Personal weather station (PWS) networks have the potential to supply precipitation data at high spatial and temporal resolution for urban hydrological modeling. Past research has shown promising results on the quality of PWS data, for example from Netatmo gauges, but studies on other PWS brands are limited. This thesis assesses the quality of precipitation measurements from the Alecto WS-5500 personal weather station. During a controlled experimental setup in an urban environment, the Alecto was found to overestimate rainfall due to incomplete emptying of the tipping bucket. Correcting this mechanical error by a 10 percent reduction factor lowered the relative bias to 0.00 or 0.06, when comparing the station to official KNMI gauge or KNMI gauge-adjusted radar, respectively. Correlations were high between stations with non-faulty setups, but at the 5 minute resolution, correlations were substantially lowered by sampling errors caused during the data transfer to PWS data platforms. A quality control method from de Vos et al. (2019) was adapted and applied to data from a citizen science project in Delft, the Netherlands, which had a 12-month period of measurements for 20 stations, and a 3-month period of measurements from 40 stations. The filtering of faulty zero measurements was improved by applying the filter on hourly accumulations, and the bias correction was stabilized. The variation over individual PWSs, however, remained high due to setup differences. The complex installation process for citizens and issues with software and data accessibility are limiting factors and warrant further research to improve the usability of PWS data for urban hydrological applications.

Due to rising temperatures, rainfall patterns around the world are being affected, causing extreme precipitation events to become more frequent and intense, resulting in an increased probability of severe flash floods. Thailand is no exception to the increased risk of these hazards, which is why Early Warning Systems are being set up. Since flash floods occur within a few hours after the triggering precipitation event, timely and accurate precipitation observations are critical, to enable timely warning and evacuation of inhabitants to mitigate the risk. One solution for this is the use of rain radar, which provides rain data with high spatial and temporal resolutions.

An analysis of this technique was chosen to form the basis of this research in Northeastern Thailand's Lam Takhlong basin. The objective was to investigate the importance of using distributed precipitation data. In this research, the hydrological response of the catchment of interest was studied. The ability of the physically-based, conceptual and distributed CALEROS model to capture this response was assessed. Different modelling strategies and sources of precipitation input were analysed, and the additional value of using distributed precipitation data was determined. The examination of multiple hydrological characteristics in the study area shows great heterogeneity of the catchments response to precipitation. A clear differentiation can be made between the hydrological response of the upstream and downstream part of the catchment. Additionally, the results from the catchment characterisation indicate great spatial variability of the precipitation patterns in the study area. This is confirmed when using the CALEROS model to recreate the catchments hydrology.

Four modelling strategies were used, varying by spatial and temporal constancy. Calibration performed using uniform precipitation showed to be incapable of capturing the discharge trends in the catchment, failing to properly catch the observed peaks as well as the base flow. Runs performed using distributed precipitation maps obtained by Inverse Distance interpolation of rain gauge data showed significantly better results, adequately capturing the trend of the discharge observed in the catchment. Differences in parameterisation only had limited effect on the outcome, making the precipitation input the most important parameter. Runs using synthetic precipitation data demonstrated the importance of properly capturing the precipitation pattern and movement across the catchment, as it greatly influences the timing of occurring discharge peaks.

A comparison between rain gauge data and rain radar data shows that, although data quality of the rain gauges seems acceptable, rain gauges are not capable of properly capturing rainfall patterns, while rain radar does. Precipitation patterns were found to be the most crucial parameter for modelling of flash floods in the area of interest, signifying the importance of using rain radar data for accurate flash flood forecasting.

Experiments using fiber-optical set-ups for distributed temperature sensing (DTS) were conducted in the LIAISE (Land surface Interactions with the Atmosphere over the Iberian Semi-arid Environment) field campaign during 15-30 July 2021 in the north-east of Spain. Three DTS set-ups were installed to measure temperature profiles along varying vertical scales; 1.6 - 40 m in the atmosphere, 0 - 1 m into the rapidly-growing alfalfa canopy and -0.5 - 0 m in the soil. Measurements were conducted at 5 s and 25.4 cm resolutions using a 1.6 mm Kevlar-reinforced fiber. The preliminary data of these three set-ups are described in the first part of this thesis, which display the potential of using DTS in a land surface campaign to capture vertical temperature structure in great detail.
A fourth fiber-optic set-up was installed with a horizontal extent of 70 m, measuring at four heights between 0.40 m and 2.05 m height. A thinner 0.5 mm cable was used here in an effort to obtain the fastest possible time response in order to measure temperature turbulence parameters using DTS. Measurements were made at 1 Hz and 12.7 cm resolution, however the actual sampling frequency appeared to be 0.15 Hz in the temperature spectrum, likely because of the long response time of the cable.
Despite the limited 0.15 Hz sampling rate it was possible to obtain turbulence information through the use of the structure parameter of temperature, C2T
. This parameter indicates the intensity of temperature fluctuations and was calculated over time, as is conventional. In a novel approach, it was also calculated over space, using the spatio-temporal dataset as obtained by DTS. Both the definition of C2T and the inertial range of the temperature spectrum were used to determine C2T. The spatial C2T obtained through
the definition method was found to have the best correlation with a sonic anemometer reference, with an R2 of 0.88. The temporal C2T lack the structure that is shown in the spatial C2T, which is likely due to 30-min averaged data for horizontal wind speed from the sonic anemometer or to Taylor’s frozen turbulence hypothesis not being a suitable assumption within the dimensions of this research. Determining C2T through the turbulent spectrum was successful for limited data points for the time series, and is currently inconclusive for the spatial series.
Recommendations for further research for using DTS in turbulence analysis are to investigate the effect of instrument noise and the limited sampling rate. Also a critical look into the current DTS calibration routines for atmospheric is recommended. This work provides a first step towards using DTS in capturing
turbulent information along spatial temperature series.

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

Torrential rain from tropical cyclones can have a devastating impact, causing loss of life and billions in damages. To better understand the risk faced by coastal communities, it is important to estimate how often a tropical cyclone could occur and how much rainfall it will produce. One way to do this is by analyzing past storms and building parametric models of rainfall rates during tropical cyclone events. While many parametric precipitation models –such as the Bader model– exist, their accuracy remains limited and many challenges still need to be overcome. The most important challenges are output overestimation and a poor representation of rainfall over land. Therefore, this thesis aims to reduce
these biases by answering the following research question:

"How can the bias in the radial rainfall distributions of a tropical cyclone in Bader’s
parametrized model be reduced and be used for reliable rainfall estimates both above land and the ocean?"

To answer this question, several new data sources were introduced from the TRMM/GPM satellites and Stage IV to improve the Bader model. While this original model only predicted precipitation based on maximum wind speed (vmax), the updated model also considers pressure deficit ΔP. The results suggest that ΔP can be a useful parameter to reduce bias and improve accuracy. However, it also
leads to larger uncertainty ranges. Next, four precipitation profiles were proposed. A profile where precipitation is constant for low maximum precipitation values based on the predicted total rainfall (area under the graph) was selected for further exploration.

The new models are explored during a case study of Hurricane Florence. Both the ΔP and vmax based models produced satisfactory results, compared to the benchmark IPET model. Moreover, an alternative fit above land has been proposed, where the highest precipitation is simulated at the eye. The proposed land fit improved the median of the predictions based on both vmax and ΔP. The ΔP
based model performed the best in the case study, however, no definitive conclusion could be reached upon which model is most suitable overall as more case studies would be required.

Finally the updated model has been compared to the original Bader model. The new data ensured better representation over land, the overestimation of precipitation was reduced, and the model was applied with more confidence outside of the training data set. Consequently, results showed an improvement
on its prediction capabilities. As a concluding remark, this research project highlights the importance of having insightful data to enhance the decision-making and risk management of natural hazards: a model that accurately quantifies uncertainty and the risks associated with a TC, representing a valuable tool for better understanding flood risk. Nonetheless, there are still several ways to further improve the modeled profiles (e.g., by including more data, introducing asymmetry or adding temporal autocorrelation).

Protecting forests from agricultural expansion and wildfires while the world population is growing and the climate is warming remains one of the biggest challenges humanity currently faces. While global modelling and regional observation based studies have found significant effects from deforestation on precipitation, leading mostly to drying precipitation trends and shorting rainy seasons, this study represents the first global estimate of first order deforestation effects on precipitation. Using a recently developed precipitationshed database and actual deforestation data, a new measure is developed to quantify potential deforestation impact per grid cell which in turn is related to annual precipitation trends as well as seasonal differences in tropical regions. In seven regions analysed, a majority of subregions suggested a relationship between deforestation impact and a relative drying precipitation trend in the 2001-2018 study period compared to the long term average. While these results provide further evidence of deforestation contributing to a downwind drying precipitation trend across different continents and climate regions, five other regions studied showed no significant relation or suggest relative wetting related to deforestation impact. One of this regions is the South America Tropical (SAT) region, the region most well-known for its widespread and intense Amazonian deforestation. The two regions downwind of the SAT region however are highly impacted by SAT deforestation and experience most relative drying in the areas impacted most impacted by deforestation, suggesting strong teleconnecting effects. In the seasonal analysis, only two out of four tropical regions studied show more subregions linking deforestation impact to relative drying in the first wet month compared to the wettest month. While these results provide new insights into the global influence deforestation can have on moisture availability, more research needs to be done into the indirect and feedback effects related to deforestation. Additionally, a more robust way of including other factors influencing precipitation trends like large scale oscillations could further enhance the understanding of this important issue.

The assessment of potential evaporation or reference combined evaporation and transpiration is among the most important components for many hydro-climatic applications such as irrigation design and management, water balance assessment studies, and assessment of aridity classification indices. Aridity classification indices such as UNEP, Thornthwaite and others are usually employed at large scale applications and require respective estimations of potential or reference combined evaporation and transpiration. The major problem in such applications is not only the limited availability of stations per se but also the limitation of many stations to provide data for a complete set of parameters (i.e., precipitation, temperature, solar radiation, wind speed, humidity). A complete set of climate parameters is prerequisite for accurate estimations of potential or reference combined evaporation and transpiration using the most advanced methods, which are expressions of energy balance (e.g., ASCE-standardized method, successor method of Penman-Monteith FAO-56). Unfortunately, large scale applications of aridity indices suffer from this limitation and the common solution is to use temperature-based formulas. The most popular and historical temperature-based formula is the one of Thornthwaite, which was developed to support the respective aridity classification index. The popularity of this formula is based on the minimum requirement of mean monthly temperature and latitude at the location of interest. Considering the above, this study aims to develop a global database of local correction factors for the original Thornthwaite formula that will better support all hydro-climatic applications but mostly to support large scale applications of aridity indices, which are highly prone to data limitations. The hypothesis that is tested in this work is that a local correction factor that integrates the local mean effect of wind speed, humidity and solar radiation can improve the performance of the original Thornthwaite formula and to convert it at the same time to a formula of reference combined evaporation and transpiration for short reference crop. The global database of local correction factors was developed using gridded climate data of the period 1950-2000 at 30 arc-sec resolution (~1 km at the equator) from freely available climate geodatabases. The correction factors were produced as partial weighted averages of monthly ratios between the benchmark ASCE-standardized method for short reference crop versus the original formula of Thornthwaite by giving more weight to the warmer months and by excluding colder months of Epr<45 mm month-1 where monthly ratios are highly unstable with unrealistic values. The validation of the correction factors was made using raw data from 525 stations of Europe, California-USA and Australia that cover periods mostly after 2000 and up to 2020. The validation procedure showed significant improvement in the estimations of reference combined evaporation and transpiration using the corrected Thornthwaite formula that led to a 19.4% reduction of RMSE for monthly and a 55% reduction of RMSE for annual estimations compared to the original formula. The variation of the correction factor was also investigated in different major Köppen climate classes and it was found that tends to increase in drier and warmer territories. The five major Köppen groups were ordered as follows B > C > A > D > E considering the magnitude of the correction factors values. The corrected and original Thornthwaite formulas were also evaluated by their use in UNEP and Thornthwaite aridity indices using as a benchmark the respective indices estimated by the ASCE-standardized method. The analysis was made using the validation data of the stations and the results showed that the corrected Thornthwaite formula increased by 18.3% the accuracy of detecting identical aridity classes with ASCE-standardized method for the case of UNEP classification, and by 10.4% for the case of Thornthwaite classification in comparison to the original formula. The performance of the corrected formula was extremely improved especially in the case of non-humid classes of both aridity indices. The overall results showed that the correction factors produced in this study can improve the performance of the original Thornthwaite formula providing better estimations of the aridity classification indices.

Rainfall-runoff modelling is essential for short- and long-term decision-making in the water management sector. The accuracy of streamflow predictions of hydrologic models increases with the availability of and the access to streamflow observations. Therefore, one of the key challenges in the field of hydrology is to produce Predictions in Ungauged Basins (PUB), where observations are lacking. Recent research has shown the potential of deep learning neural networks as an alternative approach to conceptual and process-based hydrologic models for this purpose. 
In this study, the existing Multi-Timescale LSTM (MTS-LSTM) architecture is used to investigate if such a deep learning network is able to learn universal hydrologic behaviour. Therefore, a MTS-LSTM is trained on a large variety of >500 US catchments and subsequently tested outside the US, in the European Meuse river basin. The model is not retrained or finetuned to simulate an ungauged situation and to asses whether the streamflow predictions can compete with those from the uncalibrated distributed model wflow_sbm. 
Results indicate that the MTS-LSTM trained on US data cannot compete with wflow_sbm in the Meuse catchments. The simulated streamflow time series can be unrealistically shifted and scaled compared to the time series of observed streamflow due to sensitivity regarding static model input. This means, the values for catchment characteristics cannot be extremer in the testing data than in the training data. Therefore, it is recommended to select catchments for the model training such that the most extreme conditions are covered. In the case that the MTS-LSTM is trained for a specific region, results clearly compete with or outperform the distributed model. For the Meuse test catchments, the neural network achieves Nash-Sutcliffe Efficiency (NSE) values>0.46 where the application of wflow_sbm is problematic and yields negative NSE values. To exploit the potential of an LSTM, the model should be trained on all available data of the entire Meuse basin instead of on the subset of catchments used here. For some water management applications it is important to accurately predict high flow events. Training the MTS-LSTM with the Mean Quadrupled Error (M4SE) loss function showed that the peak flow representation can improve for catchments where the use of a NSE loss already leads to good predicting performance. Thus – for and ungauged catchments – an implementation of a combined loss function appears a valuable follow-up research. 
For ungauged catchments, these results imply that the global neural network model as tested in this study should be supplemented with finetuning. Thereby, the global model could not yet be applied everywhere, however, in regions where only few years of streamflow records are available. Alternatively, a regional neural network model trained on nearby catchments could be applied for PUB, if streamflow observations are accessible in the surrounding area. Finally, it is of high importance to maintain and extend the network of streamflow gauging stations globally, and to ensure easy access to the data.

Radar rainfall nowcasting stands for the prediction of rainfall amounts and intensities over the next 6 hours by means of statistical extrapolation of radar measurements. It is the principal ingredient for modern flood forecasting and early warning systems. Radar forecasts are generated by identifying and tracking rainfall cells in radar images and extrapolating them along the main direction of motion based on the assumption of Lagrangian persistence. The latter means that the rainfall cells do not physically evolve over time. Therefore the key to improving predictions is to anticipate the growth and decay of rainfall cells during the next few hours. Studies have shown that estimating growth and decay based on past radar images is nearly impossible. Therefore, in this thesis, a new approach for predicting growth and decay based on physical guidance from a numerical weather prediction model is investigated. The idea is that numerical weather prediction models are better at anticipating changes in the atmosphere than radar and thus, should also contain information about the growth and decay of rain cells. To test this hypothesis, two simple machine learning models have been trained to learn the relationship between the observed growth and decay in radar images and the output parameters from the Dutch mesoscale numerical weather prediction model HARMONIE. The models were trained on data in the summer of 2019 around the city of Rotterdam for two area sizes, $20\times 20$ km$^2$ and $60\times 60$ km$^2$. The first model is static and assumes that all predictions from HARMONIE are correct. Unfortunately, this turns out to be too optimistic as HARMONIE often places the rain cells at the wrong locations with the wrong intensities and trends. As a result, no meaningful relationship between HARMONIE outputs and growth/decay in radar could be learned. The dynamic model overcomes this issue with the help of an additional classifier. The classifier predicts whether the information from HARMONIE can be trusted or not. Then, a regression model predicts the magnitude of growth and decay (only for the trusted cases). Out of 308 analysed cases, the classifier labelled 237 HARMONIE predictions as being untrustworthy, thereby removing a lot of bad cases. However, 98 of these were false negatives, meaning that they could have been used to predict growth and decay. On the other hand, only 71 cases were labelled as containing useful information, from which 27 cases were false positives. Despite these errors, the dynamic model was able to improve the root mean square error on the predicted growth/decay by $27$\% for these 71 cases. Also, the correlation between the predicted and actual growth/decay increased from near zero to $0.488$. This is encouraging and clearly highlights the potential of this approach. Still, some important challenges remain. In particular, results show that the dynamic model often underestimates the magnitude of growth and decay. Also, when the entire validation data set is considered, including all unusable cases, the improvement with respect to the static model is only $6$\%. This is mainly caused by the poor performance of HARMONIE and the large number of unusable cases. However, performance could be improved further by building a better classifier capable of identifying all good cases with a low number of false negatives.

The current intensification of the hydrologic cycle, in combination with expanding settlements in flood prone areas, makes an increasing share of the global population exposed to flood risks. Many parts of the world are, however, still lacking the data needed for flood risk management and risk reduction. The recent development of information and communication technologies has remarkably lowered the costs to collect data for flood resilience, which has accommodated the rise of community mapping projects to fill data gaps in resource-strained environments. This thesis utilises drainage data collected by the Ramani Huria community mapping project in Dar es Salaam, Tanzania, to investigate if community mapped drainage data can improve flood predictions on neighbourhood scale. A coupled 1D-2D hydrodynamic model is developed of Kijitonyama ward, and is run with and without Ramani Huria’s drainage data implemented in the 1D schematisation. The simulated flood depth for the scenarios is validated with citizen’s observations on flood depth during a rain event on 3 March, 2019. The developed model is then applied to investigate the impact of solid waste accumulation in the drainage system on floods, by closing the drainage segments that were recorded as blocked in Kijitonyama ward by Ramani Huria staff the morning after the simulated event. An experimental scenario is also run, to examine the impact of blocked culverts. The results show that community mapped drainage data indeed can enhance the performance of hydrodynamic models, as the model output corresponds better with the validation data when implementing Ramani Huria’s drainage data in the 1D schematisation, compared with a scenario run with only a 2D schematisation. The scenarios run with solid waste blockages do not influence the model output when comparing with citizen’s observations, but increase the water level in the drainage segments located upstream of the blockages.

Sub-Saharan Africa is one of the most vulnerable regions in the world to climate change. This is largely driven by the dependence on rain fed agriculture for food production. At the same time African climate observational networks have been in decline since the 1990s. A new kind of rainfall sensor (the intervalometer), which counts the arrival of drops at a piezo electric element, is tested during the Tanzanian monsoon season alongside tipping buckets and an impact disdrometer. Rainfall rates are derived from rainfall arrival rates using Marshall and Palmer’s (1948) exponential parameterisation. This parameterisation is defined independently of a notion of scale and therefore implicitly assumes that rainfall is a homogeneous Poisson process. Testing of the Poisson assumption shows that 22.5% of the total drops observed can reasonably be considered Poisson and that the main reason for Poisson deviations are non compliance with the stationarity criterion (36.7%) and the presence of correlations between drop counts (14.3%), particularly at higher arrival rates (ρa > 500 [m−2.s−1]). The total rainfall amount [mm] calculated from intervalometer measurements overestimates the tipping bucket value by a factor of approximately three. The overestimate is most likely due to poor calibration of the minimum detectable drop size (Dmin). A correction is applied to constrain the overestimates of mean drop size by the intervalometer parameterisation to the observed disdrometer measurements. The correction results in an improvement in the estimate of the total rainfall amount to within 10% of tipping bucket measurements. The total rainfall amount [mm] calculated from disdrometer rainfall arrival rates is within 5% of co-located tipping bucket measurements. The form of the mean drop size relation with arrival rate appears stable in time and space. The intervalometer shows good potential for use as a rainfall measurement instrument or to derive estimates of mean drop sizes.

Repeat-pass acquisitions with coherent Synthetic Aperture Radar (SAR) systems, preserving both phase and amplitude, are more readily available than ever (Bruzzone, 2016). Phase measurements from SAR systems have seen widespread use in the Interferometric Synthetic Aperture (InSAR) technique to measure deformations and elevations since the late 1980’s (Hanssen, 2002). Since the late 1990’s an increase in radar-based change detection is observed, mainly relying on amplitude measurements (Ajadi et al., 2016; Dekker, 1998). The unpredictive multiplicative noise-like speckle, inherent to coherent SAR,makes change detection in SAR imagery difficult (Bamler, 2015). However, the advantages in the all-weather mapping capabilities and object penetrating properties of SAR make it a suitable remote sensing technique for certain applications, such as natural disaster damage assessment (Bruzzone and Prieto, 2000). Broadly speaking, change detection in SAR-based images usually consists of applying an operator on two spatially filtered SAR images to create a difference image (DI), which is then analysed for change points by thresholding and/or clustering (e.g. Alphonse and Biju, 2015). However, such an approach completely neglects the long-term stability of a pixel. When taking the temporal evolution of a pixel into account, the steep increase in data volume puts an emphasis on finding an optimal (’best practice’) approach to the multitemporal change point detection problem. Here it is shown that change point detection methods that properly take the temporal evolution of a pixel into account can provide good segmentation results in multi-temporal SAR data stacks, even in unfiltered stacks that preserve the complete spatial resolution and without considering the spatial context in a pixel’s direct neighbourhood. Moreover, it is found thatmore sophisticated change point detection algorithms don’t
necessarily yield superior segmentation results for various discontinuity functions. This means algorithm selection has to be application driven. The results demonstrate that the suitability of algebraic methods in heterogeneous areas is limited, whereas proper time series analysis yields fairly consistent results over different land covers within the same image. Often, little effort is spend on finding an optimal approach; neglecting data selection and storage, a sensitivity analysis and/or the post-processing analysis procedure, all of which are shown or known to increase the success rate, efficiency and understanding of the segmentation results. It is anticipated that change point detection in SAR imagery will shift away from the classical bi-temporal methodology and multi temporal approaches will become the norm, be it by decomposing multi-temporal stacks or time series analysis.

Our climate is changing. In the Netherlands, higher rainfall intensities and longer periods of drought are expected in the coming years. Urban environments have to be adapted in order to maintain (or improve) the living standards that we have set for today. Models of the urban drainage system play a vital role in the identification of flood-prone areas, participation of stakeholders and the design of effective measures. This research focuses on the modelling of the urban drainage system to assess pluvial flooding. A method is selected to compare the main modelling concepts and differences between the most often used modelling concepts are assessed.

Nowcasting, synonymous for short term (0-6h) forecasting of precipitation with high detail in terms of location, timing and intensity, is a dynamic field of research in weather forecasting that is relevant in mitigating the impact of severe weather events. Along with the development of remote sensing instruments, the first methods that were developed to nowcast precipitation are based on the extrapolation of radar echoes, which are still widely employed in mobile and web applications today. However, the forecasting skill of these extrapolation based methods degrades rapidly with time. Especially in the case of convective rainfall extrapolation methods are not able to forecast the development of these precipitation events that have a more turbulent nature, primarily because they lack the description of the physical processes that govern the formation and dissipation of convective precipitation. For that reason, efforts in nowcasting are shifting towards numerical weather prediction (NWP) based methods. “High resolution” (<4km) NWP models covering a limited area are employed for their ability to describe large scale convection. Still, especially for accurately forecasting precipitation, these high resolution models require a detailed initial state which is generally not achieved by conventional initialization from global models having a much coarser resolution (>25km). To obtain a more accurate and detailed description of the initial state so called data assimilation methods have been developed, which use detailed information from recent observations to update the model. In this study it is investigated whether the precipitation forecasting skill of a regional Weather Research and Forecasting (WRF) model covering the Benelux can be improved by means of variational data assimilation. To that end, a variety of atmospheric observations is assimilated into a 4-km resolution WRF model using the accompanying WRFDA system with a 3DVAR rapid update cycling strategy. Besides conventional observations from surface weather stations, airports and soundings, observations from remote sensing platforms like radar (reflectivity and radial velocity) and zenith total delay (ZTD) from GPS are also assimilated. In a case study featuring a large squall line it is shown that assimilation of every single type of observation results in a notable improvement of forecasting skill, except for the radar assimilation of hydrometeors whose impact on the forecast is negligible after having precipitated to the surface. Especially assimilation of estimated humidity based on radar reflectivity yields a significant improvement during the full 6 hours of the forecast. The ability of the other types of observations to positively impact forecasting skill is attributed the large scale forcing by wind convergence along the cold front that is driving the formation of precipitation. Overestimation of precipitation quantity in the scenario without assimilation however is aggravated when radar derived humidity is assimilated, most likely as a consequence of the rapid update cycling strategy.