Search

The exceptional flood of July 2021 in central Europe impacted Belgium severely. As rainfall was the triggering factor of this event, this study aims to characterize rainfall amounts in Belgium from 13 to 16 July 2021 based on two types of observational data. First, observations recorded by high-quality rain gauges operated by weather and hydrological services in Belgium have been compiled and quality checked. Second, a radar-based rainfall product has been improved to provide a reliable estimation of quantitative precipitation at high spatial and temporal resolutions over Belgium. Several analyses of these data are performed here to describe the spatial and temporal distribution of rainfall during the event. These analyses indicate that the rainfall accumulations during the event reached unprecedented levels over large areas. Accumulations over durations from 1 to 3 d significantly exceeded the 200-year return level in several places, with up to 90 % of exceedance over the 200-year return level for 2 and 3 d values locally in the Vesdre Basin. Such a record-breaking event needs to be documented as much as possible, and available observational data must be shared with the scientific community for further studies in hydrology, in urban planning and, more generally, in all multi-disciplinary studies aiming to identify and understand factors leading to such disaster. The corresponding rainfall data are therefore provided freely in a supplement. [ABSTRACT FROM AUTHOR]

The radar-based estimation of intense precipitation produced by convective storms is a challenging task and the verification through comparison with gauges is questionable due to the very high spatial variability of such type of precipitation. In this study, we explore the potential benefit of using a superconducting gravimeter as a new source of in-situ 10 observations for the evaluation of radar-based precipitation estimates. The superconducting gravimeter used in this study is installed in Membach (BE), 48 m underneath the surface, at 85 km distance from a C-band weather radar located in Wideumont (BE). The 15-year observation record 2003-2017 is available for both gravimeter and radar with 1-min and 5-min time steps, respectively. The gravimeter integrates soil water in a radius of about 400 m around the instrument. This allows capturing rainfall at larger spatial scale than traditional rain gauges. The precision of the gravimeter is a few nm/s²; 1 nm/s² corresponding 15 to 2.6 mm of water. The comparison of reflectivity and gravity time series for short duration intense rainfall events shows that reflectivity peaks larger than 40 dBZ are associated with a rapid decrease of the underground measured gravity. A remarkable correspondence between radar and gravimeter time series is found. The precipitation amounts derived from gravity measurements and from radar observations are further compared for 505 rainfall events. A correlation coefficient of 0.58, a mean bias (radar/gravimeter) of 1.24 and a mean absolute difference (MAD) of 3.19 mm are obtained. A better agreement is 20 reached when applying a hail correction by truncating reflectivity values to a given threshold. No bias, a correlation coefficient of 0.64 and a MAD of 2.3 mm are reached using a 48-dBZ threshold. The added value of underground gravity measurements as verification dataset is discussed. The two main benefits are the spatial scale at which precipitation is captured and the interesting property that gravity measurements are directly influenced by water mass at ground no matter the type of precipitation: hail or rain. [ABSTRACT FROM AUTHOR]

In Belgium, only rain gauge time-series have been used so far to study extreme precipitation at a given location. In this paper, the potential of a 12-year quantitative precipitation estimation (QPE) from a single weather radar is evaluated. For the period 2005-2016, independent sliding 1h and 24h rainfall extremes from automatic rain gauges and collocated radar estimates are compared. The extremes are fitted to the exponential distribution using regression in QQ-plots with a threshold rank which minimises the mean squared error. A basic radar product used as reference exhibits unrealistic high extremes and is not suitable for extreme value analysis. For 24h rainfall extremes, which occur partly in winter, the radar-based QPE needs a bias correction. A few missing events are caused by the wind drift of convective cells and strong radar signal attenuation. Differences between radar and gauge values are caused by spatial and temporal sampling, gauge rainfall underestimations and radar errors due to the relation between reflectivity and rain rate. Nonetheless the fit to the QPE data is within the confidence interval of the gauge fit, which remains large due to the short study period. A regional frequency analysis is performed on radar data within 20km of the locations of 4 rain gauges with records from 1965 to 2008. Assuming that the extremes are correlated within the region, the fit to the two closest rain gauge data is within the confidence interval of the radar fit, which is small due to the sample size. In Brussels, the extremes on the period 1965-2008 from a rain gauge are significantly lower than the extremes from an automatic gauge and the radar on the period 2005-2016. For 1h duration, the location parameter varies slightly with topography and the scale parameter exhibits some variations from region to region. The radar-based extreme value analysis can be extended to other durations. [ABSTRACT FROM AUTHOR]