km2. Based on a spatial variance analysis between these two types of citizen observatories and a professional station (KNMI) in the region, weighing factors are determined in order to merge and interpolate the data from the citizen observatories into a single rainfall map. This interpolated map is then compared to the radar rainfall maps, which are bias-corrected based on a network of professional ground stations (Overeem et al., 2016), and provide rainfall data with a 1x1 km spatial resolution. The baseline for improvement comes from the assumption that the rainfall measured by the KNMI station is uniform over the entire study area. By comparing the differences between KNMI and radar and citizen observatories and radar, it can be assessed whether the citizen observatories are an improvement. The results show that the interpolated rainfall maps are better at capturing the spatial structure of the rainfall event, both on the small-scale (10-minute timestep) and event-scale (total accumulation). When it comes to the actual rainfall values, it became clear that rainfall intensities within a 4 km radius from the KNMI station were better represented by the KNMI station while after a distance of 8 km, the citizen stations clearly helped give a better representation.

The northern Ghana climatological conditions are favourable for agricultural production. The annual average precipitation of 800-1250 mm/y is theoretically sufficient for farmers to be year-round self-sufficient. However, the majority of precipitation falls in a 4-month wet season spanning from late May to October. As a consequence, the region is subjected to both seasonal flooding and long periods of drought. The dry season agriculture is of moderate intensity, takes place at small-scale and is groundwater dependent. In the near future it is not unthinkable that extraction exceeds natural recharge and groundwater withdrawal is no longer sustainable in the northern Ghana regions. The small-holder farmers' use of Aquifer Storage and Recovery (ASR) systems can potentially contribute to the continued sustainable use of groundwater in northern Ghana. An ASR system acts as a seasonal bridge. The system recharges flood water and extracts groundwater in periods of drought. A feasibility study on the sustainable use of an ASR system in northern Ghana has been performed by taking present conditions and multiple system improvements into account.

Aquifer tests are carried out at five study sites in northern Ghana to determine local geohydrological conditions. The TTim analytic element modelling environment is used to analyze the obtained groundwater drawdown data and derive parameters for subsurface characteristics. TTim allows for the inclusion of additional model parameters (e.g. borehole storage, well skin resistance and multiple model layers) and outperforms the analytic Theis method in this research. Although some uncertainties are present in the derived subsurface parameters, plausible values for transmissivity (T) and storativity (S) are suggested to be present in the ranges of respectively 1 to 100 (m2/d) and 1e-3 to 1e-2 (-).

The year-round performance of a northern Ghana single ASR system is studied with a MODFLOW model. The potential types of ASR system improvements that are examined are (a) the extension of daily pumping time, (b) the enlargement of the borehole diameter, and (c) the reduction of the well skin resistance. The ASR systems sensitivities to changing environmental conditions are explored by (a) the degradation of well depth by clogging, (b) the shortening of the wet season inundation time, and (c) the reduction of the wet season inundation levels. Research results show that well maintenance is key for the performance of existing (and new) ASR systems. The recharge and discharge volumes can be improved by cleaning of the borehole depth and well screen. In the case of a new ASR system, the performance can positively be influenced by an enlargement of the borehole diameter. Furthermore, the construction of a proper permeable well skin (screen and gravel-pack around the well) can also result in increased system capacities. Despite the imposed options of system modifications, the geographic position of an ASR system remains of utmost importance for system performance. The construction of an ASR system at a location sensitive to flooding (riverbank overtopping or rainfall based) can be beneficial from a sustainable perspective. Recharge volumes are normative for the sustainable use of an ASR system. The recharges are (approximately linear) dependent on the time-span and levels of inundation. Moreover, the research contains soil scenarios, and demonstrates that the ASR system performs significantly better in regions with higher transmissivity (T) values.

To give insight on some financial aspects of an operational ASR system, the obtained (improved) ASR system discharge capacities are transformed to agricultural and financial yields. A subdivision of the dry season into a tomato and a groundnut cropping season demonstrates that financial yields are crop type dependent. The ASR system revenues are dominantly affected by the choice in crop type(s) and crop-specific market prices. The yields are compared to the ASR system pumping costs. The importance of pump selection is demonstrated by the implementation of the Pedrollo 4" submersible pump efficiencies. The use of a pump that is tuned to local conditions can be beneficial for the operational costs of an ASR system. Although no distinctive conclusion on the financial feasibility can be drawn, the examined system improvements are substantially beneficial for the revenues of a northern Ghana ASR system.

Quantifying the anthropogenic impacts such as reservoir characterization is a big challenge in the field of water management. In this work, a computer graphics based geometric approach is presented which can predict the underlying topology of large-scale reservoirs. The proposed algorithm uses freely available, satellite based landscape data of the surrounding regions to predict reservoir characteristics. The premise of the presented approach is that the slope of the surrounding landscape is an important determinant to understand the underlying landscape of the reservoirs. This method outperforms the existing state-of-the-art techniques used to estimate the storage capacities drastically, both in terms of estimated maximum volume stored and estimated volume area curves. Evaluation of the geometric model presented is done on 28 reservoirs using the HydroSHEDs data which was developed using the Shuttle RADAR Topography Mission conducted by NASA. This HydroSHEDs data was obtained in 2000 which acts as ground truth data for the reservoirs built after 2000. Further, model parameters are introduced to improve the modeling capabilities of the reconstructed reservoirs. This approach further intensifies the case of using computer graphics techniques for raster based analysis and provides a platform for further research in the field of water management.

What are the national and provincial economic potentials of ocean thermal energy conversion in relation to the provincial electricity demand fulfilment in Indonesia? When we are talking about the economic potential of OTEC in Indonesia, we are talking about potential as a result of different possible configurations. On the national level, the economic potentials of both cost assumptions are more than enough to fulfil the total electricity demand of the country. However, Indonesia seizes an immensely broad area of land and sea. A big part of the potential lies outside the provincial boundaries, far away from demand centres. Still, when only the provincial potentials are considered, the total economic potential is higher than the demand. This does not necessarily mean that the available resource is accessible by the demand centres, though. The available resources of OTEC are not distributed evenly throughout the country. Some areas are blessed with remarkably good resources in abundance while the other areas are not so lucky with the resources. Hence, the stations are selected based on the proximity to the demand centres. It turns out that based on the results of the thesis, it is not possible to fully supply the electricity demand on the provincial level in Indonesia solely by OTEC electricity production of the selected stations. It is indeed possible for some areas to do that, but in other areas, there is no economic potential at all. However, it is not likely that a single resource will ever be used to cover 100% of supply. In this sense, there is a huge opportunity for OTEC to enter the energy landscape and become the provider of baseload electricity

In Manica Province in Central Mozambique, agricultural activities consist for a significant part of farmer-led irrigated agriculture. These farmers construct and maintain their own irrigation systems with local inputs and have a commercial intent. Organization is often individual and farmers receive zero to minimal external support from donors, government or non-governmental organizations. This type of agriculture seems to work quite well. It is increasing quite fast and has a high production rate per hectare. In addition, it can boost agricultural production and contributes to higher food security, poverty alleviation and economic growth. However, it is unclear what the actual extent of irrigated agriculture is. A way to obtain this might be making use of satellite imagery and remote sensing. However, the feasibility of this depends on various factors and is not proved yet. Therefore, this study aims to provide insight in the possibilities and limitations of remote sensing regarding the identification and mapping of farmer-led irrigated agriculture in Central Mozambique, by using optical satellite imagery combined with ground data. This study consists of three parts: ground data collection, classification and additional analysis. Ground data is collected during fieldwork in Central Mozambique in three catchments. Classification is performed with a Maximum Likelihood classifier and uses optical satellite imagery acquired by Sentinel 2. Additional analysis consists of terrain analysis with a Height Above Nearest Drainage raster, examination of distances to streams for land uses throughout the research area, determination of the reach of irrigation canals and looking into the possibilities of thermal remote sensing. The classification results are mixed and inconsistent, especially regarding irrigated fields and light seasonal vegetation. A more detailed analysis of spectral signatures and scatterplots shows spectral overlap between these land uses. Analysis on field level or with a more detailed division in land uses shows that spectral responses reflect agricultural practices in some cases, but in general results are unreliable. Results of the additional analysis show similarities between irrigated fields and light vegetation as well for Height Above Nearest Drainage and thermal remote sensing. Also, distance to streams is not suitable as an indicator for irrigation, because irrigation canals increase the reach of the streams. It can be concluded that optical remote sensing as applied by this study does not give accurate results regarding the identification and mapping of farmer-led irrigated agriculture in the study area, because of similarities in the spectral responses of irrigated fields and light vegetation. This low inter-class separability is mainly a result of the heterogeneity of the area and the flexibility of agricultural practices. Due to these diverse practices, the agricultural plots show different and unique patterns both over time and over space, which makes it hard to generalize and classify them. However, even though an accurate substantiation of the extent of farmer-led irrigated agriculture is not feasible, valuable information obtained by this study contributes to better grasping the presence of irrigated agriculture in Ruaca, Chirodzo and Godi catchments.

In deep polders in the Netherlands the dominant mechanism of salinization is via intense seepage through boils. These are local connections between the surface water and a deep saline aquifer. The study area of this research is the Lissertocht. This catchment, which is located inside the Haarlemmermeerpolder, contains a lot of boils that can cause problems for the main land use in this catchment agriculture. During summer the salinity is lowered by flushing the catchment with fresh water from outside the polder.

The objective of this thesis is to measure and model the salt concentrations and their sources in order to optimise the placement of sensors in a salinity monitoring network for optimal flushing in the Lissertocht catchment.

This thesis has provided insights into the detection of boils using different methods (Distribute Temperature Sensing (DTS), EC routing and visual inspection). The DTS method was the only method that is able to detect boils located at the bottom of the bigger main channels. Moreover, DTS is able to characterize the boil flux on a qualitative level and also on a quantitative level for a boil that is located near a pump; both have been done in this research. The boil flux is calculated with an energy balance and an 1D advection-diffusion model, the results correspond to values found in literature.

Data from CTD divers and sensitivity analysis have shown the influence of various processes on the spatial and temporal distribution of the salinity in the catchment. The sensitivity analysis in SOBEK also shows the areas which are sensitive to flushing, the flushing speed and the main pathways which the water from the inlets take. The complexity of the spatial and temporal distribution of the salinity is reduced to a small system of linear components using the Principal Component Analysis (PCA). For this case study, it was shown that over 90\% of the total variance in the system can be captured with three orthogonal linear components. A predictive model is built based on the PCA, this model is able to test the performance of a layout of sensors by reconstructing the principal components with virtual measurements. This shows that with a small number of sensors, a salinity monitoring network can be built that is able to predict the spatial and temporal distribution of the salinity in the catchment. Based on the knowledge from the measurements and modelling, three categories of key locations to monitor can be identified.

The optimisation of the sensor locations shows that a greedy algorithm is a very efficient way to solve the sensor placement problem. When more sensors are added with the greedy algorithm the performance of the monitoring network improves, but after six sensors the additional gain of an extra sensor is small in this catchment. The robustness analysis against disturbances shows that a monitoring network with more than three sensors is more robust. Moreover, the robustness analysis shows that a monitoring network is most sensitive to the locations of the boils.