The Greenland Ice Sheet (GrIS), which stores freshwater equal to more than seven meters of potential sea level rise, strongly interacts with the global, Arctic and North Atlantic climate. In a warming climate, the GrIS has been losing mass and is projected to lose mass at an increasing rate. The interactions between the GrIS and the climate have the potential to amplify or dampen GrIS mass balance responses to a CO2 forcing. We investigate the impact of ice sheet-climate interactions on the mass balance and climate of the GrIS using the Community Ice Sheet Model version 2 (CISM2) coupled to the Community Earth System Model version 2 (CESM2). We compare idealized simulations with a non-evolving and evolving ice sheet topography in which we apply an annual 1 % increase until we reach four times pre-industrial (PI) CO2 concentrations. Furthermore, we analyze an idealized simulation in which we first apply a 4x PI CO2 forcing and thereafter annually reduce atmospheric CO2 by 5 % until PI concentrations are reached. By comparison of a 1- and 2-way coupled simulation, we find significant changes in atmospheric blocking, precipitation and cloud formation over Greenland as the GrIS topography evolves, acting as negative feedback mechanisms on mass loss. Besides, we find that a uniform temperature lapse rate misrepresents temperature changes in the ablation area, leading to an overestimation of the positive melt-elevation and melt-albedo feedback in 1-way coupled simulations, resulting in an overestimation of mass loss. During a 350 year 4xPI CO2 forcing period, the ice sheet loses a total mass of 1.1 m sea level equivalent, and part of its margins retreat land inward. When applying an annual 5 % decrease in CO2 to 1xPI CO2 concentrations, melt reduces rapidly. The small discharge concerned with the retreated state of the ice sheet allows for halting the GrIS mass loss, despite a surface mass balance that is only slightly positive under a remaining global warming of 2 K. During a complex transitional phase towards a colder climate, the GrIS, Arctic and North Atlantic ocean strongly interact, causing the area south of the GrIS to transition from a ’warming hole’ towards a ’cooling hole’. Elevated atmospheric temperatures, larger ocean heat transport and a poorer state of the snowpack, compared to the initial pre-industrial state, result in limited regrowth of the ice sheet under reintroduced PI CO2 conditions.

Fire both shapes and destroys forests. Forest fires are therefore essential to ecological processes and vegetation as we know them. With climate change, forest fires are expected to increase in severity and frequency. To maintain the functioning of the forests, it is important to understand the vegetation response to and recovery from forest fires. Within this field of study, remote sensing techniques are common, and optical indices such as the NDVI are most prevalent.

With recent developments, the ASCAT variables slope and curvature have become of increasing interest in structural vegetation monitoring. These variables are a second-order Taylor polynomial’s first and second derivatives used to normalise the ASCAT backscatter-incidence angle relationship. This study explored the possibility to use these novel variables in forest fire research.
The focus was to discover to what extent the variables responded to a major forest fire.

To do so, grid points affected during the 2009 Australian Black Saturday Fires are compared to unaffected control grid points utilizing Z-scores. These control grid points have been selected based on time series similarity in the two years before the fire. Time series from 2007 to 2021 are used to investigate fire impact and recovery. The ASCAT variables are compared to a similar
NDVI time series to aid in interpreting the results.

The findings show that both slope and curvature are sensitive to the major forest fire. Both variables show an impact shortly after the fire, which can be explained by the loss of scattering elements in the vegetation due to the fire. In the following years, there is a notable recovery which can be explained by the vegetation regrowth forming new scatterers. The NDVI showed similar
behaviour but the recovery was differently timed, suggesting that the signal recovery is driven by something else or that the regrowth of leaves is different from the regrowth of the structural elements that the ASCAT variables represent.

The results help in understanding the ASCAT variables and their interpretation in terms of vegetation scatterers. Especially for the curvature, the clear change in signal deflects the discussion of whether the variable has information potential. Nevertheless, for the ASCAT variables to be applicable in forest fire research, they need to be better understood and additional research is necessary. Suggestions are for example a ground validation study or a global forest fire study, which would suit the coarse resolution of the ASCAT variables better and improve the understanding of the interaction between forest fires and the variables.

Although using the ASCAT variable for forest fire research is in its infancy, the results about its suitability are promising, both for ASCAT product development, as well as for the forest fire research field. Exploring the possibilities further is worthwhile, especially considering the fact that the slope and curvature time series cover a long continuous period starting as early as 1991. This long time series makes the ASCAT variables suitable for long-term forest fire monitoring, and the daily nature of the data might also make them interesting for short-term monitoring.

This thesis developed a forward model for Sentinel-1 C-band co-pol and cross-pol backscatter and coherence using crop biophysical variables including leaf area index, tops weight, surface soil moisture and root zone soil moisture as inputs for sugarbeet. These input variables are simulated using a crop model called Decision Support System for Agrotechnology Transfer (DSSAT). The prediction of SAR signals is conducted using random forest regression model across all the sugarbeet fields in Noord-Brabant, the Netherlands. The correlation between simulated variables and the C-band SAR observables is investigated, as well as an evaluation of the effect of different feature combinations.

Empirical-statistical rainfall landslide initiation thresholds are popularly used for early warning systems to discriminate between the occurrence and non-occurrence of rainfall-induced landslides. However, the few studies that have derived landslide initiation thresholds for landslide-prone and data-scarce Rwanda rely solely on the limited in situ data. Therefore, our objective is to explore the feasibility of using satellite data and hydrological model derived data to derive both trigger and trigger-cause thresholds for landslides in Rwanda. We firstly evaluated seven precipitation products (TRMM 3B42v7, CHIRPS, PERSIANN-CDR, GLDAS 2.1, CFSv2, IMERG, and ERA5) using the rain gauge data as a reference and found that IMERG was the most suitable product for obtaining rainfall triggering conditions. We then studied the added value of incorporating the antecedent soil moisture from both a high spatial satellite data and from a distributed hydrological model following the trigger-cause framework. The results showed that the event precipitation volume E, the event duration D and the bilinear threshold E-D are the landslide initiation thresholds that accurately predict the highest number of landslide events while keeping the false and the failed alarms low. Including the antecedent soil moisture products as the causal variables -expected to account for the hillslope hydrologic processes predisposing the slopes to near failure- did not lead to any improvement with respect to the trigger only thresholds for predicting landslides in Rwanda.

Smallholder farmers cultivate more than 75% of the available agricultural land in Africa and therefore this form of agriculture is crucial to the global food supply. At present, little is known about these smallholder agricultural and related irrigation practices, yet the increasing availability and accessibility of remotely sensed data provides significant opportunities to assess the status quo of these practices. However, these characteristic agricultural landscapes can create complexity in identifying land use with the help of remote sensing. The cultivated plots are small and consist of intercropping systems with dynamic spatio-temporal practices concerning planting, irrigation and harvesting. 
This research aims to provide insight into the usefulness of remotely sensed passive Sentinel-2 Level-1C and active Sentinel-1 SAR data for land use classification of these complex landscapes with a focus on irrigated agriculture, using a case study in Central Mozambique. For this purpose, an open source-code is written that uses open-source satellite data from Google Earth Engine to execute the supervised image classification methodology, using confusion matrices as an assessment method. 
The results of this research appear to show that a nonparametric RF classifier ( = 88.0%) is preferred over a parametric ML classifier ( = 85.0%) for processing the data that is high in variability, in which classifications based on the chlorophyll sensitive Red Edge and SWIR bands provide the highest overall accuracies (>88.0%). However, the classifier overestimates the amount of irrigated areas by a factor of 1.5 in the first and a factor of 3 in the second irrigation season. The opportunistic sampling method appears to cause inflated accuracy outcomes and an optimistic bias towards classification of the main class in training. Spectral analysis of the temporal behavior of various S-2 bandwidths does not provide insight into the underlying mechanisms on which the algorithm performs classification. Although it appears that irrigated agriculture with S-2 data can be identified on the basis of an increase in vegetation biomass and that the classifier benefits from more information through the use of multiple bands. Research into the use of Sentinel-1 SAR data appears to have potential for identifying irrigation. Time series of the VV backscatter signal show a difference between the irrigation class and the classes non irrigated and light seasonal vegetation in irrigation season 2. However, high standard deviations do reflect the high intra variability of the data, and classification accuracies in this period, do not exceed an overall accuracy of 64.1%. The main confusion as identified by the confusion matrices, comes from classes that are often identified as irrigated, whereas they are not, overestimating the amount of irrigated areas as with the use of S-2 data.The results of this research show that the used method and data collections do not provide accurate information for the intended classification goal. This research demonstrates in several ways the complexity of supervised image classification in complex agricultural landscapes: the unbalanced and variable reference data of different land uses, which often consist of only a few satellite pixels, make it difficult to identify characteristics of land classes, from which the classifier can derive information. In which Sentinel-1 as added and used in this research, offers no additional insights. Therefore, in order to improve the identification of farmer-led irrigated agriculture in Manica, other technologies for smart agriculture can be explored in addition to deploying satellite data, such as citizen science. Further research is recommended on the field of using S-1 and S-2 data for classification of complex agricultural landscapes. This may include more advanced methods of performing image classification and accuracy assessement with imbalance datasets, such as: the use of a weighted confusion matrix for accuracy assessment or exploring the use of spatial-spectral instead of pixelwise random forest algorithms. These algorithms seem to be better at handling spatial dependencies and intrinstic heterogeneity which is characteristic of these complex agricultural landscapes. Lastly, it is strongly recommended to assess the use of speckle filters when using SAR data for small target objects. These filters make use of a buffer zone, consisting of a few pixels in size and about the same size of the target object. The main challenge with be to balance the need of speckle reduction and class specific information preservation.

Many conceptual hydrological models do not include any information on plant phenology, which represents the dynamic behavior of vegetation, to partition evaporation in a catchment. Such models often use potential evaporation to quantify evaporation from the interception reservoir and what remains afterward to quantify transpiration, or the other way around depending on the modeler’s decisions. Such defined orders are the opposite of what is perceived in reality. Vegetation is alive, continuously developing, and impacts the partitioning of evaporation through its ability to transpire. Hence, vegetation is able to constantly change this defined order and thus the ratio of transpiration over interception evaporation over time.

This study aimed to include information from plant phenology, representing the dynamic behavior of vegetation, to partition evaporation in a catchment in conceptual hydrological models. This study hypothesized that the conceptual hydrological models that do include this information provide more reliable hydrological responses. To test this hypothesis, this study used two conceptual hydrological models, which are the lumped FLEX and GR4J models, to simulate daily values streamflow and evaporation for two distinct catchments in the United States.

To include information from plant phenology, this study applied three separate modifications to the conventional structures of the two conceptual hydrological models. The first modified structure uses a part of the Jarvis model, in which stomata respond to temperature, to partition evaporation. The second modified structure uses a method similar to the crop evaporation method of the Food and Agriculture Organization. Lastly, the third modified structure uses a combination of the other two structures.

This study concludes that conceptual hydrological models that include information from plant phenology, as in the three modified model structures, are able to provide streamflow simulations as good as models that do not include plant phenology. The streamflow simulations of conceptual hydrological models, such as the three modified model structures, are therefore not necessarily more reliable. However, conceptual hydrological models that do include information from plant phenology might still benefit hydrological studies that take into account the effects of long and short-term changes to the vegetation in a catchment, such as the effects of climate change, land-use change, and forest fires. Before plant phenology should be used in the models of such studies, challenges such as the uncertainties in the simulations of the evaporation components need to be addressed.

Accurate estimates of terrestrial water storage variations (TWSV) are critical for a variety of applications, e.g., model calibration and climate studies. This study aims to find the added value of river run-off data for regularizing GRACE mascon solution, from which TWSV can be estimated. Most subbasins of the Mississippi Basin show an exponential relationship between run-off and TWSV with moderate to good predictive value. The mean value of the explained variance (R^2) for the models is 0.4, when excluding the Lower Mississippi and Middle Missouri subbasins the mean value increases to 0.5. The Lower Mississippi and Middle Missouri subbasins are the only subbasins with no clear correlation between run-off and storage, most likely due to aquifer depletion.

Since ready-to-used GRACE mascon solutions cannot be modified, the mascon method developed at the Geoscience and Remote Sensing Department of the Delft University of Technology is used in this study. This study proposes modifications to tailor this method to the Mississippi Basin and mass anomalies of hydrological origin. After applying standard Tikhonov regularization, the exponential relationship is used to determine the most probable model outcome for each subbasin, adding a new term to the estimator. The resulting GRACE mascon solutions (Tikhonov and run-off regularized solutions) have a root-mean-square deviation (RMSD) of around 3.5 cm when compared to ready-to-use mascon solutions from the Jet Propulsion Laboratory and Goddard Space Flight Center. The run-off regularized solutions show smaller deviation (on average 7 %) than the Tikhonov regularized solutions with respect to an independent validation data set. The largest difference between the Tikhonov and run-off regularized solutions occur in the dryer subbasins (Platte, Middle, and Upper Missouri, and Upper Mississippi subbasin). These areas show a stronger inter-annual trend in TWSV, these trends are captured better by the run-off regularized solutions. Additional research, co-estimating the regularization bias, shows that the run-off regularized solutions induce less bias into the solutions. This study shows that using run-off data when processing GRACE data could be of added value, especially in semi-humid to arid areas, since the TWSV in these areas is more susceptible to inter-annual variations.

The validation of satellite retrieved soil moisture over a tropical region (Bago, Myanmar) has been conducted in this research. The downscaled soil moisture products of VanderSat using the Land Parameter Retrieval Model, were compared with an in-situ soil moisture network that operated between 2017 and 2018 and a newly placed network installed during a fieldwork trip early 2020. Soil moisture is a key variable in the hydro- logical cycle, but the understanding of soil moisture content in tropical areas like Myanmar is substandard. This study provides new insights, being one of the first of its kind in the tropics. The reliability of both in-situ soil moisture networks was analysed and indicated that the 2017 network could be used for the validation, and the 2020 network for a visual comparison, with a focus on the soil temperature as a source of error. The results of the validation showed high correlation with Pearson’s R ranging between 0.70 and 0.89. The highest correlations were found for the L-band, between 0.78 and 0.89 and for the location Hpayargyi. The root mean square difference (RMSD), unbiased RMSD and bias indicated that the accuracy and precision need improve- ment. For the ubRMSD values between 0.059 m3/m3 and 0.129 m3/m3 are found. The best result is found for the L-band at Hpayargyi, which is the only location/band combination that meets the target accuracy of 0.06 m3/m3 that is often set for SMAP satellite retrieval. The visual comparison of the 2020 network showed good agreement for the soil moisture measurement. When the in-situ soil temperature measurements were compared with the satellite retrieved land surface temperature measurement, however, this resulted in a large bias of approximately 15 degrees Celsius. To evaluate possible improvements for the validation result, several sources of errors were explored. Two sources concerning the in-situ network indicated the challenges of an in-situ network in a remote tropical area. The first showed that re-calibration of the sensor is necessary to improve representation of the ground truth. The second discussed the spatial mismatch between the point scale in-situ measurement and the downscaled satellite products, with a pixel size of 100x100m. The sparse in-situ network with four stations cannot represent the entire area and can therefore result in errors. The other sources of errors that were discussed originated from the retrieval algorithm of the Land Parameter Re- trieval Model. A parameterisation was performed to assess the sensitivity of the ubRMSD to the polarisation mixing factor (Q), the single scattering albedo (!) and the roughness parameters (h1 and h2). The parame- ters were optimised to achieve the lowest ubRMSD, the result indicated that an improvement of the ubRMSD up to 40% can be achieved. However the evidence that linked the optimised parameters to the tropics was not found and it was concluded that the optimising could be covering the large temperature bias, that is a direct input in the Land Parameter Retrieval Model. The large bias in the temperature was evaluated as a source of error. The bias can partially be explained by the masking of water bodies, where the dynamic ef- fect of the tropical monsoon on the water bodies is not taken into account. Another explanation for the bias could potentially be found in the emissivity that is determined by optimising the ¢T , which can cause im- perfections in the prediction of the temperature. It was concluded that the satellite retrieved soil moisture products show good correlation but need improvement in accuracy and precision. This can be accomplished by updating the in-situ soil moisture network, creating new water masks, solving the retrieval algorithm for the temperature bias and then optimising tropical specific location parameters.

Agriculture is an important source of income for many countries in the Global South, where it may account for as much as 25% of GDP. Precipitation is crucial for agriculture in countries like Ghana, where ~95% of farming is rainfed. Accurate rainfall observations are limited in Ghana. The sparse rain gauge network and the lack of weather radars make remote sensing methods a potentially attractive alternative source of rainfall data. Radar satellites, such as Sentinel-1, emit radiation that passes through the atmosphere and is scattered back to the satellite by the Earth’s surface. The backscatter measured by the satellite is correlated with the wetness of the soil but the existence of vegetation hinders straightforward quantification of soil moisture. By choosing sites with a simple and, more or less, constant phenology, it may be possible to eliminate the effect of vegetation on backscatter. Soccer field may qualify as sites with such a simple and constant phenology. The main objective of this study is to use the Sentinel-1 data over soccer fields and assess them as rainfall detectors. A machine learning approach will be used to reach this objective. This research assessed the stability and the generalization capabilities of a classification model (rain/no rain). The model was trained with and applied to different locations and periods (2019 & 2020). Ground observations from 53 Ghanaian (TAHMO) and 1 Greek stations were used. Soccer fields in Ghana and Greece were selected and their suitability as rainfall detectors was checked based on the correlation between modeled soil moisture and backscatter strength. The rain/no rain classification of the soccer fields was made with a stacked classifier that was trained and validated with both spaceborne and ground data. The classifier was tested on six different datasets from Greece and Ghana 2019 and 2020. The stability of the model was assessed by a Leave-p out cross-validation approach. The generalization in space was tested by using different environments. The generalization in time was tested by using different time periods. The results showed that the classification was stable. The minimum and maximum performances for the different testing datasets were 0.43 to 0.85. The median performance of the algorithm in Ghana for 2020 is 67%. The stacked classifier was found to have the best performance compared to other classifiers. Finally, the performance of the stacked classifier was competitive in comparison with the performance of the well-known IMERG algorithm. The study showed that there is a potential for using radar backscatter from suitable fields to detect rainfall. The classifier is stable and can be generalized in time and space under certain conditions.

Human interventions in the White Volta River in Ghana increase the flood risk in living spaces of the local communities, causing great challenges in the area. Burkina Faso has built a hydropower dam near the border with Ghana which has a large impact on the discharge extremes of the White Volta River. As a result of poor water management of this dam, water is spilled during the wet season, causing more floods in northern Ghana. Several communities, living near the White Volta River, depend on the river water for their drinking water demand and use the sand in the river bed as construction material. This puts them in a vulnerable position when changes in the natural flow regime of the river occur.

This thesis studies the impact of constructing multiple small dams in the White Volta river in order to decrease the flood risk in villages located close to the river and create an opportunity for controlled sand mining as a result of sedimentation.

Tamale is, with 672.000 inhabitants, the largest city in this area and struggles with the consequences of the floods as well. The drinking water company has trouble meeting the drinking water demand during both the wet and the dry season. When the area is flooded, the water intake point shuts down, and during low water levels, the pumps can get clogged. This clogging occurs even more often as a result of increased illegal sand mining from the river banks by the local communities. This has broadened the river, resulting in lower water levels and a higher turbidity.

A possible solution to ensure the water intake in Tamale and reduce the flood risk in northern Ghana is the construction of multiple small dams in the river bed to flatten the discharge peaks and slow down the water in the White Volta River. At the same time, these dams can create an additional advantage by causing upstream sedimentation that provides a possibility for controlled sand mining in the river bed. To research the effectiveness of this solution, the impact of the dams on flood risk and sedimentation, near seven villages in northern Ghana, was modelled.

A base case of the flood of 2003 was compared to a scenario in which seven dams were implemented at locations close to villages that are often exposed to floods. This was done by building a hydromorphological model of the White Volta River, using the D-HYDRO 1D2D software.
The results of the morphological simulation show that in total, the sedimentation upstream of the dams can fill 81 trucks with sand per day. However, extracting the sediment at the most upstream dams decreases the possibilities for sedimentation near the downstream dams. The results of the 1D2D hydraulic model show that the proposed dams are not able to reduce the flood extent in villages and can even increase inundation depths, resulting in more flood damage.

It is recommended to increase the 2D grid of the hydraulic model in order to receive more realistic backwater curves.

Subsidence can be observed at various locations in the Netherlands. While it can be due to both shallow and deep subsurface processes, the shallow subsidence is mainly a result from compaction, oxidation and/or groundwater drainage. Moreover, for grasslands on organic soils, in particular pastures on drained peat soils, the vertical position of the ground level is subject to significant temporal variability. Temporal scales of the vertical motion are expected to vary between days and centuries, and spatial scales between millimeters and kilometers. Unfortunately, performing precise, reliable, and representative geodetic measurements of shallow subsidence processes is very difficult for soils, as fixed benchmarks do not exist. Here we propose an insitu measurement procedure for grass-covered soils using laser scanning, both terrestrial and airborne, using a vertically fixed reference platform. We show that for terrestrial laser scanning (TLS) it is possible to detect changes in the vertical position of soils with a systematic error of up to 2.6 mm and a standard deviation of 0.4 mm. Given a predefined level of significance of 훼 =0.05 (confidence level of 95%) and a detectability power of 훾 =90%, we achieve a minimal detectable vertical deformation (MDD) of 21.1 mm in our study area. We show that the results are influenced by the grass density and length, the incidence angle of the laser beam, as well as other settings of the laser scanner. We find that the parameter settings of the method for estimating soil surface, and subsequently the subsidence, has an influence on the results and related statistics. For airborne laser scanning (ALS), using a precisely leveled reference platform, we find that the quality of elevation estimates is still limited, requiring further considerations on the design of data acquisition surveys and the reference platforms. Our results demonstrate the ability of laser scanning technology for investigating shallow subsurface motion of grass-covered soils relative to a benchmark on a local scale. Based on the quality assessment, the detection of vertical ground level change is better understood in terms of time and probability. In future research, the factors affecting terrestrial laser scanning technology to accurately identify soils affected by vegetation, environment, and device condition, should be further studied.

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.

Europe is a continent with diverse climatic conditions. The dominant climates are the Oceanic, the Mediterranean and the Continental ones. The western part of Europe has an oceanic climate, southern Europe has a Mediterranean climate and eastern Europe has a continental climate. Because of such heterogeneities, a vast range of extreme climatic events might occur in different areas. We define extreme climatic events the droughts, floods and heavy snowfall. Those events will be generically referred to in this research as anomalies. The purpose of this study is the identification of these extreme climatic events in the area of Europe, with the use of Special Sensor Microwave Imager (SSM/I) data at 37GHz frequency. The data that are used are Brightness Temperature (TB) values. The detection of the events will be achieved with the Polarization Difference Brightness Temperature (PDBT). The PDBT values can be related to changes to surface wetness and the surface geometry. It could be used as an indicator of an anomaly, because the higher the values of PDBT the higher the surface wetness. The methodological steps of the work consist in a statistical analysis of the SSM/I time-series, in the design of a detection algorithm of the anomalies under investigation and on the debate of its performance. The analysis of the temporally long SSM/I data will provide a first understanding of the data sensitivity to events under investigation and of their distribution for the statistical modelling of the Normalized Polarization Difference Brightness Temperature (NPDBT) indicator. The calculation of the NPDBT exploits the same principles as the well-known z-score index. The detection of the anomalies will be then achieved through thresholding the NPDBT index. Further information for the detection of anomalies is provided by the soil moisture time series from the Soil Moisture Active Passive (SMAP) sensor and the precipitation data from the Global Satellite Mapping of Precipitation (GSMAP). The soil moisture data appear to be more useful for the dry events, whereas the precipitation data for the flooding and the heavy snowfall events.

Snow in mountainous areas is of great importance for the water supply in many catchments. To get data on snow cover, ground station data is not enough and, in many catchments, not available. Therefore, satellite data is used to measure snow cover. In this thesis the MODIS daily snow cover dataset (MOD10A1) is used. These images are obstructed by clouds. In order to create a complete dataset, the Regional Snowline Elevation Method is used which uses the elevation of the snowline to interpolate over the missing data. This method is accurate but is computationally demanding. Using Google Earth Engine, it is attempted to improve the method. The method developed combines grid cells with daily images and computes the RSLE for each cell, for each day. The results are exported to a CSV file, reducing the downloaded data from 150GB to 41.6 MB for the 0.50° resolution and 168MB for the 0.25° resolution. The computation time however was not improved with this method. The developed code was used on the Caucasus area. After the data from Google Earth Engine was downloaded, trends on yearly snow cover duration were computed using the Mann-Kendall test. It followed that 17% of the trends in both resolutions were significant and, except for one location, were all decreasing trends. The decreasing trends show a decline of snow cover duration of 1-4 days per year. Looking at regional differences it becomes clear the greatest number of trends can be found in the south-west. The Google Earth Engine code was able to compute the required data however, it took a long time doing so. Therefore, a more sophisticated code has been developed, making used of the ability to reduce the resolution of an image, and computing the value of pixels at the same time. This code runs quicker, but is at this moment unusable, due to problems with the thresholds and export. Being unable to export an image collection is one of the shortcomings of Google Earth Engine. Others include: download tasks that run out after twelve days without raising an error when starting the task, limits to the amounts of bands used, and sensitivity of the computation time to busyness on servers. Things that need to be improved and need further research are: the filters applied to smoothen the dataset, the use of other trend analyses, the effect of snow cover trends on the area, and the resolution in combination with the cloud threshold value. The analyses took six hours to run, which can be improved by using function-based programming, instead of process-based.
In the end the goal to develop a more efficient method has partly been met by decreasing the amount of downloaded data significantly, even though the running time has not been improved. Using the data from the developed method, decreasing trends were found in snow cover duration over the entire Caucasus but mainly in the South-West, which can greatly influence the water supply to a large part of the Caucasus and its surrounding areas.

MOSES DSS web-platform aims to assist stakeholders such as governments and farmers in order to manage water irrigation distribution in a higher efficiency and sustainability. The constructed algorithms are focused on forecast using weather models, data, as well as satellite multispectral observations in such a way that a 7-day ahead crop water requirement estimation is generated. The current drawback of the system in using the available and free satellite products such as Landsat 8 and Sentinel 2, is that it assumes that the crops are under standard conditions, e.g. there is no water stress, diseases etc.. The current work investigates how possible errors due to this assumption can be potentially tackled in the future by comparing Crop Water Requirement (CWR) with S2REP VI and/or with water stress index and see the discriminative power of the latter. Furthermore, a comparison between several discrepancies between S2 and L8 (e.g. AC and co-registration) are studied since it is a crucial issue especially in temporal applications such as MOSES. On the one hand, the results showed that a harmonization of the two products is certainly needed. On the other hand, it seems that S2REP is capable of revealing crop stress information based on the methodologies of this work, thus it could potentially give more information compared to NDVI which is not sensitive to crop stress.