The exchange of energy and mass between the ocean and atmosphere plays a crucial role in shaping oceanic and atmospheric circulation patterns. However, accurately representing these air-sea fluxes remains a challenge for current weather and climate models. Improving the accuracy of bulk flux pa- rameterizations is crucial to improve the quality of weather forecasts and climate predictions, as these parameterizations play a fundamental role in estimating the air-sea fluxes. This study aims to evaluate the performance of the ECUME and ECUME6 parameterizations in simulating air-sea fluxes by utilizing in situ observations obtained from R/V Ron Brown and R/V Meteor, and conducting a comparison with the COARE3.6 parameterization.

To evaluate the ECUME and ECUME6 parameterizations, surface flux diagnostics are established, which illustrate how air-sea fluxes vary with changes in the respective atmospheric variables. By comparing the surface flux diagnostics of the in situ observations with those of the parameterizations, sources of error are identified. The analysis reveals that both ECUME and ECUME6 tend to overestimate the heat fluxes in comparison to EC observations and the COARE3.6 parameterization, with ECUME6 exhibiting a larger overestimation. The degree of overestimation becomes more pronounced as wind speeds increase. Concerning the momentum flux, the parameterizations exhibit an underestimation, with the discrepancy becoming more significant at elevated wind speeds.

By employing an offline model for ECUME and COARE3.6, the iteratively obtained parameters are compared. This analysis demonstrates that the air-sea fluxes derived from the parameterizations strongly depend on the determined neutral transfer coefficients. Addressing these sources of error and refining the parameterization methodology can improve the accuracy of the parameterizations and enhance their applicability for estimating air-sea exchange between the Earth’s surface and the atmosphere.

The wind turbine wake is a downstream region of kinetic energy and velocity deficit, higher turbulence and has a complex helical vortex structure. The wind turbine wake’s stability depends mainly on the ambient turbulence and the wind turbine tip- speed ratio. Researchers have developed several wake control techniques to mitigate this issue, such as static axial induction control by torque or pitch control, dynamic axial induction control by pitch control and actuating flaps or other actively controlled add- on devices. Some gaps are present in these methods, in terms of applicability of these techniques in terms of full scale validation and practicality to have the devices as add on to existing blades.

To address these research gaps, this study focuses on designing and evaluating segmented Gurney flaps (in line with the ECN (now TNO Wind Energy) patent by Edwin Bot and Arne Van Garrel) for wind turbine blades to enhance the wake recovery by inducing turbulence in the wake to excite the tip vortices. 4 Gurney flaps were attached in the tip region of each blade of a GE 3.8MW research wind turbine. Field tests were conducted in this study for the wind turbine wake (using a scanning LiDAR to scan a sector up to 5.5D downstream at different altitudes; with a scan time of roughly 3 minutes) and performance analysis (using 10 minute averaged measurement data). Free vortex wake simulations were conducted to validate the faster wake breakdown by the change in lift distribution upon addition of segmented Gurney flaps. Simulations using dynamic blade element momentum theory with IEC NTM inflow conditions from cut in to cut out wind speed were conducted to assess the structural impacts on the retrofitted wind turbine.

The field tests’ wake analysis was quantified with different wind speed, turbulence intensity, wind shear, wind direction conditions. Post processing of LiDAR data involved filtering, creating bin averaged data set, using Gaussian process regression and retrieval of the required wind component. The results show a consistent increase in wake recovery, generally at all downstream distances. The retrofitted configuration results are associated with a higher standard error because of a shorter testing period (due to increased noise levels) than the baseline configuration. The wake simulations indicate an earlier wake break down position, by roughly 2D.

The simulations indicate AEP increase for the retrofitted wind turbine to be roughly +0.2% (+50MWh), at the expense of increased blade and tower structural loads. The wind speed weighted damage equivalent blade flapwise bending moment was found to increase by roughly +2% to +4%; with peaks observed in partial load region of the wind turbine, associated with the operating angle of attack in this region.

Satellite data, such as optical and Synthetic Aperture Radar imagery, can provide information about the location and level of destruction caused by natural hazards. This information is essential to optimise the rescue mission logistics by humanitarian aid organisations and save people in need. Currently, many Automatic Damage Assessment (ADA) methods exist, designed explicitly for one data type with corresponding spatial resolution. However, the weather and satellite coverage conditions can hinder rapid and complete data acquisitions after large events. Therefore, it is important to identify the limits and capabilities of novel methodologies testing various data availability scenarios and adjusting them to become robust and widely deployable.
In this research, the Convolutional Neural Network Caladrius of 510 ­ an organisation of the Red Cross Netherlands is selected to perform experiments. Initially, the model was designed to input high-­resolution imagery and based on the Siamese Architecture, including two Inception­-V3 modules fol­lowed by three connected layers. The multiple experiments are based on single­, dual­, and cross­mode scenarios, representing data characteristics with varying resolutions, satellite sources and observation sensor types. The xBD dataset provides pre- and post-­event high­-resolution optical imagery of numer­ous disasters with corresponding validated damage labels of the included buildings. Subsequently, this dataset is replicated in three down­sampled versions and using Sentinel-­2 1C and Sentinel­-1 GRD data. With the use of the Macro F1-­score and the Cohen’s Kappa coefficient the performances are compared and the predictions’ reliability is determined in operational situations.
The results indicate that a lower resolution of the input data has a negative effect on the correct classi­fied buildings. A linear relation does not express the loss in performance, as most damage propertiesare captured between 0.5­ and 2.5-meter. Consequently, this implies that the Sentinel 10­-meter res­olution datasets provide little recognisable features. The Sentinel­-2 1C experiment outperforms the Sentinel­-1 GRD, which equals the output of a random classifier. However, no final conclusion is drawn between the true prediction rate of the model compared to the input data type; optical and SAR im­agery due to the non-­optimal experiment circumstances and limited included datasets. Furthermore,the results from the dual-­mode mapping showcase the importance of identical data characteristics be­tween train and test datasets. Conversely, with the use of the cross-­mode experiments, it is found not essential to match the pre-­ and post­-event resolution imagery. This latter is very promising for the Red Cross and creates flexibility to construct datasets quickly after the disaster has struck.

Providing farmers with the tools to monitor their crops continuously and reliably can aid the scaling of global food production to meet the ever-growing demand. The optical Normalised Difference Vegetation Index (NDVI) is commonly used to monitor crop greenness as an indicator for biomass, but it is limited by clouds and signal saturation. Synthetic Aperture Radar (SAR) imagery, which is not hampered by clouds, can be used to complement the NDVI. The Biomass Proxy (BP) combines the NDVI and SAR data, but spatial biomass estimations still strongly depend on the NDVI and therefore face the same limitations.
This study aims to assess the potential value of spatial SAR data to approximate the in-field biomass distribution. The SAR signal from Sentinel-1 and the NDVI signal from Sentinel-2 were analysed temporally and spatially for fields of maize, barley, oat, and spring wheat in the Dutch province of Flevoland. It was assumed that consistent SAR patterns in the spatial signal correspond to biophysical changes in the monitored crops. A framework to detect these patterns and include them in the BP was created based on combining cluster detection with spatial autocorrelation. The components of this framework demonstrate that backscatter intensity, phenological stage and crop type influence the probability of consistent patterns and that consistent patterns could not be observed from the spatial NDVI signal. Moreover, it was found that the BP’s sensitivity to the input signals depends on crop type. With the knowledge of when and where consistent patterns occur, targeted research can be done to understand the spatial SAR signal better and, thereby, optimally use all available information.

Satellite radar interferometry (InSAR) is a powerful technique for monitoring deformation phenomena. While deformation phenomena occur in a three-dimensional (3D) world, one of the limitations of the InSAR phase observations is that they are only sensitive to the projection of the 3D displacement vector onto the radar line-of-sight (LoS) direction. To uniquely estimate the three displacement components, we would require at least three sets of spatiotemporally coinciding independent (STCI) LoS observations, (i.e., scatterers on an object that is not subject to internal deformations, observed at the same time) available over the same Region of Uniform Motion (RUM). More importantly, the system of equations needs to have a full rank coefficient matrix. Unfortunately, in most practical situations at most two sets of STCI LoS observations are available, resulting in an underdetermined system with an infinite amount of possible solutions.

Within the InSAR literature we encounter different approaches to address the underdeterminancy problem, unfortunately often with either mathematical or semantic flaws. We concluded that the InSAR community has no uniform way of addressing the underdeterminancy problem. We developed a taxonomy for the different fallacious approaches that can help by evaluating InSAR results and reviewing InSAR papers.

Moreover, using the east-north-up (ENU) reference frame for decomposing the LoS observations provides results that are not tuned to the needs of the end-user of an InSAR product. Therefore, we developed an alternative solution to the underdetermined problem, in the form of a `strap-down' approach, which uses a local strap-down reference system that is fixed to the deformation phenomenon with transversal, longitudinal, and normal (TLN) components. For many practical cases, such as line-infrastructure, landslides, or subsidence bowls, analysis of the main driving forces supports the assumption that significant deformations in the longitudinal direction are unlikely.

We found that using the strap-down approach gives physically more relevant estimates. Moreover, it results in more relevant estimates since it properly includes all uncertainties. We can further conclude that the conventional way of communicating (PS)-InSAR results by means of a `dot distribution map' is sub-optimal when considering the quality of the estimates. For many cases, `vector arrow maps', or traditional geodetic vector-based visualizations, including error ellipses are a viable and more optimal alternative.

To effectively mitigate environmental changes and to manage coastal environments, a good under- standing of nearshore bathymetry and its evolution is required. The coastal zone is expected to be vulnerable in the foreseeable future due to the combined effects of increased coastal impacts and a growing coastal population. The use of spaceborne remote sensing methods has been developed in recent decades to estimate nearshore bathymetry. Compared to traditional methods, spaceborne remote sensing techniques offer a relatively large spatial coverage as well as frequent temporal mon- itoring. The Sentinel-2 mission (EC/ESA) is potentially able to provide regular bathymetry estima- tions, mainly due to its significant revisit time and freely accessible data.

This research focuses on the development of a spatio-temporal cross-correlation model in order to construct a one-minute wave-representing video. The video is used to show the potential for wave- derived bathymetry estimation, exploiting a DMD-based DIA to invert depth. The model is ap- plied to two distinct datasets: a synthetic FUNWAVE dataset and Sentinel-2 imagery. The Sentinel-2 imagery covers research sites in Saint-Louis, Senegal and Capbreton, France. An application frame- work related to the model is furthermore developed by analysing a set of synthetic sinusoidal waves.

Three model requirements are created based on the research objectives. The model should show wave propagation for at least one minute and, moreover, the waves in the video should accurately represent the wave field. Both requirements are demanded by the use of a DMD-based DIA. Lastly, the resulting video should enable an accurate bathymetry estimation. Based on the requirements, the developed spatio-temporal cross-correlation model includes four model parts: pre-processing, image resolution augmentation, wave characteristics estimation and video construction.

The research has led to a range of insights. A discrepancy between the quality of constructed videos and related bathymetry estimations is observed. The videos as constructed by the model generally show a good representation of average wave propagation for a sufficiently long duration, while the bathymetry estimations are less accurate. The low quality of bathymetry estimations is explained by three main error sources: celerity estimation errors, the applied filtering methods and the way in which the video is constructed. These three error sources together lead to videos that lack pixel-wise detail and therefore decrease the bathymetry estimation quality. The developed application frame- work shows that estimating wave characteristics from Sentinel-2 imagery by means of the model is at the edge of possibilities. In temporal sense, relatively large celerity estimation errors are expected for wave periods lower than 5 s and higher than 7 s. The model is less sensitive for spatial parame- ters: as long as wavelengths are larger than circa 150 m the celerity estimation error is acceptable.

All together, the developed model and related video constructions offer added value, although not for the purpose of a bathymetry estimation by means of a DMD-based DIA. The constructed videos represent wave propagation in an average sense and can therefore be exploited for wave-related purposes, e.g. obtaining wave spectra, estimating dominant wave direction and estimating wave climates. In general, the video offers a way to enlarge the temporal range of Sentinel-2 imagery. It is furthermore concluded that the model is probably more suitable for use in combination with other types of imagery, including satellite imagery with larger burst duration and increased spatial resolution as well as standard video imagery.

ISAR imaging is a well known technique which uses range and Doppler information from a radar to generate an image of a moving target. Research performed on the combination of a large receive array with airborne ISAR is scarce. This thesis is aimed to reach the highest possible image quality for AMBER, a radar with a 24 element receive array which was developed by TNO. Optimizing the image quality was split up in two parts, motion compensation and clutter suppression. A literature study is performed to research the existing methods. A performance evaluation is performed on different motion compensation techniques based on a simulation which is set up to resemble the radar parameters of AMBER. Based on the results of this evaluation, keystone formatting combined with image contrast maximization is concluded to be the best fitting approach. For clutter suppression a similar approach is followed with real measured data. Apart from the existing techniques, MVDR, DPCA and ODPCA, a new technique is proposed which filtered a target from clutter based on motion. It is referenced to as the IDPCA. The techniques show better image quality than regular beam forming. To enhance the image quality more, a method is searched to combine different clutter suppression techniques. This leads to a unique method which is proposed in this thesis. The new technique exploits the circular phase variance of pixels in the range Doppler image between different sub arrays. With this approach, strong clutter can be filtered in the range Doppler domain based on its angle of arrival. The newly proposed technique is applied as a mask, as it does not contain amplitude information of the target. Combined with the ODPCA method, the final image is generated. The combination of techniques shows clear improvements from the other discussed approaches.

Land subsidence is a complicated phenomenon, relevant in the Holocene coastal-plain of the Netherlands, that can be triggered by many different mechanisms and components. To cope with the effects of land subsidence, detailed information on which mechanisms are causing subsidence and its variation over time and space are needed. Subsidence studies predominantly focus on grass and agricultural areas with organic soils above the groundwater level. In this study, subsidence is also evaluated for dikes and the urban environment. It is attempted to model subsidence behaviour of four different projects, grass plots within the Krimpenerwaard polder, locations along the N3 Dordrecht, an embankment constructed for the A5 Badhoevedorp and locations along the Markermeerdikes, and to verify these model results with subsidence measurements available. Existing one-dimensional models for different subsidence components are used in the evaluations. For grass and agricultural areas it is found that subsidence behaviour could be modelled and the results lie within the range of uncertainty of the subsidence measurements used. However, the equations used are often empirical and do not include all relevant influencing factors and couplings. Compression by degradation of organic material as a response to drainage of organic soil is the main subsidence component in these areas. Anaerobic degradation of organic material is also included in the modelling approach, a component that is often neglected in other subsidence studies. Two infrastructure related projects provided insight to subsidence components in the urban environment. Compression by loading is the main subsidence component relevant in this area type, and thus settlement models are suitable to approximate the subsidence behaviour. However, a slight underestimation of creep shows that the existence of a small additional subsidence component for a situation with soft soil layers pushed below the water table could not be excluded. Unravelling subsidence behaviour from dikes showed to be more complicated, where an interplay between different components is relevant. Based on the evaluation in this study compression by loading and compression by an oxidation or shrinkage component are indicated the main subsidence components. Many uncertainties and assumptions used in this evaluation influenced the results. For all area types disentangling subsidence into contributions of separate components based on a total measured subsidence signal causes large uncertainties, as modelled subsidence contributions for individual components could not be verified.

Soil Moisture is a key hydrological variable since it controls the interactions between the atmosphere, biosphere and hydrosphere. It is responsible for the partitioning of precipitation into evaporation, transpiration, percolation and run-off. Soil Moisture monitoring is used to indicate droughts in vegetated areas and is an important parameter to early warning systems for flood. Therefore, throughout the years human population attempted to monitor and control it. The advent of Remote Sensing, during the past decades, enormously influenced soil moisture research by enabling acquisition of large scale data. Many Remote Sensing systems were developed exclusively to study this variable. Land subsidence, triggered both by human activity and natural processes, is another phenomenon whose monitoring is crucial for the environment and the human populationa and is thus an essential variable which needs to be continuously monitored in vulnerable areas, since it can damage buildings’ foundations. Remote Sensing and more specifically Microwave remote sensing has been pivotal in studying land deformation and subsidence in near real time. Detecting and monitoring land subsidence and deformation with InSAR method has been meticulously researched however there are still obstacles to overcome such as vegetation. The aim of this research is to study whether InSAR closure phases can be used to detect moisture changes. The idea of using closure phases for soil moisture estimation was proposed by De Zan, Parizzi, et al. 2013 and further studied by Zwieback, Hensley, and Hajnsek 2015a. The closure phase inversion model of De Zan and Gomba 2018 is implemented in this thesis using Sentinel-1 C-band for the inversion and SMAP data for the evaluation of the results. The results support the idea that this method has potential over bare soil and low vegetated areas but struggles to overcome vegetation due to its limited penetration capability. Furthermore, soil moisture changes may introduce a systematic error to land subsidence measurements. For this reason, the idea was to make use of the generated soil moisture data to produce interferometric phase corrections over the study areas. However, the results are inconclusive due to the poor quality of the data over vegetated areas.

In this thesis, the joint DOA-range estimation of stationary targets is investigated using multiple FMCW MIMOs with super-resolution capability. To address the low azimuth resolution problem of single small MIMO, a novel topology of array is used, which consists of multiple MIMOs arranged along the azimuth to increase azimuth resolution by extending the effective aperture size. According to such topology, signal models are formulated using FMCW waveform. To accurately model the scenarios, targets are considered as near-field objects for the system, but they are treated as far-field targets for each MIMO.

After formulating signal models, two algorithms are investigated and tested to localize targets in the observing domain. The generalized 2D-MUSIC algorithm is applicable for both multi-static and mono-static configurations of the system. The FBSS technique is used to tackle highly correlated signals. Though this algorithm provides super-high resolutions, it requires prior knowledge of the number of targets (model order). The performance would drop significantly by incorrect estimation of model order. To avoid this limitation, an augmented Lagrangian method is introduced for the first time to address the localization problem, which is named extended C-SALSA. This method casts target localization problem as a sparse representation problem, and then the problem is transferred from estimating targets' locations to the problem of sparse spectrum estimation. It utilizes variable splitting and augmented Lagrangian to handle objective functions. For both algorithms, with the accurate positions of sensors in the system, geometrical constraints of the system can be maintained by applying the same search grid to all virtual arrays, consequently, data association is avoided.

The feasibility of both proposed methods are analyzed with numerical simulations of point targets and electromagnetic simulations of an extended target. MATLAB simulation results demonstrate that the azimuth resolution is increased using multiple MIMOs with both proposed algorithms. Besides the resolution, the accuracy of the generalized 2D-MUSIC is also compared with the derived CRLB. Moreover, CRLB is used to analyze the potential accuracy for the estimation results of the mono-static configuration. In spite of the requirement of model order, the generalized 2D-MUSIC outperforms the extended C-SALCA for extended targets and is more robust for off-grid targets.

Traditionally, a geographical area can be monitored via land based communication systems or spaceborne remote systems for disaster prevention. However both methods have their respective limitations in terms of resolution, accuracy and real time observation. A new suborbital constellation concept offers an independent unit device which will be attached to an aircraft and use commercial flights as means of monitoring forest fires or forecasting solar energy. The shape of the fuselage can be used to redirect the creeping wave generated from the antenna away from the aircraft’s surface, towards the satellites for communication purposes. This research aims to determine an optimum antenna placement on the electrically large body in order to satisfy efficient communication with ground and satellites from the same location on the aircraft.
The quarter wavelength monopole operating at 1.575 GHz in L1 band for the GPS communication and a patch antenna with dielectric constant of 2.2 for the substrate operating at 800 MHz for the GSM communication are selected. Various algorithms and approaches are investigated to perform comparison of accuracy for results and computational resources between different simulation techniques namely Multi-Level Fast Multipole Method (MLFMM), Physical Optics (PO) and Uniform Theory of Diffraction (UTD) using a modern computational electromagnetic numerical platform FEKO. The simulation results show that the optimum antenna location on a Boeing 737 is found to be at the bottom of the aircraft in the conical section of the tail 30.95 m away from the tip of the nose in the longitudinal direction. In addition to that, this work provides a unique perspective to system engineering for using aircraft as a land monitoring station.

The North Sea is known for its marine activities, which need ocean data for safe operations. The ocean surface is described using the two-dimensional wave spectrum. This ocean wave spectrum can be estimated using wave buoys, wave models or remote sensing techniques. Wave buoys and numerical models are known to contain inaccuracies, which stresses the importance of other observation techniques. The solution of this thesis is using a remote sensing technique from Sentinel-1: Synthetic Aperture Radar.

In total 11 case studies were analyzed in the North sea and 3 case studies in Portugal. Sentinel-1 SAR cross-spectra were verified and validated using wave buoy measurements and cross-spectra from simulated SAR images from OCEANSAR. In total, 6 North Sea case studies showed a positive result, where a swell peak was visible and the peak matched spectral data from wave buoys. Case studies from the Portuguese coast showed the best results, where SAR cross-spectra agreed well with buoy measurements.

Further improvements include the removal of non-linear contribution from cross-spectra, mapping of wave buoy measurements on the quasi-linear swell spectrum and calculating the full Modulation Transfer Function. This thesis showed it is possible to measure low-frequency waves in the North Sea using Sentinel-1 Synthetic Aperture Radar. SAR images lead to a better understanding of the movements of swells and provide additional information for marine activities. A combination of wave buoy data, Sentinel-1 SAR data and OCEANSAR data showed measuring swell waves is possible with this state-of-the-art remote sensing technique.