Observations of precipitation with X-band and C- band polarimetric radars in Piedmont region (Italy)

Transcription

1 Observations of precipitation with X-band and C- band polarimetric radars in Piedmont region (Italy) R.Cremonini 1, L Baldini 2, E. Gorgucci 2, V. Romaniello 2, R.Bechini 1, V. Campana 1 1 Arpa Piemonte, Sistemi Previsionali, Torino, Italia, r.cremonini@arpa.piemonte.it 2 CNR Istituto di Scienze dell'atmosfera e del Clima, Roma, Italia 1. Introduction Roberto Cremonini Networks of meteorological radar are used in many countries to provide reliable estimates with high spatial and temporal resolution. Networks are designed to provide seamless coverage with the lowest number of radar and minimizing the region of overlapping. However, there are specific cases where the coverage of radar networks is not adequate. At farther distance a radar samples at increasing heights, due both to earth curvature and to the need of using low but greater than zero elevation angle, and with broad resolution volumes (Chandrasekar et al. 2004). Moreover, many regions, especially in Europe, are characterized by complex orography and therefore the coverage of operational C-band networks is often inadequate or even non-existent due to partial or total blocking of the radar beam. This problem can be partially mitigated by resorting to sampling precipitation at higher elevation, determining an increase in the decorrelation between rainfall at the ground and precipitation sampled by radar. X-band radars can offer a viable solution both to fill observational gaps of these networks in mountainous regions to improve the observation where spatial sampling is not adequate or also to improve the observation of phenomena that needs higher time and space resolution than that provided by national radar networks. Although less expensive and easily transportable, they are affects by attenuation due to propagation through precipitation that determines important errors in radar rainfall estimation obtained from parameters derived from power measurements. The use of dualpolarization techniques in X-band radar systems has provided solutions to mitigate the impact of attenuation. This fact has revived in the recent years the interest of scientific and operational community on these systems.. One advantage of X-band is the improved sensitivity, especially as far as differential phase measurements are concerned, that allows building up reliable algorithms for rainfall estimation based on differential phase shift, a parameter based on phase of radar return, immune to attenuation caused by propagation, and at some extent, to beam blocking. This feature increases usefulness of X-band polarimetric systems in estimating precipitation in mountainous area. Using opportunities of several EU-funded or National projects, (namely FRAMEA, Alcotra programme to test polarimetric Doppler X-band radar in Alpine regions; ForeAlps, Alcotra programme , to test low-cost X-band microradar), ARPA Piemonte manages the mobile X-band polarimetric radar ARX since The radar has been sited in Carmagnola (TO), 25 km far from Turin over the plains, and will be moved to an Alpine area within CRISTAL project activities, to collect data in mountainous context. The X-band radar has a good visibility towards Bric della Croce (TO), the operational polarimetric Doppler C-band radar of ARPA Piemonte. This paper illustrates the synergy between C-band and X-band observations by the analysis of the evolution of a daughter cell within a multicell hailstorm that interested the Western Po valley on July 1st 2009 exploiting the high resolution volumes and vertical observations provided by the X-band radar. 2. A case study in the Po valley, NW Italy On July 1st 2009, a stable anticyclonic synoptic configuration started to become increasingly unstable due to cooler air advection from the East. The 12 UTC Milano sounding reveals a moderate convective instability (CAPE of 1200 J/kg). The typical daily convective activity over the Alps foothills began to spread over adjacent plains in the early afternoon and around 16 UTC (18 local time) several heavy rain and hail cells were moving mainly South- Eastwards, with the mean mid-tropospheric flow. The Cuneo sounding also reveals a weak East-South-Easterly flow in the lower levels (2 m/s average intensity between the surface and 2 km MSL) turning to North-Westerly in the mid and upper troposphere, with increasing intensity with height (6 m/s at 5 km MSL, 20 m/s at 9 km MSL). The resulting moderate wind-shear in the cloud layer is approximately 3x10-3 s -1, which is a typical value of environment condition for multicell storms Marwitz (1972).

2 2.1. Data The radar data used in this study are from two polarimetric radar. One is an operational system at C-band and the second is the ARX system. The Bric della Croce radar is a C-band system located on the top of the Torino hill (736 m MSL), providing continuous monitoring over North-Western Italy with a range of 170 km, range resolution of 340 m and beam angular resolution of 1.0 deg. The new transportable X-band polarimetric system (for system characteristics, see Bechini et al 2008) was deployed over the plains (235 m MSL), just 16 km South of the C-band radar for test purposes, and provided measurements up to 50 km range, with high spatial resolution: 125 m range resolution, 1.3 deg azimuth and elevation resolution. Both systems provide measurements of reflectivity, Doppler velocity and spectrum width, differential reflectivity, correlation coefficient and differential phase shift. Fig 1 illustrates an estimate of bias correction obtained from a self-consistency based calibration (Gorgucci et al. 1992) over a complete scan using data collected at an elevation of 4-degrees. The NW sector ( deg azimuth) is blanked by purpose (radiation switched off), due to the presence of a close building. Beams with peaks of bias correction bias indicate heavy blocking from nearby obstacles (trees, high tension power lines). Both radars performed volume scans with a 5-minute update, using a set of 11 elevations between 0 and 30 deg. ARX performed an additional vertical pointing scan (90 deg elevation) within the 5 minutes loop. The radar is transportable and can operate with or without radome (Bechini et al., 2008); during the event considered it was operating with the radome, which implies that during heavy precipitation over the radar, radome wetting can cause an important attenuation (Kurri and Huuskonen, 2008; Bechini et al., 2010), in addition to the typical beam propagation loss. The attenuation due to the wet radome is estimated by comparison with the disdrometer calculated reflectivity and the vertical pointing measurements are corrected accordingly. The path attenuation is corrected using the ZPHI algorithm (Testud et al., 2000). At the X-band radar site a fixed meteorological station including a tipping bucket rain-gauge (0.2 mm resolution) is available. In addition, a laser disdrometer ( was installed close to the radar trailer. The disdrometer provides particle counts in 22 classes of diameters (0.16 mm to > 8 mm) and 20 classes of velocity (0.2 m/s to 20 m/s). During the most intense precipitation phase at the site (17:23 UTC), the X-band radar shut-down following a power supply interruption. The disdrometer also switched off, while the rain-gauge kept operating with the auxiliary batteries. The radar system eventually rebooted at 17:45 UTC. FIG. 1. Bias correction as a function of azimuth at 4.0 deg elevation

3 3. Mesoscale evolution and precipitation As it is often the case in the region around the city of Torino, cumulus convection was orographically triggered on the North-Western flank of the Torino hill (740 m MSL, roughly m above the surrounding plains). Figure 2 shows a series of PPIs plotted on a reduced 16 km range, to enhance the spatial details of the cells evolution. As the Torino cell (denoted A) started to move Southward, a new cell (denoted B) developed just West of the X-band radar around 17 UTC, likely generated by the combined effect of the cold pools associated with the Torino storm and the Southernmost convective cells. The high resolution X-band data allows to clearly distinguish two well defined line-shaped daughter cells associated with the gust fronts of cells A and B (fig. 3, 17:00 and 17:05 UTC). The daughter cell of A develops on the right forward flank respect to the main core and propagates South- Eastwards, while the daughter cell of B forms on the forward flank and propagates Eastwards. The two cells A and B eventually merged roughly at the X-band location between 17:10 and 17:15 UTC. At 17:13 precipitation started to reach the surface at the radar site. Fig. 3 shows the time series of the rain-gauge and disdrometer data. A first hail burst was observed with the disdrometer at 17, as highlighted by the jump to 60 dbz in the reflectivity calculated from the disdrometer. The rain-gauge did not record anything yet at this time, partly due to the intrinsic delay in the tipping-bucket mechanism (no tips until 0.2 mm accumulation is reached), and because this first burst was mainly composed of hail. In the following minutes, between 17:16 and 17:20, precipitation mainly in the form of rain, with a minor component of small hail occurred, with the rain rate reaching the maximum intensity of 180 mm/h at 17:20 as confirmed by the one-minute disdrometer DSD (particles diameter < 6.5 mm, synop code 88/89: ice pellet/light hail showers); and from the good agreement between the disdrometer rain rate estimates, which is calculated integrating over all the observed particles, and the rain-gauge measurements. FIG. 2. PPI at 6.9 deg elevation, range 16 km, from the X-band radar from 16:50 until 17:15 UTC. Letters A and B mark the two cells approaching the radar site.

4 For the last three available minutes before power failure (17:21-17:23) the disdrometer indicates hail showers, with relevant concentration of particles with D > 6.5 mm, and the important ice phase contribution to precipitation is also evident in the increasing deviation of the disdrometer-based rain rate from the rain-gauge data (Fig. 3). Overall, the rain-gauge measured 40 mm of rainfall in less than 30 minutes at the radar site. 3. X-band vs. C-band observations: daughter cell growth The radar measures the Doppler velocity, which at vertical incidence is given by V D = w W T, where w is the air vertical velocity and W T is the particles terminal velocity. In order to get an estimate of the updraft velocity, an empirical relation (Joss and Waldvogel, 1970) has been applied to calculate the terminal fall velocity from the reflectivity: W T = 2.6 Z (1) where the reflectivity Z is expressed in mm 6 m -3. The relation (1) is derived for liquid particles, so its application to the whole vertical cloud extent may imply significant estimation errors. In fact while the liquid drops terminal fall speed varies in the range 0-9 m/s, ice particles have fall speed much weaker (~0-2 m/s) and hailstones can reach over 20 m/s. In fig. 4 a height-time diagram from the vertical pointing X-band reflectivity (colour) and estimated updraft velocity (contour) is presented. Between 16:30 and 16:45 the anvil cloud from the approaching storm is visible in the 6-10 km height layer. The upward motion in the anvil (~2 m/s) is likely overestimated, due to the presence of ice particles with lower terminal speed than liquid drops. More noticeable is the onset of the updraft around cloud base between 16:45 and 16:55. The updraft with intensity > 5 m/s extends from 2 to 4 km height, slightly above the freezing level, represented by the blue horizontal line at 3.7 km. Five minutes later, at 17:00, the marked increase in reflectivity associated with the developing daughter cloud (Z > 40 dbz at 4.4 km height) is accompanied by a temporary decrease in the updraft intensity, which is now between 0 and 3.7 m/s in the km layer, just above the reflectivity maximum. This phase in the storm evolution above the radar appears central for the future development of hail. In fact a relatively weak updraft is crucial to maintain for a sufficient time the hailstone embryos in the mixed-phase region of the cloud (Foote, 1984). At 17:05 the updraft reached a maximum estimated intensity of 24 m/s at 4.7 km height, corresponding to an environment temperature of -7 C. This would correspond to a forecast of pea to grape hail size, according to the nomogram of Renick and Maxwell (1977). FIG. 3. Time series of disdrometer and rain-gauge data. The vertical light blue thin sectors indicate the time and duration of the vertical pointing X-band radar scan, for comparison.

5 Figure 5 shows vertical cuts at 17:00 and 17:05, during the initial radar echo increase associated with the daughter cell, for both radars, along the black line in fig. 2 (upper-left panel). Those vertical cuts are obtained from the volume scan, interpolating in elevation at a given azimuth (192 deg for the C-band radar; two-side cut for the X-band: 192 deg and 12 deg). Since the data at increasing elevations are collected at different times (spanning a total of 4'30'') a correction for the intervening advection has been applied on each elevation, based on an estimate of the cell motion of 6.5 m/s Southwards. The vertical pointing scan of the X-band radar is graphically expanded for display purposes in fig. 4 (right column). Although the C-band radar (left column) provides a good coverage in the region around the X-band system, the latter allows to resolve fine details at lower levels and in the vertical (right column). The horizontal black line in these and in the following vertical cuts marks the height of the 0 C level (3.7 km MSL), while the height of all the vertical cuts is 12 km MSL. The vertical line in the middle of the C-band cuts corresponds to the location of the X-band site and the purple thick line in the lower part of the plots indicates the position of the gust front, as inferred from the analysis of the radar Doppler velocity. The reflectivity seen by the vertical pointing X- band increase from >40 dbz to >50 dbz between 17:00 and 17:05 (reflectivity from the C-band is slightly lower due to path attenuation). The C-band scan allows to assess that the daughter cloud is developing within the overhang of the main storm, whose core is still around 12 km to the North. The gust front approaching the X-band site is located 4 km North at 17:00 and 3 km North at 17:05. The succeeding storm evolution is represented in fig. 6, showing the vertical cuts from the C-band system at 17:15 and 17:25. The gust front reached the X-band site around 17:15, when heavy precipitation began to be observed at the surface. The most intense precipitation phase with hail and rain rate above 120 mm/h went on until 17:25, time when the core of the storm is seen to have moved further Southwards. In the same sequence (fig. 6) it is possible to clearly see the dissipation of the older cell to the North and the rapid intensification of the daughter cell, which eventually became the main core around 17:15. The maximum intensity (Z > 60 dbz) is reached at 17:20 aloft and at 17:25 in the lower levels. FIG. 4. Time-height diagram from 16:30 to 17:15 UTC of reflectivity (color) and estimated updraft (contours: thick lines above 5 m/s, dashed lines for negative values), see text for details. The times on the x-axis indicate the nominal scan time (beginning of the volume scan), the actual time of the vertical pointing measurements is about 4 minutes later. The horizontal blue line indicates the height of the 0 C level.

6 FIG. 5. Vertical cuts of reflectivity from the C-band radar (left panels) and from the X-band (right panels) at 17:00 and 17:05 UTC. The vertical scan from the X-band system is expanded for display purposes. The position of the gust front, derived from the Doppler data is marked in purple. The height of these and all following vertical cuts is 12 km MSL. As in many multicell storms reported in the literature (Foote, 1984; Kennedy and Detwiler, 2003), one may speculated that the hail embryos originated from ice particles in the anvil cloud (negative Doppler at vertical incidence above 7 km height) or by horizontal advection in the shelf cloud at mid-levels, and then grew mainly through accretion of supercooled drops. The role of supercooled drops can be understood considering Z dr from the X-band radar. Columns of enhanced Z dr extending above the ambient 0 C level are considered a signature of an updraft in thunderstorms (Bringi et al. 1991; Brandes et al. 1995; Loney et al. 2002). In fact positive Z dr and relatively low reflectivity imply the presence of oblate hydrometeors, either in the form of liquid drops in low concentration or drops with ice cores. In fig. 7, representing a West-East vertical cut through the cell named B in fig. 2, a Z dr column is extending until approximately 6 km height. The significance of this Z dr signature is confirmed by the Doppler velocity (fig. 7, lowerright panel), showing a split updraft arising from the South-Easterly inflow: the Westernmost positive Doppler velocity signature is co-located with the Z dr column. This means that supercooled drops until 6 km height were not uncommon in the storm updraft. And the 4-6 km height layer is also were the strongest daughter cell development was observed in the initial phase, between 17:00-17:05 (fig. 3). Acknowledgment This work have been co-funded by European Union - European funds for regional development - IT-F ALCOTRA Programme project CRISTAL n 008.

7 Figure 6. Vertical cuts of reflectivity from the C-band radar at 17:15 and17:20 UTC. 16:55 16:55 FIG. 7. Upper-left: vertical cut of reflectivity along a West- East section at 16:55 UTC obtained from the X-band radar. Upper-right: differential reflectivity Z dr. Lower-right: Doppler velocity. No correction for advection is applied in this case, since during this initial phase the cell appeared to be mostly stationary. 16:55 References Battan, L. J., 1975: Doppler radar observations of a hailstorm. J. Appl. Meteor., 14, Bechini R., Cremonini, R., Campana, V., Tomassone, L. and V. Chandrasekar, 2008: A Transportable X-Band polarimetric radar in Italy for deployment in complex terrain: results of the first year measurement campaign. Proceedings from the Fifth European Conference on Radar in Meteorology and Hydrology ERAD 2008, Helsinki, Finland, Bechini R., Chandrasekar V., Cremonini R., Lim S., 2010: Radome attenuation at X-band radar operations, Proceedings from the Sixth European Conference on Radar in Meteorology and Hydrology ERAD 2010, Sibiu, Romania,2010. Brandes, E. A., J. Vivekanandan, J. D. Tuttle, and C. J. Kessinger, 1995: A study of thunderstorm microphysics with multiparameter radar and aircraft observations. Mon. Wea. Rev., 123, Bringi, V. N., D. A. Burrows, and S. M. Menon, 1991: Multiparameter radar and aircraft study of raindrop spectral evolution in warm-based clouds. J. Appl. Meteor., 30, Chandrasekar, V., S. Lim, N. Bharadwaj, W. Li, D. McLaughlin, V. N. Bringi, and E. Gorgucci, 2004b: Principles of networked weather radar operation at attenuating frequencies. Proc. ERAD 2004, Gotland, Sweden, ERAD, Gorgucci, E., G. Scarchilli, and V. Chandrasekar, 1992: Calibration of radars using polarimetric techniques. IEEE Trans. Geosci. Remote Sens., 30, Joss, J. and A. Waldvogel. In Preprints of 14th Rad. Meteor. Conf., 1970, pp Kurri, M. and A. Huuskonen, 2008: Measurements of the transmission loss of a radome at different rain intensities. J. Atmos. Oceanic. Technol., 25, Loney, M. L., D. S. Zrnić, J. M. Straka, and A. V. Ryzhkov, 2002: Enhanced polarimetric radar Zdr signatures above the melting level in a supercell storm. J. Appl. Meteor., 41, Renick, J. H., and J. B. Maxwell, 1977: Forecasting hailfall in Alberta. Hail: A review of hail science and hail suppression, Meteor. Monogr., No. 38, Amer. Meteor. Soc., Testud J., E. Le Bouar, E. Obligis, M. A.-Mehenni, 2000: The Rain Profiling Algorithm Applied to Polarimetric Weather Radar. Journal of Atmospheric and Oceanic Technology, 17:3,