Royal Meteorological Institute Research Articles

Quality control and correction method for air temperature data from a citizen science weather station network in Leuven, Belgium

Abstract. The growing trend toward urbanisation and the increasingly frequent occurrence of extreme weather events emphasise the need for further monitoring and understanding of weather in cities. In order to gain information on these intra-urban weather patterns, dense high-quality atmospheric measurements are needed. Crowdsourced weather stations (CWSs) could be a promising solution to realise such monitoring networks in a cost-efficient way. However, due to their nontraditional measuring equipment and installation settings, the quality of datasets from these networks remains an issue. This paper presents crowdsourced data from the “Leuven.cool” network, a citizen science network of around 100 low-cost weather stations (Fine Offset WH2600) distributed across Leuven, Belgium (50∘52′ N, 4∘42′ E). The dataset is accompanied by a newly developed station-specific temperature quality control (QC) and correction procedure. The procedure consists of three levels that remove implausible measurements while also correcting for inter-station (between-station) and intra-station (station-specific) temperature biases by means of a random forest approach. The QC method is evaluated using data from four WH2600 stations installed next to official weather stations belonging to the Royal Meteorological Institute of Belgium (RMI). A positive temperature bias with a strong relation to the incoming solar radiation was found between the CWS data and the official data. The QC method is able to reduce this bias from 0.15 ± 0.56 to 0.00 ± 0.28 K. After evaluation, the QC method is applied to the data of the Leuven.cool network, making it a very suitable dataset to study local weather phenomena, such as the urban heat island (UHI) effect, in detail. (https://doi.org/10.48804/SSRN3F, Beele et al., 2022).

Impact of bias nonstationarity on the performance of uni- and multivariate bias-adjusting methods: a case study on data from Uccle, Belgium

Abstract. Climate change is one of the biggest challenges currently faced by society, with an impact on many systems, such as the hydrological cycle. To assess this impact in a local context, regional climate model (RCM) simulations are often used as input for rainfall-runoff models. However, RCM results are still biased with respect to the observations. Many methods have been developed to adjust these biases, but only during the last few years, methods to adjust biases that account for the correlation between the variables have been proposed. This correlation adjustment is especially important for compound event impact analysis. As an illustration, a hydrological impact assessment exercise is used here, as hydrological models often need multiple locally unbiased input variables to ensure an unbiased output. However, it has been suggested that multivariate bias-adjusting methods may perform poorly under climate change conditions because of bias nonstationarity. In this study, two univariate and four multivariate bias-adjusting methods are compared with respect to their performance under climate change conditions. To this end, a case study is performed using data from the Royal Meteorological Institute of Belgium, located in Uccle. The methods are calibrated in the late 20th century (1970–1989) and validated in the early 21st century (1998–2017), in which the effect of climate change is already visible. The variables adjusted are precipitation, evaporation and temperature, of which the former two are used as input for a rainfall-runoff model, to allow for the validation of the methods on discharge. Although not used for discharge modeling, temperature is a commonly adjusted variable in both uni- and multivariate settings and we therefore also included this variable. The methods are evaluated using indices based on the adjusted variables, the temporal structure, and the multivariate correlation. The Perkins skill score is used to evaluate the full probability density function (PDF). The results show a clear impact of nonstationarity on the bias adjustment. However, the impact varies depending on season and variable: the impact is most visible for precipitation in winter and summer. All methods respond similarly to the bias nonstationarity, with increased biases after adjustment in the validation period in comparison with the calibration period. This should be accounted for in impact models: incorrectly adjusted inputs or forcings will lead to predicted discharges that are biased as well.

High temperatures trigger suicide mortality in Brussels, Belgium: A case-crossover study (2002–2011)

BackgroundTemperature may trigger the risk of suicide, however, the extent and shape of the associations show geographical variation. Here, we investigate the short-term effects of temperature on suicide deaths occurring in Brussels between January 1st, 2002 and December 31st, 2011. MethodsWe conducted a bidirectional time-stratified case-crossover study with cases being suicide deaths occurring among Brussels residents aged 5 years or older. Cases were matched by day of the week with control days from the same month and year. The exposure was the daily average temperature measured at the Uccle station (Brussels) and obtained from the Belgian Royal Meteorological Institute. We combined conditional logistic regression with distributed lag non-linear models (DLNM) to obtain one week (lag 0–6) cumulative risk ratios (RR) and their 95% confidence intervals (CI) for the effects of moderate and extreme cold (5th and 1st percentiles of temperature, respectively) and moderate and extreme heat (95th and 99th percentiles of temperature, respectively), relative to the median temperature. ResultsIn total, 1891 suicide deaths were included. The median temperature was 11.6 °C, moderate and extreme cold temperatures were 0 and -3.1 °C, respectively, and moderate and extreme high temperatures were 20.9 and 24.4 °C, respectively. The cumulative risk of suicide mortality was almost twice higher among lags 0 to 6 for both moderate and extreme heat, relative to the period median temperature (e.g. moderate heat RR = 1.80 CI:1.27–2.54). No statistically significant associations were observed for cold temperatures. ConclusionsIn Brussels, a western European city with temperate climate, high temperatures may trigger suicide deaths up to one week later. In the context of climate change, adaptation strategies must take into consideration the effects of temperature on mental health.

The impacts of Storm Odette in eastern England, 24–26 September 2020

An area of low pressure over the southern North Sea, named Storm Odette by the Royal Meteorological Institute of Belgium, produced an extended period of disruptive winds, exceptional rainfall and notably cold temperatures to parts of the United Kingdom between 24–26 September 2020. Several sites recorded their coldest September daily maximum temperature on record, while parts of Norfolk experienced one of the longest duration of gales to affect the area during the past 30 years.

Short-term associations between Legionnaires' disease incidence and meteorological variables in Belgium, 2011–2019

The number of reported cases with Legionnaires' disease (LD) is increasing in Belgium. Previous studies have investigated the associations between LD incidence and meteorological factors, but the Belgian data remained unexplored. We investigated data collected between 2011 and 2019. Daily exposure data on temperature, relative humidity, precipitation and wind speed was obtained from the Royal Meteorological Institute for 29 weather stations. Case data were collected from the national reference centre and through mandatory notification. Daily case and exposure data were aggregated by province. We conducted a time-stratified case-crossover study. The 'at risk' period was defined as 10 to 2 days prior to disease onset. The corresponding days in the other study years were selected as referents. We fitted separate conditional Poisson models for each day in the 'at risk' period and a distributed lag non-linear model (DLNM) which fitted all data in one model. LD incidence showed a yearly peak in August and September. A total of 614 cases were included. Given seasonality, a sequence of precipitation, followed by high relative humidity and low wind speed showed a statistically significant association with the number of cases 6 to 4 days later. We discussed the advantages of DLNM in this context.

SmarterRoutes – Data-driven road complexity estimation for level-of-detail adaptation of navigation services

Abstract. SmarterRoutes aims to improve navigational services and make them more dynamic and personalised by data-driven and environmentally-aware road scene complexity estimation. SmarterRoutes divides complexity into two subtypes: perceived and descriptive complexity. In the SmarterRoutes architecture, the overall road scene complexity is indicated by combining and merging parameters from both types of complexity. Descriptive complexity is derived from geospatial data sources, traffic data and sensor analysis. The architecture is currently using OpenStreetMap (OSM) tag analysis, Meten-In-Vlaanderen (MIV) derived traffic info and the Alaro weather model of the Royal Meteorological Institute of Belgium (RMI) as descriptive complexity indicators. For the perceived complexity an image based complexity estimation mechanism is presented. This image based Densenet Convolutional Neural Network (CNN) uses Street View images as input and was pretrained on buildings with Bag-of-Words and Structure-from-motion features. The model calculates an image descriptor allowing comparison of images by calculation of the Euclidean distances between descriptors. SmarterRoutes extends this model by additional hand-labelled rankings of road scene images to predict visual road complexity. The reuse of an existing pretrained model with an additional ranking mechanism produces results corresponding with subjective assessments of end-users. Finally, the global complexity mechanism combines the aforementioned sub-mechanisms and produces a service which should facilitate user-centred context-aware navigation by intelligent data selection and/or omission based on SmarterRoutes’ complexity input.

Global Radiative Flux and Cloudiness Variability for the Period 1959–2010 in Belgium: A Comparison between Reanalyses and the Regional Climate Model MAR

The use of regional climate models (RCMs) can partly reduce the biases in global radiative flux (Eg↓) that are found in reanalysis products and global models, as they allow for a finer spatial resolution and a finer parametrisation of surface and atmospheric processes. In this study, we assess the ability of the MAR («Modèle Atmosphérique Régional») RCM to reproduce observed changes in Eg↓, and we investigate the added value of MAR with respect to reanalyses. Simulations were performed at a horizontal resolution of 5 km for the period 1959–2010 by forcing MAR with different reanalysis products: ERA40/ERA-interim, NCEP/NCAR-v1, ERA-20C, and 20CRV2C. Measurements of Eg↓ from the Global Energy Balance Archive (GEBA) and from the Royal Meteorological Institute of Belgium (RMIB), as well as cloud cover observations from Belgocontrol and RMIB, were used for the evaluation of the MAR model and the forcing reanalyses. Results show that MAR enables largely reducing the mean biases that are present in the reanalyses. The trend analysis shows that only MAR forced by ERA40/ERA-interim shows historical trends, which is probably because the ERA40/ERA-interim has a better horizontal resolution and assimilates more observations than the other reanalyses that are used in this study. The results suggest that the solar brightening observed since the 1980s in Belgium has mainly been due to decreasing cloud cover.

Solar irradiation from the energy production of residential PV systems

Abstract Considering the dense network of residential photovoltaic (PV) systems implemented in Belgium, the paper evaluates the opportunity of deriving global horizontal solar irradiation data from the electrical energy production registered at PV systems. The study is based on one year (i.e. 2014) of hourly PV power output collected at a representative sample of roughly 1500 residential PV installations. Validation is based on ground-based measurements of solar radiation performed within the network of radiometric stations operated by the Royal Meteorological Institute of Belgium and the method's performance is compared to the satellite-based retrieval approach. Our results indicate that the accuracy of the derived solar irradiation data depends on a number of factors including the efficiency of the PV system, the weather conditions, the density of PV systems that can be used for the tilt to horizontal conversion, other data sources that can be accessed to complement the PV data. In particular, the computed solar irradiation data degrade as the information about the orientation and tilt angles of the PV generator becomes more inaccurate. It is also found that there are certain sun positions (i.e. low solar elevations) for which the method fails to produce a valid estimation.

An automatic DI-flux at the Livingston Island geomagnetic observatory, Antarctica: requirements and lessons learned

Abstract. The DI-flux, consisting of a fluxgate magnetometer coupled with a theodolite, is used for the absolute manual measurement of the magnetic field angles in most ground-based observatories worldwide. Commercial solutions for an automated DI-flux have recently been developed by the Royal Meteorological Institute of Belgium (RMI), and are practically restricted to the AutoDIF and its variant, the GyroDIF. In this article, we analyze the pros and cons of both instruments in terms of its suitability for installation at the partially manned geomagnetic observatory of Livingston Island (LIV), Antarctica. We conclude that the GyroDIF, even if it is less accurate and more power demanding, is more suitable than the AutoDIF for harsh conditions due to the simpler infrastructure that is necessary. Power constraints in the Spanish Antarctic Station Juan Carlos I (ASJI) during the unmanned season require an energy-efficient design of the thermally regulated box housing the instrument as well as thorough power management. Our experiences can benefit the geomagnetic community, which often faces similar challenges.

Tilt to horizontal global solar irradiance conversion: An evaluation at high tilt angles and different orientations

Many transposition models have been proposed in the literature to convert solar irradiance on the horizontal plane to that on a tilted plane. The inverse process, i.e., the conversion from tilted to horizontal is investigated here based on six months of in-plane global solar irradiance measurements recorded on the roof of the Royal Meteorological Institute of Belgium's radiation tower in Uccle (Latitude 50.79°N, Longitude 4.35°E). Up to three pyranometers mounted on inclined planes of different tilts and orientations were involved in the inverse transposition process. Our results indicate that (1) the tilt to horizontal irradiance conversion is improved by using a multi-pyranometers approach and (2) the improvement from using an isotropic model to anisotropic models in the inverse transposition problem is not significant.

The Belgian Space Weather Observatory in Dourbes

AbstractThe paper outlines the space-weather-related observations, instrumentation, research and service developments carried out at the Geophysical Centre of the Royal Meteorological Institute (RMI) in Dourbes, Belgium.

Respiratory and circulatory deaths attributable to influenza A & B

ObjectiveTo estimate mortality attributable to influenza adjusted for othercommon respiratory pathogens, baseline seasonal trends and extremetemperatures.IntroductionAssigning causes of deaths to seasonal infectious diseases is difficultin part due to laboratory testing prior to death being uncommon. Sinceinfluenza (and other common respiratory pathogens) are thereforenotoriously underreported as a (contributing) cause of death in death-cause statistics modeling studies are commonly used to estimate theimpact of influenza on mortality.MethodsUsing primary cause of death (Statistics Netherlands) we modeledweekly timeseries of1) respiratory deaths (ICD10 codes J00-J99) and2) circulatory deaths (ICD10 codes I00-I99).We used regression models with an identity link and Poissonerror to relate mortality to counts of influenza A & B diagnoses.We adjusted for other common respiratory pathogens (all pathogendata was at population level from the national laboratory surveillance),temperature (from the Dutch Royal Meteorological Institute), andbaseline linear and cyclical (i.e. seasonal) trends. To account forthe yearly variation in the severity of the main circulating influenzaA strain we used time dependent variables for influenza A (fixedat lag 0 – assuming a direct effect of influenza. For influenza Band the confoundig pathogens we considered a 0 tot -4 time lag(thus allowing infection to precede death for up to 4 weeks).We performed the analyses separately per death cause group and by3 different age groups (0-64, 65-74,75+ years) over a 14-year time-period (mid 1999-mid 2013, thus 14 complete winter seasons).ResultsIn the Netherlands on average 2,636 all cause deaths occurper week varying by season (lower in summer min: 2,219 and higherin winter max: 3,564) with yearly incidence ranging from 20/10,000in 0-64 year olds to 885/10,000 in 75-plus year olds.Circulatory mortality (31% of total deaths) was higher thanrespiratory mortality (10% of total deaths) and both showed clearseasonality in all age-groups. Overall, 0.14% of all deaths wereactually coded as influenza deaths.Preliminary model estimates showed that the proportion ofrespiratory deaths attributable to influenza A were quite similar for 0-64and 65-74 year olds but higher in 75+ (5.1%, 5.7%, 7.0% respectively)while this proportion was stable across age-groups for circulatorydeaths (approximately 1.5% in all agegroups for influenza A).Influenza B was significantly associated with respiratory deathsand circulatory deaths in the oldest age group of 75+ years(with proportions of 0.7% and 0.2% respectively) while in the65-74 year olds it was associated only with circulatory deaths (0.2%).Influenza B was not significantly associated with either respiratory orcirculatory mortality in the 0-64 year age group.On average, yearly in the 75+ age group 70/10,000 respiratorydeaths and 39/10,000 circulatory deaths were attributable to influenzaA. For influenza B the incidences were 7 to 10 fold lower (7/10,000and 6/10,000 respectively).ConclusionsInfluenza A was significantly associated with respiratory andcirculatory mortality in all age groups while influenza B wassignificantly associated with respiratory and circulatory mortality inthe elderly only.

Evaluation of various procedures transposing global tilted irradiance to horizontal surface irradiance

Abstract. Many transposition models have been proposed in the literature to convert solar irradiance on the horizontal plane to that on a tilted plane. The inverse process, i.e. the conversion from tilted to horizontal is investigated here based upon seven months of in-plane global solar irradiance measurements recorded on the roof of the Royal Meteorological Institute of Belgium's radiation tower in Uccle (Longitude 4.35° E, Latitude 50.79° N). Up to three pyranometers mounted on inclined planes of different tilts and orientations were involved in the inverse transposition process. Our results indicate that (1) the tilt to horizontal irradiance conversion is improved when measurements from more than one tilted pyranometer are considered (i.e. by using a multi-pyranometer approach) and (2) the improvement from using an isotropic model to anisotropic models in the inverse transposition problem is not significant.

Season of birth is associated with first-lactation milk yield in Holstein Friesian cattle

The aim of the present research was to assess factors associated with first-lactation milk yield in dairy heifers, including maternal and environmental factors, factors related to the development of the heifer and factors related to its offspring such as gender of the calf. In addition, the potential underlying mechanism, in particular metabolic adaptations, was further explored. Data on body growth, reproduction and milk yield of 74 Holstein Friesian heifers on three herds in Flanders (Belgium) were collected. At birth, body measurements of the heifers were recorded and blood samples were taken (in order) to determine basal glucose and insulin concentrations. Body measurements were assessed every 3 months until first calving, and gender and weight of their first calf were recorded. Information on fertility and milk yield of the heifer and its dam were collected from the herd databases. Daily temperature and photoperiod were recorded from the database of the Belgian Royal Meteorological Institute. Linear mixed models were run with herd as a random factor, to account for differences in herd management. Heifers grew 867±80.7 g/day during their first year of life and were inseminated at 14.8±1.34 months. First calving took place at 24.5±1.93 months, at a weight of 642±61.5 kg and heifers produced 8506±1064 kg energy corrected milk during their first 305-day lactation. Regression models revealed that none of the maternal factors such as milk yield and parity, nor the growth of the heifer during the 1st year of life were associated with milk yield during first lactation. Age, and to a lesser extent BW at first parturition were positively associated with first-lactation milk yield. In addition, the season of birth, but not calving, had a significant influence on milk yield, with winter-born heifers producing less than heifers born in any other season. The lower yielding winter-born heifers had higher insulin concentrations at birth, whereas glucose concentrations were similar, the latter being suggestive for lower insulin sensitivity of the peripheral tissues. Furthermore, environmental temperature at the end of gestation was negatively correlated with neonatal insulin concentrations. In conclusion, results of the present study suggest heifers born during the hotter months are born with a higher peripheral insulin sensitivity, finally leading to a higher first-lactation milk yield.

Postprocessing of Medium Range Hydrological Ensemble Forecasts Making Use of Reforecasts

A hydrological ensemble prediction system is running operationally at the Royal Meteorological Institute of Belgium (RMI) for ten catchments in the Meuse basin. It makes use of the conceptual semi-distributed hydrological model SCHEME and the European Centre for Medium Range Weather Forecasts (ECMWF) ensemble prediction system (ENS). An ensemble of 51 discharge forecasts is generated daily. We investigate the improvements attained through postprocessing the discharge forecasts, using the archived ECMWF reforecasts for precipitation and other necessary meteorological variables. We use the 5-member reforecasts that have been produced since 2012, when the horizontal resolution of ENS was increased to the N320 resolution (≈30 km over Belgium). The reforecasts were issued weekly, going back 20 years, and we use a calibration window of five weeks. We use these as input to create a set of hydrological reforecasts. The implemented calibration method is an adaption of the variance inflation method. The parameters of the calibration are estimated based on the hydrological reforecasts and the observed discharge. The postprocessed forecasts are verified based on a two-and-a-half year period of data, using archived 51 member ENS forecasts. The skill is evaluated using summary scores of the ensemble mean and probabilistic scores: the Brier Score and the Continuous Ranked Probability Score (CRPS). We find that the variance inflation method gives a significant improvement in probabilistic discharge forecasts. The Brier score, which measures probabilistic skill for forecasts of discharge threshold exceedance, is improved for the entire forecast range during the hydrological summer period, and the first three days during hydrological winter. The CRPS is also significantly improved during summer, but not during winter. We conclude that it is valuable to apply the postprocessing method during hydrological summer. During winter, the method is also useful for forecasting exceedance probabilities of higher thresholds, but not for lead times beyond five days. Finally, we also note the presence of some large outliers in the postprocessed discharge forecasts, arising from the fact that the postprocessing is performed on the logarithmically transformed discharges. We suggest some ways to deal with this in the future for our operational setting.

Seasonal variation in cutaneous melanoma incidence, link with recent UV levels: a population-based study in Belgium (2006-2011).

Our objective was to test the hypothesis of a short-term late-promoting effect of ultraviolet (UV) exposure on the development of cutaneous melanoma as an explanation for the summer peak in melanoma incidence. Therefore, we studied seasonal variation in melanoma incidence in relation to recent UV levels by direct UV measurements. Data from the Belgian Cancer Registry on invasive cutaneous melanoma diagnosed during 2006-2011 were used for analysis. Daily data on UV measurements in Belgium were obtained from the Royal Meteorological Institute. Simple and multiple negative binomial regression models were used to investigate the influence of recent UV levels on melanoma incidence. The sum of the mean UV doses in the 2 months before diagnosis was used as a proxy for recent UV exposure in the population. To include variable sunburn risks during the year, the categorical variable 'semester' was created. The incidence of melanoma in Belgium shows a distinct seasonal variation, with peaks in June or July. We found that part of this variation could be explained by the variation in dermatologic activity and, therefore, used this as an offset in our models. We found a linear relationship between melanoma incidence and UV dose in the 2 months preceding the diagnosis. UV levels had more impact in the first semester. The effect of UV levels was not modified by sex nor age. The interaction between anatomical site and UV levels was significant (P=0.002) and showed a higher effect on the upper and lower limbs compared with the head and neck and trunk.

Validation of the ALARO-0 model within the EURO-CORDEX framework

Abstract. Using the regional climate model ALARO-0, the Royal Meteorological Institute of Belgium and Ghent University have performed two simulations of the past observed climate within the framework of the Coordinated Regional Climate Downscaling Experiment (CORDEX). The ERA-Interim reanalysis was used to drive the model for the period 1979–2010 on the EURO-CORDEX domain with two horizontal resolutions, 0.11 and 0.44°. ALARO-0 is characterised by the new microphysics scheme 3MT, which allows for a better representation of convective precipitation. In Kotlarski et al. (2014) several metrics assessing the performance in representing seasonal mean near-surface air temperature and precipitation are defined and the corresponding scores are calculated for an ensemble of models for different regions and seasons for the period 1989–2008. Of special interest within this ensemble is the ARPEGE model by the Centre National de Recherches Météorologiques (CNRM), which shares a large amount of core code with ALARO-0. Results show that ALARO-0 is capable of representing the European climate in an acceptable way as most of the ALARO-0 scores lie within the existing ensemble. However, for near-surface air temperature, some large biases, which are often also found in the ARPEGE results, persist. For precipitation, on the other hand, the ALARO-0 model produces some of the best scores within the ensemble and no clear resemblance to ARPEGE is found, which is attributed to the inclusion of 3MT. Additionally, a jackknife procedure is applied to the ALARO-0 results in order to test whether the scores are robust, meaning independent of the period used to calculate them. Periods of 20 years are sampled from the 32-year simulation and used to construct the 95 % confidence interval for each score. For most scores, these intervals are very small compared to the total ensemble spread, implying that model differences in the scores are significant.

High-resolution ionospheric observations and modeling over Belgium during the solar eclipse of 20 March 2015 including first results of ionospheric tilt and plasma drift measurements

The ionospheric behavior over Belgium during the partial solar eclipse of 20 March 2015 is analyzed based on high-resolution solar radio flux, vertical incidence sounding, and GPS TEC measurements. First results of ionosonde-based ionospheric plasma drift and tilt observations are presented and analyzed, including some traveling ionospheric disturbances caused by the eclipse. Also, collocated ionosonde and GPS measurements are used to reconstruct the time evolution of the vertical electron density distribution using the Royal Meteorological Institute (RMI) ionospheric specification system, called Local Ionospheric Electron Density profile Reconstruction (Liedr).

Quality control of the RMI's AWS wind observations

Abstract. Wind observations are important for a wide range of domains including among others meteorology, agriculture and extreme wind engineering. To ensure the provision of high quality surface wind data over Belgium, a new semi-automated data quality control (QC) has been developed and applied to wind observations from the automated weather stations operated by the Royal Meteorological Institute of Belgium. This new QC applies to 10 m 10 min averaged wind speed and direction, 10 m gust speed and direction, 2 m 10 min averaged wind speed and 30 m 10 min averaged wind speed records. After an existence test, automated procedures check the data for limits consistency, internal consistency, temporal consistency and spatial consistency. At the end of the automated QC, a decision algorithm attributes a flag to each particular data point. Each day, the QC staff analyzes the preceding day's observations in the light of the assigned quality flags.

Development and verification of a real-time stochastic precipitation nowcasting system for urban hydrology in Belgium

Abstract. The Short-Term Ensemble Prediction System (STEPS) is implemented in real-time at the Royal Meteorological Institute (RMI) of Belgium. The main idea behind STEPS is to quantify the forecast uncertainty by adding stochastic perturbations to the deterministic Lagrangian extrapolation of radar images. The stochastic perturbations are designed to account for the unpredictable precipitation growth and decay processes and to reproduce the dynamic scaling of precipitation fields, i.e., the observation that large-scale rainfall structures are more persistent and predictable than small-scale convective cells. This paper presents the development, adaptation and verification of the STEPS system for Belgium (STEPS-BE). STEPS-BE provides in real-time 20-member ensemble precipitation nowcasts at 1 km and 5 min resolutions up to 2 h lead time using a 4 C-band radar composite as input. In the context of the PLURISK project, STEPS forecasts were generated to be used as input in sewer system hydraulic models for nowcasting urban inundations in the cities of Ghent and Leuven. Comprehensive forecast verification was performed in order to detect systematic biases over the given urban areas and to analyze the reliability of probabilistic forecasts for a set of case studies in 2013 and 2014. The forecast biases over the cities of Leuven and Ghent were found to be small, which is encouraging for future integration of STEPS nowcasts into the hydraulic models. Probabilistic forecasts of exceeding 0.5 mm h−1 are reliable up to 60–90 min lead time, while the ones of exceeding 5.0 mm h−1 are only reliable up to 30 min. The STEPS ensembles are slightly under-dispersive and represent only 75–90 % of the forecast errors.