Search

Accurate and precise KDP estimates are essential for radar-based applications, especially in quantitative precipitation estimation and radar data quality control routines. The accuracy of these estimates largely depends on the post-processing of the radar's measured Φ DP , which aims to reduce noise and backscattering effects while preserving fine-scale precipitation features. In this study, we evaluate the performance of several publicly available KDP estimation methods implemented in open-source libraries such as PyArt and Wradlib, and the method used in the Vaisala weather radars. To benchmark these methods, we employ a polarimetric self-consistency approach that relates KDP to reflectivity and differential reflectivity in rain, providing a reference self-consistency KDP (K DPSC) for comparison. This approach allows for the construction of the reference KDP observations that can be used to assess the accuracy and robustness of the studied KDP estimation methods. We assess each method by quantifying uncertainties using C-band weather radar observations where the reflectivity values ranged between 20 and 50 dBZ. Using the proposed evaluation framework we could define optimized parameter settings for the methods that have user-configurable parameters. Most of such methods showed significant reduction in the estimation errors after the optimization with respect to the default settings. We have found significant differences in the performances of the studied methods, where the best performing methods showed smaller normalized biases in the high reflectivity values (i.e., ≥ 40 dBZ) and overall smaller normalized root mean squared errors across the range of reflectivity values. [ABSTRACT FROM AUTHOR]

Because of their high sensitivity to hydrometeors and their high vertical resolutions, space-borne radar observations are emerging as an undeniable asset for Numerical Weather Prediction (NWP) applications. The EUMETSAT (European Organisation for the Exploitation of Meteorological Satellites) NWP SAF (Satellite Application Facility) released an active sensor module within version 13 of the RTTOV (Radiative Transfer for TOVS) software with the goal of simulating both active and passive microwave instruments within a single framework using the same radiative transfer assumptions. This study provides an in-depth description of the radar simulator available within this software. In addition, this study proposes a revised version of the existing melting layer parametrization scheme of Bauer (2001) within the RTTOV-SCATT v13.1 model to provide a better fit to observations below the freezing level. Simulations are performed with and without melting layer schemes for the Dual precipitation radar (DPR) instrument onboard GPM using the ARPEGE (Action de Recherche Petite Echelle Grande Echelle) global NWP model running operationally at Météo-France for two different one-month periods (June, 2020 and January, 2021). Results for a case study over the Atlantic ocean show that the revised melting scheme produces more realistic simulations as compared to the default scheme both at Ku (13.5 GHz) and Ka (35.5 GHz) frequencies and these simulations are much closer to observations. A statistical assessment using more samples show significant improvement of the first-guess departure statistics with the revised scheme compared to the existing melting scheme. As a step further, this study showcases the use of melting layer simulations for the classification of precipitation (stratiform, convective and transition) using the Dual Frequency Ratio algorithm (DFR). The classification results also reveal a significant overestimation of the rain reflectivities in all hemispheres, which can either be due to a tendency of the ARPEGE model to produce a too large amount of convective precipitation, or to a mis-representation of the convective precipitation fraction within the forward operator. [ABSTRACT FROM AUTHOR]

In recent years, the application of compact and cost-effective deployable X-band polarimetric radars has gained in popularity, particularly in regions with complex terrain. The deployable radars generally use a radome constructed by joining multiple panels using metallic threads to facilitate easy transportation. As a part of the ESPOIRS project, Laboratoire de l'Atmosphεave;re et des Cyclones has acquired an X-band meteorological radar with four panel radome configuration. In this study, we investigated the effect of the radome on the measured polarimetric variables, particularly differential reflectivity and differential phase. Our observations reveal that the metallic threads connecting the radome panels introduce power loss at vertical polarization, leading to a positive bias in the differential reflectivity values. To address the spatial variability bias observed in differential reflectivity and differential phase, we have developed a novel algorithm based on the Discrete Fourier Transform. The algorithm's performance was tested during an intense heavy rainfall event caused by the Batsirai cyclone on Reunion Island. The comparative and joint histogram analysis demonstrates the algorithm's effectiveness in correcting the spatial bias in the polarimetric variables. [ABSTRACT FROM AUTHOR]

Given that the Gaussianity of observation error distribution is the fundamental principle of some data assimilation and machine learning algorithms, the error structure of radar reflectivity becomes increasingly important with the development of high resolution forecasts and nowcasts of convective systems. This study examines the error distribution of radar reflectivity and discusses what give rise to the non-Gaussian error distribution by using 6 month observations minus backgrounds (OmBs) of composites of vertical maximum reflectivity (CVMRs) in mountainous and hilly areas. By following the symmetric error model in all-sky satellite radiance assimilation, we unveil the error structure of CVMRs as a function of symmetric rain rates, which is the average of observed and simulated rain rates. Unlike satellite radiance, the error structure of CVMRs shows a sharper slope in light precipitations than moderate precipitations. Thus, a three-piecewise fitting function is more suitable for CVMRs. The probability density functions of OmBs normalized by symmetric rain rates become more Gaussian in comparison with the probability density function normalized by the whole samples. Moreover, the possibility of using third-party predictor to construct the symmetric error model are also discussed in this study. The Gaussianity of OmBs can be further improved by using a more accurate precipitation observations. According to the Jensen-Shannon divergence, a more linear predictor, the logarithmic transformation of rain rate, can provide the most Gaussian error 20 distribution in comparison with other predictors. [ABSTRACT FROM AUTHOR]

Radar based microphysical retrievals of cloud and droplet properties are vital for informing model parameterisations of clouds and precipitation but these retrievals often do not capture the details of small droplets in light rain or drizzle. A state-of-the-art G-band radar is used here to demonstrate improvements to microphysical retrievals in a case study featuring light rain. Improvements are seen, as compared to W-band radar, in the retrieval of vertical wind speed, due to the location of Mie minima at smaller droplet sizes with the G-band radar. This, in turn, has an impact on the retrieval of the drop size distribution, allowing for better accuracy in the retrieval of the characteristic drop diameter and improvements in the retrieval of the number of concentration of small droplet sizes. The differential Doppler velocity between Ka- and G-bands shows increased dynamic range compared to the Ka-W pairing, particularly for instances presenting small characteristic drop diameters. The increased attenuation experienced at G-band enables improved retrievals of liquid water content and precipitation rate when paired with W-band or Ka-band as compared to the W-band and Ka- band pairing. This is particularly noticeable in periods of light rain where the W-band and Ka-band radars receive negligible attenuation while the attenuation at G-band is much greater. [ABSTRACT FROM AUTHOR]

Rain cells are the most common units in the natural precipitation system. Enhancing the understanding of these rain cell characteristics can significantly improve the cognition of the precipitation system. Previous studies have mostly analyzed rain cells from a single radar data. In this study, we merged the precipitation parameters measured by the Tropical Rainfall Measuring Mission (TRMM) precipitation radar (PR) with the muti-channel cloud-top radiance measured by the visible and infrared scanner (VIRS) and the muti-channel brightness temperature measured by the TRMM microwave imager (TMI). The rain cells were identified within the PR orbit, and the swath truncation effect was eliminated. We used two methods for rain cell identification: the minimum bounding rectangle (MBR) method and the best fit ellipse (BFE) method, and compared the differences between these two methods in describing the rain cell characteristics. The results indicate that both methods can better reflect the geometric characteristics of rain cells. Compared with the MBR method, the BFE method can obtain a smaller rain cell area, and the filling ratio is better. However, the MBR method can simplify the data storage volume. Consequently, we employed the MBR method to analyze the precipitation structure of two typical rain cell precipitation cases. The results show that the new rain cell dataset can be used for the analysis of rain cell precipitation parameters and visible/infrared and microwave signals, which provides valuable data for comprehensive studies on the rain cell structural characteristics and furthers the understanding of precipitation mechanisms. The data which were used in this paper are freely available at https://doi.org/10.5281/zenodo.8352587 (Wu et al., 2023). [ABSTRACT FROM AUTHOR]

Weather radar is a crucial tool in rainfall estimation, providing high-resolution estimates in both space and time. Despite this, radar rainfall estimates are subject to many error sources – including attenuation, ground clutter, beam blockage and the drop-size distribution – with the true rainfall field unknown. A flexible stochastic model for simulating errors relating to the radar rainfall estimation process is implemented, inverting standard weather radar processing methods, imposing path-integrated attenuation effects, a stochastic drop-size distribution field, along with sampling and random errors. This can provide realistic weather radar images, of which we know the true rainfall field, and the corrected 'best guess' rainfall field which would be obtained if they were observed in the real-world case. The structure of these errors is then investigated, with a focus on frequency and behaviour of 'rainfall shadows'. Half of simulated weather radar images have at least 3 % of significant rainfall rates shadowed and 25 % had at least 45 km2 containing rainfall shadows, resulting in underestimation of potential impacts of flooding. A model framework for investigating the behaviour of errors relating to the radar rainfall estimation process is demonstrated, with the flexible and efficient tool performing well at generating realistic weather radar images visually, for a large range of event types. [ABSTRACT FROM AUTHOR]

'Train effect' is defined as a cloud system in which several convective cells pass over the same place in a short time. Trains produce large amount of rainfall, frequently leading to flash floods, reported mainly over North America during spring and summer. Thirty train events were identified, using radar images, calibrated by rain-gauges, for four winters, all associated with Cyprus Lows (CL). The dynamic factors responsible for their formation in Israel were examined, utilizing the ECMWF Integrated Forecast System of 0.1° resolution. Seventeen out of the 30 events share common features. Each one was found at the cold sector in the southern periphery of a CL at its occluded stage, and located in the left flank of a maximum wind belt, where cyclonic shear vorticity exists. The trains cross the Israeli coast near 32.2°N, with a mean length of 45 km, last 1-3 hrs, and yield ~35 mm rainfall. The maximum wind belts right of the trains were found to delineate the limit of the precipitative region of the CLs. Unlike classical trains, activated by thermal or frontal forcing, the EM trains that develop in cold air-mass, can be entitled 'cold trains', rather than the classical 'warm trains'. [ABSTRACT FROM AUTHOR]

This study assimilates for the first time polarimetric C-band radar observations from the German meteorological service (DWD) into DWD's convective-scale model ICON-D2 using DWD's ensemble-based KENDA assimilation framework. We compare the assimilation of conventional observations (CNV) with the additional assimilation of radar reflectivity Z (CNV+Z), with the additional assimilation of liquid or ice water content (LWC or IWC) estimates below or above the melting layer instead of Z where available (CNV+LWC/Z or CNV+IWC/Z, respectively). Hourly quantitative precipitation forecasts (QPF) are evaluated for two stratiform and one convective rainfall event in the summers of 2017 and 2021. With optimized data assimilation settings (e.g., observation errors), the assimilation of LWC mostly improves first guess QPF compared to the assimilation of Z alone (CNV+Z), while the assimilation of IWC does not, especially for convective cases, probably because of the lower quality of the IWC retrieval in these situations. Improvements are, however, notable for stratiform rainfall in 2021, for which the IWC estimator profits from better specific differential phase estimates due to a higher radial radar resolution compared to the other cases. The assimilation of all radar data sets together (CNV+LWC+IWC+Z) yields the best first guesses. All assimilation configurations with radar information consistently improve deterministic nine-hour QPF compared to the assimilation of only conventional data (CNV). Forecasts based on the assimilation of LWC and IWC retrievals on average slightly improve FSS and FBI compared to the assimilation of Z alone (CNV+Z), especially when LWC is assimilated for the 2017 convective case and when IWC is assimilated for the high-resolution 2021 stratiform case. However, IWC assimilation again degrades forecast FSS for the convective cases. Forecasts initiated using all radar data sets together (CNV+LWC+IWC+Z) yield the best FSS. The development of IWC retrievals more adequat for convection constitutes one next step to further improve the exploitation of ice microphysical retrievals for radar data assimilation. [ABSTRACT FROM AUTHOR]

The European climatological high-resolution gauge-adjusted radar precipitation dataset, EURADCLIM, addresses the need for an accurate (sub)daily precipitation product covering 78 % of Europe at a high spatial resolution. A climatological dataset of 1 and 24 h precipitation accumulations on a 2 km grid is derived for the period 2013 through 2020. The starting point is the European Meteorological Network (EUMETNET) Operational Program on the Exchange of Weather Radar Information (OPERA) gridded radar dataset of 15 min instantaneous surface rain rates, which is based on data from, on average, 138 ground-based weather radars. First, methods are applied to further remove non-meteorological echoes from these composites by applying two statistical methods and a satellite-based cloud-type mask. Second, the radar composites are merged with the European Climate Assessment & Dataset (ECA&D) with potentially ∼ 7700 rain gauges from National Meteorological and Hydrological Services (NMHSs) in order to substantially improve its quality. Characteristics of the radar, rain gauge and satellite datasets are presented, as well as a detailed account of the applied algorithms. The clutter-removal algorithms are effective while removing few precipitation echoes. The usefulness of EURADCLIM for quantitative precipitation estimation (QPE) is confirmed by comparison against rain gauge accumulations employing scatter density plots, statistical metrics and a spatial verification. These show a strong improvement with respect to the original OPERA product. The potential of EURADCLIM to derive pan-European precipitation climatologies and to evaluate extreme precipitation events is shown. Specific attention is given to the remaining artifacts in and limitations of EURADCLIM. Finally, it is recommended to further improve EURADCLIM by applying algorithms to 3D instead of 2D radar data and by obtaining more rain gauge data for the radar–gauge merging. The EURADCLIM 1 and 24 h precipitation datasets are publicly available at 10.21944/7ypj-wn68 and 10.21944/1a54-gg96. [ABSTRACT FROM AUTHOR]

Raindrop size distributions (DSDs) play a crucial role in quantitative rainfall estimation using weather radar. Thanks to dual polarization capabilities, crucial information about the DSD in a given volume of air can be retrieved. One popular retrieval method assumes that the DSD can be modeled by a constrained gamma distribution in which the shape (μ) and rate (Λ) parameters are linked together by a deterministic relationship. In the literature, μ – Λ relationships are often taken for granted and applied without much critical discussion. In this study, we take another look at this important issue by conducting a detailed analysis of μ – Λ relations in stratiform rain and quantifying the accuracy of the associated DSD retrievals. Crucial aspects of our research include the sensitivity of μ – Λ relations to the temporal aggregation scale, drop concentration, inter-event variability, and adequacy of the gamma distribution model. Our results show that μ – Λ relationships in stratiform rain are surprisingly robust to the choice of the sampling resolution, sample size, and adequacy of the gamma model. Overall, the retrieved DSDs are in a rather decent agreement with ground observations (correlation coefficient of 0.57 and 0.74 for μ and Dm). The main sources of errors and uncertainty during the retrievals are calibration offsets in reflectivity (Zhh) and differential reflectivity (Zdr). Measurement noise and differences in scale between radars and disdrometers also play a minor role. The raindrop concentration (NT) remains the most difficult parameter to retrieve, which can be off by several orders of magnitude. After careful data filtering and removal of problematic Zhh/Zdr pairs, the correlation coefficient for the retrieved NT values remained low, only slightly increasing from 0.12 into 0.24. [ABSTRACT FROM AUTHOR]

The European climatological high-resolution gauge-adjusted radar precipitation dataset, EURADCLIM, addresses the need for an accurate (sub-)daily precipitation product covering 78% of Europe at high spatial resolution. A climatological dataset of 1-h and 24-h precipitation accumulations on a 2-km grid is derived for the period 2013 through 2020. The starting point is the European Meteorological Network (EUMETNET) Operational Program on the Exchange of weather RAdar Information (OPERA) gridded radar dataset of 15-min instantaneous surface rain rates, which is based on data from, on average, 138 ground-based weather radars. First, methods are applied to further remove non-meteorological echoes from these composites by applying two statistical methods and a satellite-based cloud type mask. Second, the radar composites are merged with the European Climate Assessment & Dataset (ECA&D) with potentially ~7700 rain gauges from National Meteorological and Hydrological Services (NMHS) in order to substantially improve its quality. Characteristics of the radar, rain gauge and satellite datasets are presented, as well as a detailed account of the applied algorithms. The clutter removal algorithms are effective, while removing few precipitation echoes. The usefulness of EURADCLIM for quantitative precipitation estimation (QPE) is confirmed by comparing against rain gauge accumulations employing scatter density plots, statistical metrics, and a spatial verification. These show a strong improvement with respect to the original OPERA product. The potential of EURADCLIM to derive pan-European precipitation climatologies and to evaluate extreme precipitation events is shown. Specific attention is given to remaining artefacts in and limitations of EURADCLIM. Finally, it is recommended to further improve EURADCLIM by applying algorithms to 3D instead of 2D radar data, and by obtaining more rain gauge data for the radar-gauge merging. The EURADCLIM 1-h and 24-h precipitation datasets are publicly available at https://doi.org/10.21944/7ypj-wn68 (Overeem et al., 2022a) and https://doi.org/10.21944/1a54-gg96 (Overeem et al., 2022b). [ABSTRACT FROM AUTHOR]

The U.S. Department of Energy Atmospheric Radiation Measurement program Tropical Western Pacific site hosted a C-band polarization (CPOL) radar in Darwin, Australia. It provides 2 decades of tropical rainfall characteristics useful for validating global circulation models. Rainfall retrievals from radar assume characteristics about the droplet size distribution (DSD) that vary significantly. To minimize the uncertainty associated with DSD variability, new radar rainfall techniques use dual polarization and specific attenuation estimates. This study challenges the applicability of several specific attenuation and dual-polarization-based rainfall estimators in tropical settings using a 4-year archive of Darwin disdrometer datasets in conjunction with CPOL observations. This assessment is based on three metrics: statistical uncertainty estimates, principal component analysis (PCA), and comparisons of various retrievals from CPOL data. The PCA shows that the variability in R can be consistently attributed to reflectivity, but dependence on dual-polarization quantities was wavelength dependent for 110mmh-1. Rainfall estimates during these conditions primarily originate from deep convective clouds with median drop diameters greater than 1.5 mm. An uncertainty analysis and intercomparison with CPOL show that a Colorado State University blended technique for tropical oceans, with modified estimators developed from video disdrometer observations, is most appropriate for use in all cases, such as when 110mmh-1 (deeper convective rain). [ABSTRACT FROM AUTHOR]

A new method to determine the melting layer height using a micro rain radar (MRR) is presented. The MRR is a small vertically pointing frequency-modulated continuous-wave radar that measures Doppler spectra of precipitation. From these Doppler spectra, various variables such as Doppler velocity or spectral width can be derived. The melting layer is visible due to higher reflectivity and an acceleration of the falling particles, among others. These characteristics are fed to a neural network to determine the melting layer height. To train the neural network, the melting layer height is determined manually. The neural network is trained and tested using data from two sites that cover all seasons. For most cases, the neural network is able to detect the correct melting layer height well. Sensitivity studies show that the neural network is able to handle different MRR settings. Comparisons to radiosonde data and cloud radar data show a good agreement with respect to the melting layer heights. [ABSTRACT FROM AUTHOR]

This study presents two methods for evaluating the reflectivity calibration of W-band cloud radars. Both methods use natural rain as a reference target. The first approach is based on a self-consistency method of polarimetric radar variables, which is widely used in the precipitation radar community. As previous studies pointed out, the method cannot be directly applied to higher frequencies where non-Rayleigh scattering effects and attenuation have a nonnegligible influence on radar variables. The method presented here solves this problem by using polarimetric Doppler spectra to separate backscattering and propagational effects. New fits between the separated radar variables allow one to estimate the absolute radar calibration using a minimization technique. The main advantage of the self-consistency method is its lower dependence on the spatial variability in radar drop size distribution (DSD). The estimated uncertainty of the method is ±0.7 dB. The method was applied to three intense precipitation events, and the retrieved reflectivity offsets were within the estimated uncertainty range. The second method is an improvement on the conventional disdrometer-based approach, where reflectivity from the lowest range gate is compared to simulated reflectivity using surface disdrometer observations. The improved method corrects, first, for the time lag between surface DSD observations and the radar measurements at a certain range. In addition, the effect of evaporation of raindrops on their way towards the surface is mitigated. The disdrometer-based method was applied to 12 rain events observed by vertically pointed W-band radar and showed repeatable estimates of the reflectivity offsets at rain rates below 4 mm h -1 within ±0.9 dB. The proposed approaches can analogously be extended to Ka-band radars. Although very different in terms of complexity, both methods extend existing radar calibration evaluation approaches, which are inevitably needed for the growing cloud radar networks in order to provide high-quality radar observation to the atmospheric community. [ABSTRACT FROM AUTHOR]

Accurate estimation of the Freezing Level (FL) is essential in radar rainfall estimation to mitigate the bright band enhancement, to classify hydrometeors, to correct for rain-attenuation and to calibrate radar measurements. Here we present a novel and robust FL estimation algorithm that can be applied to either Vertical Profiles (VPs) or Quasi-Vertical Profiles (QVPs) built from operational polarimetric weather radar scans. The algorithm depends only on data collected by the radar itself, and it is based on the detection of strong gradients within the profiles and relies on the combination of several polarimetric variables. The VPs and QVPs of ZH showed a good similarity in the profiles (r ≈ 0.7) even though the QVPs are built from low-elevation angles. The algorithm is applied to one year of rainfall events and validated using measured FLs from radiosonde data. The results demonstrated that combining the profiles of ZH, ΡHV and the gradient of the velocity V showed the best FL estimation performance when using VPs, whereas combining the profiles of ZH, ΡHV and ZDR showed the best FL estimation performance when using QVPs. The VP computed from the gradient of the velocity showed to be extremely valuable in the FL estimation when using VPs. The errors in the FL estimation using either VPs or QVPs are within 250 m. [ABSTRACT FROM AUTHOR]

In stratiform rainfall, the melting layer (ML) is often visible in radar observations as an enhanced reflectivity band, the so-called bright band. Despite the ongoing debate on the exact microphysical processes taking place in the ML and on how they translate into radar measurements, both model simulations and observations indicate that the radar-measured ML properties are influenced by snow microphysical processes that take place above it. There is still, however, a lack of comprehensive observations to link the two. To advance our knowledge of precipitation formation in ice clouds and provide new insights into radar signatures of snow growth processes, we have investigated this link. This study is divided into two parts. Firstly, surface-based snowfall measurements are used to develop a new method for identifying rimed and unrimed snow from X- and Ka-band Doppler radar observations. Secondly, this classification is used in combination with multifrequency and dual-polarization radar observations collected during the Biogenic Aerosols – Effects on Clouds and Climate (BAECC) experiment in 2014 to investigate the impact of precipitation intensity, aggregation, riming and dendritic growth on the ML properties. The results show that the radar-observed ML properties are highly related to the precipitation intensity. The previously reported bright band "sagging" is mainly connected to the increase in precipitation intensity. Ice particle riming plays a secondary role. In moderate to heavy rainfall, riming may cause additional bright band sagging, while in light precipitation the sagging is associated with unrimed snow. The correlation between ML properties and dual-polarization radar signatures in the snow region above appears to be arising through the connection of the radar signatures and ML properties to the precipitation intensity. In addition to advancing our knowledge of the link between ML properties and snow processes, the presented analysis demonstrates how multifrequency Doppler radar observations can be used to get a more detailed view of cloud processes and establish a link to precipitation formation. [ABSTRACT FROM AUTHOR]

We investigate the depth-duration relationship of maximum rainfall over the whole of Germany based on 16 yrs of radar derived Quantitative Precipitation Estimates (namely, RADKLIM-YW, German Meteorological Service) with a space-time resolution of 1 km and 5 min. Contrary to the long-term historic records that identified a smooth power law scaling behavior between the maximum rainfall depth and duration, our analysis revealed three distinct scaling regimes of which boundaries are approximately 1.5 h and 1 d. Few extraordinary events dominate a wide range of durations and deviate to the usual power law. Furthermore, the shape of the depth-duration relationship varies with the sample size of randomly selected radar pixels. A smooth scaling behavior were identified when the sample size is small (e.g. 10 to 100), but the original three distinct scaling regimes became more apparent as the sample size increases (e.g. 1000 to 10 000). Lastly, a pixel wise classification of the depth-duration relationship of the maximum rainfall at all individual pixels in Germany revealed three distinguishable types of scaling behavior, clearly determined by the temporal structure of the extreme rainfall events at a pixel. Thus, the relationship might change with longer time series and can be improved once available. [ABSTRACT FROM AUTHOR]

The U.S. Department of Energy Atmospheric Radiation Measurement program Tropical Western Pacific site hosted a C-band POLarization (CPOL) radar in Darwin, Australia. It provides two decades of tropical rainfall characteristics useful for validating global circulation models. Rainfall retrievals from radar assume characteristics about the droplet size distribution (DSD) that vary significantly. To minimize the uncertainty associated with DSD variability, new radar rainfall techniques use dual polarization and specific attenuation estimates. This study challenges the applicability of several specific attenuation and dual-polarization based rainfall estimators in tropical settings using a 4-year archive of Darwin disdrometer datasets in conjunction with CPOL observations. This assessment is based on three metrics: statistical uncertainty estimates, principal component analysis (PCA), and comparisons of various retrievals from CPOL data. The PCA shows that over 99 % of the variability in estimated rainfall rate R can be explained by radar reflectivity factor for rainfall rates 1 < R <10 mm hr-1. These rates primarily originate from stratiform clouds and weak convection (median drop diameters less than 1.5 mm). The dual-polarization specific differential phase increases in usefulness for rainfall estimators in times with R > 10 mm hr-1. Rainfall estimates during these conditions primarily originate from deep convective clouds with median drop diameters greater than 1.5 mm. Using specific attenuation for estimating R generally does not provide additional skill beyond other metrics for Darwin. An uncertainty analysis and intercomparison with CPOL show that a CSU-blended technique for tropical oceans, with modified estimators developed from VDIS observations, is most appropriate for use in all cases, such as when 1 < R < 10 mm hr-1 (stratiform rain), and when R > 10 mm hr-1 (deeper convective rain). [ABSTRACT FROM AUTHOR]

Weather radar has become an invaluable tool for monitoring rainfall and studying its link to hydrological response. However, when it comes to accurately measuring small-scale rainfall extremes responsible for urban flooding, many challenges remain. The most important of them is that radar tends to underestimate rainfall compared to gauges. The hope is that by measuring at higher resolutions and making use of dual-polarization radar, these mismatches can be reduced. Each country has developed its own strategy for addressing this issue. However, since there is no common benchmark, improvements are hard to quantify objectively. This study sheds new light on current performances by conducting a multinational assessment of radar's ability to capture heavy rain events at scales of 5 min up to 2 h. The work is performed within the context of the joint experiment framework of project MUFFIN (Multiscale Urban Flood Forecasting), which aims at better understanding the link between rainfall and urban pluvial flooding across scales. In total, six different radar products in Denmark, the Netherlands, Finland and Sweden were considered. The top 50 events in a 10-year database of radar data were used to quantify the overall agreement between radar and gauges as well as the bias affecting the peaks. Results show that the overall agreement in heavy rain is fair (correlation coefficient 0.7–0.9), with apparent multiplicative biases on the order of 1.2–1.8 (17 %–44 % underestimation). However, after taking into account the different sampling volumes of radar and gauges, actual biases could be as low as 10 %. Differences in sampling volumes between radar and gauges play an important role in explaining the bias but are hard to quantify precisely due to the many post-processing steps applied to radar. Despite being adjusted for bias by gauges, five out of six radar products still exhibited a clear conditional bias, with intensities of about 1 %–2 % per mmh -1. As a result, peak rainfall intensities were severely underestimated (factor 1.8–3.0 or 44 %–67 %). The most likely reason for this is the use of a fixed Z – R relationship when estimating rainfall rates (R) from reflectivity (Z), which fails to account for natural variations in raindrop size distribution with intensity. Based on our findings, the easiest way to mitigate the bias in times of heavy rain is to perform frequent (e.g., hourly) bias adjustments with the help of rain gauges, as demonstrated by the Dutch C-band product. An even more promising strategy that does not require any gauge adjustments is to estimate rainfall rates using a combination of reflectivity (Z) and differential phase shift (Kdp), as done in the Finnish OSAPOL product. Both approaches lead to approximately similar performances, with an average bias (at 10 min resolution) of about 30 % and a peak intensity bias of about 45 %. [ABSTRACT FROM AUTHOR]

Raindrop size distribution (DSD) information is fundamental in understanding the precipitation microphysics and quantitative precipitation estimation, especially in complex terrain or urban environments which are known for complicated rainfall mechanism and high spatial and temporal variability. In this study, the DSD characteristics of rainy seasons in the Beijing urban area are extensively investigated using 5-year DSD observations from a Parsivel 2 disdrometer located at Tsinghua University. The results show that the DSD samples with rain rate < 1 mm h -1 account for more than half of total observations. The mean values of the normalized intercept parameter (log⁡10Nw) and the mass-weighted mean diameter (Dm) of convective rain are higher than that of stratiform rain, and there is a clear boundary between the two types of rain in terms of the scattergram of log⁡10Nw versus Dm. The convective rain in Beijing is neither continental nor maritime, owing to the particular location and local topography. As the rainfall intensity increases, the DSD spectra become higher and wider, but they still have peaks around diameter D∼0.5 mm. The midsize drops contribute most towards accumulated rainwater. The Dm and log⁡10Nw values exhibit a diurnal cycle and an annual cycle. In addition, at the stage characterized by an abrupt rise of urban heat island (UHI) intensity as well as the stage of strong UHI intensity during the day, DSD shows higher Dm values and lower log⁡10Nw values. The localized radar reflectivity (Z) and rain rate (R) relations (Z=aRb) show substantial differences compared to the commonly used NEXRAD relationships, and the polarimetric radar algorithms R(Kdp) , R(Kdp , ZDR) , and R(ZH , ZDR) show greater potential for rainfall estimation. [ABSTRACT FROM AUTHOR]

In this paper, we study the impact of lightning and radar reflectivity factor data assimilation on the precipitation VSF (very short-term forecast, 3 h in this study) for two severe weather events that occurred in Italy. The first case refers to a moderate and localized rainfall over central Italy that occurred on 16 September 2017. The second case occurred on 9 and 10 September 2017 and was very intense and caused damages in several geographical areas, especially in Livorno (Tuscany) where nine people died. The first case study was missed by several operational forecasts, including that performed by the model used in this paper, while the Livorno case was partially predicted by operational models. We use the RAMS@ISAC model (Regional Atmospheric Modelling System at Institute for Atmospheric Sciences and Climate of the Italian National Research Council), whose 3D-Var extension to the assimilation of radar reflectivity factor is shown in this paper for the first time. Results for the two cases show that the assimilation of lightning and radar reflectivity factor, especially when used together, have a significant and positive impact on the precipitation forecast. For specific time intervals, the data assimilation is of practical importance for civil protection purposes because it changes a missed forecast of intense precipitation (≥40 mm in 3 h) to a correct one. While there is an improvement of the rainfall VSF thanks to the lightning and radar reflectivity factor data assimilation, its usefulness is partially reduced by the increase in false alarms, especially when both datasets are assimilated. [ABSTRACT FROM AUTHOR]

Tropical Rainfall Measuring Mission (TRMM) precipitation radar measurements are used to examine the variation in vertical structure of precipitation with sea surface temperature (SST) over the Arabian Sea (AS) and Bay of Bengal (BOB). The variation in reflectivity and precipitation echo top with SST is remarkable over the AS but small over the BOB. The reflectivity increases with SST (from 26 to 31 ∘ C) by ∼1 and 4 dBZ above and below 6 km, respectively, over the AS, while its variation is <0.5 dBZ over the BOB. The transition from shallow storms at lower SSTs (≤27 ∘ C) to deeper storms at higher SSTs is strongly associated with the decrease in stability and mid-tropospheric wind shear over the AS. In contrary, the storms are deeper at all SSTs over the BOB due to weaker stability and mid-tropospheric wind shear. At lower SSTs, the observed high aerosol optical depth (AOD) and low total column water (TCW) over AS results in the small cloud effective radius (CER) and weaker reflectivity. As SST increases, AOD decreases and TCW increases, leading to a large CER and high reflectivity. The changes in these parameters with SST are marginal over the BOB and hence the CER and reflectivity. The predominance of collision–coalescence process below the bright band is responsible for the observed negative slopes in the reflectivity over both the seas. The observed variations in reflectivity originate at the cloud formation stage over both the seas, and these variations are magnified during the descent of hydrometeors to the ground. [ABSTRACT FROM AUTHOR]

Cloud Feedback Model Intercomparison Project Observational Simulator Package (COSP) has been widely used to diagnose model performance and physical processes via an apple-to-apple comparison to satellite measurements. Although the COSP provides useful information about clouds and their climatic impact, outputs that have a subcolumn dimension require large amounts of data. This can cause a bottleneck when conducting sets of sensitivity experiments or multiple model intercomparisons. Here, we incorporate two diagnostics for warm-rain microphysical processes into COSP2, the latest version of the simulator. The approach used here employs existing diagnostic methodologies that probe how the warm-rain processes occur using statistics constructed from simulators of multiple satellite instruments along with their subcolumn information. The new diagnostics are designed to produce statistics online during the COSP execution, eliminating the need to output subcolumn variables. Users can also readily conduct regional analysis tailored to their particular research interest (e.g., land-ocean differences), using an auxiliary post-process package after the COSP calculation. This inline tool also generates global maps of the occurrence frequency of warm-rain regimes (i.e., non-precipitating, drizzling, and precipitating) classified according to CloudSat radar reflectivity, putting the warm-rain process diagnostics into the context of geographical distributions of precipitation. The inline diagnostics are applied to the MIROC6 GCM to demonstrate how known biases common among multiple GCMs relative to satellite observations are revealed. The inline multisensor diagnostics are intended to serve as a tool that facilitates process-oriented model evaluations in a manner that reduces the burden on modelers for their diagnostics effort. [ABSTRACT FROM AUTHOR]

Erosive rainfall varies pronouncedly in time and space. Severe events are often restricted to a few square kilometers. Radar rain data with high spatiotemporal resolution enable this pattern of erosivity to be portrayed with high detail. We used radar data with a spatial resolution of 1 km 2 over 452 503 km 2 to derive a new erosivity map for Germany and to analyze the seasonal distribution of erosivity. The expected long-term regional pattern was extracted from the scattered pattern of events by several steps of smoothing. This included averaging erosivity from 2001 to 2017 and smoothing in time and space. The pattern of the resulting map was predominantly shaped by orography. It generally agrees well with the erosivity map currently used in Germany (Sauerborn map), which is based on regressions using rain gauge data (mainly from the 1960s to 1980s). In some regions the patterns of both maps deviate because the regressions of the Sauerborn map were weak. Most importantly, the new map shows that erosivity is about 66 % larger than in the Sauerborn map. This increase in erosivity was confirmed by long-term data from rain gauge stations that were used for the Sauerborn map and which are still in operation. The change was thus not caused by using a different methodology but by climate change since the 1970s. Furthermore, the seasonal distribution of erosivity shows a slight shift towards the winter period when soil cover by plants is usually poor. This shift in addition to the increase in erosivity may have caused an increase in erosion for many crops. For example, predicted soil erosion for winter wheat is now about 4 times larger than in the 1970s. These highly resolved topical erosivity data will thus have definite consequences for agricultural advisory services, landscape planning and even political decisions. [ABSTRACT FROM AUTHOR]

This study presents an empirical method for optimizing polarimetric variables in order to improve the accuracy of dual-polarization radar rainfall estimation using data derived from radars operated by different agencies. The empirical method was developed using the Yong-In Testbed (YIT) radar operated by the Korea Meteorological Administration (KMA). The method is based on the determination of relations between polarimetric variables. Relations for Z - ZDR and Z - KDP are derived from the measurements of a two-dimensional video disdrometer installed about 30 km away from the YIT radar. These relations were used to adjust the polarimetric variables of the dual-polarization constant altitude plan position indicator (CAPPI) at a height of 1.5 km. The CAPPI data with the adjusted polarimetric variables were used to estimate rainfalls using three different radar rainfall estimation algorithms. The first algorithm is based on Z, the second on Z and ZDR, and the third on Z, ZDR, and KDP. The accuracy of the radar-estimated rainfall was then assessed using raingauge observations. Three rainfall events with more than 40 mm of maximum hourly rainfall were shown to have the best estimation when the method using Z, ZDR, and KDP was used. However, stratiform precipitation events were better estimated by the algorithm using Z and ZDR. The method was also applied to the data of three radars that belong to KMA and the Ministry of Land, Infrastructure, and Transport. The evaluation was done for six months (May-October) in 2015. The results show an improvement in radar rainfall estimation accuracy for stratiform, frontal, and convective precipitation from approximately 50 % to 70 %. [ABSTRACT FROM AUTHOR]

Three methods for determining the reflectivity bias of single polarization radar using dual polarization radar reflectivity and disdrometer data (i.e., the equidistance line, overlapping area, and disdrometer methods) are proposed and evaluated for two low-pressure rainfall events that occurred over the Korean Peninsula on 25 August 2014 and 8 September 2012. Single polarization radar reflectivity was underestimated by more than 12 and 7 dB in the two rain events, respectively. All methods improved the accuracy of rainfall estimation, except for one case where drop size distributions were not observed, as the precipitation system did not pass through the disdrometer location. The use of these bias correction methods reduced the RMSE by as much as 50 %. Overall, the most accurate rainfall estimates were obtained using the overlapping area method to correct radar reflectivity. [ABSTRACT FROM AUTHOR]

Fine-scale rainfall observations for modelling exercises are often not available, but rather coarser data derived from a variety of sources are used. Effectively using these data sources in models often requires the probability distribution of the data at the applicable scale. Although numerous models for scaling distributions exist, these are often based on theoretical developments, rather than on data. In this study, we develop a model based on the α-stable distribution of rainfall fields, and tested on 5 min radar data from a Belgian weather radar. We use these data to estimate functions that describe parameters of the distribution over various scales. Moreover, we study how the mean of the distribution and the intermittency change with scale, and validate and design functions to describe the shape parameter of the distribution. This information was combined into an effective model of the distribution. [ABSTRACT FROM AUTHOR]

Recently, a small rain radar in the 24GHz band is under development in Korea. This instrument is intended for the measurement of quantitative rainfall within a radius of 1 Km. Quantitative rainfall measurements for small watersheds are predicated on flood forecasting. The purpose of this device development was focused on the amount of rainfall reaching the ground rather than the amount of rainfall in the atmosphere. The device is currently being developed as a prototype and is undergoing various performance verification with the aim of commercialization.In this presentation, we show the comparison result of performance verification of small rain radar and show the performance superiority. In order to verify the performance of this device, it took a long time from the selection of the comparative verification site to the design and installation of a pit gauge. The comparative site was equipped with 2 parsivels, one pit gauge(composed of one weighing type of rain gauge and 12 tipping bucket type rain gauges) and 17 tipping bucket type gauges. The comparative analysis method of the measurement result converts the scanning result of this device into the reflectance and compares it with the verification device. [ABSTRACT FROM AUTHOR]

Flash flood disaster occur frequently, causing major economic losses to society. The distribution of rainfall in mountainous areas is uneven and mostly is local heavy rains, which leads to inaccurate monitoring of rainfall data at ground rainfall stations and in turn leads to inaccurate flash flood forecasting. In order to study the application prospect of radar quantitative precipitation estimation in flash flood forecasting, taking the Guanshan River Basin as an example, Xin'anjiang model, SCS model and TOPMODEL model are used to simulate floods in the case of rainfall station rainfall or radar quantitative precipitation estimation as input conditions. The model parameters were calibrated by the 10 observed flood events from 2009-2011, and the 10 observed floods events from 2012-2015 were used to validate the performance of the model. The results show that under the condition of the radar quantitative precipitation estimation as input conditions, the simulation accuracy of Xin'anjiang model, SCS model and TOPMODEL model is improved. The Nash-Sutcliffe efficiency coefficient of the three models during the verification period increased from 45.28%, 42.37%, and 43.46% to 63.25%, 58.91%, and 61.34%, respectively. Therefore, the radar quantitative precipitation estimation can be considered as input condition for the flash flood forecasting of the Guanshan River Basin. [ABSTRACT FROM AUTHOR]

Quantitative precipitation estimation (QPE) from weather radar data is crucial forhydrological applications, which benefit of the high spatio-temporal resolution of X-bandradar imagery. Nonetheless, radar rainfall retrieval process involves a challenging correctionof reflectivity, which is particularly difficult for single polarized technology. In addition, highmountain topography as the Andean cordillera deeply increases the complexity of radar QPE.This study explores a Random Forest (RF), state-of-the-art data-driven model, for optimizingthe radar rainfall derivation of the highest X-band radar in the world (4450 m a.s.l.) by meansof single Plan Position Indicator (PPI) scans. Several features were derived from rawreflectivity to account for the evolution of rainfall along the beam. Performance of the RFmodel was evaluated in comparison with the well-known step-wise approach. Forthis, the Marshall-Palmer and a site-specific Z-R relationship were used. The biasadjustment step was omitted from the step-wise correction chain due to the lack ofevenly distributed gauge networks in mountain regions. Results showed that thestep-wise approach slightly improved when using the site-specific instead of theMarshall-Palmer Z-R relationship. However, both models performed poorly for radarQPE (R2 &#x003C; 0.1). Conversely, the RF model remarkably outperformed the othermodels in all different testing locations and rainfall events (R2 up to 0.85). This is thefirst time a QPE random forest model is applied on single polarized X-band radarimagery. The results are outstanding and provide a simplified model to overcome thecomplexity of deriving QPE from single polarized radars. Further work will focus onblending techniques to enhance the radar rainfall maps obtained using this RF model. [ABSTRACT FROM AUTHOR]

Improving the accuracy of flood forecasting is critical in reducing the damage that can occur during flood disasters. In Japan, the flood prediction system is in operation in the first class rivers – relatively large and important rivers. However, small rivers, especially urban rivers, are so difficult to predict the river stage because of the quick rainfall-runoff responses. Such rivers are regarded as difficult to deliver the real time information on flood risk.Recently in Japan, X-band radar rainfall data can be used in real-time prediction for operation. X-band radar is highly resolution in both spatially and temporally. Furthermore, deep learning model has been applied to the river stage prediction and confirmed to have a good accuracy. By applying the X-band radar data to the deep learning model, river-stage prediction could be possible to the urban small rivers.In this study, we developed the deep learning river stage prediction model using the x-band radar rainfall data. We applied the model to the small urban rivers in Tokyo and confirmed the good prediction result. [ABSTRACT FROM AUTHOR]

The Global Precipitation Measurement (GPM) mission provides a new generation of precipitation measurements from space. In particular, Dual-frequency Precipitation Radar (DPR), on board the GPM core observatory, has the advantage in precipitation retrievals compared to its single-frequency counterparts on Tropical Rainfall Measurement Mission (TRMM). However, it is still challenging for DPR to obtain instantaneous rainfall rate with higher accuracy, especially for widely varying rainfall types (e.g., hurricane, storms, etc). In this study, we propose a hierarchal Bayesian approach for improving the DPR rainfall retrievals and perform an initial experiment in Southeastern United States (US), where the US Weather Service Dual-Polarization Radar network is used to obtain locally accurate rainfall estimates for each event. The analysis presented here demonstrates the feasibility of this method for enhancing the DPR instantaneous rainfall product. In summary, this approach demonstrates the potential for enhancing the GPM rainfall product. [ABSTRACT FROM AUTHOR]

This work is part of the EU-funded FloodCitiSense project, the aim of which is to develop an urban pluvial flood forecasting and warning service for and by the citizens. The specific focus of this work is to characterise convective storm features over one of the project pilot cities – Birmingham (UK) – and in turn to provide useful 'local' knowledge that can contribute to the development of the pluvial flood forecasting service. The proposed work comprises two stages: 1) development of a regional convective rain cell database, and 2) characterisation of storm cells' features in relation to the pilot. In the initial development of the database, individual convective rain cells and the traces between any two successive cells were extracted from the high-resolution UK Met Office (Nimrod) radar rainfall archive for the period of 2005-2017 over a 500 × 500 km2 area centred at Birmingham city. These cells and traces were identified using a new object-based convective storm cell tracking algorithm developed by Muñoz et al. (2018). Incorporating an optical flow-based rain-field tracking and a multi-threshold object identification techniques, this new algorithm enables the accurate isolation and tracking of convective rain cells from high-resolution radar reflectivity data. The extracted rain cells and traces were then stored in an open-source database, called TimescaleDB (http://www.timescale.com/). In the second stage, the specific properties of the identified cells and traces and their relationship to the pilot city were extracted from the database for further analyses. These include characterising statistical features of rain cells' intensities, extents, lifespans and their inter-relationships, and mapping the locations of rain cells, traces and given properties over the pilot city. Preliminary results of rain cell and trace mapping enable us to locate the 'hotspots' that are prone to the initiation and maturing of convective cells. In addition, the mapping of the traces suggests the existence of certain storm 'corridors' in the southern part of Birmingham city. This provides valuable 'local' knowledge to the development of urban pluvial flood forecasting and warning services. [ABSTRACT FROM AUTHOR]

The melting layer height is especially important for aviation meteorology because it can help determining the precipitation type and assessing the risk of freezing rain. A new method using the Micro Rain Radar (MRR) to determine the melting layer height is presented. The MRR is a small vertically pointing FMCW Radar which measures Doppler Spectra of precipitation. From these Doppler Spectra, various variables such as Doppler velocity or spectral width can be derived. The melting layer is visible through a higher reflectivity and an acceleration of the falling particles, among others. These characteristics are fed to a neural network to determine the melting layer height. For the training of the neural network, the melting layer height is determined manually. The neural network is tested using data from two sites covering all seasons. For most cases, it is well able to detect the correct melting layer height. [ABSTRACT FROM AUTHOR]

Heavy precipitation events (HPE) are a significant hydrometeorological phenomenon in thehighly variable climate of the eastern Mediterranean (EM). They can cause flash-floods, andare closely-related to water availability. The formation of a HPE requires specific atmosphericconditions that result in relatively high precipitation efficiency, and causes distinctspatiotemporal rainfall patterns. Analyses of rainfall patterns during HPEs traditionally utilizeatmospheric or hydrological indices, Intensity-Duration-Frequency curves based onrain gauges, high-resolution meteorological simulation of case studies, or coarserresolution climatological simulations. However, these are insufficient to characterize thespatiotemporal rainfall patterns in HPEs, and high resolution simulation of large number ofHPEs are scarce. To characterize the rainfall patterns in HPEs in the EM, and toevaluate the ability to simulate them using a weather model we (a) identified HPEsin the EM using a uniquely long calibrated weather radar data, consisting of 24years, (b) simulated the same HPEs using a convection-permitting, high-resolutionnumerical weather research and forecasting model (WRF). Then we analyzed jointly therainfall patterns derived both from the radar record and the WRF, and evaluated themodel&#x2019;s ability to simulate the observed patterns through fractional skill scores,structure-amplitude-location analysis, temporal and spatial autocorrelation of rain-cells,among other techniques. Pixel-based (1X1 km2) identification of HPEs from the radar record yielded 41 eventsthat were later simulated by the WRF. Model simulated rainfall patterns are in goodagreement with the radar data, except for two highly-localized, short duration HPEs. EMrainfall in HPEs, derived both from the radar and the model, consists of small (∼10 kmdecorrelation distance) and intense rain-cells, oriented WSW-ENE, covering a smallportion of the region at each stage of the event (the median fraction of the areacovered by at least 50 mm for 30 min is less than 5%). However, given enough time,frequent arrival of rain-cells accounts for the accumulation of high rainfall depthsover increasingly wider area (&#x003E;80% of the area is covered by &#x003E;150 mm for 24h). The methodology and analyses presented above, of high-resolution, event-basedsimulations, is highly valuable in representing the specifics of rainfall patterns during HPEsand allow the investigation of causative meteorological ingredients for HPEs. Thus, it will befurther used in our future research to study and simulate rainfall patterns under a changingclimate. [ABSTRACT FROM AUTHOR]

The weather radar has the advantage of providing a very high spatial resolution of rainfall information. Recently, radar precipitation has been widely used in the field of hydrology especially for flood warnings and flash flood in urban areas. Radar rainfall is routinely estimated by using the existing reflectivity-rainfall intensity relationship so that there may be biases in their measurements. In this study, a generalized linear model (GLM) for calibrating the radar rainfall is proposed within a Bayesian framework. The biased corrected radar rainfall from the GLM can lead to more accurate rainfall estimates than the existing mean field biased correction method. Moreover, we explored the variability in the bias correction parameters corresponding to the geomorphological characteristics (e.g. latitude, longitude, altitude and separation distance) and further proposed a regionalized formula of the bias correction factors with the height and the separation distance. KEYWORDS: Radar rainfall, Bias correction, Bayesian, Generalized linear model, RegionalizationAcknowledgementThis research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2018R1A6A3A11047264). [ABSTRACT FROM AUTHOR]

The ratio between point and areal precipitation, called the Areal Reduction Factor (ARF), has been identified to vary significantly depending on a variety of factors. Here, we highlight the influence of internal spatial variability of storms on the ARF value. For this, we employ a storm identification algorithm on the radar composite data to identify a total of 54,758 elliptically-shaped extreme storms over the six-year study period. Then, we investigate the relationship between the various storm characteristics to their ARF value. Our findings are as follow: First, we confirm a widely-accepted notion that ARF generally increases with the duration, and it is inversely related to the storm area. Second, we discover that spatial variability within storm, e.g., the coefficient of variation of radar image pixel rainfall values, is a very strong predictor of the ARF value along with area and duration. Last, the difference of ARF values between storms that have elliptical shapes and those that are circular over the same area is about 20 % on average. These findings inform that the current design framework of areal rainfall estimation will be improved by incorporating the information on the rainfall spatial variability and storm shape. [ABSTRACT FROM AUTHOR]

The short term forecast of rainfall intensities for fine temporal and spatial resolutions, has always been challenging due to the chaotic nature of rainfall. Commonly at such scales, radar data are employed to track and extrapolate rainfall storms in the future. Studies have shown that for short lead times linear extrapolation of rainfall storms produces reliable results, however for higher lead times the error increases considerably, limiting so the predictability of rainfall for urban application to 25-30min. Therefore, the aim of this study, is to investigate if this boundary can be pushed further on by including better radar input, and considering non-linear extrapolation as learned by past observed rainfall storms. First the quality of radar data is improved by merging on real time the station data with raw radar data at the 5min and 1 km2 resolution. These data are later fed into an existing linear extrapolation forecast (HyRaTrac) and the importance of the radar quality on forecast performance is assessed. To account for the non-linear motion and transformation of the rainfall storms captured by the radar data, a nearest neighbour method is integrated in HyRaTrac which uses certain predictors to estimate similar patterns in the past and uses these patterns to predict rainfall intensities of the current storm from 5min up to 2 hours horizons. The benefit of including such non-linear transformations, is discussed by comparing directly the two forecast methods both on Lagrangian (area, intensity, direction and velocity of storm) and Eulerian scale (intensities at the recording gauge). The study area is the Hannover radar range (Lower Saxony, Germany) where several storms from the period 2002-2018 are selected in order to train, validate and compare the forecasts method. The data used in the study include the C-band radar data (1km2 and 5min) and 81 recoding stations (5min) provided by DWD. The results show that the radar-station merging cannot increase significantly the predictability of rainfall, however for lead times up to 30 min it can improve the forecast performance (at point scales) considerably. On the other side, the applied nearest neighbour method improves the rainfall forecast for lead times higher than 15 min (up to 1 hour). Overall the methods employed exhibit better forecast skill than the linear extrapolation of raw radar data at such small scales, however as they are triggered by the recognition of storms from radar data, the predictability of rainfall (especially of convective storms) is still bounded by the birth of storms. [ABSTRACT FROM AUTHOR]