Hailstone Characteristics in Northeast Italy from 29 Years of Hailpad Data

Agostino Manzato aARPA Friuli Venezia Giulia–OSMER, Palmanova, Udine, Italy

Search for other papers by Agostino Manzato in
Current site
Google Scholar
PubMed
Close
,
Andrea Cicogna aARPA Friuli Venezia Giulia–OSMER, Palmanova, Udine, Italy

Search for other papers by Andrea Cicogna in
Current site
Google Scholar
PubMed
Close
,
Massimo Centore aARPA Friuli Venezia Giulia–OSMER, Palmanova, Udine, Italy

Search for other papers by Massimo Centore in
Current site
Google Scholar
PubMed
Close
,
Paolo Battistutta aARPA Friuli Venezia Giulia–OSMER, Palmanova, Udine, Italy

Search for other papers by Paolo Battistutta in
Current site
Google Scholar
PubMed
Close
, and
Mauro Trevisan aARPA Friuli Venezia Giulia–OSMER, Palmanova, Udine, Italy

Search for other papers by Mauro Trevisan in
Current site
Google Scholar
PubMed
Close
Free access

Abstract

Although hail is a well-known meteorological hazard, it is hard to find long records of hail observed at the ground with high spatial resolution. Most hail climatologies are based on remote sensing observations or inhomogeneous networks of human observers. In the plain of Friuli Venezia Giulia (northeast Italy), a hailpad network of 367 stations has operated since 1988. During the 1988–2016 warm seasons, 7782 hailpads were impacted by hailstones and more than one million dents were observed and automatically analyzed, even though only 63% of them were associated with valid hailstone dents. In this work, this large quantity of direct hail observations is used to build a hail climatology in terms of hailstone size, areal density, and flux of kinetic energy. From the empirical distributions of data collected, it is possible to fit statistical distributions to the different hailstone/hailpad behaviors. In particular, it is also possible to find an approximate estimation of the flux of kinetic energy based only on the largest hail diameter observed on the hailpad. Last, temporal and spatial distributions of these characteristics are investigated. Hailstones are larger along a southwestern-to-northeastern alley, which is parallel to the main pre-Alpine crest, with the very largest sizes being more frequent on the southwestern corner. The only hail climate change signal that one can infer from the analysis of these multidecadal trends is that, in more recent years, hailstorms seem to produce fewer and larger hailstones, on average.

© 2022 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Agostino Manzato, agostino.manzato@osmer.fvg.it

Abstract

Although hail is a well-known meteorological hazard, it is hard to find long records of hail observed at the ground with high spatial resolution. Most hail climatologies are based on remote sensing observations or inhomogeneous networks of human observers. In the plain of Friuli Venezia Giulia (northeast Italy), a hailpad network of 367 stations has operated since 1988. During the 1988–2016 warm seasons, 7782 hailpads were impacted by hailstones and more than one million dents were observed and automatically analyzed, even though only 63% of them were associated with valid hailstone dents. In this work, this large quantity of direct hail observations is used to build a hail climatology in terms of hailstone size, areal density, and flux of kinetic energy. From the empirical distributions of data collected, it is possible to fit statistical distributions to the different hailstone/hailpad behaviors. In particular, it is also possible to find an approximate estimation of the flux of kinetic energy based only on the largest hail diameter observed on the hailpad. Last, temporal and spatial distributions of these characteristics are investigated. Hailstones are larger along a southwestern-to-northeastern alley, which is parallel to the main pre-Alpine crest, with the very largest sizes being more frequent on the southwestern corner. The only hail climate change signal that one can infer from the analysis of these multidecadal trends is that, in more recent years, hailstorms seem to produce fewer and larger hailstones, on average.

© 2022 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Agostino Manzato, agostino.manzato@osmer.fvg.it

1. Introduction

Despite hail being a severe socioeconomical problem (Changnon et al. 2009; Púčik et al. 2019) that causes a huge economic loss (e.g., Kunz et al. 2009, 2018), regular and precise hail observations are still a challenging problem, as is hail forecasting (Brimelow et al. 2002; Martius et al. 2018). Hail at the surface is commonly observed by human spotters, for example those reporting hail events in the European Severe Weather Database (ESWD; https://eswd.eu; e.g., Groenemeijer et al. 2017), or is derived from insurance loss data. While it is still possible to build climatologies of hail from these reports, it is very likely that the results are biased toward the most populated areas (spatial inhomogeneity) and that they include artificial temporal bias, varying with the number of available observers at any given time (Schuster et al. 2005; Cecil 2009; Taszarek et al. 2020).

For these reasons, hail climatologies are often based on remote sensing detection, like hail probability or hail size estimated from radar (e.g., Nisi et al. 2016; Murillo et al. 2021) or from satellite (e.g., Cecil and Blankenship 2012; Punge et al. 2017; Bang and Cecil 2019). However, hail detection from remote sensing has also drawbacks: first, because it is an indirect observation of hail that relies on predetermined relationships between the mean observations of some radar measurements of an atmospheric volume and the size distribution of the hailstones inside it; second, because that volume (e.g., the radar bin) is very large in comparison with hailstones, hence the spatial resolution is not high enough to resolve the characteristics of individual hailstones. Moreover, remote sensing detection usually estimates hail at a certain altitude above surface; that is, it neglects the effect of melting happening in the lowest part of the troposphere, and therefore it cannot measure the “ground truth.”

Hailpads (e.g., Towery et al. 1976) are cheap polystyrene rectangles that are exposed outdoors and are manually collected when they are hit by hailstones. However, the spatial density of hailpad stations is usually not sufficient to cover the very high spatial variability of hail as observed at the ground (because hailstreaks are generally on the order of 10 km long but only on the order of 0.5–2 km wide; Morgan and Towery 1975; Changnon 1977; Morgan and Summers 1986). Hailpads have also other limitations, such as the fact that hailstones are assumed to be spheres, falling at their terminal velocity, and, when there is a high areal density of hailstones, it is possible that some dents overlap (Palencia et al. 2011). If the areal density is extremely large, then the hailpad can ultimately saturate. Despite their limitations, hailpads are the most efficient and cost-effective way to observe hail at the ground, apart from the—inhomogeneous—manual human observations or a network of expensive automatic hail sensors (also called “hail disdrometers”), like those that estimate hail size from the sound produced when the hailstones hit the sensor (e.g., “HailSense”; see Löffler-Mang et al. 2011; Sommer HDI, https://www.sommer.at/en/products/wind-weather/hail-sensor-hdi).

Discussing the pros and cons of hail observed by hailpads or estimated by radar it is interesting to recall that Waldvogel et al. (1978b) compared the kinetic energy of hailstones derived from an S-band radar with that observed by the corresponding hailpads. The network of hailpads (one every 3.8 km2) was located less than 30 km away from the radar, where the radar beam is still relatively small. The agreement of this particular combination of radar and near hailpads showed that the kinetic energies were different by less than ±20%. They concluded that a combination of both hailpad network and radar would be the best solution to observe hail. However, it is unrealistic to install an S-band radar every 60 km plus a very dense hailpad network on large areas affected by hail, like, for example, the Italian Po Valley. Unfortunately, hailpads are operated only in few parts of the world, mostly located in South America and Europe (see Sánchez et al. 2009 and references therein). For an in-depth discussion of different hailpad networks operated in Europe the reader is directed to the comprehensive discussion of Punge and Kunz (2016).

Datasets of homogeneous and high-spatial-resolution observations of hail are very useful also for other kinds of studies, besides building hail climatologies. In fact, high-density observations with fine-scale details of hail on the ground are required to validate hail-model forecasts (Kumjian et al. 2021). Moreover, hail-growth models, like WRF–HAILCAST (Jewell and Brimelow 2009; Adams-Selin and Ziegler 2016), or the model developed by Kumjian and Lombardo (2020), would greatly benefit from direct comparison between simulated and surface-observed hail size distributions. For example, Manzato et al. (2020) compared the forecast maximum hailstone size from WRF–HAILCAST with observations from the hailpad network located in Friuli Venezia Giulia, hereinafter FVG. They found that some of the simulations made with WRF–HAILCAST provide useful information on the severity of the event, both in terms of maximum hailstone diameter and of reflectivity shape.

Hail is a relatively frequent event in the Po Valley (northern Italy), as shown by Morgan (1973), Prodi (1976), and Cacciamani et al. (1995). For this reason, in 1988 Ente Regionale per la Promozione e lo Sviluppo dell’Agricoltura (ERSA, now named Agenzia Regionale per lo Sviluppo Rurale) FVG started a hailpad network in the plain of FVG (Morgan 1992; Giaiotti et al. 2001). In fact, Fig. 8 in Punge et al. (2014) and Fig. 6 of Punge et al. (2017) show how, from satellite data, one can derive that northeastern Italy is a hot spot for hailstorms in Europe, as is the case for lightning flashes (Manzato et al. 2022; Taszarek et al. 2020). This is probably due to FVG’s close proximity to the Marano and Grado Lagoons and the Adriatic Sea to the south, which are a source of low-level moisture, and the pre-Alps and Alps to the west, north, and east, which provide orographic lift and low-level convergence (Davolio et al. 2016; Miglietta et al. 2016).

Manzato (2012) prepared a climatology of the number of hailpads impacted by hail in FVG for 1992 through 2009. The aim of this study is to extend the previous work to the climatology of hailstone sizes (and derived quantities) from hailpad analysis, using data from 1988 through 2016, in order to better characterize the FVG hailstorms and to offer a climatology of these quantities from a long record of surface observed hailstones. Section 2 shows the data and fields studied in this work. Section 3 presents the statistical distributions of all the studied variables, and sections 4 and 5 show the temporal and spatial distributions, respectively. Section 6 summarizes the results and conclusions.

2. Data and methods

a. Hailpad network and semiautomatic analysis

Figure 1 shows the plain of the FVG region (northeastern Italy), which has the Alps to the north and the Adriatic Sea and lagoons to the south. The FVG hailpad network studied herein has been operated by 367 volunteers since 1988 and covers an area of about 4500 km2. The mean distance between stations is about 3–4 km. Hailpads are installed/deployed from 1 April until 30 September every year. The network, coordinated by the regional meteorological service [Osservatorio Meteorologico Regionale (OSMER)–Agenzia Regionale per la Protezione dell’Ambiente (ARPA) FVG], has already been described in Morgan (1992), Giaiotti et al. (2003), and Manzato (2012), where additional details are given.

Fig. 1.
Fig. 1.

Locations of hailpad stations in the FVG plain (northeastern Italy). Each number shows the total number of hailpads collected at the station during the study period. Seven stations with only one hailpad are not displayed. The 15 km × 15 km grid boxes are used in section 6.

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

The numbers shown in Fig. 1 identify the locations of the 360 hailpad stations with the total number of hailpads collected at each of them during April–September 1988–2016. The total number of studied hailpads is 7782, including seven hailpads collected by stations that have only one hailpad (which are not shown in Fig. 1). The station with the maximum number of hailpads has 85 of them and is found in the northeast corner of the network (close to a village called Torreano). This maximum is much larger than the mean number of total hailpads collected per station, that is, 21. Not all volunteers apply the same care when monitoring the hailpads,1 and many volunteers have been replaced during the 29 years of operation; when a volunteer is replaced, we attempt to identify a new one no more than a few kilometers away. This partly explains the high spatial variability shown in Fig. 1 for the total number of hailpads, which is not only due to the large natural variability of the hailswaths (Morgan and Towery 1975). For this reason, when we study the spatial distribution of hail (section 6), we regrid the data onto a larger 15 km × 15 km grid, and we remove stations that have collected less than 15 hailpads during the study period (29 years).

After being collected, the hailpads are blackened with a roller (filled with typographical ink) to reveal the hail dents as white areas, since hailpads were covered with white latex paint before being exposed. Next, they are scanned and analyzed using a custom version of the NCAR TITAN program (Dixon and Wiener 1993),2 that fits each white dent with an ellipse. More details on this analysis can be found in appendix A of Giaiotti et al. (2001). Here we reiterate that this semiautomatic analysis is subjective and that this can be a potential source of error. In fact, when the hailpad has many dents, it often happens that they overlap (Palencia et al. 2011). To address this problem, a human operator manually separates the dents by drawing in a black line, before performing the TITAN analysis. Although this reduces the impact of overlap, it is a subjective process.

To better illustrate this problem, Fig. 2 shows the same hailpad analyzed 2 times with the TITAN program. The first time (Fig. 2a) the hailpad was analyzed as it was archived years ago by an operator: the maximum hail dent diameter is estimated to be about 49 mm. However, a recent inspection of this hailpad revealed that such a large dent was in reality due to the overlapping of several hailstone dents. Figure 2b shows the newly remade analysis, where the large dent has been divided into smaller dents, caused by five different hailstones. Of course, the output of the hailpad TITAN program is very different in the two cases, and this illustrates how human dependent this analysis can be in specific cases.

Fig. 2.
Fig. 2.

A hailpad observed at about 0200 UTC 11 Aug 2001 near Pasiano di Pordenone (45.83°N, 12.57°E) analyzed with the custom hailpad version of the TITAN program (Dixon and Wiener 1993). (a) In the original handmade correction, some overlapping dents were not well separated (leading to a big Ddent of about 49 mm), whereas (b) in the recent reanalysis, the largest hailstone has Ddent of 39 mm (corresponding to Dhail = 41 mm).

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

To avoid the “wind blown” effect (Changnon 1973; Dieling et al. 2020), only the minor axis of the fitting ellipse is taken to represent the true dent of the hailstone. That choice can induce an underestimation of the maximum hail size, which can be particularly significant for the largest hailstones, given the increasingly ellipsoidal nature of larger hailstones (Shedd et al. 2021).

b. Converting dent diameters to hailstone diameters

The minor axis of the fitted ellipse is taken to estimate the hailstone-dent diameter Ddent that is then converted into an estimated hail diameter Dhail. This conversion is done using the calibration equation specific for the hailpad material, which in our case is the Dow Chemical “Floormate 300” [2-or 3-cm-thick panels of extruded polystyrene (XPS) foam with a compressive strength at 10% deformation ≥ 300 kPa]. In this work we used the calibration already computed for this kind of material as shown in appendix A of Giaiotti et al. (2001). In this approach, hailstones are supposed to be spherical and there is the following empirical relation between the diameter of the falling hailstone and the diameter of the dent that it leaves on the hailpad:
{Dhail=0.962Ddent+3.262if Ddent2.8mmDhail=2.100Ddent if Ddent<2.8mm.
Note that this calibration curve has been derived using steel balls (having the same kinetic energy of a hailstone of that diameter at its terminal velocity) with diameters varying from 6 to 28 mm, and outside this range the calibration fit has been linearly extrapolated. In particular, for dents below 2.8 mm, produced by steel balls of 6 mm, the linear fit has been made toward the (0, 0) point. For this reason, for dents smaller than 2.8 mm the confidence in the estimated hailstone size (<6 mm) is low.

To investigate this issue, Fig. 3 shows the distribution of all of the (minor axes Ddent of the fitting ellipses for the) 1 189 632 observed hailstone dents from the 7782 hailpads. Note that, since we are particularly interested in the lower values, the abscissa is plotted on a logarithmic scale. Together with the estimated hailstone diameters (black continuous line), also three percentiles of the aspect ratio [here defined as the ratio between major and minor axes, that is, the inverse of that defined in Shedd et al. (2021)] of the fitting ellipse are shown, as dark-gray lines.

Fig. 3.
Fig. 3.

The frequency distribution of all of the minimum-axis dents (black continuous line) observed on all the hailpads. The dark-gray lines show, for each class of the hail dents, the values of three percentiles (10, 50, and 90, respectively) of the aspect ratio between the maximum and minimum axes of the ellipse fitting the hail dents. The light-gray dashed lines mark the thresholds used to discriminate the valid hail dents from spurious dents.

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

It is well known that small hailstones tend to be more spherical (aspect ratio ≈ 1.2), whereas the largest hailstones tend to be more “oblate,” having a larger aspect ratio (up to 1.8 for hailstones of about 40–50 mm, following Fig. 6 of Shedd et al. 2021). In our data, 90% of the aspect ratios are less than 2 when Ddent ≥ 2.8 mm. However, for Ddent < 2.8 mm the median of the aspect ratio (50% of its distribution) increases exponentially and is always larger than 2.0 for Ddent smaller than approximately 1.9 mm, whereas for Ddent ≥ 2.8 mm this median is almost constant (approximately 1.25). In practice, it was understood that the large quantity of dents with very low diameters and very high aspect ratio (that is unusual for small hailstones) are in reality thin “line cuts,” due to imperfections of the material or to accidental marks produced managing the hailpad, and likely do not represent true hailstone signatures.

Given the fact that the smallest calibration ball (6 mm) produced dents of only 2.8 mm and that below 2.8 mm there are a lot of spurious thin marks having a very large aspect ratio, it was decided to consider as “valid” a hailstone dent when the minimum diameter was at least 2.8 mm (hailstones of at least 6 mm) and with an aspect ratio (major/minor) between 1 and 2. Incidentally, we note that the World Meteorological Organization Manual of Codes number 306 (WMO 1988) defines as hail a frozen precipitation of at least 5 mm in diameter; thus we are neglecting hailstones in the range between 5 and 6 mm. With these two filters (represented by the two light-gray dashed lines in Fig. 3) only 747 759 hailstone dents (63% of the original) are retained as “valid hailstone dents” and are converted to their equivalent hailstone diameter using the calibration Eq. (1).

Last, it is worth noting that for the dent bins centered at 2.25 and 2.75 mm there seems to be a cutoff in the peak of the diameter distribution. This could be an effect due to hailstone melting that does not affect equally all sizes but impacts particularly the hailstones reaching ground with sizes smaller than 6 mm (leaving dents smaller than 2.8 mm), as has been already hypothesized in appendix B of Giaiotti et al. (2001).

c. Converting hailstone diameter to kinetic energy

The classical method (e.g., Waldvogel et al. 1978a) to compute the kinetic energy associated with a falling hailstone involves the following hypothesis:

  • The hailstone is homogeneous with a constant density ρice. Here we use the hailstone density experimentally found in the Po Valley by Vittori and di Caporiacco (1959), that is, ρice ≈ 0.9 (g cm−3).

  • The hailstone shape is spherical; hence its mass m is given by
    m=ρice43π(Dhail2)311000(kg),

where Dhail is expressed in centimeters.

  • The air drag is constant and can be estimated by the coefficient cD ≈ 0.6.

  • The hailstone hits the pad at its terminal velocity Vterminal that can be estimated by the Weickmann (1953) equation as
    Vterminal=11004gρiceDhail3cDρdry14Dhail(ms1),

where g is the acceleration due to gravity (981 cm s−2) and ρdry is the density of dry air (0.001 g cm−3). Following Eq. (3) of Manzato et al. (2020), Vterminal is computed in meters per second, whereas Dhail is expressed in centimeters for convenience.

With these assumptions the hailstone kinetic energy Kehailstone is simply given by
Kehailstone=12mVterminal20.046Dhail4(J).
For example, a hailstone with a diameter of 1 cm has a classical kinetic energy of 0.05 J; if the diameter is 4 cm, then its kinetic energy is 11.8 J.

Since the hailstone density ρice can vary depending on the conditions and diameters (e.g., Heymsfield et al. 2018) and the hailstone shape is often not spherical, Heymsfield et al. (2018, 2020) proposed a different approach, based on the “best Reynolds number.” In their approach, both the hail density and terminal velocity are functions of the hailstone diameter Dhail so that the kinetic energy is proportional to Dhail3.5, instead of to its fourth power. For the majority of diameters in our database the difference between the two methods is very small, since only the tail of the largest Ke values is substantially reduced by the new approach, with respect to the old one. Hence, for the largest diameters of our database, the classical Ke computed by Eq. (4) is more biased (larger values) than the method based on the “Reynolds number” and the largest classical Ke values computed on our dataset must be considered with care. However, in this work we will study the correlations between the different hailpad behaviors (section 3b), and, since the linear correlation between the classical Ke values and those obtained using the Heymsfield et al. (2020) approach gives a Pearson coefficient as high as R = 0.985 (on our data), hereinafter we will show only the Ke values computed with the classical method.

d. Hailpad characteristics

While hailstone diameter and kinetic energy are specific characteristics of each hailstone, it is interesting to also define some characteristics of each hailpad. For this reason, we will also study some parameters of the distribution of all the hailstones hitting each hailpad, such as the maximum diameter Dmax or the median diameter D50. Moreover, it is worthwhile to also estimate the impact energy by territory and to calculate the sum of the kinetic energy of all the hailstones hitting a hailpad.

On average, the hailpads have an area of Ahailpad ≈ 0.291 × 0.395 m ≈ 0.115 m2, but the dimensions can slightly vary from one hailpad to another. Moreover, when part of that area is ruined by marks considered false dents (by the two filters introduced in the previous section), the effective area for hail observation is reduced. In general, each hailpad may have some Nnotvalid dents that are considered not-valid dents, and their areas must be removed from the total hailpad area to estimate the area available for observing the valid hailstones. Hence, for each hailpad we define its “valid area” as
Valid_Area=Ahailpadi=1Nnot validAreaEllipse(Ddenti,aspecti)i(Ddenti<2.8or aspecti>2)(m2).
If a hailpad has no false dents (Nnotvalid = 0), then its Valid_Area ≈ 0.115 m2. Given that Nvalid is the number of valid hailstones impacted on a hailpad, then we define the hailpad “areal density” as Density = Nvalid/Valid_Area (No. m−2).
Last, it is possible to estimate from each hailpad the “flux of kinetic energy” (KeFlux, also called “accumulated kinetic energy”) with the following equation:
KeFlux=i=1NstonesKeiValid_Area(J m2),
where Kei is the individual kinetic energy of the ith hailstone hitting the hailpad. KeFlux is an important cumulative parameter used for evaluating the hail damage potential. However, it must be noted that the same intensity of KeFlux can be obtained in different ways: for example, with only a few large hailstones (small areal density but large Dmax, being very problematic for the automotive market) or with many small hailstones (e.g., large N and Density but small D50, being very damaging for crops), while the kind of damage in the two cases would be very different.

3. Statistical distributions

In this section the statistical behaviors of the valid hailstones will be presented. The first part will show the results for the total hailstone population, while the second part will show the results characterizing the hailpad population.

a. Hailstone statistical distributions

Figure 4 shows the probability distribution3 for all the 747 759 valid hailstone diameters (black line), using a bin width of 0.5 mm, and for their kinetic energies (gray line). The minimum size is 6.0 mm (because of the filter of the previous section), the mean value of all studied hailstones is 8.0 mm, while the maximum diameter is 46 mm. The mean value of Kehailstone is 0.03 J, the maximum value is 21 J, and the 99% quantile is 0.32 J.

Fig. 4.
Fig. 4.

The probability distribution from the 747 759 valid hailstone diameters (mm; black continuous line), with a logarithmic axis for probability (note that the convention of e±XX indicates that the preceding value is multiplied by 10±XX). The light-gray dashed line shows the exponential law of Eq. (7), which fits very well the majority of diameters in the 6–20-mm range. The dark-gray line is the probability distribution of all of the hailstone kinetic energies (J), whose scale is shown on the top of the panel.

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

The gray dashed line shows an exponential-law fit of the hailstone size using the following equation:
P(Dhail)=3.55×100.2Dhail().
(the empty brackets meaning that probability has no units). The fit works reasonably well in the majority of cases, since 99% of hailstones have a diameter ≤ 16.3 mm, but strongly underestimates the largest hailstones, for example, >20 mm. A better fit can be reached with a more complex distribution, such as the two-parameter log-logistic (dotted line in Fig. 4):
P(Dhail)=(β/α)(Dhail/α)β1[1+(Dhail/α)β]2()withα=6.4,β=5.8,
even if it slightly overestimates the diameters in the 7–10-mm range.

The dark-gray line in Fig. 4 shows the probability distribution of the kinetic energy computed by Eq. (4) from all the 747 759 valid hailstones. Of course, because the kinetic energy depends on the fourth power of the diameter, its distribution is much more skewed than that of the diameters, since 98.5% of data are inside the first bin. In this case a simple exponential law is not a good fit.

b. Hailpad-derived statistical distributions

Figure 5 shows the probability distribution for the total number of hailstones and for the areal density of each of the 7782 studied hailpads. In general, the two distributions have the same shape, as expected, with a scale factor of 10. The mean value of Density is 815 hailstones per meter squared, and the 99% quantile is 5650 hailstones per meter squared.

Fig. 5.
Fig. 5.

The probability distribution for the number of valid hailstones (unit label # = No.) per each of the 7782 hailpads (black continuous line), with probability on a logarithmic axis. The gray line is the probability distribution of the areal density (No. m−2) per each hailpad, whose scale is shown on the top of the panel.

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

On the other hand, Fig. 6 shows the probability distribution for the maximum value Dmax and for the 50% percentile D50 of the hailstone size distribution in each hailpad. The D50 probability peaks at 6.8 mm (mode), with a mean value of 7.4 mm; its maximum value is 24 mm, and the 99% quantile is 11.3 mm. The probability distribution for the maximum hailstone diameter of each hailpad peaks at 9.5 mm, with a mean value of 12.0 mm; the largest value is 46 mm, and the 99% quantile is 29 mm.

Fig. 6.
Fig. 6.

The probability distribution of the maximum hailstone diameter (mm) in each of the 7782 hailpads (black continuous line). Note the logarithmic probability axis. The gray line is the probability distribution of the 50th percentile of the hailstone size distribution (mm) of each hailpad.

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

From these results one may think that hail in FVG plain is relatively small when compared with other regions, where hail is reported by human observers. However, it must be considered that observers (as those reporting hail in the ESWD) are attracted by the largest hailstone available in the whole area surrounding them. In comparison, the sampling area of the hailpad is very small, being only 0.115 m2, and the probability that the largest hailstone in the area hits the hailpad is very low. Smith and Waldvogel (1989) suggested that the largest hailstone in the vicinity of a hailpad would be about 2 times that of the maximum hailstone hitting the hailpad. Grieser and Hill (2019) tried to improve this empirical rule with the following exponential fit:
Dhailpad=82.16[1e0.01319(Dvicinity1.123)](mm)

In practice, the largest hailstones observed by the FVG network of hailpads are in the 40–46-mm range, which corresponds to hailstones “in the vicinity of the hailpad” possibly having diameters of 80–92 mm for the Smith and Waldvogel (1989) rule, or of about 52–64 mm for the Grieser and Hill (2019) approach. Both values seem reasonable, since many observers in the past have already shown pictures of hailstones in FVG with diameter of about 50 or 60 mm, whereas very recently (on 1August 2021) two independent observers documented hailstones in the western part of FVG up to 90 mm (in a village called Azzano Decimo, located 11 km to the south of the city of Pordenone). Thus, in FVG there is also large hail, but in 99% of cases the maximum hailstone in their vicinity should be smaller than approximately 34 mm for Grieser and Hill (2019) or 58 mm following the Smith and Waldvogel (1989) rule, since the 99% of Dmax is 29 mm.

Figure 7 shows the probability distribution for the flux of kinetic energy of each hailpad, KeFlux. It peaks at the minimum value (∼10 J m−2) and has a strongly decreasing trend up to about 500 J m−2, above which only few spikes are observed, with an absolute maximum of 1624 J m−2. However, 99% of this distribution lie below 317 J m−2. Its mean value is 27 J m−2. As a reference, Strong and Lozowski (1977) found that cereal crop damage was already evident when the accumulated kinetic energy exceeded 50 J m−2.

Fig. 7.
Fig. 7.

The probability distribution of the flux of kinetic energy for each of the 7782 hailpads (J m−2; black continuous line), with a logarithmic probability axis. The gray line is the probability distribution obtained in converting the Dmax distribution with Eq. (10).

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

The specific hailpads for which the first four absolute values (highlighted in boldface type) of Density, Dmax, and KeFlux have been observed are listed in Table 1, reporting their date and location. In general, some of the hailpads having a maximum in Dmax have also a maximum in KeFlux, while hailpads having the largest value of areal density are typically different. That is because large KeFlux values are usually associated with large hailstones, whereas large values of Density are mostly related to a large number of relatively small hailstones.

Table 1

List of hailpads having the four maximum values (highlighted in boldface type) for the maximum diameter, flux of kinetic energy, and areal density.

Table 1

It is also interesting to study the linear correlations between the different hailpad characteristics. Apart from an obvious, very high, correlation (R = 0.997) between the number of hailstones and Density, the second-best correlation is found between Dmax and KeFlux (R = 0.686), closely followed by the correlation between Dmax and D50 (R = 0.680). The linear correlations between KeFlux and the number of hailstones or the Density are both of the order of R = 0.58, whereas the correlation between D50 and KeFlux is only R = 0.51.4 That happens because the kinetic energy varies with the fourth power of the diameter and hence the flux of kinetic energy is more sensitive to the largest hailstone diameter than to the median of the whole distribution.

Because of this, we investigate whether the linear correlation between KeFlux and Dmax for each hailpad would increase when Dmax is elevated to a power larger than one. All exponents between 1.1 and 4.0, with steps of 0.1, are tested. It is found that the linear correlations reach a maximum of R = 0.753 when the exponent is equal to 2.6, that is, with the following fit:
KeFlux0.042Dmax2.68(J m2),
where Dmax is expressed in millimeters. To show the goodness of this approximation, the gray line in Fig. 7 compares the probability distribution of KeFlux with the one obtained converting the Dmax distribution with Eq. (10). Even if the probability for the lowest values of KeFlux is not well matched by the transformed Dmax distribution (that does not peak at the lowest value as KeFlux does), above 40 J m−2 the two distributions overlap reasonably well.

Fraile et al. (2003) studied hailpad data over southwestern France from 1987 to 2001 and calculated the total kinetic energy for each hailpad, which is proportional to KeFlux if all hailpads have the same area. Similarly, Grieser and Hill (2019) applied a power law to the Fraile et al. (2003) dataset, finding exactly the same result shown above. That is, the total kinetic energy accumulated on the hailpads in southwest France is proportional to the maximum hailstone to the power of 2.60 [see Eq. (11) of Grieser and Hill 2019]. Thus, it seems reasonable that Eq. (10) is a good approximation of the flux of kinetic energy in a relatively small area, knowing only the maximum hailstone found inside that area. This is a useful result, whose validity should be tested also in other regions where hailpad, or in situ hail sensor data, are available.

Figure 8 shows, as a 2D density plot, the scatterplot between Dmax and KeFlux, and the black line shows the linear fit of Eq. (10), between Dmax2.6 and KeFlux. While, for the majority of hailpads, KeFlux is very low, when the maximum hail diameter increases above 2 cm then KeFlux increases exponentially. For example, if Dmax = 3 cm then KeFlux ≈ 280 J m−2, while if Dmax = 4 cm then KeFlux ≈ 600 J m−2.

Fig. 8.
Fig. 8.

The density distribution of the 2D scatterplot between Dmax and KeFlux for all hailpads. Light-gray squares show the areas with the highest density of scatterplot points, corresponding to the lowest values of KeFlux. The black line is the linear fit between KeFlux and Dmax2.6 [Eq. (10)].

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

4. Temporal distributions

In the previous section, the statistical distribution of some hailpad-derived characteristics was analyzed without any regard to when the 7782 hailpads are impacted by hail. In this section, the same data will be stratified with regard to year, day of year, and hour of day to study the hailstorm temporal trends.

The climatology of the “number of hailpads” was already discussed in detail in Manzato (2012), even if using a shorter dataset, so here only an update of the temporal distribution is shown for this field (Fig. 9). Figure 9a shows a strong variability for the number of hailpads during these 29 years. The years with most hailpads (>500) are 2008, 2007, and 1998. During the more recent years the number of hailpads is lower than before and, in particular, the very low number of hailpads collected in the last two years (2015 and 2016) could more likely be a failure in the network collection efficiency rather than a realistic drop in hailstorm frequency. Figure 9b shows a seasonal cycle very similar to that of Fig. 6 of Manzato (2012), with the maximum number of hailpads collected between the end of June and mid-July. The distribution of the diurnal cycle (Fig. 9c) shows a peak at 1500 UTC (1700 local time or 1600 solar time), and the minimum number of hailpads is observed between 0200 and 0600 UTC (0400–0800 local time). This result is consistent with the diurnal cycle of cloud-to-ground lightning for the Po Valley, as shown in Fig. 3d of Feudale and Manzato (2014).

Fig. 9.
Fig. 9.

(a) Temporal distributions of the total number of hailpads during the studied years, (b) day of the year, after a ±15-days moving average and with shaded area covering ±1 std dev, and (c) number of hailpads by hour of the day.

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

Figure 10 shows the variation, during these 29 years, for the areal density, the maximum and median hailstone diameters, and the flux of kinetic energy. Note that in this figure both “intensity” (solid line, showing the mean value of that bin) and “probability” of exceeding a given threshold (filled areas, showing the probability of exceeding the 50th, 75th, and 90th quantiles of the entire distribution, respectively) are shown. In general, there are no noteworthy trends during the studied period, apart from Density (Fig. 10a), which has an increasing trend from 1988 to 1995, whereas—after that—there is a clear negative trend. If one applies a linear regression to the mean areal density versus years, one finds a decrease with time (−58 hailstones per meter squared per year), with a correlation coefficient of R = 0.68 that is statistically significant (p = 3 × 10−5). The only other relatively high correlation (R = 0.50) is found for the annual trend of the mean D50 (Fig. 10c), which has a positive slope (+0.06 mm yr−1) and a relatively high statistical significance (p = 3 × 10−3). Also, the mean flux of kinetic energy has a decreasing trend, but the correlation with years is only R = 0.38, with a relatively low statistical significance (p = 0.02).

Fig. 10.
Fig. 10.

Temporal trends during the 29 studied years. The black line is the mean value (scale on the right side), and filled grays show the probability to reach the 50th, 75th, and 90th quantiles of the entire distribution (scale on the left side).

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

In conclusion, it seems that during 1995–2016 the hailpads were hit by a lower number of hailstones (lower Density), but this is accompanied by an increase in the median of the hail diameters. This positive trend in D50 is in agreement with similar results of Eccel et al. (2012), who found a positive trend for the 90th percentile of the hailstone diameter during 1975–2009 for a hailpad network in the Trentino region (northern Italy). However, Fig. 10b does not show a clear positive trend for Dmax, having a large interannual variability, likely because data for only a single hailstone value per hailpad are much noisier than calculating the median of all of the hailstones that hit that hailpad.

Figure 11 shows the annual cycle, after applying a ±15-day moving average of the daily values. The annual (seasonal) cycle has a clearer trend than those found for the different years (Fig. 10) or for the hour of the day (Fig. 12). They are all bell shaped, apart from the areal density (Fig. 11a), which has a relative maximum at the end of June and an absolute maximum in mid-September.5

Fig. 11.
Fig. 11.

As in Fig. 10, but for the annual cycle for each day of year, after a ±15-day moving average.

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

Fig. 12.
Fig. 12.

As in Fig. 10, but for the daily cycle of four hailpad characteristics.

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

The annual cycle of Dmax reaches a relatively flat peak from mid-June to the beginning of August, when there is the highest probability of observing a Dmax above the 90th quantile (dark solid area in Fig. 11b). Very similar behavior is found for D50 (Fig. 11c), which has a high probability (≈0.55) of reaching the 50th quantile of its distribution (7.1 mm) from mid-June to mid-August (top of the midgray shade), but reaches the maximum probability (≈0.1) of exceeding its 90th percentile (8.6 mm) at the beginning of August (top of the black shaded area). This pattern is similar to the probability for the number of hailpads shown in Fig. 6 of Manzato (2012). KeFlux has a peculiar annual cycle, since the 50th and 75th percentiles peak at the end of June, while the mean value and the 90th percentile peaks at the beginning of August (Fig. 11d), as also happens to the mean value of Dmax and D50. In conclusion, from June to September hailstones are likely larger than in May and—in particular—April, but the largest hailstones are more likely observed at the beginning of August. A similar pattern is visible for the average KeFlux, while the Density reaches its maximum values in September. From these results and considering the fact that the last 15 days of September are lost because of the moving average, extension of the hail observation period to the month of October should be considered in the future.

Last, Fig. 12 shows the diurnal cycle. In general, this cycle is not as well defined as the previous one, in particular for Density, which has a large variability during the day (Fig. 12a). For Dmax and D50 (Figs. 12b,c) it looks like there is a relative maximum, meaning larger hailstone diameters, during the night, in particular between 2200 and 0300 UTC. Moreover, Figs. 12b, 12c, and 12d show a peak between 0600 and 0700 UTC that has apparently no justification. One may think that at 0600–0700 UTC people go to check the hailpad and if they found it being impacted then they could erroneously assign that time to hailstorms that happened during night (when it is more difficult for volunteers to check the hailpad). However, that would produce a corresponding peak also in the number of hailpads, which is not present in Fig. 9c. Figure 9c implies that the 0200–0700 UTC period is the less sampled one and hence even a few cases can have a large impact on the mean value. In fact, if one considers the 61 hailpads fallen between 0600 and 0700 UTC, then D50 (for example) has a mean value of 7.78 mm, but just removing the four largest cases (between 11 and 15 mm) will decrease the mean value to 7.44 mm, that is, similar to the mean value between 0400 and 0500 UTC. In practice, the peak between 0600 and 0700 UTC in Figs. 12b, 12c, and 12d is not an “error,” but it is due to the large variability (e.g., σ = 1.64 mm for the mean D50) of a relatively small population.

5. Spatial distributions

Figure 1 shows all the stations having at least two hailpads hit by hail during the study period. To get a spatial distribution with a robust statistical significance, the full domain has been divided in 15 km × 15 km grid and the stations reporting a small number of hailpads (less than 15 hailpads in the 1988–2016 period) have been removed from the following analysis. Last, only the 24 grid boxes having at least three of the remaining stations have been retained. This means that the domain for this spatial analysis is smaller than that covered by all the squares shown in Fig. 1. Consequently, only 6692 hailpads collected from 241 different stations are used in this spatial analysis. There are an average of 10 stations and 279 hailpads per grid box. The grid box with the least hailpads has a total of only 59 of them, while the grid box with most hailpads has 577.

Considering only these 6692 hailpads/241 stations, the mean value of the number of hailpads collected per station in each grid box shows that, on average, hail is locally observed once per year (Fig. 13a). Only one grid box in the northeastern corner of the network receives an average of about 1.5 hailpad per station per year, while the lowest frequency is observed in the southeastern corner, with only 0.7 hailpad per year per station. About two-thirds of the grid boxes in the northwestern and western regions experience a mean value per station of ≥1.0 of hail events per year. This is consistent with the locations of the maximum hail occurrence shown in Fig. 3 of Manzato (2012).

Fig. 13.
Fig. 13.

The spatial distribution of the (a) mean number of hailpads per year (No. hailpad yr−1) and (b) mean areal density (No. hailstones m−2).

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

Figure 13b shows the spatial mean value of the areal density, with a mean value of 790 hailstones per meter squared. The maximum values of Density are clustered on the northern side of the domain, meaning that hailstorms seem to produce, on average, a larger number of hailstones toward the foothills than in the plain and coastal areas.

Figure 14a shows the spatial mean value of the D50, which has an average value (for all grid boxes) of 7.4 mm. This spatial distribution is very peculiar in that it has the maximum values clustered in the southwestern corner, that is, on the plain area along the border between the FVG and Veneto regions of Italy. Thus, while hailstorms are typically more likely in the northwestern corner of the studied area (plus a hot spot also in the northeastern corner), hailstones seem to have larger diameters in the southwestern corner of the FVG plain.

Fig. 14.
Fig. 14.

The spatial distribution of the mean value of (a) D50 (mm) and (b) Dmax (mm).

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

Figure 14b shows the spatial distribution of Dmax, which has a mean value of 11.9 mm and a maximum value of 13.7 mm, again in the southwestern corner. However, looking at all the grid boxes with Dmax > 12 mm, there seem to be a “large-hail alley” extending from the southwestern corner toward northeast. The alley is almost parallel to the southwest-to-northeast pre-Alpine chain,6 but shifted by about 35 km toward the plain. Hence, the hail size spatial distribution could be related to the particular orography, which forms a barrier to the southerly flow of moist air from the Adriatic Sea, as explained in Davolio et al. (2016). Moreover, Pucillo et al. (2020) described a possible mechanism where a cold flow, moving down from the pre-Alps, could play a role in enhancing the severity of storms passing on the plain, ahead of the southwest-to-northeast steep barrier. Additional studies should be done to see if such a mechanism could also be responsible for the increase in average hail size inside the large-hail alley.

Figure 15a shows the spatial distribution of the absolute maximum diameter collected inside each of the grid boxes. Of course, this parameter is statistically less robust than what shown in Fig. 14, because it is based on a single hailstone observation per grid box. However, apart from the absolute maximum value (a hailstone diameter of 46 mm) which is located on the northwestern corner, all the other hailstones with diameters larger than 40 mm are located in the southwestern–northeastern large-hail alley.

Fig. 15.
Fig. 15.

The spatial distribution of the (a) absolute maximum diameter (mm) and (b) mean flux of kinetic energy (J m−2).

Citation: Journal of Applied Meteorology and Climatology 61, 11; 10.1175/JAMC-D-21-0251.1

Last, Fig. 15b shows the spatial distribution of KeFlux, which has a mean value of 27 J m−2. The largest fluxes of kinetic energy are mainly observed on the western and northern sides of the domain, probably showing a combined effect of the grid boxes with the largest Dmax and those with the largest Density.

6. Summary and conclusions

Hailstone data from hailpads collected from April to September (1988–2016) for 367 stations located over the plain of the Friuli Venezia Giulia region (northeast Italy), were analyzed using a semiautomatic algorithm, based on the hailpad TITAN program. Despite the fact that these data are managed by volunteers, interesting insights on the hailstone size distribution and the spatiotemporal distribution of hail characteristics can be derived with a relatively good reliability. This was possible because of the large sample size involved: 7782 hailpads with a total of 747 759 valid hailstones.

Analyzing the hail dents and their aspect ratios it is found that a reliable threshold for detecting small hail is 2.8 mm (hail detection “sensitivity” of the hailpad). This corresponds to the dent left by a spherical hailstone with diameter of 6 mm. Fitting the distribution of all the hailstones with a diameter equal or larger than 6 mm, it is found that a simple exponential law like Eq. (7) describes 99% of the hail size distribution but underestimates the largest diameters, above 20 mm. For the tail of diameters larger than 20 mm the log-logistic distribution of Eq. (8) represents a much better fit, even if it slightly overestimates the distribution of diameters below 10 mm. In practice, one can use the probability mass function of Eq. (7) below 10 mm and that of Eq. (8) above that threshold. The empirical distribution and also the maximum hailstone observed are in agreement with the hailstone distributions found by Sánchez et al. (2009) when analyzing hailpads in Spain, France, and Argentina, where the maximum diameters observed were in the 44–52-mm range.

Although the maximum estimated diameters observed in our study may seem small, this is because a hailpad only detects hail over a very small area. Consequently, the probability of a hailpad detecting the largest hailstone from a hailstorm is very small (Smith and Waldvogel 1989; Grieser and Hill 2019). In fact, in the plain of FVG human observers has frequently reported pictures of large (50–60 mm) hailstones, and on 1 August 2021 some hailstones up to 90 mm were collected on the western part of the FVG plain.

Another interesting result of this work is the high linear correlation between Dmax and KeFlux (R = 0.69), which increases to R = 0.75 if Dmax is elevated to the power of 2.6 [Eq. (10) and Fig. 8]. This is consistent with Eq. (11) in Grieser and Hill (2019), which was developed using the Fraile et al. (2003) dataset. In practice, that approximation is an easy way to estimate the flux of kinetic energy on an area, given only information on the maximum hailstone diameter inside that area. However, to verify if the relation also holds for areas larger than a small hailpad, additional field experiments are required.

The interannual cycle for the number of hailpads (Fig. 9a) and for Dmax (Fig. 10b) do not show any clear trend. The interannual cycle of D50 (Fig. 10c) shows an increase with time, with a linear Pearson correlation coefficient of R = 0.50. In contrast, Density (Fig. 10a, but partially also KeFlux in Fig. 10d), shows a decrease with time, with a linear correlation of R = 0.68. In practice, in the last studied years hailstorms seem to produce fewer hailstones but with a larger size, on average.

The day-of-year cycles from 15 April to 15 September show a peak between mid-June and mid-July for the number of hailpads (Fig. 9b), whereas the average values for Dmax, D50, and KeFlux all peak in the first half of August (Figs. 11b,c,d). Given that the maximum frequency of cloud-to-ground lightning in the “plain subarea” of northeast Italy peaks at the beginning of August (Fig. 6c of Feudale and Manzato 2014), this result confirms the strong relation between hail and lightning flashes (e.g., Takahashi 1978; Lopez 2016; Nisi et al. 2020). A unique seasonal cycle is found for Density (Fig. 11a), with a bimodal distribution and the absolute maximum found at the end of the period (mid-September), suggesting that—at least for that characteristic—hail should be monitored also during October.

The diurnal cycle has a very clear trend for the number of hailpads (Fig. 9c), but it is not so for other hailstone characteristics (Fig. 12). Despite the fact that the peak at 0600 UTC is probably a nonmeteorological signal, Figs. 12b, 12c, and 12d still show a weak indication that during the night hail could be larger than during the rest of the day, on average.

Last, to study the spatial distribution, only 6692 hailpads are used in order to provide a robust spatial sampling on twenty-four 15 km × 15 km grid boxes. The hailpad areal density (Fig. 13b) tends to increase toward the foothills (northern side), while the average Dmax per grid box (Fig. 14b) tends to be larger within a southwestern to northeastern orientated “hail alley,” which is parallel to the pre-Alpine crest (the orographic line located on the left of the northwestern border of the hailpad network in Fig. 1). However, the largest values for the grid box mean of D50 and Dmax are observed over the southwestern corner of the network. This suggests that also the plain area to the west of the FVG (i.e., the plain of the Veneto region) likely experiences large hail events with considerable frequency.

Since many of the results found are in good agreement with previous studies, like for example the climatology of cloud-to-ground lightning in northeastern Italy, one can use these data with confidence. For example, these data can be used for testing hail-forecasting models or remote sensing hail-detection algorithms.

1

The volunteer should inspect visually the hailpad after each thunderstorm, looking for hailstone marks, but not all volunteers exercise the same amount of diligence or have the same ability to detect marks, in particular since the small ones can be very hard to see.

2

The “hailpad” version of TITAN should not be confused with the TITAN program used to analyze radar images because it is an adapted version of it made by M. Dixon to study black-and-white hailpad images.

3

Note that these are probability mass distributions and do not need to be integrated over an interval to give the probability value, as happens with the probability density functions.

4

All of these correlations are statistically significant, having a p value that is less than 2 × 10−16.

5

The last 15 days of September are not shown because data end on 30 September and there is the ±15-day moving-average filter.

6

This southwest-to-northeast orographic line corresponds to the crest of the pre-Alps located on the left of the northwestern side of the hailpad network, shown in Fig. 1. Manzato et al. (2022) have shown that the southwest-to-northeast pre-Alpine crest is a hot spot of both cloud-to-ground lightning density and convective-initiation events.

Acknowledgments.

First of all, the authors thank the “hail volunteers,” who have managed the hailpad stations for many years. The same applies to the ARPA personnel who worked on this project during all these years. Thanks are given to Julian Brimelow (Environment and Climate Change Canada) for fruitful discussion on an earlier version of this work. Gabriele Fasano (OSMER–ARPA FVG) and, in particular, Rich Rotunno (NCAR) helped to improve the text. The director of OSMER–ARPA FVG (Fulvio Stel) strongly supported this work. The authors kindly thank the two reviewers for all their precious suggestions. This work is dedicated to the memory of Griffith Morgan (1934–2017), who established the hail network in FVG.

Data availability statement.

Because data used in this work are collected by volunteers and are subject to different kinds of errors, access to individual hailpad data is not available. However, an aggregated version of this dataset is freely available (https://www.meteo.fvg.it/grandine.php), as explained in the appendix.

APPENDIX

Hailpad-Based Database for the Scientific Community

As has been explained in this work, but also in Manzato (2012), hailpad data collected by volunteers are far from being a “perfect dataset” and should not be used as “pointwise” values. For example, if in a given area of FVG plain at a given time there are no impacted hailpads it is not certain that hail did not really occur, because it may be the case that volunteers in that area were not able to check the hailpads on that occasion. For this reason, these data have been used up to now only to derive cumulative behaviors, as in this climatological work or in the case study, having more than 100 impacted hailpads, that was discussed in Manzato et al. (2020).

On the other hand, for the scientific community to have the so-called hail ground truth, as data collected from these hailpads, would give it a unique source for validating possible forecasting models or remote sensing hail detection algorithms, as those derived from radar (e.g., Waldvogel et al. 1979; Witt 1998; Bechini and Chandrasekar 2015; Nisi et al. 2016) or from satellite (e.g., Punge et al. 2017; Bang and Cecil 2019; Laviola et al. 2020). For this reason, in recent years ARPA FVG–OSMER has received many requests from the scientific community to share these data.

What has been done (with a huge effort to preserve a decent data quality) to answer this request is to develop a dataset that is intermediate between the original pointwise data and the totally aggregated climatology. In particular, on the ARPA FVG website (https://www.meteo.fvg.it/grandine.php), it will be possible to access the data used for this work, presented in a time- and space-aggregated format. In practice, the FVG plain has been divided into four big subareas, and, for each day from April to September 1988–2016, it is reported for each subarea the total number of hailpads, the average Density, the average D50, the average KeFlux, and the absolute maximum Dmax. Even if this dataset is aggregated on relatively large areas (encompassed by a 45 km × 30 km rectangle) and relatively long time intervals (24 h) there are still some uncertainties. For example, considering all four subareas together and defining as “FVG hail day” all the days with at least one hailpad in the FVG plain, then one gets a mean value of 46 hail days per year, whereas using at least two hailpads (which is probably a more reliable signal) causes this frequency to drop to only 27 hail days per year. The database is freely available since December 2021. ARPA FVG–OSMER must be cited as data provider, although it will not assume any responsibility of the use of these data.

REFERENCES

  • Adams-Selin, R. D., and C. L. Ziegler, 2016: Forecasting hail using a one-dimensional hail growth model within WRF. Mon. Wea. Rev., 144, 49194939, https://doi.org/10.1175/MWR-D-16-0027.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bang, S. D., and D. J. Cecil, 2019: Constructing a multifrequency passive microwave hail retrieval and climatology in the GPM domain. J. Appl. Meteor. Climatol., 58, 18891904, https://doi.org/10.1175/JAMC-D-19-0042.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bechini, R., and V. Chandrasekar, 2015: A semisupervised robust hydrometeor classification method for dual-polarization radar applications. J. Atmos. Oceanic Technol., 32, 2247, https://doi.org/10.1175/JTECH-D-14-00097.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brimelow, J. C., G. W. Reuter, and E. R. Poolman, 2002: Modeling maximum hail size in Alberta thunderstorms. Wea. Forecasting, 17, 10481062, https://doi.org/10.1175/1520-0434(2002)017<1048:MMHSIA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cacciamani, C., F. Battaglia, P. Patruno, L. Pomi, A. Selvini, and S. Tibaldi, 1995: A climatological study of thunderstorm activity in the Po Valley. Theor. Appl. Climatol., 50, 185203, https://doi.org/10.1007/BF00866116.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cecil, D. J., 2009: Passive microwave brightness temperatures as proxies for hailstorms. J. Appl. Meteor. Climatol., 48, 12811286, https://doi.org/10.1175/2009JAMC2125.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cecil, D. J., and C. B. Blankenship, 2012: Toward a global climatology of severe hailstorms as estimated by satellite passive microwave imagers. J. Climate, 25, 687703, https://doi.org/10.1175/JCLI-D-11-00130.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Changnon, S. A., 1973: Hail sensing and small-scale variability of windblown hail. J. Wea. Modif., 5, 3042.

  • Changnon, S. A., 1977: The scales of hail. J. Appl. Meteor., 16, 626648, https://doi.org/10.1175/1520-0450(1977)016<0626:TSOH>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Changnon, S. A., D. Chagnon, and S. D. Hilberg, 2009: Hailstorms across the nation: An atlas about hail and its damages. Illinois State Water Survey Contract Rep. 2009-12, 101 pp., https://www.isws.illinois.edu/pubdoc/cr/iswscr2009-12.pdf.

    • Search Google Scholar
    • Export Citation
  • Davolio, S., A. Volonté, A. Manzato, A. Pucillo, A. Cicogna, and M. E. Ferrario, 2016: Mechanisms producing different precipitation patterns over northeastern Italy: Insights from HyMeX-SOP1 and previous events. Quart. J. Roy. Meteor. Soc., 142, 188205, https://doi.org/10.1002/qj.2731.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dieling, C., M. Smith, and M. Beruvides, 2020: Review of impact factors of the velocity of large hailstones for laboratory hail impact testing consideration. Geosciences, 10, 500, https://doi.org/10.3390/geosciences10120500.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dixon, M., and G. Wiener, 1993: TITAN: Thunderstorm Identification, Tracking, Analysis, and Nowcasting—A radar-based methodology. J. Atmos. Oceanic Technol., 10, 785797, https://doi.org/10.1175/1520-0426(1993)010<0785:TTITAA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Eccel, E., P. Cau, K. Riemann-Campe, and F. Biasioli, 2012: Quantitative hail monitoring in an alpine area: 35-year climatology and links with atmospheric variables. Int. J. Climatol., 32, 503517, https://doi.org/10.1002/joc.2291.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Feudale, L., and A. Manzato, 2014: Cloud-to-ground lightning distribution and its relationship with orography and anthropogenic emissions in the Po Valley. J. Appl. Meteor. Climatol., 53, 26512670, https://doi.org/10.1175/JAMC-D-14-0037.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fraile, R., C. Berthet, J. Dessens, and J. L. Sánchez, 2003: Return periods of severe hailfalls computed from hailpad data. Atmos. Res., 67–68, 189202, https://doi.org/10.1016/S0169-8095(03)00051-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Giaiotti, D., E. Gianesini, and F. Stel, 2001: Heuristic considerations pertaining to hailstone size distributions in the plain of Friuli Venezia Giulia. Atmos. Res., 57, 269288, https://doi.org/10.1016/S0169-8095(01)00080-1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Giaiotti, D., S. Nordio, and F. Stel, 2003: The climatology of hail in the plain of Friuli Venezia Giulia. Atmos. Res., 67–68, 247259, https://doi.org/10.1016/S0169-8095(03)00084-X.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grieser, J., and M. Hill, 2019: How to express hail intensity—Modeling the hailstone size distribution. J. Appl. Meteor. Climatol., 58, 23292345, https://doi.org/10.1175/JAMC-D-18-0334.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Groenemeijer, P., and Coauthors, 2017: Severe convective storms in Europe: Ten years of research and education at the European Severe Storms Laboratory. Bull. Amer. Meteor. Soc., 98, 26412651, https://doi.org/10.1175/BAMS-D-16-0067.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Heymsfield, A., M. Szakáll, A. Jost, I. Giammanco, and R. Wright, 2018: A comprehensive observational study of graupel and hail terminal velocity, mass flux, and kinetic energy. J. Atmos. Sci., 75, 38613885, https://doi.org/10.1175/JAS-D-18-0035.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Heymsfield, A., M. Szakáll, A. Jost, I. Giammanco, R. Wright, and J. Brimelow, 2020: Corrigendum. J. Atmos. Sci., 77, 405412, https://doi.org/10.1175/JAS-D-19-0185.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jewell, R., and J. Brimelow, 2009: Evaluation of Alberta hail growth model using severe hail proximity soundings from the United States. Wea. Forecasting, 24, 15921609, https://doi.org/10.1175/2009WAF2222230.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kumjian, M. R., and K. Lombardo, 2020: A hail growth trajectory model for exploring the environmental controls on hail size: Model physics and idealized tests. J. Atmos. Sci., 77, 27652791, https://doi.org/10.1175/JAS-D-20-0016.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kumjian, M. R., K. Lombardo, and S. Loeffler, 2021: The evolution of hail production in simulated supercell storms. J. Atmos. Sci., 78, 34173440, https://doi.org/10.1175/JAS-D-21-0034.1.

    • Search Google Scholar
    • Export Citation
  • Kunz, M., J. Sander, and C. Kottmeier, 2009: Recent trends of thunderstorm and hailstorm frequency and their relation to atmospheric characteristics in southwest Germany. Int. J. Climatol., 29, 22832297, https://doi.org/10.1002/joc.1865.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kunz, M., U. Blahak, J. Handwerker, M. Schmidberger, H. J. Punge, S. Mohr, E. Fluck, and K. M. Bedka, 2018: The severe hailstorm in southwest Germany on 28 July 2013: Characteristics, impacts and meteorological conditions. Quart. J. Roy. Meteor. Soc., 144, 231250, https://doi.org/10.1002/qj.3197.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Laviola, S., V. Levizzani, R. R. Ferraro, and J. Beauchamp, 2020: Hailstorm detection by satellite microwave radiometers. Remote Sens., 12, 621, https://doi.org/10.3390/rs12040621.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Löffler-Mang, M., D. Schön, and M. Landry, 2011: Characteristics of a new automatic hail recorder. Atmos. Res., 100, 439446, https://doi.org/10.1016/j.atmosres.2010.10.026.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lopez, P., 2016: A lightning parameterization for the ECMWF Integrated Forecasting System. Mon. Wea. Rev., 144, 30573075, https://doi.org/10.1175/MWR-D-16-0026.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Manzato, A., 2012: Hail in northeast Italy: Climatology and bivariate analysis with the sounding-derived indices. J. Appl. Meteor. Climatol., 51, 449467, https://doi.org/10.1175/JAMC-D-10-05012.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Manzato, A., V. Riva, A. Tiesi, and M. M. Miglietta, 2020: Observational analysis and simulations of a severe hailstorm in northeastern Italy. Quart. J. Roy. Meteor. Soc., 146, 35873611, https://doi.org/10.1002/qj.3886.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Manzato, A., S. Serafin, M. M. Miglietta, D. Kirshbaum, and W. Schulz, 2022: A pan-Alpine climatology of lightning and convective initiation. Mon. Wea. Rev., 150, 22132230, https://doi.org/10.1175/MWR-D-21-0149.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Martius, O., A. Hering, M. Kunz, A. Manzato, S. Mohr, L. Nisi, and S. Trefalt, 2018: Challenges and recent advances in hail research. Bull. Amer. Meteor. Soc., 99, ES51ES54, https://doi.org/10.1175/BAMS-D-17-0207.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Miglietta, M. M., A. Manzato, and R. Rotunno, 2016: Characteristics and predictability of a supercell during HyMeX SOP1. Quart. J. Roy. Meteor. Soc., 142, 28392853, https://doi.org/10.1002/qj.2872.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Morgan, G. M., Jr., 1973: A general description of the hail problem in the Po Valley of northern Italy. J. Appl. Meteor., 12, 338353, https://doi.org/10.1175/1520-0450(1973)012<0338:AGDOTH>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Morgan, G. M., Jr, 1992: Some results of aircraft investigation of internal properties of thunderstorm cloud systems in northeastern Italy with an interpretation for hail prevention. Atmos. Res., 28, 259269, https://doi.org/10.1016/0169-8095(92)90012-Y.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Morgan, G. M., Jr, and N. G. Towery, 1975: Small-scale variability of hail and its significance for hail prevention experiments. J. Appl. Meteor., 14, 763770, https://doi.org/10.1175/1520-0450(1975)014<0763:SSVOHA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Morgan, G. M., Jr, and P. W. Summers, 1986: Hailfall and hailstorm characteristics. Thunderstorms: A Social and Technological Documentary, E. Kessler, Ed., Vol. 2, 2nd ed. University of Oklahoma Press, 363408.

    • Search Google Scholar
    • Export Citation
  • Murillo, E. M., C. R. Homeyer, and J. T. Allen, 2021: A 23-year severe hail climatology using GridRad MESH observations. Mon. Wea. Rev., 149, 945958, https://doi.org/10.1175/MWR-D-20-0178.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nisi, L., O. Martius, A. Hering, M. Kunz, and U. Germann, 2016: Spatial and temporal distribution of hailstorms in the Alpine region: A long-term, high resolution, radar-based analysis. Quart. J. Roy. Meteor. Soc., 142, 15901604, https://doi.org/10.1002/qj.2771.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nisi, L., A. Hering, U. Germann, K. Schroeer, H. Barras, M. Kunz, and O. Martius, 2020: Hailstorms in the Alpine region: Diurnal cycle, 4D-characteristics, and the nowcasting potential of lightning properties. Quart. J. Roy. Meteor. Soc., 146, 41704194, https://doi.org/10.1002/qj.3897.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Palencia, C., A. Castro, D. Giaiotti, F. Stel, and R. Fraile, 2011: Dent overlap in hailpads: Error estimation and measurement correction. J. Appl. Meteor. Climatol., 50, 10731087, https://doi.org/10.1175/2010JAMC2457.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Prodi, F., 1976: Hailstorms in the Po Valley of northern Italy. Riv. Ital. Geofis., 3, 2131.

  • Púčik, T., C. Castellano, P. Groenemeijer, T. Kühne, A. T. Rädler, B. Antonescu, and E. Faust, 2019: Large hail incidence and its economic and societal impacts across Europe. Mon. Wea. Rev., 147, 39013916, https://doi.org/10.1175/MWR-D-19-0204.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pucillo, A., M. M. Miglietta, K. Lombardo, and A. Manzato, 2020: Application of a simple analytical model to the study of a bow echo–like storm in northeastern Italy. Meteor. Appl., 27, e1868, https://doi.org/10.1002/met.1868.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Punge, H. J., and M. Kunz, 2016: Hail observations and hailstorm characteristics in Europe: A review. Atmos. Res., 176–177, 159184, https://doi.org/10.1016/j.atmosres.2016.02.012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Punge, H. J., K. M. Bedka, M. Kunz, and A. Werner, 2014: A new physically based stochastic event catalog for hail in Europe. Nat. Hazards, 73, 16251645, https://doi.org/10.1007/s11069-014-1161-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Punge, H. J., K. M. Bedka, M. Kunz, and A. Reinbold, 2017: Hail frequency estimation across Europe based on a combination of overshooting top detections and the ERA-INTERIM reanalysis. Atmos. Res., 198, 3443, https://doi.org/10.1016/j.atmosres.2017.07.025.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sánchez, J. L., and Coauthors, 2009: Characterization of hailstone size spectra in hailpad networks in France, Spain, and Argentina. Atmos. Res., 93, 641654, https://doi.org/10.1016/j.atmosres.2008.09.033.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schuster, S. S., R. J. Blong, and M. S. Speer, 2005: A hail climatology of the greater Sydney area and New South Wales, Australia. Int. J. Climatol., 25, 16331650, https://doi.org/10.1002/joc.1199.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shedd, L., M. R. Kumjian, I. Giammanco, T. Brown-Giammanco, and B. R. Maiden, 2021: Hailstone shapes. J. Atmos. Sci., 78, 639652, https://doi.org/10.1175/JAS-D-20-0250.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Smith, P. L., and A. Waldvogel, 1989: On determinations of maximum hailstone sizes from hailpad observations. J. Appl. Meteor. Climatol., 28, 7176, https://doi.org/10.1175/1520-0450(1989)028<0071:ODOMHS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Strong, G. S., and E. P. Lozowski, 1977: An Alberta study to objectively measure hailfall intensity. Atmosphere, 15, 3353, https://doi.org/10.1080/00046973.1977.9648429.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Takahashi, T., 1978: Riming electrification as a charge generation mechanism in thunderstorms. J. Atmos. Sci., 35, 15361548, https://doi.org/10.1175/1520-0469(1978)035<1536:REAACG>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Taszarek, M., J. T. Allen, T. Púčik, K. A. Hoogewind, and H. E. Brooks, 2020: Severe convective storms across Europe and the United States. Part II: ERA5 environments associated with lightning, large hail, severe wind, and tornadoes. J. Climate, 33, 10 26310 286, https://doi.org/10.1175/JCLI-D-20-0346.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Towery, N. G., S. A. Changnon, and G. M. Morgan, 1976: A review of hail measuring instruments. Bull. Amer. Meteor. Soc., 57, 1132–1141, https://doi.org/10.1175/1520-0477(1976)057<1132:AROHMI>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vittori, O., and G. di Caporiacco, 1959: The density of hailstones. Nubila, 2, 5157.

  • Waldvogel, A., W. Schmid, and B. Federer, 1978a: The kinetic energy of hailfalls. Part I: Hailstone spectra. J. Appl. Meteor., 17, 515520, https://doi.org/10.1175/1520-0450(1978)017<0515:TKEOHP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Waldvogel, A., B. Federer, W. Schmid, and J. F. Mezeix, 1978b: The kinetic energy of hailfalls. Part II: Radar and hailpads. J. Appl. Meteor., 17, 16801693, https://doi.org/10.1175/1520-0450(1978)017<1680:TKEOHP>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Waldvogel, A., B. Federer, and P. Grimm, 1979: Criteria for the detection of hail cells. J. Appl. Meteor., 18, 15211525, https://doi.org/10.1175/1520-0450(1979)018<1521:CFTDOH>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weickmann, H. K., 1953: Observational data on the formation of precipitation in cumulonimbus clouds. Thunderstorm Electricity, H. R. Byers, Ed., University Chicago Press, 66138.

    • Search Google Scholar
    • Export Citation
  • Witt, A., M. D. Eilts, G. J. Stumpf, J. T. Johnson, E. D. W. Mitchell, and K. W. Thomas, 1998: An enhanced hail detection algorithm for the WSR-88D. Wea. Forecasting, 13, 286303, https://doi.org/10.1175/1520-0434(1998)013<0286:AEHDAF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • World Meteorological Organization, 1988: Manual on codes. Vol. 1, WMO Tech Doc. 306, 203 pp.

Save
  • Adams-Selin, R. D., and C. L. Ziegler, 2016: Forecasting hail using a one-dimensional hail growth model within WRF. Mon. Wea. Rev., 144, 49194939, https://doi.org/10.1175/MWR-D-16-0027.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bang, S. D., and D. J. Cecil, 2019: Constructing a multifrequency passive microwave hail retrieval and climatology in the GPM domain. J. Appl. Meteor. Climatol., 58, 18891904, https://doi.org/10.1175/JAMC-D-19-0042.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bechini, R., and V. Chandrasekar, 2015: A semisupervised robust hydrometeor classification method for dual-polarization radar applications. J. Atmos. Oceanic Technol., 32, 2247, https://doi.org/10.1175/JTECH-D-14-00097.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brimelow, J. C., G. W. Reuter, and E. R. Poolman, 2002: Modeling maximum hail size in Alberta thunderstorms. Wea. Forecasting, 17, 10481062, https://doi.org/10.1175/1520-0434(2002)017<1048:MMHSIA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cacciamani, C., F. Battaglia, P. Patruno, L. Pomi, A. Selvini, and S. Tibaldi, 1995: A climatological study of thunderstorm activity in the Po Valley. Theor. Appl. Climatol., 50, 185203, https://doi.org/10.1007/BF00866116.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cecil, D. J., 2009: Passive microwave brightness temperatures as proxies for hailstorms. J. Appl. Meteor. Climatol., 48, 12811286, https://doi.org/10.1175/2009JAMC2125.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cecil, D. J., and C. B. Blankenship, 2012: Toward a global climatology of severe hailstorms as estimated by satellite passive microwave imagers. J. Climate, 25, 687703, https://doi.org/10.1175/JCLI-D-11-00130.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Changnon, S. A., 1973: Hail sensing and small-scale variability of windblown hail. J. Wea. Modif., 5, 3042.

  • Changnon, S. A., 1977: The scales of hail. J. Appl. Meteor., 16, 626648, https://doi.org/10.1175/1520-0450(1977)016<0626:TSOH>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Changnon, S. A., D. Chagnon, and S. D. Hilberg, 2009: Hailstorms across the nation: An atlas about hail and its damages. Illinois State Water Survey Contract Rep. 2009-12, 101 pp., https://www.isws.illinois.edu/pubdoc/cr/iswscr2009-12.pdf.

    • Search Google Scholar
    • Export Citation
  • Davolio, S., A. Volonté, A. Manzato, A. Pucillo, A. Cicogna, and M. E. Ferrario, 2016: Mechanisms producing different precipitation patterns over northeastern Italy: Insights from HyMeX-SOP1 and previous events. Quart. J. Roy. Meteor. Soc., 142, 188205, https://doi.org/10.1002/qj.2731.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dieling, C., M. Smith, and M. Beruvides, 2020: Review of impact factors of the velocity of large hailstones for laboratory hail impact testing consideration. Geosciences, 10, 500, https://doi.org/10.3390/geosciences10120500.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dixon, M., and G. Wiener, 1993: TITAN: Thunderstorm Identification, Tracking, Analysis, and Nowcasting—A radar-based methodology. J. Atmos. Oceanic Technol., 10, 785797, https://doi.org/10.1175/1520-0426(1993)010<0785:TTITAA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Eccel, E., P. Cau, K. Riemann-Campe, and F. Biasioli, 2012: Quantitative hail monitoring in an alpine area: 35-year climatology and links with atmospheric variables. Int. J. Climatol., 32, 503517, https://doi.org/10.1002/joc.2291.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Feudale, L., and A. Manzato, 2014: Cloud-to-ground lightning distribution and its relationship with orography and anthropogenic emissions in the Po Valley. J. Appl. Meteor. Climatol., 53, 26512670, https://doi.org/10.1175/JAMC-D-14-0037.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fraile, R., C. Berthet, J. Dessens, and J. L. Sánchez, 2003: Return periods of severe hailfalls computed from hailpad data. Atmos. Res., 67–68, 189202, https://doi.org/10.1016/S0169-8095(03)00051-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Giaiotti, D., E. Gianesini, and F. Stel, 2001: Heuristic considerations pertaining to hailstone size distributions in the plain of Friuli Venezia Giulia. Atmos. Res., 57, 269288, https://doi.org/10.1016/S0169-8095(01)00080-1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Giaiotti, D., S. Nordio, and F. Stel, 2003: The climatology of hail in the plain of Friuli Venezia Giulia. Atmos. Res., 67–68, 247259, https://doi.org/10.1016/S0169-8095(03)00084-X.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grieser, J., and M. Hill, 2019: How to express hail intensity—Modeling the hailstone size distribution. J. Appl. Meteor. Climatol., 58, 23292345, https://doi.org/10.1175/JAMC-D-18-0334.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Groenemeijer, P., and Coauthors, 2017: Severe convective storms in Europe: Ten years of research and education at the European Severe Storms Laboratory. Bull. Amer. Meteor. Soc., 98, 26412651, https://doi.org/10.1175/BAMS-D-16-0067.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Heymsfield, A., M. Szakáll, A. Jost, I. Giammanco, and R. Wright, 2018: A comprehensive observational study of graupel and hail terminal velocity, mass flux, and kinetic energy. J. Atmos. Sci., 75, 38613885, https://doi.org/10.1175/JAS-D-18-0035.1.

    • Crossref