Wind has a significant impact on precipitation measurement instruments, including disdrometers, by inducing aerodynamic disturbances around their bodies. These airflow features divert trajectories of falling hydrometeors, often reducing the amount of precipitation detected when compared to  windless conditions. Furthermore, the shape of disdrometers, which is non-radially symmetric, makes the wind-induced bias dependent on wind direction. Traditionally, field experiments have been used to develop corrections for the wind-induced bias. However, Computational Fluid Dynamics (CFD) simulations offer a more versatile approach for studying wind-induced bias on different instrument designs under varying climatic conditions. In this work a wind tunnel experimental campaign was conducted to show the interaction between wind and disdrometers and to validate a suitable CFD model by providing detailed data on drop trajectories. Full-scale tests were conducted in the high-speed test section of the Wind Tunnel facility available at Politecnico di Milano. The chamber (4m wide, 3.8m high and 6m long) is characterized by a nearly laminar flow and a narrow boundary layer. The disdrometers were fixed to the ground on a rotating plate to facilitate alignment with the flow direction. Furthermore, a specially designed drop generator – attached to a moving gantry – was used to release water drops into the wind flow, allowing precise control of drop diameter, release height and timing. Finally, a high-speed camera, operating at 1000 fps, recorded the trajectories of the drops approaching the sensing areas of the disdrometers. Images were processed to identify each drop, calculate their velocity and track their movement through the camera field of view. The study focused on two disdrometer models, the Thies CLIMA LPM and the OTT Parsivel², which use an optical method to measure drop size and velocity. The experiments were conducted for wind speeds of 10 m/s, drop diameters ranging from 1.0 to 1.2 mm, and three wind directions (0°, 45°, and 90°). Results showed that wind significantly alters drop trajectories, often diverting them away from the sensing area or causing them to collide with the instrument body. A numerical model - already used in e.g., Chinchella et al., (2024) – was validated by simulating the experimental conditions and comparing the results against observations. Validation shows that the numerical approach is suitable for developing adjustment curves to correct disdrometer measurements under windy conditions. This work further highlights the importance of addressing wind effects in precipitation measurements, by applying correction curves (see e.g., Chinchella et al., 2024) to enhance the accuracy of rainfall measurements obtained from disdrometers like the Thies CLIMA LPM or the OTT Parsivel².

ACKNOWLEDGMENTS

The wind tunnel campaign on disdrometers was carried out within the framework of the Italian national projects PRIN2022MYTKP4 “Fostering innovation in precipitation measurements: from drop size to hydrological and climatic scales”.

References:

Chinchella E., Cauteruccio, A., & Lanza, L. G. (2024). Quantifying the wind-induced bias of rainfall measurements for the Thies CLIMA optical disdrometer. Water Resources Research, 60(10), e2024WR037366. https://doi.org/10.1029/2024WR037366   

How to cite: Lanza, L. G., Chinchella, E., Calamelli, F., Cauteruccio, A., and Rocchi, D.: Wind tunnel experiments to evaluate the wind-induced bias on disdrometer measurements, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-17651, https://doi.org/10.5194/egusphere-egu25-17651, 2025.

Disdrometers are in-situ, non-catching devices capable of measuring the size and fall velocity (for most models) of each individual hydrometeor (solid or liquid) that enters their measurement volume. These devices are primarily used for research purposes, and their data have applications in fields such as meteorology, climatology, and hydrology. However, their measurements can be influenced by the presence of wind. In this context, one of the objectives of the PRIN project titled “Fostering innovation in precipitation measurements: from drop size to hydrological and climatic scales” is to quantify the accuracy of disdrometers. In this regard, data collected from a Thies Clima disdrometer and wind sensors installed in the city of Pescara serve as a valuable resource for: i) characterizing precipitation, ii) conducting a joint analysis of atmospheric conditions, including wind directionand speed, and iii) evaluating the effect of wind on disdrometer measurements. The dataset covers the period from July 2021 to August 2024, although it includes significant interruptions. This study presents the main characteristics of the site in terms of wind and rain distributions, as well as their joint distributions. Additionally, the effects of wind on disdrometer measurements are quantified in terms of the associated bias on on DSD (Drop Size Distribution) estimation. Results indicate that wind-corrected DSDs differ, on average, by 136.41m⁻³ ·mm⁻¹ in terms of root mean square error compared to uncorrected DSDs. Subsequently, since we do not have a DSD from the rain gauge, we hypothesize that it has the form of an exponential αe^β, and we interpolate these parameters from the disdrometer data. Then this parametrs are used to apply corrections to nearby rain gauge measurements, and the corrected and uncorrected values are compared. These differences are found to be statistically significant. Furthermore, twenty-six stations in Calabria, equipped with rain gauges and anemometers, are analyzed using the same DSD parameters derived from the Pescara dataset. Precipitation amounts obtained from corrected and uncorrected DSDs are compared with corresponding corrected and uncorrected rain gauge data, revealing statistically significant differences. These findings provide insight into the effects of the applied correction on rain rate measurements.
Acknowledgments
This work was carried out within the framework of the ongoing Italian national project PRIN2022MYTKP4 “Fostering innovation in precipitation measurements: from drop size to hydrological and climatic scales”.

How to cite: Adirosi, E., Parasporo, L., Baldini, L., Cauretuccio, A., Chinchella, E., Caloiero, T., and Lanza, L.: The wind effects on disdrometer and rain gauges measurements: results from a 4-year long rain series data-set in Pescara and a 10-year long rain series data-set in Calabria (Italy), EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-15409, https://doi.org/10.5194/egusphere-egu25-15409, 2025.

Rainfall has been monitored with microwave links opportunistically for nearly 20 years. So far, most studies have focused on retrieving rainfall rates using the mean received signal, based on the power-law relation between specific attenuation and rainfall rate. However, theories and measurements have indicated that the power spectral density (PSD) of received signal contains extra information about rainfall. The drop size distribution (DSD) and the motion of raindrops both play a role in determining the scintillation spectrum of rain. To evaluate the feasibility of making use of rain spectra for retrieving information about DSDs, measurements from different experimental datasets are investigated. Initial results indicate that some information about rainfall (e.g. rainfall rate) is indeed retained in the spectra measured by a radio link at 26 GHz and a scintillometer at 160 GHz. Furthermore, a simulation of the PSD of the received voltage during rain is made to gain understandings of its behavior. The simulation, based on Ishimaru’s work (1978), allows for the customization of various settings (e.g., wavelength, geometry, antenna gain functions) of radio links, as well as the DSD at different locations along the links. It is shown that large raindrops have more influence on the PSD of received voltage than smaller raindrops. A theoretical method to retrieve DSD from the PSD of the received voltage is proposed and its performance is assessed by simulation. Results show that the concentration of the tiniest raindrops is hard to retrieve because of their minor impacts on PSD. In the simulation, the concentration of larger raindrops can be relatively reliably retrieved, even when a large variation of DSDs is present along the microwave link.

How to cite: Wang, P., Droste, A., Schleiss, M., and Uijlenhoet, R.: Rain scintillation spectra from microwave links: A potential source of information for raindrop size distributions, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-15697, https://doi.org/10.5194/egusphere-egu25-15697, 2025.

In hydrological modeling and forecasting, rainfall data is a key factor in determining the model’s accuracy. The higher the accuracy of the estimated rainfall, the more accurate the model’s predictions can be. Rain gauges can be utilized to estimate the amount of rainfall within a catchment area but their effectiveness is often limited by the sparse distribution of rain gauges and the lack of sufficient spatial information they provide for comprehensive distributed hydrological simulations. Weather radar serves as an alternative source of rainfall data, capable of providing remotely sensed rainfall estimates with high temporal and spatial resolution. However, conventional radar quantitative precipitation estimation (QPE) is subject to uncertainties, primarily arising from variations in the drop size distribution (DSD) of hydrometeors and variations in vertical profile reflectivity (VPR). Those variations are typically influenced by the local climate and weather conditions and their impacts on the performance of QPE remains a subject of research especially in tropical regions. Therefore, this study aims to investigate relationships between weather conditions and the performance of radar QPE using statistical and machine learning approaches at different time scales. In Thailand, the radar-based rainfall data is derived with a standard fixed power law relationship between radar reflectivity and rain rate, from three weather radars located in different parts of the country. The rainfall estimates from this radar rainfall product are investigated with weather conditions from ERA5 reanalysis datasets and local observations in the period of 2022-2024. The findings help us to identify the key factors influencing the accuracy of radar rainfall estimation, which can be used to improve radar rainfall estimation, for example through finding adequate predictors for the construction of a dynamic Z-R relationship in tropical conditions. Future studies could expand this analysis by integrating these impact factors into radar QPEs and implementing improved estimated rainfall products in hydrological models.

How to cite: Luangdilok, N., Imhoff, R., Brauer, C., and Weerts, A.: Assessing the Impact of Weather Conditions on Radar-Based Rainfall Estimation in the Tropics: A Case Study in Thailand, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-17854, https://doi.org/10.5194/egusphere-egu25-17854, 2025.

Dual-polarization radar rainfall rate estimates are based on scaling laws involving the horizontal reflectivity Z_h and the ratio between horizontal and vertical reflectivity Z_DR. Scaling law parameters obtained from disdrometric observations are highly dependent on the data set used. As a consequence ZR scaling laws do not generalize well. Using the jargon of data science, a ZR scaling law has an accpetable training accuracy and a poor validation accuracy. 

To overcome this limitation, we propose the Formula-R algorithm based on the adoption of the data science parametrization of drop size distributions and its universal shape factors [https://doi.org/10.1175/JHM-D-21-0211.1]. We show, using a worldwide catalog of disdrometric observations, how the Formula-R outperforms the ZR scaling law both in training and validation accuracy. 

The Formula-R algorithm could be used as the foundation of a universal remote sensing retrieval algorithm making the question "which ZR-relationship should we use?" a question of the past.

How to cite: Ignaccolo, M. and De Michele, C.: A novel algorithm for remote sensing rainfall retrieval, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-3311, https://doi.org/10.5194/egusphere-egu25-3311, 2025.

Estimating the distribution of hail sizes is crucial for assessing related weather hazards and potential damage to buildings, vehicles and agriculture. In this study, we present a novel technique for estimating the hail size number distribution (HSND) using polarimetric C-band radar data. A generalized additive model (GAM) is employed to estimate two empirical moments of the HSND, which is then reconstructed using double-moment normalization. This approach capitalizes on the relative invariance of the double-moment normalized HSND. The model is trained on data from the Swiss network of automatic hail sensors, spanning from September 2018 to August 2024 and covering three regions of Switzerland particularly prone to hail. Several polarimetric features are extracted from a 3D radar composite that combines observations from all operational Swiss radars. Among the various extracted features, the model selects the echo-top height of 50 dBZ reflectivity value at vertical polarization and the volume of the region with a cross-correlation coefficient rhoHV below 0.97, as these provided the best predictive performance. Radar-derived HSND estimates show good agreement with independent hail sensor observations. Additionally, the model is evaluated through comparisons with photogrammetric drone surveys and crowd-sourced reports of hail. This technique enables high spatio-temporal resolution (1 km and 5 minutes) retrievals of HSND and related metrics, such as kinetic energy. Further ground observations, particularly drone-based, are essential for more comprehensive evaluation of the retrieved HSND.

How to cite: Guidicelli, M., Ferrone, A., Ghiggi, G., Gabella, M., Germann, U., and Berne, A.: Retrieval of the hail size number distribution from polarimetric C-band weather radar using double-moment normalization, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-11476, https://doi.org/10.5194/egusphere-egu25-11476, 2025.

 Accurate precipitation estimation is crucial for hydrological modeling and flood forecasting in the Yangtze River Basin (YRB), China. This study explores the use of machine learning (ML) and deep learning (DL) methods to fuse multi-source precipitation data, including satellite, radar, and ground-based observations. We apply models such as Random Forest (RF), Support Vector Machines (SVM), Convolutional Neural Networks (CNN), and Long Short-Term Memory (LSTM) networks to improve precipitation estimation accuracy. Performance is evaluated using metrics like Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). Our results demonstrate that deep learning models, particularly CNNs and LSTMs, outperform traditional ML methods in terms of accuracy and spatial consistency. This work provides a robust approach to multi-source data fusion, enhancing precipitation monitoring and hydrological applications in the YRB.

How to cite: Chen, T.: Machine Learning and Deep Learning for Multi-Source Precipitation Integration in the Yangtze River Basin, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-1507, https://doi.org/10.5194/egusphere-egu25-1507, 2025.

Spatial and temporal interpolation methods are generally used for estimation of missing data. Objective selection of control points (sites) with available data in a region for use in spatial interpolation to estimate missing data in space and time is always a challenge. The numerical weights derived through spatial and temporal interpolation approaches attached to data available at different sites have an impact of the estimation of missing data. Parsimonious and robust interpolation models can be developed using schemes that objectively select optimal number of sites and methodologies that eliminate redundant sites and regulate the weights. In this study regularization schemes, mathematical programming model formulations and different feature selection methods used in machine learning field are developed and evaluated for optimal and objective selection of sites for estimation of missing precipitation records. Variants of regularization schemes such as ridge regression, lease absolute shrinkage selection operator (LASSO) and elastic net are experimented. Mixed integer nonlinear optimization programming (MINLP) models with binary variables and multiple feature selection methods are adopted in this work. A case study using precipitation data at several rain gauges in a temperate climatic region of Kentucky, USA is used to demonstrate the benefits of using regularization schemes and optimization with binary variables to select an optimal subset of control points. Results point to improved estimations when these approaches are used for estimation of missing precipitation data.

How to cite: Teegavarapu, R.: Objective and Optimal Spatial Interpolation Approaches for Imputing Missing Precipitation Records, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13245, https://doi.org/10.5194/egusphere-egu25-13245, 2025.

Localized and Intense Rainfall Events (LIREs) can cause significant societal and economic damages. Typically developing over very small spatial and temporal scales, the accurate characterization and forecasting of such events remains, however, particularly challenging. By collecting distributed space-time observations, weather radars can provide useful information for the analysis of such events. In this study we take advantage of the experimental high-resolution radar data from the MeteoSwiss operational radar network available at 83 m radial resolution, every 5 minutes, over 20 different elevations to analyze the small-scale spatial variability associated with the extreme Lausanne LIRE of June 11^th, 2018, leading to the largest ever recorded 10-min rain gauge accumulation in Switzerland (41 mm). First, investigating the large-scale processes associated with this extreme event, a synoptic and dynamic analysis was conducted. This revealed the presence of a moderately unstable maritime tropical airmass which aided in the formation of a multicell thunderstorm which produced a wet microburst right over the city of Lausanne pouring an enormous quantity of water over very small spatial and temporal scales and leading to considerable localized flood and wind damage. Then, relying on the high-resolution radar data, the variability at small scale was measured by comparing rain rate values derived at different resolutions. More specifically, starting from the 83 m radar data, different existing hydrometeor-specific Z-R / Z-S relationships were used to compute an equivalent rain rate value at the gate level. Those were then compared against the corresponding rain rate values integrated at coarser radial resolutions of 500 m and 1000 m, and the difference across resolutions was derived as an indicator of small-scale spatial variability. With 1.5%, 0.41% and 0.18% of the total extracted and pre-processed gate volume showing differences larger than 25, 50 and 75 mm/hr between the 83 m and the 500 m data, a few but extreme small-scale rainfall variability peaks were observed, mostly associated with intensity peaks. Although most of these peaks were located above or within the melting layer, several of them were still observed below the melting layer, at proximity to the ground, and potentially decisive for hydrological applications. Converting this 3D information into 2D maps of sub-grid variability, a significant variability at the 5 min / 1km² resolution was observed highlighting not only the highly dynamic evolution of this event but also and the added value of high-resolution radar data to capture small-scale peaks associated with this extreme LIRE. By providing complementary insights on rainfall variability peaks, the retrieved sub-grid information can help improve the characterization of LIRE and enrich existing rainfall products.

How to cite: Liernur, A., Peyraud, L., Gabella, M., Germann, U., and Berne, A.: Small-scale spatial rainfall variability during the extreme convective rain event of June 11th, 2018, over the city of Lausanne, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-15743, https://doi.org/10.5194/egusphere-egu25-15743, 2025.

High resolution rainfall data are essential to quantify small scale and fast hydrological processes. The objective of the paper is to determine temporal variability and spatial patterns of precipitation statistic of one-minute resolution rainfall across Germany. The German Weather Service (DWD) started in 1993 to deploy rain gauges that achieve 1 minute temporal and 0.01 mm volumetric resolution by combining tipping buckets with weigthing (rain[e]H3 by LAMBRECHT meteo GmbH and OTT Pluvio by OTT Hydromet). 345 of those stations all over Germany have data with more than 10 years. For each station empirical cumulative distribution functions (eCDF) of precipitation intensity and dry periods were derived. Data were then aggregated to lower resolutions ranging from 2 min to 4 months. For all aggregation levels we fitted power law, log-normal and Weibull distribution functions and compared the goodness of fit. To determine spatial correlations between stations we extracted intensity and dry period duration at a given frequency from the empirical distribution function and applied a correlation analysis with station longitude, latitude, elevation and total rainfall. Annual and diurnal variations were analysed by fitting a power law to a moving window of data. A 60d segment of the yearly cycle (combining data of all years) and a 4h segment of the daily cycle (combining data of all days) were used. Similar the dependence of the power-law coefficient on temperature was analysed with a moving window of 2.5K width, shifted between -10 to 30°C.

We show that rainfall intensity measured at 1 minute resolution shows a distinct power-law distribution for all stations. The dry period durations instead are not purely power-law distributed. When aggregated, the distribution of the data transitions to lognormal distribution at 15 min aggregation level and to a Weibull distribution from 6 hours onwards. This has significant implication for estimating flood risk and deriving design storm properties as each temporal resolution requires a different statistical distribution to be fitted. We conclude that the mixing of the intensity and dry-period statistic creates this effect. While total rainfall in Germany clearly varies, with high totals in the north-west and lower values in the east, the intensity distribution does not reflect that. We find no significant correlation with longitude, latitude, elevation nor total station rainfall. But the dry-period statistic correlates well. This leads to the conclusion that rainfall intensity statistic is very similar in all of Germany and the difference in recurrence intervals and total rainfall is mostly defined by the dry periods between rain events. The power-law exponent varies annually with a sine curve from -1 to -2 in phase with the annual temperature cycle. It also shows a clear diurnal cycle. It can be expected that those cycles are driven by a strong dependence on temperature. The power-law exponent is close to -3 at 0°C and -1 at 25°C, creating higher intensities at higher temperatures.

How to cite: Frechen, N. and Hinz, C.: One-minute rainfall data reveal temperature dependend seasonal and diurnal variability of the power-law distribution for Germany, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-6463, https://doi.org/10.5194/egusphere-egu25-6463, 2025.

Rainfall is known to exhibit extreme variability over wide range of space and time scale, which makes it challenging to characterize, model and even measure. Rainfall measurement devices have observation scales very different from one another ranging from roughly 20 cm in space and few tens seconds (or few minutes) in time for punctual measurement such as disdrometers (or rain gauge), to few hundreds meters in space and few minutes in time for operational weather radars, and up to few kilometres in space and few tens of minutes for satellite data. This very significant observation scale gap between these devices creates a challenge in the comparion simply because of the intrinsic variability of rainfall, even without considering instrumental biases associated to each device.

This work focuses on the impact of the intrinsic rainfall variability on the comparison between punctual (disdrometer or rain gauge) and weather radar rainfall measurement. In order to achieve this, the physically based and mathematically robust framework of Universal Multifractals will be used. It relies on the assumption that rainfall is generated through an underlying multiplicative process. In such framework, the rain rate field can be written as the resolution (defined as the ratio between the outer scale of the phenomenon and the observation scale) to the power of a singularity. This singularity is preserved through scales.

Rainfall data collected in UK and Taiwan are used. These include high-resolution radar composite products and ground gauge records. In the UK, C-band radar composite, Nimrod, at 5-min and 1-km resolutions is used to compare with 1-min rainfall records derived from tipping bucket gauge records, while, in Taiwan, S-band radar composite, QPESUM, at 10-min and 1-km resolutions is used to compare with 10-second disdrometer rainfall records.

The concept of singularity is used to suggest an innovative comparison approach between rainfall measurement devices. More precisely, the local singularity along with the associated uncertainty is assessed using radar data on the range of available space time scales and then compared with the one of disdrometer or rain gauge accounting for the ratio between the observation scales. Results and interpretation of this novel comparison method on the available data will be discussed.

Authors acknowledge the France-Taiwan Ra2DW project (supported by the French National Research Agency – ANR-23-CE01-0019-01 and Taiwan’s National Science and Technology Council – 113-2923-M-002-001-MY4) for partial financial support.

How to cite: Gires, A. and Wang, L.-P.: Multifractal singularity to bridge the scale gap between various rainfall measurement devices, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13715, https://doi.org/10.5194/egusphere-egu25-13715, 2025.