Abstract

This paper presents an analysis of a hail event that occurred 27 May 2012 over Brignoles, located in southeastern France. The event was observed by an X-band polarimetric radar located in Mont Maurel, 75 km northeast of the hailstorm. Lightning data from the French national network (owned and operated by Météorage) are also used in the study. The analysis highlights that the lightning and radar data provide complementary information that may allow a better microphysical interpretation of the hailstorm and potentially increase the probability of its detection.

1. Introduction

The potential of polarimetry for better detection and discrimination of weather phenomena has been thoroughly established by the radar community [see Zrnić and Ryzhkov (1999) for a review]. However, most studies also highlight the difficulty of discriminating between hailing and nonhailing events at the ground (see, e.g., Tabary et al. 2010). There are multiple reasons for this. First, because hail is usually a very localized phenomenon, different hydrometeors may coexist within the radar resolution volume. Second, there is large overlap between the membership functions of hail and intense rain. Third, radar measurements (particularly at long ranges) are at high altitudes and small- and medium-sized hail may melt before reaching the ground. Therefore, it seems necessary to use a multisensor approach to better discriminate between hailing and nonhailing storms. Because of the different intervening mechanisms, lightning activity may vary significantly between hail- and nonhail-producing events. Articles showing the correlation between polarimetric radar data and lightning activity abound (see, e.g., Bruning et al. 2007; Tessendorf et al. 2007; Lund et al. 2009).

Recent studies have also shown that polarimetric variables may exhibit significantly different responses at different wavelengths in the presence of hail because of the different scattering mechanisms at each frequency. For example, both Anderson et al. (2011) and Picca and Ryzhkov (2012) showed that differential reflectivity Zdr values tend to be much larger, and copolar correlation coefficient ρhv values tend to be much smaller in the core of the hailstorm at C band than at S band. X-band weather radars are increasingly used by weather services for specific purposes (i.e., urban hydrology). However, relatively few in-depth analyses of polarimetric data at X band in cases of hailstorms exist in the literature. For example, Cremonini et al. (2010) presented a case study where a storm was observed by a C-band nonpolarimetric radar and an X-band polarimetric radar. Unfortunately the X-band radar suffered a power failure during the event and only data from the onset of the hailstorm could be recorded. Suzuki et al. (2012) examined several hailstorms at X band. They concluded that X-band polarimetric data are suitable to detect pure hail because, in its presence, the polarimetric variables present distinct behavior from that in the presence of rain. They state that Zdr may be high but the specific differential phase Kdp remains low, and thus they propose methods based on Kdp for hail detection. Most recently, Matrosov et al. (2013) recommended the use of Kdp to minimize the overestimation of rainfall caused by hail after examining a rain–hail mixed precipitation event. To further expand the record of observations, this paper reports on a hailstorm event that occurred 27 May 2012 in Brignoles (southeastern France) and that was observed by an X-band radar used as “gap filler” in the French Alps. The region is often affected by flash floods and hail events. However, the large extension of the damaged area and the intensity of this particular event were rather exceptional.

Unlike in the previously mentioned papers dealing with X-band radar, in this paper lightning data are also used to obtain a better insight into the microphysics of the event. As it will be shown, the evolution of the polarimetric signatures is in good agreement with the evolution of the lightning activity. Most notable is the existence of large pockets of negative specific differential phase at high altitude, which is what drew our attention to the event in the first place. The article is structured as follows: section 2 provides an overview of the instrumentation and section 3 discusses the synoptic situation, the in situ data, and the evolution and extent of the storm according to local witnesses. Section 4 analyses the lightning and radar observations, and section 5 discusses the relation between radar and lightning data and provides a tentative relation to the microphysics of the storm. General conclusions are summarized in section 6.

2. Instrumentation and data processing

Météo-France is currently deploying a network of four X-band polarimetric weather radars in the framework of the Risques Hydrométéorologiques en Territoires de Montagnes et Mediterranéens (RHYTMME) project [see Kabeche et al. (2012) for details]. One of the radars already operational is installed on top of Mont Maurel (latitude 44°00′46″N, longitude 06°31′45″E, altitude 1778 m MSL), that is, approximately 75 km northeast of Brignoles. It is a SELEX-Gematronik Meteor 50DX radar transmitting 75 kW of peak power. The antenna's one-way 3-dB beamwidth is 1.3° and the pulse width is 2 μs, which corresponds to a nominal range resolution of 300 m. The data are oversampled to a polar grid of 240 m × 0.5°. The raw data (horizontal reflectivity Zh, Zdr, ρhv, and copolar differential phase φdp) are processed on site by an in house–built radar processor [Chaîne d'Acquisition, de Surveillance et de Traitement des Observations Radar 2 (CASTOR2)] and transmitted to a server in Toulouse, France, that completes the data processing and produces derived products such as rainfall rate and reflectivity composite. The signal processing is discussed in detail in Figueras i Ventura et al. (2012, 2013). It includes the filtering of the raw φdp using a moving median filter of 25 range bins (6 km) to reduce the influence of phase noise and the backscatter copolar correlation phase δco, Kdp is estimated from the filtered φdp using a linear regression, and Zh and Zdr are corrected for precipitation-induced attenuation using a linear relation between attenuation and filtered φdp. The quantitative precipitation estimation (QPE) is performed using a hybrid algorithm whereby Kdp is used to estimate rainfall rate in intense rain (Kdp above 1° km−1) and hail, and Zh corrected for attenuation is used otherwise.

The lightning sensor network is the French national network owned and operated by Météorage and is also called Météorage. It is composed of 24 lightning sensors, LS7001, from Vaisala. In addition, data are exchanged with the neighboring countries, so that the detection level is uniform all over the country. The sensor detects electromagnetic emissions produced by lightning at frequencies from 0.4 to 350 kHz. Each sensor is provided with a precise GPS-synchronized clock, which allows dating the event at temporal resolutions of microseconds. The sensor provides in real time the date, the angle of arrival, the intensity, and the polarity of the event to a central processor model [Total Lightning Processor (TLP)] from Vaisala that calculates the exact location of the event by both the time of arrival method and the direction finding method [see Cummins and Murphy (2009) and Cummins et al. (1998) for a review of the methodology]. The detection efficiency is at least 95% over France, with a precision smaller than 500 m for the cloud-to-ground (CG) flashes. For the cloud-to-cloud (CC) flashes, the detection efficiency is much lower. Over the southeast of France, the ratio of CC to CG is 30% on average, while for a typical storm there are on the order of 6–7 times more CC than CG flashes.

3. Synoptic situation and storm evolution

A very violent hailstorm occurred on 27 May 2012 in the area of Brignoles, in the Var region. According to the local newspaper Var Matin, it was rather sunny until 1600 (1400 UTC) local time when it became very cloudy. Around 1750 (1550 UTC) was the onset of the hailstorm, lasting for 15–20 min. The hailstones had diameters of up to 3 cm, and the hail accumulation on the ground was up to 8 cm. A posteriori evaluations signaled around 50–60 km2 of damaged vineyards. The hailstorm also affected traffic on the A8 highway, north of the village. Figure 1 shows the approximate extent of the damaged area.

Fig. 1.

Approximate extent of the hailstorm damage.

Fig. 1.

Approximate extent of the hailstorm damage.

There was no automatic weather station located within the area most affected by the hailstorm. However, the weather stations of Montfort-sur-Argens, just north of the area, recorded an hourly precipitation accumulation of 37 mm at 1600 UTC (0.5 mm in the next hour), whereas the weather station of Collobrières, to the south, registered 30.9 mm at 1700 UTC (none in the previous hour). Other stations in the area recorded much more modest values. Thus, it confirms that the hailstorm was a very localized phenomena, moving from north to south. At the same time, there was a sudden drop in temperature. The maximum temperature registered between 1500 and 1600 UTC in Montfort was 24.4°C and dropped to 15.0°C from 1600 to 1700 UTC. In Collobrières the maximum dropped from 20.5° to 16.6°C from 1600 to 1700 UTC. There was also a large increase in local wind speed. Local press reported tornado-like winds. Weather stations surrounding the area, though, measured only a modest increase in wind speed with instantaneous wind speeds of up to 44.3 km h−1 registered from 1500 to 1600 UTC in Entrecasteaux (north of the area) and 50 km h−1 from 1600 to 1700 UTC in Cuers (south of the area).

4. Observations

a. Lightning observations

A total of 432 return strokes were detected in an area of 15 km centered around latitude 43.40°N, longitude 6.17°E between 1520 and 1640 UTC. The lightning stroke positions progressed from north to south. Of these, 283 were CG, with 33 positive and 250 negative; and 149 were CC, with 120 positive and 29 negative. Figure 2 shows the evolution of the number of return strokes of each type in 5-min steps. During the period, there were two peaks of lightning activity in the area: at 1540 and at 1620 UTC (which roughly correspond to the beginning and end of the hailstorm). The ratio of positive-to-negative polarity was less than unity for most of the period; however, at 1550 UTC, there was a peak of positive activity to the northwest of the area caused by an increase in positive CC return strokes. CC activity was most intense between 1535 and 1615 UTC.

Fig. 2.

Number of detected lightning return strokes in 5-min steps.

Fig. 2.

Number of detected lightning return strokes in 5-min steps.

b. Radar observations

The radar images showed a set of intense precipitation cells rapidly moving from northwest to southeast. According to numerical weather prediction models, the Iso-0°C altitude was roughly 3300 m MSL in the area of interest. Figure 3 shows the evolution of the 5-min rainfall accumulation in the area from 1550 to 1625 UTC in 5-min intervals. This period roughly corresponds to the duration of the hail event. It should be noticed that because of the large filter window used in the filtering of φdp, the peak intensity of the convective cells may be underestimated. At 1550 UTC, an intense precipitation cell with three very strong nuclei entered the area. Significant positive lightning activity was observed ahead of the cell. At 1600 UTC, another precipitation cell appeared to the west of the area. At 1605 UTC, a new cell started developing to the south of the latter. This new cell grew in intensity and at 1615 UTC the two later cells merged. In this merged cell, a lot of negative CG lightning activity occurred. The two distinct cells traveled from northwest to southeast until they left the area of interest at 1630 UTC. The two cells yielded maximum intensity between 1610 and 1625 UTC. During this same period, there was an increase in lightning activity within the area of the precipitating cells.

Fig. 3.

Selected 5-min rainfall accumulation images (X = CG, * = CC, red = positive, yellow = negative).

Fig. 3.

Selected 5-min rainfall accumulation images (X = CG, * = CC, red = positive, yellow = negative).

The mean value in the area of interest (i.e., a region of 40 km × 40 km centered 30 km to the east and 70 km to the south of the radar) of some of the polarimetric variables was examined (see Fig. 4). At low elevation (0.4°, roughly 2500 m MSL, just below the Iso-0°C), the mean reflectivity increased rapidly from 1520 to 1545 UTC. It then remained at a value close to 35 dBZ until 1625 UTC, when it started to decrease. At high elevation (5.5°, roughly 9000 m MSL and therefore well above the Iso-0°C), the average reflectivity value reached a rather high value (around 30 dBZ) from 1525 to 1615 UTC. The mean Kdp value at low elevation had two distinctive peaks, at 1545 and at 1605 UTC. More remarkable, at high elevation it was significantly negative from 1535 to 1610 UTC. The Zdr at low altitude was slightly negative over the entire period. This may be because of an insufficient correction of the differential attenuation or to miscalibration. Another significant fact is that Zdr at high altitude was rather negative during the same period, when Kdp was negative as well. The period of predominantly negative Zdr and Kdp correlated rather well with the period with the most intense CC lightning activity, which is consistent with prior literature (Ryzhkov and Zrnić 2007).

Fig. 4.

Evolution of the average value of the polarimetric variables in the hailstorm area for elevations of (a),(c),(e) 0.4° and (b),(d),(f) 5.5° for (top) Zh, (middle) Kdp, and (bottom) Zdr.

Fig. 4.

Evolution of the average value of the polarimetric variables in the hailstorm area for elevations of (a),(c),(e) 0.4° and (b),(d),(f) 5.5° for (top) Zh, (middle) Kdp, and (bottom) Zdr.

Figure 5 shows plan position indicator (PPI) data for some variables at both high and low elevations at 1550 UTC. This time corresponds to one of the minima of the average Kdp values. It can be seen that, at high altitude, there was a very negative Kdp cell in an area with rather high Zh values. The Zdr showed radial signatures alternating between rather high values and rather low ones. Such behavior has been attributed by Ryzhkov and Zrnić (2007) to the presence of electrified ice crystals. Note that although Zdr may be insufficiently corrected for attenuation and bias, such huge variations of values from one beam to the next one can only be explained by the presence of ice crystals that are depolarizing the signal. At low elevation, there was a rather high reflectivity cell. Within this cell, Kdp had two distinctive behaviors: strongly positive values to the northeast of the cell and strongly negative values to the southwest of the cell. Interestingly, the area with negative values corresponded to an area where a lot of positive lightning activity had been detected. In this same area, a distinctive negative Zdr pattern could be seen.

Fig. 5.

As in Fig. 4, but for PPI of polarimetric variables at 1550 UTC.

Fig. 5.

As in Fig. 4, but for PPI of polarimetric variables at 1550 UTC.

Figure 6 shows the pseudo-RHI of the polarimetric variables at 206.5° from north of the same cycle as in Fig. 5. A rather large reflectivity column reaching up to 10 000 m is evident. The far end of the column corresponds to an area of very negative Kdp. In the same region, there is a dramatic drop in ρhv to levels below 0.7. The Zdr signature is rather noisy but is overall negative.

Fig. 6.

Pseudo-RHI at 1550 UTC corresponding to a cut of 206.5° from the north: (from left to right and from top to bottom) Zh, Zdr, Kdp, and ρhv.

Fig. 6.

Pseudo-RHI at 1550 UTC corresponding to a cut of 206.5° from the north: (from left to right and from top to bottom) Zh, Zdr, Kdp, and ρhv.

5. Discussion

The mean reflectivity in the area shown in Figs. 4a and 4b suggests that a large number of ice crystals have reached the cloud top, likely because of a large updraft, around 1535 UTC. These ice crystals then begin to precipitate and grow by collision, such that the reflectivity at low levels increases sharply, reaching a peak between 1545 and 1550 UTC, right at the time when local observers report the start of the hailstorm. The presence of a large number of ice crystals at the top of the cloud and graupel at the bottom leads to an increase of electrical activity within the cloud. As a consequence of this activity, the ice crystals on top of the cloud become vertically aligned, which produce a negative Kdp in the high-altitude radar measurements. In contrast, the large size of the hailstones reaching the ground results in fewer collisions, which is the cause of the reduction of the CG lightning activity experienced during the hailstorm, as suggested by Soula et al. (2004). This process reaches a peak at 1605 UTC. According to the local observers, this is the approximate end of the hail event. At this point, the number of ice crystals on top of the cloud is reduced, causing a sharp reduction of the reflectivity and from this point onward the CC activity is reduced and the average Kdp at high altitude increases toward 0 because fewer ice crystals remain vertically aligned. The average reflectivity at low level (situated just below the Iso-0°C) decreases at a lower rate and the average Zdr increases. This suggests that hailstones at this time are smaller in size and are melting at a higher rate. The reduced dominance of hailstones appears as an increase of CG activity, which reaches a peak at 1620 UTC and then diminishes sharply. It should be noted that, as it can be seen in Fig. 5, when hail first reaches the ground, there is a pocket of intense positive CG lightning activity in the area of the cell, which translates into a negative Kdp.

6. Conclusions

This paper has analyzed a hailstorm event that was observed by an X-band polarimetric radar. X-band radars are increasingly being used in meteorological applications. However, few reports on X-band polarimetric observations of hail exist and it is likely that polarimetric signatures of hail at X band vary significantly from those at S and C bands because of the different scattering mechanisms involved. Moreover, to our knowledge no report so far has attempted to correlate X-band polarimetric data with lightning data.

In the examined event, it is particularly noteworthy that the report of hail on the ground was preceded by a drastic increase in the number of ice particles at high altitude as inferred from polarimetric observations (notably, a marked increase in Zh and decrease in Kdp). This increase was likely due to a strong updraft, which has been deemed a condition sine qua non for the formation of hail (see, e.g., Tessendorf et al. 2007). Because the reports signal hail on the ground from 1550 UTC, whereas the drop in Kdp at high altitude starts already at 1535 UTC, which is a nonnegligible lead time, this suggests that at X band the value of Kdp at high altitude could be used as a predictor for severe storms. Moreover, the analysis also highlights the importance of considering the entire radar volume and not only the lower elevations, because in this case there was no distinctive hail signature at the lowest elevations.

Polarimetric radar data alone cannot provide reliable hail detection. However, as shown in numerous papers (among others, those cited in this article), hail-producing storms have a distinct lightning pattern. Therefore, it is suggested that future hail detection algorithms rely on volumetric polarimetric radar data as well as on lightning data in order to potentially improve the probability of detection and reduce the false alarm rate.

Acknowledgments

The financial support for this study was provided by the European Union, the Provence–Alpes–Côte d'Azur region, and the French Ministry of Ecology, Energy, Sustainable Development and Sea through the RHYTMME project. The lightning data were provided by Météorage.

REFERENCES

REFERENCES
Anderson
,
M. E.
,
L. D.
Carey
,
W. A.
Petersen
, and
K. R.
Knupp
,
2011
: C-band dual-polarimetric radar signatures of hail. Electron. J. Oper. Meteor., 2011-EJ02. [Available online at http://www.nwas.org/ej/pdf/2011-EJ2.pdf.]
Bruning
,
E. C.
,
W. D.
Rust
,
T. J.
Schuur
,
D. R.
MacGorman
,
P. R.
Krehbiel
, and
W.
Rison
,
2007
:
Electrical and polarimetric radar observations of a multicell storm in TELEX
.
Mon. Wea. Rev.
,
135
,
2525
2544
.
Cremonini
,
R.
,
L.
Baldini
,
E.
Gorgucci
,
V.
Romaniello
,
R.
Bechini
, and
V.
Campana
,
2010
: Observations of precipitation with X-band and C-band polarimetric radars in Piedmont region (Italy). Proc. Sixth European Conf. on Radar in Meteorology and Hydrology (ERAD 2010), Sibiu, Romania, EUMETSAT, 7 pp. [Available online at http://www.erad2010.org/pdf/oral/thursday/xband/05_ERAD2010_0338.pdf.]
Cummins
,
K. L.
, and
M. J.
Murphy
,
2009
:
An overview of lightning locating systems: History, techniques, and data uses, with an in-depth look at the U.S. NLDN
.
IEEE Trans. Electromagn. Compat.
,
51
,
499
518
.
Cummins
,
K. L.
,
M. J.
Murphy
,
E. A.
Bardo
,
W. L.
Hiscox
,
R. B.
Pyle
, and
A. E.
Pifer
,
1998
:
A combined TOA/MDF technology upgrade of the U.S. National Lightning Detection Network
.
J. Geophys. Res.
,
103
(D8),
9035
9044
.
Figueras i Ventura
,
J.
, and
P.
Tabary
,
2013
:
The new French operational polarimetric radar rainfall rate product
.
J. Appl. Meteor. Climatol.
,
52
,
1817
1835
.
Figueras i Ventura
,
J.
,
A.-A.
Boumahmoud
,
B.
Fradon
,
P.
Dupuy
, and
P.
Tabary
,
2012
:
Long-term monitoring of French polarimetric radar data quality and evaluation of several polarimetric quantitative precipitation estimators in ideal conditions for operational implementation at C-band
.
Quart. J. Roy. Meteor. Soc.
,
138
,
2212
2228
.
Kabeche
,
F.
,
J.
Figueras i Ventura
,
B.
Fradon
,
A.-A.
Boumahmoud
,
S.
Frasier
, and
P.
Tabary
,
2012
: Design and test of an X-band optimal rain rate estimator in the frame of the RHYTMME project. Proc. Seventh European Conf. on Radar in Meteorology and Hydrology (ERAD 2012), Toulouse, France, Météo-France, 6 pp. [Available online at http://www.meteo.fr/cic/meetings/2012/ERAD/extended_abs/QPE_161_ext_abs.pdf.]
Lund
,
N. R.
,
D. R.
MacGorman
,
T. J.
Schuur
,
M. I.
Biggerstaff
, and
W. D.
Rust
,
2009
:
Relationships between lightning location and polarimetric radar signatures in a small mesoscale convective system
.
Mon. Wea. Rev.
,
137
,
4151
4170
.
Matrosov
,
S. Y.
,
R.
Cifelli
, and
D.
Gochis
,
2013
: Measurements of heavy convective rainfall in presence of hail in flood-prone areas using an X-band polarimetric radar. J. Appl. Meteor. Climatol.,52, 395–407.
Picca
,
J.
, and
A.
Ryzhkov
,
2012
:
A dual-wavelength polarimetric analysis of the 16 May 2010 Oklahoma City extreme hailstorm
.
Mon. Wea. Rev.
,
140
,
1385
1403
.
Ryzhkov
,
A. V.
, and
D. S.
Zrnić
,
2007
:
Depolarization in ice crystals and its effect on radar polarimetric measurements
.
J. Atmos. Oceanic Technol.
,
24
,
1256
1267
.
Soula
,
S.
,
Y.
Seity
,
L.
Feral
, and
H.
Sauvageot
,
2004
:
Cloud-to-ground lightning activity in hail-bearing storms
.
J. Geophys. Res.
,
109
, D02101, doi:10.1029/2003JD003669.
Suzuki
,
S.
,
K.
Iwanami
,
T.
Maesaka
,
S.
Shimizu
,
N.
Sakurai
, and
M.
Maki
,
2012
: Observations of hailstorms by X-band dual polarization radar. Weather Radar and Hydrology, R. J. Moore, S. J. Cole, and A. J. Illingworth, Eds., IAHS Publ. 351, 415–420.
Tabary
,
P.
, and
Coauthors
,
2010
: Hail detection and quantification with C-band polarimetric radars: Results from a two-year objective comparison against hailpads in the south of France. Proc. Sixth European Conf. on Radar in Meteorology and Hydrology (ERAD 2010), Sibiu, Romania, EUMETSAT, 5 pp. [Available online at http://www.erad2010.org/pdf/oral/tuesday/radpol2/2_ERAD2010_0046.pdf.]
Tessendorf
,
S. A.
,
S. A.
Rutledge
, and
K. C.
Wiens
,
2007
:
Radar and lightning observations of normal and inverted polarity multicellular storms from STEPS
.
Mon. Wea. Rev.
,
135
,
3682
3706
.
Zrnić
,
D. S.
, and
A. V.
Ryzhkov
,
1999
:
Polarimetry for weather surveillance radars
.
Bull. Amer. Meteor. Soc.
,
80
,
389
406
.

Footnotes

*

Current affiliation: MeteoSwiss, Locarno, Switzerland.