APRIL 2015 V A N D E N B R O E K E A N D V A N D E N B R O E K E 329 Polarimetric Radar Observations from a Waterspout-Producing Thunderstorm MATTHEW S. VAN DEN BROEKE Department of Earth and Atmospheric Sciences, University of Nebraska—Lincoln, Lincoln, Nebraska CYNTHIA A. VAN DEN BROEKE Lincoln, Nebraska (Manuscript received 20 September 2014, in final form 20 January 2015) ABSTRACT A family of four waterspouts was produced by a convective cell over western Lake Michigan on 12 September 2013. This storm initiated along a boundary north of a mesolow in a low-level cold-air advection regime, and developed supercell characteristics once the second waterspout was in progress. Polarimetric characteristics of the storm, and of the development of supercell character, are presented. These observations represent the first documented polarimetric radar observations of waterspout-producing convection in the Great Lakes region. Unusually high differential reflectivity values accompanied this storm and its initiating boundary. The high values along the boundary are partially explained by a high density of dragonflies. High differential reflectivity values were present through much of the storm of interest despite very low aerosol concentration at low levels in the lake-influenced air mass. Finally, this case illustrates the importance of environmental awareness on waterspout-favorable days, especially when boundaries are nearby to serve as a potential source of enhanced environmental vertical vorticity. 1. Introduction and motivation Conditions favorable for waterspout development in- clude low-level instability, low-level shear, and possibly A waterspout is defined as ‘‘any tornado over a body of slow-moving or intersecting gust fronts (Simpson et al. water’’ (Glickman 2000), and waterspouts display all the 1986). In addition, waterspout-producing cloud lines diversity in behavior, appearance, and origin of their kin typically develop under weak synoptic disturbances in over land. In North America, waterspouts most com- the presence of differential heating or sea surface tem- monly occur in the Florida Keys (50–500 waterspouts per perature gradients (Golden 1974a; Simpson et al. 1986). year) and along the southeast coast of Florida (Golden There were 46 waterspouts per year on average from 1977) but have been observed on the Great Lakes (e.g., 1994 to 2010 over the Great Lakes (Sioutas et al. 2013). Gay 1921; Hurd 1928), the Great Salt Lake (Simpson Waterspouts were observed on every lake, though Lake et al. 1991), and even Lake Tahoe (Grotjahn 2000). Erie had the highest annual occurrence (Sioutas et al. Golden (1974b) first proposed a five-stage waterspout 2013). On Lake Michigan, 173 waterspouts were ob- life cycle based on observations of Florida Keys water- served from 1993 to 2013 (W. Szilagyi 2014, personal spouts. Most of these waterspouts occurred in a tropical communication). Most waterspouts on the Great Lakes, environment, developed in cloud lines, and were non- including Lake Michigan, occur in the months of August supercellular in origin. In fact, much of the waterspout and September when the water surface temperature is literature based on larger field projects has examined relatively warm (Szilagyi 2004; Sioutas et al. 2013). storms in the tropics or subtropics, and waterspouts Sioutas et al. (2013) identified other conditions favor- forming through primarily nonsupercell processes (e.g., able for waterspout outbreaks on the Great Lakes, such Golden 1974a; Leverson et al. 1977; Simpson et al. 1986). as a 500-hPa long-wave trough or a closed low over the region, increased instability from cold advection, and in Corresponding author address: Matthew S. Van Den Broeke, 306 some cases a land breeze. Bessey Hall, Lincoln, NE 68588-0340. On the afternoon of 12 September 2013, a series of E-mail: [email protected] four waterspouts developed over western Lake Michigan DOI: 10.1175/WAF-D-14-00114.1 Ó 2015 American Meteorological Society Unauthenticated | Downloaded 10/07/21 09:27 AM UTC 330 WEATHER AND FORECASTING VOLUME 30 (NWS 2013). The waterspout-producing storm, which of rhv and ZDR were consistent with these expectations was well observed from the polarimetric Weather Sur- when averaged over several points at each of five times veillance Radar-1988 Doppler (WSR-88D) at Milwaukee, examined between 1759 and 1840 UTC (not shown), Wisconsin (KMKX), appeared to become more super- indicating no consistent, substantial ZDR bias. cellular in nature while the waterspouts were in progress. Radar data were supplemented by additional observa- Using these radar data in conjunction with environmental tions, including routine upper-air and surface observa- and aerosol data, this study provides the following: tions. Rapid Refresh (RAP) model data at 1800 UTC 12 September 2013 were obtained from the National Cli- 1) the first published polarimetric radar observations of maticDataCenter(NCDC).Thesedatawereusedto a waterspout-producing storm in the Great Lakes estimate the sounding and storm-relative helicity (SRH) region, near the waterspout-producing storm. Maps of Lake 2) a chronology of polarimetric features associated with Michigan water surface temperature, estimated using an the transition to supercell convection, Advanced Very High Resolution Radiometer (AVHRR) 3) in situ observations of biological scatterers contrib- satelliteborne instrument, were obtained from the Great uting to high differential reflectivity Z values DR Lakes Environmental Research Laboratory (GLERL). along a boundary, and These data were limited by patchy cloud cover over 4) the potential occurrence of a drop size distribution southern Lake Michigan, but portions of the lake offshore (DSD) biased toward unusually large liquid drops from Wisconsin and northeastern Illinois were cloud free. despite very low observed aerosol concentrations. Aerosol data, including particulate matter with a diameter This case is of particular interest given the small number less than 10 mm (PM10), were obtained from the Envi- of prior observational studies of waterspout-producing ronmental Protection Agency (EPA) for a station near the storms in the Great Lakes region, and given the poten- Lake Michigan shoreline (indicated as red star in Fig. 1). tial for substantial human impacts had the storm been displaced only a small distance toward the land. 3. Overview of the synoptic- and local-scale environment 2. Data and methods A long-wave trough over the Great Lakes character- A radar dataset was analyzed from KMKX, which was ized the environment at 1200 UTC 12 September 2013. upgraded to polarimetric capability in April 2012. This The trough axis was located from Hudson Bay through dataset extended from the time a linear reflectivity central Ontario and Wisconsin, just west of Lake maximum first appeared east of KMKX (1455 UTC) Michigan (Fig. 2). Two jet streaks were evident at until the storm of interest moved well southeast of 300 hPa: the first on the west side of the trough axis over KMKX (2029 UTC). The storm of interest was within Minnesota and the second to the east over lower Mich- 120 km of KMKX throughout this period, minimizing igan (Fig. 2). As the trough moved eastward through data quality issues inherent at long range. All heights the region, model output indicated that by 1800 UTC noted in this paper are above radar level (ARL). Po- the trailing jet streak was in a favorable location for larimetric radar variables utilized included ZDR, which southeastern Wisconsin to experience synoptic-scale lift, affords an estimate of the reflectivity-weighted mean with strong northwest flow at 300 hPa (Fig. 3a). The axis ratio of scatterers in a sample volume, and copolar eastward-moving trough brought strong 850-hPa cold- cross-correlation coefficient rhv, which provides an in- air advection to the western Great Lakes (Fig. 2). At the dication of scatterer diversity, orientation, and phase surface, a cold front had passed through the region (e.g., Bringi and Chandrasekar 2001). overnight and by 1800 UTC was located from central Very high ZDR values observed within storms Illinois through north-central Indiana (Fig. 2). Much of throughout the KMKX domain on this day were initially the region was dominated by northwest surface flow and suspected of being in error, so a scatterer-based cali- a gradual northward surface temperature decline. These bration procedure was implemented to ensure no large conditions, especially the long-wave trough and surface ZDR bias. It was assumed that most hydrometeors cold front with attendant northwesterly flow, are typical should be dry snow aggregates ;1.5 km above the of a Great Lakes waterspout outbreak environment melting level. Typical values of radar variables in such (Szilagyi 2004; Sioutas et al. 2013). Surface-based in- hydrometeors include radar reflectivity factor at a hori- stability was relatively weak at 1800 UTC, with typical 21 zontal polarization ZHH between 20 and 35 dBZ, rhv . values of 200–300 J Kg across southeastern Wisconsin 0.97 (often .0.99), and ZDR averaging 0.1–0.2 dB and western Lake Michigan (Fig. 3b), according to the (Ryzhkov et al. 2005a; Picca and Ryzhkov 2012). Values 1800 UTC RAP initialization. Unauthenticated | Downloaded 10/07/21 09:27 AM UTC APRIL 2015 V A N D E N B R O E K E A N D V A N D E N B R O E K E 331 FIG. 1. Mesoscale features present at 1800 UTC 12 Sep 2013, when convection was already in progress. Station plots include temperature and 2 dewpoint (8C) and wind [knots (kt; 1 kt 5 0.51 m s 1); full barb 5 10 kt; half barb 5 5 kt]. Dewpoint is color shaded, and black contours represent surface pressure (contour interval 5 1 hPa). White star represents location of KMKX, and red star represents location of Chicago aerosol monitoring site. Square represents location of RAP sounding in Fig. 3. Triangle shows location of the IBSP Hawk Watch site.