Using Disdrometer, Radar, Lightning, and Model Data to Investigate

Severe Thunderstorm Microphysics

by

EVAN ANTHONY KALINA

B.S., Florida State University, 2010

A thesis submitted to the

Faculty of the Graduate School of the

University of Colorado in partial fulfillment

of the requirement for the degree of

Doctor of Philosophy

Department of Atmospheric and Oceanic Sciences

2015

This thesis entitled: Using disdrometer, radar, lightning, and model data to investigate severe thunderstorm microphysics written by Evan Anthony Kalina has been approved for the Department of Atmospheric and Oceanic Sciences.

______

Katja Friedrich

______

John Cassano

Date______

The final copy of this thesis has been examined by the signatories, and we find that both the content and the form meet acceptable presentation standards of scholarly work in the above mentioned discipline. Kalina, Evan Anthony (Ph.D., Department of Atmospheric and Oceanic Sciences)

Using disdrometer, radar, lightning, and model data to investigate severe thunderstorm microphysics

Thesis directed by Assistant Professor Katja Friedrich

Dual-polarization radar, disdrometer, lightning, and model data are analyzed to determine 1) the usefulness and accuracy of disdrometer and attenuation-corrected X-band mobile radar data from severe thunderstorms, 2) the effect of cloud condensation nuclei (CCN) concentration on idealized supercell thunderstorms, and 3) the synoptic weather, dual-polarization radar, and lightning characteristics of

Colorado plowable hailstorms.

The results in Chapter 2 demonstrate that the best agreement (1 dB in reflectivity Z and 0.2 dB in differential reflectivity ZDR) between the disdrometer and X-band radar data was obtained when the radar signal quality index (SQI) was at least 0.8 and large hail was not present. Disagreement in Z (ZDR) increased to 6 dB (1.6 dB) and 13 dB (0.6 dB) in large hail and SQI < 0.8, respectively. Since better agreement was obtained under these conditions when the disdrometer measurements were compared to S- band radar data, the X-band attenuation-correction scheme was likely responsible for the disagreement.

In Chapter 3, results from idealized supercell thunderstorm simulations in which the CCN concentration was varied from 100-10 000 cm-3 for several different environmental soundings are presented. Changes in the microphysical process rates saturated at CCN ≈ 3000 cm-3. In heavily polluted conditions (CCN = 10 000 cm-3), supercell thunderstorms formed up to 30% larger rain and 3% larger hail particles, produced up to 25 mm more precipitation near the updraft, and tracked more poleward. The area and size of the cold pool were also sensitive to the CCN concentration, especially when the low-level relative humidity was fairly dry (~60%).

Chapter 4 analyzes the synoptic weather, radar, and lightning characteristics from four severe thunderstorms that produced “plowable” hail accumulations of 15-60 cm along the Colorado Front Range.

Westerly flow at 500 hPa at slow speeds (5-15 m s-1), combined with moist upslope low-level flow, iii accompanied each hailstorm. The accumulated hail mass derived from the radar data pinpointed the times and locations of deep hail, with estimated hail depths of greater than 5 cm (less than 1.5 cm) in areas with plowable (non-plowable) hail. An increase in lightning flash rate also preceded deep hail accumulations.

iv Acknowledgements

I greatly appreciate the helpful feedback that I received from my thesis committee, Dr. George Bryan, Dr.

John Cassano, Dr. Katja Friedrich, Dr. Jeffrey Thayer, and Dr. Owen Brian Toon, throughout my time in graduate school. For their insight and contributions to the work that we published together, I would also like to thank the coauthors of my papers: Dr. George Bryan, Dr. Donald Burgess, Dr. Wiebke Deierling,

Dr. Scott Ellis, Dr. Katja Friedrich, Dr. Hugh Morrison, Brian Motta, Nezette Rydell, and Dr. Geoffrey

Stano. I am especially grateful to my thesis adviser, Dr. Katja Friedrich, who funded a portion of my graduate work, made me a part of several landmark field campaigns, including the second Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX2), and carefully read and provided thoughtful feedback on every draft of my proposals, papers, and thesis. Most importantly, Katja demonstrated to me the importance of quickly becoming an independent researcher and allowed me to pursue my own research interests. I would also like to thank my professors at Florida State University, particularly Dr. Henry Fuelberg and Dr. Robert Hart, for mentoring me as an undergraduate and for encouraging me to attend graduate school. Finally, I greatly appreciate the teaching, support, and mentorship that I continue to receive from Dr. Joseph Cione, whom I have known for 15 years as a scientist and a friend.

The work presented here would not have been possible without the generosity of the CU

Department of Atmospheric and Oceanic Sciences, which provided me with a Teaching Assistantship during my first semester, funded me to present at the AMS Cloud Physics Conference in Boston, MA in

July 2014, and provided me with the instruction and tools that I needed to complete my work. I would also like to thank the National Science Foundation, which provided me with a Graduate Research

Fellowship that fully funded the final three years of my graduate work, an exceptional benefit that allowed me to focus on my research and see it through to its completion.

v I would also like to acknowledge the incredible support that I received from several of my friends and colleagues. I was deeply inspired by my officemate, Dr. James Rudolph, who finished graduate school in four years and published four papers, all while raising a son. Josh Aikins and Brian

Vanderwende hiked frequently with me and reminded me of the importance of taking needed time away from work. I am especially grateful to one of my closest friends, Dr. Stephanie Higgins, philosopher, scientist, and chef extraordinaire and a person whom I greatly admire. Stephanie also kindly assisted me in the formatting of this thesis. I would also like to thank Zak Tamurian for his friendship and unique example of “subverting the dominant paradigm.” Finally, Elizabeth Swiman cultivated my passion for environmentalism when I was a student at Florida State and has remained an excellent friend to me (her tastes in college football notwithstanding).

It is difficult for me to express how grateful I am to my parents, Donald and MaryEllen Kalina, in a way that would do justice to their exceptional contributions to who I am. My parents have always loved and supported me beyond measure and have been thrilled by my achievements for as long as I can remember. I owe them a world of gratitude for providing me with a peaceful, loving home with everything that I needed and more, for saving for my college very early on in my life, for supporting my interest in meteorology, astronomy, geology, and construction, for showing me how important it was to take my education seriously, and, last but not least, for living in a place (Miami, Florida) that had exciting weather. My parents also never once made me feel like I needed to be anyone other than who I was and made it clear that I could make my own choices in life, which I believe to be an exceptionally rare perspective and one that I appreciate immensely.

Finally, I am enormously grateful to my partner, Rochelle Worsnop, who is also a graduate student in the Department of Atmospheric and Oceanic Sciences. The intelligence, kindness, and capacity for friendship and unconditional love that Rochelle possesses make her a true inspiration to me. She has supported me tirelessly throughout graduate school, critiqued and improved my work, and enthusiastically accompanied me into the mountains on many oddly planned backpacking trips, some of

vi which actually reached their destinations. Thank you, Rochelle, for making graduate school an enjoyable and deeply meaningful experience for me!

vii Table of Contents

1 Introduction ...... 1

2 Comparison of Disdrometer and X­band Mobile Radar Observations in Convective

Precipitation ...... 5

2.0 Abstract ...... 5

2.1 Introduction ...... 6

2.2 Cases, Instruments, and Data Collection ...... 9

2.2.1 Case selection ...... 9

2.2.2 Disdrometer measurements ...... 11

2.2.3 Radar measurements ...... 11

2.3 Data Processing ...... 13

2.3.1 Disdrometer ...... 13

2.3.1.1 Quality control and hydrometeor classification scheme ...... 13

2.3.1.2 Computation of meteorological variables from disdrometer data ...... 16

2.3.2 Radar data processing ...... 16

2.3.2.1 Radar attenuation correction scheme ...... 16

2.3.2.2 Radar hydrometeor classification scheme ...... 18

2.3.3 Radar‐disdrometer comparison method ...... 18

2.4 Results and Discussion ...... 19

2.4.1 Radar and disdrometer comparison of Z and ZDR ...... 19

2.4.1.1 X‐band Radar Z and Zdr ...... 19

2.4.1.2 S‐band radar Z ...... 20

2.4.2 17 May 2010: Supercell with radar Z and ZDR larger than disdrometer values ...... 22

2.4.3 9 June 2010: Supercell thunderstorm with radar Z and ZDR less than disdrometer

values ...... 26 viii 2.4.4 12 June 2010: Squall line with radar Z and ZDR similar to disdrometer values ...... 26

2.4.5 Radar and disdrometer hydrometeor classification comparisons ...... 32

2.5 Summary and Conclusions ...... 34

2.6 Acknowledgments ...... 36

2.7 Appendix A: Sensitivity to Hailstone Characteristics in the T­Matrix Program ...... 37

2.8 Appendix B: Sensitivity to Disdrometer Hydrometeor Classification Scheme ...... 40

3 Aerosol Effects on Idealized Supercell Thunderstorms in Different

Environments ...... 42

3.0 Abstract ...... 42

3.1 Introduction ...... 43

3.2 Methods ...... 46

3.2.1 Model configuration ...... 46

3.2.2 Microphysics scheme ...... 51

3.3 Results ...... 55

3.3.1 CCN effects on hydrometeor characteristics and microphysical processes ...... 55

3.3.2 CCN effects on cold pool size and strength ...... 67

3.3.3 CCN effects on surface precipitation ...... 69

3.3.4 Comparison to simulations with rain μ set to zero...... 73

3.4 Summary and Conclusions ...... 76

3.5 Acknowledgements ...... 79

4 An Overview of Colorado Plowable Hailstorms: Synoptic Weather,

Dual­Polarization Radar, and Lightning Data ...... 80

4.0 Abstract ...... 80

4.1 Introduction ...... 81

4.2 Data and Methods ...... 85 ix 4.2.1 Overview of cases ...... 85

4.2.2 Radar data and operational soundings ...... 85

4.2.3 Lightning data ...... 89

4.3 Results and Discussion ...... 91

4.3.1 Meteorological conditions ...... 91

4.3.2 Radar analysis ...... 98

4.3.2.1 Near‐surface radar features during hail accumulation ...... 98

4.3.2.2 Time‐height evolution of radar features ...... 104

4.3.2.3 Estimating hail accumulation from radar data ...... 110

4.3.3 Lightning and ice mass analysis ...... 111

4.4 Summary and Conclusions ...... 114

4.5 Acknowledgements ...... 116

5 Overall Conclusion ...... 118

5.1 Summary of Major Findings ...... 118

5.2 Outlook ...... 121

References ...... 123

x List of Tables

2.1: Deployment details for the cases included in this analysis. All of the cases listed are supercell thunderstorms, except for the squall line of 12 June 2010...... 9

2.2: NOXP radar characteristics for the 2010 VORTEX2 field campaign...... 13

2.3: Parameters used in the T-matrix program [i.e., canting angle (CA), axis ratio (AR), bulk density (BD), fractional water content (FWC), and temperature (T)]. The mean and standard deviation are denoted by μ and σ, respectively. For comparison to NOXP radar (WSR-88D) data, calculations were performed at a radar frequency of 9.41 GHz (2.895 GHz) and a radar elevation angle of 1° (0.5°)...... 15

2.4: Mean sensitivity in disdrometer Z and ZDR to small hail fractional water content (FWC), large hail axis ratio (AR), small and large hail canting angle standard deviation (σCA), and the disdrometer hydrometeor classification scheme for two subsets of the data in Fig. 2.5: observations from disdrometer CU01 on 17 May 2010 and observations from disdrometer UF05 on 12 June 2010. All values are relative to those obtained using the default parameters listed in Tab. 2.3 and the disdrometer hydrometeor classification scheme shown in Fig. 2.4...... 38

3.1: Relative humidity, Convective Available Potential Energy (CAPE), and Richardson number for the soundings used to initialize the WRF model in the default (def), low relative humidity (loRH), high relative humidity (hiRH), and high vertical wind shear (hiWS) cases...... 50

3.2: The percent change in microphysical and thermodynamic quantities between the cleanest (CCN = 100 cm-3) and dirtiest (CCN = 10 000 cm-3) simulations (DIRTIEST – CLEANEST) at t = 120 min for all cases: default (def), high relative humidity (hiRH), low relative humidity (loRH), high vertical wind shear (hiWS), and the default sounding with μ for rain set to zero (zero μ or ZMU). Cold pool characteristics, precipitation, and hydrometeor diameters are calculated at the lowest model level (z = 170 m)...... 62

4.1: Characteristics of Colorado plowable hailstorms in 2013-2014 derived from the Community Collaborative Rain, Hail, and Snow (CoCoRaHS) network and NOAA’s Storm Events Database. Hail times and locations correspond to the plowable hail reports, and severe weather (other than large hail) includes any tornadoes or wind gusts greater than 25 m s-1...... 82

4.2: Surface-based Convective Available Potential Energy (SBCAPE), 0-6 km AGL bulk shear, and total precipitable water vapor (PWAT) derived from Denver rawinsonde soundings (Fig. 4.5) for each of the cases listed in Tab. 4.1...... 92

xi List of Figures

2.1: Plan position indicators of attenuation-corrected radar reflectivity measured by NOXP at 1° elevation angle at a) 2212 UTC on 17 May 2010, b) 2118 UTC on 19 May 2010, c) 2320 UTC on 2 June 2010, d) 0002 UTC on 8 June 2010, e) 0130 UTC on 10 June 2010, and f) 2136 UTC on 12 June 2010. Disdrometer and radar locations are denoted by open circles and filled squares, respectively. The arrow shows the storm motion direction...... 10

2.2: A photograph of an articulating disdrometer (foreground) and a stationary disdrometer (background) deployed in Artesia, NM on 17 May 2010...... 12

2.3: An idealized schematic that shows the disdrometer and radar deployment strategy for supercell thunderstorms. The disdrometers were deployed in a line that was perpendicular to the storm motion vector with an instrument spacing of 0.2-1 km. The radar was deployed ahead of the forward-flank downdraft of the thunderstorm, and was always within 45 km of the disdrometer deployments (ideally within 15 km). The location of the 40-dBZ isoline is shown in black...... 12

2.4: Fall speed vs. diameter plot depicting the quality control procedures and the hydrometeor classification scheme applied to the disdrometer data (adapted from Fig. 5 in Friedrich et al. 2013a). The white solid lines are the empirical diameter-fall speed relationships for rain (Atlas et al. 1973), graupel (Locatelli and Hobbs 1974), and hail (Knight and Heymsfield 1983)...... 15

2.5: Comparison of radar and disdrometer observations before (a, c) and after (b, d) attenuation correction for Z (a-b) and ZDR (c-d). The gray shaded region is the sampling uncertainty of the PARSIVEL disdrometer, taken from Jaffrain and Berne (2011). Uncertainties for Z > 50 dBZ and ZDR > 3 dB are outlined in green and were obtained via linear extrapolation. Observations from the hailstorm on 17 May 2010 are plotted in red, while observations with radar SQI < 0.8 are plotted in blue. All other observations are plotted in black. Note that four of the 51 observations from the hailstorm have SQI < 0.8 and are included in the hailstorm subset. The median disagreement (radar – disdrometer) for all data is shown in the upper left, while the bottom right shows the median disagreement for each subset. The number of observations in each plot is 183, consisting of cases described in section 2.2.1 and Tab. 2.1...... 21

2.6: As in Fig. 2.5b, but for unattenuated S-band WSR-88D Z...... 22

2.7: Time series data recorded by NOXP (solid lines) and disdrometer CU01 (dashed lines) from the supercell thunderstorm with large hail (d ~ 50 mm) observed on 17 May 2010: a) attenuation-corrected radar and disdrometer reflectivity, b) attenuation-corrected radar and disdrometer differential reflectivity, c) disdrometer-observed ice volume, and d) disdrometer-observed maximum hail size. The error bars represent the sampling uncertainty of the PARSIVEL disdrometer...... 24

2.8: Accumulated particle counts recorded by disdrometer CU01 on 17 May 2010, binned by the observed fall speed and diameter. The black lines represent the empirical fall speed-diameter relationships for rain, graupel, and hail that are shown in Fig. 2.4. Hail bins are outlined in red...... 25

2.9: Plan position indicators of a) attenuation-corrected radar reflectivity and b) signal quality index for the supercell thunderstorm observed by NOXP at 1° elevation angle on 10 June 2010 at 0130 UTC. Black, open circles denote disdrometer locations. The arrow shows the direction of storm motion. The distance between each labeled tick mark is approximately 8 km in X and 11 km in Y...... 28

2.10: As in Fig. 2.7, but for the supercell thunderstorm observed by disdrometer UF01 on 9 June 2010. The radar signal quality index is shown in c)...... 29

xii 2.11: As in Fig. 2.7, but for the squall line observed by disdrometer UF05 on 12 June 2010...... 30

2.12: Plan position indicators of attenuation-corrected a) radar reflectivity and b) differential reflectivity for the squall line observed by NOXP at 1° elevation angle on 12 June 2010 at 2136 UTC. The location of disdrometer UF05 is denoted by the black, open circle, and the location of NOXP is annotated. The convective cell is outlined in blue, and an area of large radar reflectivity and differential reflectivity is circled in red. The arrow indicates the storm motion direction. The distance between each labeled tick mark is approximately 18 km in X and 11 km in Y...... 31

2.13: As in Fig. 2.8, but for disdrometer UF05 on 12 June 2010...... 31

2.14: Pie chart comparing the outputs from the disdrometer and radar hydrometeor classification schemes. The area of each sector in the pie chart is proportional to the percentage of the total number of time steps (179) included in each sector. Each sector is labeled with the class assigned by the disdrometer scheme (i.e., rain, small hail, large hail) in bold, followed by a solidus (/) and the class assigned by the radar scheme (i.e., rain, hail) in italics. The number of time steps in each sector is also listed. Sectors in which the outputs from the two schemes disagree have been separated from the rest of the chart...... 34

2.15: Sensitivity of Z (red lines) and ZDR (blue lines) from disdrometer CU01 on 17 May 2010 to a) fractional water content of small hail, b) axis ratio of large hail, and c) canting angle standard deviation of small and large hail. The sensitivities are relative to the Z and ZDR obtained by using the default values of small hail fractional water content (0.5), large hail axis ratio (0.8), and small and large hail canting angle standard deviation (50°)...... 39

2.16: As in Fig. 2.15, but for a) the precipitation type of the particles in the unclassified region in Fig. 2.4 and b) the small hail region in Fig. 2.4. The sensitivities are relative to the Z and ZDR obtained by a) excluding the unclassified particles and b) including the small hail particles...... 41

3.1: Maximum daily surface CCN (at supersaturation between 0.9 and 1.1%) and condensation nuclei (CN) number concentrations at the DOE-ARM Southern Great Plains (SGP) site from 20 April to 10 June 2011. Days with convective activity (i.e., showers and/or thunderstorms) near the SGP site are indicated in red, and days with supercell thunderstorms are shown in purple. Data were obtained from the DOE- ARM online archive (http://www.arm.gov)...... 46

3.2: Skew-T log-P diagram with the soundings used to initialize the WRF model, including the default (def) sounding and the soundings used for the sensitivity tests: low relative humidity (loRH; dashed line), high relative humidity (hiRH; dotted line), and high vertical wind shear (hiWS; rightmost wind barbs). The solid red line is the temperature profile, while the dewpoint temperature profiles are shown in blue. The wind speed and direction are represented by two sets of wind barbs on the right side of the diagram: one set for the hiWS sensitivity test and one set for all other simulations (def)...... 48

3.3: Hodograph of the wind profile used to initialize the WRF model in the high wind shear case (red line; hiWS) and in all other cases (blue line; def). Each filled circle represents an individual wind vector from the skew-T log-P diagram in Fig. 3.2. The numbers and tick marks along the red and blue lines indicate the height above the surface (in km), and the numbers along the concentric circles indicate the wind speed (in m s-1)...... 49

3.4: Horizontal cross-sections of simulated radar reflectivity (assuming a 10-cm wavelength) at z = 1 km AGL at a) t = 30 min, b) t = 60 min, c) t = 90 min, and d) t = 120 min using the default (def) sounding and a CCN concentration of 10 000 cm-3...... 56

3.5: Updraft helicity (integrated over 2-5 km AGL) from t = 10 min to t = 120 min at ten minute intervals for the CCN = 10 000 cm-3 runs of a) def, b) hiRH, c) loRH, and d) hiWS...... 57 xiii 3.6: Conditional, domain-averaged vertical profiles of hydrometeor mean mass diameter at t = 120 min for cloud droplets (green lines), rain (blue lines), and hail (purple lines) for a) def, b) hiRH, c) loRH, and d) hiWS soundings. Results from the cleanest (CCN = 100 cm-3; solid lines) and dirtiest (CCN = 10 000 cm-3; dashed lines) simulations are shown...... 58

3.7: As in Fig. 3.6, but for hydrometeor number concentration...... 59

3.8: As in Fig. 3.6, but for hydrometeor mass mixing ratio...... 60

3.9: Domain-averaged vertical profiles of the rate of cloud droplet collection by rain (green lines) and rain evaporation rate (blue lines) at t = 80 min (thin lines) and t = 120 min (thick lines) for a) def, b) hiRH, c) loRH, and d) hiWS soundings. Solid (dashed) lines represent profiles from the cleanest (dirtiest) simulation...... 64

3.10: As in Fig. 3.9, but for the rate of riming hailstones with cloud droplets (green lines), rate of riming hailstones with rain (blue lines), and the melting rate of hail (purple lines)...... 65

3.11: Vertically-integrated, horizontally averaged microphysical process rates versus CCN concentration at t = 120 min for a) def, b) hiRH, c) loRH, and d) hiWS soundings...... 66

3.12: Total area (solid lines) and mean perturbation potential temperature (dashed lines) of the cold pool at the lowest model level (z = 170 m) at t = 100 min (blue lines) and t = 120 min (red lines) versus CCN concentration for a) def, b) hiRH, c) loRH, and d) hiWS soundings...... 68

3.13: Domain-averaged, accumulated surface precipitation at t = 90 min (blue line), t = 100 min (green line), t = 110 min (yellow line), and t = 120 min (red line) versus CCN concentration for a) def, b) hiRH, c) loRH, and d) hiWS soundings...... 70

3.14: Difference in accumulated surface precipitation between the dirtiest (CCN = 10 000 cm-3) and cleanest (CCN = 100 cm-3) simulations at t = 120 min (color fill) for a) def, b) hiRH, c) loRH, and d) hiWS soundings. The purple and black contours indicate the maximum updraft speeds that were simulated at z = 5 km for the duration of the cleanest and dirtiest simulations, respectively. These contours range from 10 m s-1 to 30 m s-1 at an interval of 10 m s-1. The approximate locations of the main left- and right-moving updrafts at several times during the simulations are also indicated...... 72

3.15: As in a) Fig. 3.11, b) Fig. 3.12, c) Fig. 3.13, and d) Fig. 3.14, but for simulations with the default sounding and the shape parameter μ in the raindrop size distribution set to zero...... 74

3.16: As in a) Fig. 3.9 and b) Fig. 3.10 at t = 120 min, but for simulations with the default sounding and the shape parameter μ in the raindrop size distribution set to zero...... 75

4.1: Hail being plowed in Lakewood, CO after the 9 Sept 2013 hailstorm. Reprinted with permission from http://www.thedenverchannel.com/news/hail-rain-pours-in-lakewood-wheat-ridge. Photo credit: 7NEWS Reporter Marshall Zelinger...... 82

4.2: Maps showing the locations of hail reports (diamonds), the KFTG radar (cross), COLMA stations (squares), the center of COLMA (plus sign), and the approximate storm tracks (lines) relative to a) the elevation of the topography (km MSL) and b) the height of the center of the lowest radar beam (km AGL). Dashed lines indicate areas of beam blockage along the storm tracks. The numbers indicate a) the start and end times (UTC) of the analysis periods for each case and b) the distances (km) from the plowable hail reports to the KFTG radar (cross) and to the COLMA center (plus sign), respectively...... 87

xiv 4.3: Observations at the 500-hPa pressure level at 1200 UTC: Air temperature (°C, red numbers), dewpoint temperature (°C, green numbers), geopotential height (dm, purple numbers), and wind barbs (knots, blue) on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. Temperature (dashed thin red lines) and height (black lines) are contoured at intervals of 2 °C and 6 dm, respectively. Dashed thick red lines denote the positions of trough axes...... 93

4.4: Surface observations at 1800 UTC: Air temperature (°F, red numbers), dewpoint temperature (°F, green numbers), mean sea level pressure (hPa, large tan numbers), mean sea level pressure change relative to three hours earlier (10×hPa, small tan numbers), and wind barbs (knots, blue) on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. Mean sea level pressure (brown lines) is contoured at intervals of 4 hPa. Frontal boundaries, trough axes, dry lines, and high- and low-pressure systems are denoted by their standard symbols at the surface...... 94

4.5: Skew-T log-P diagram with air temperature (solid lines), dewpoint temperature (dotted lines), and wind velocity (barbs) at KDEN on a) 0000 UTC 4 Aug 2013 (black), b) 0000 UTC 23 Aug 2013 (blue), c) 0000 UTC 10 Sept 2013 (green), and d) 1800 UTC 21 May 2014 (red)...... 95

4.6: Bar plots of a) column-integrated precipitable water vapor and b) freezing level height from KDEN rawinsondes at 1200 UTC on the morning of the plowable hailstorm (blue) and at 0000 UTC on the evening of the plowable hailstorm (red). The monthly mean climatological values of precipitable water and freezing level height are shown in green...... 97

4.7: Constant altitude plan position indicators of reflectivity at a) 2216 UTC 3 Aug 2013 at z = 3.5 km MSL, b) 2344 UTC 22 Aug 2013 at z = 3 km MSL, c) 2107 UTC 9 Sept 2013 at z = 2.5 km MSL, and d) 2028 UTC 21 May 2014 at z = 2.5 km MSL. The black lines are contours of reflectivity from 50 dBZ to 70 dBZ at intervals of 5 dBZ. The white plus signs indicate the locations of the plowable hail reports. . 100

4.8: As in Fig. 4.7, but for differential reflectivity...... 101

4.9: As in Fig. 4.7, but for correlation coefficient...... 102

4.10: As in Fig. 4.7, but for specific differential phase...... 103

4.11: Time-height plots of the maximum reflectivity for Z ≥ 50 dBZ for the hailstorms on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. Gray lines are contours of graupel mass of (1, 5, 10, 15, and 20) × 107 kg. Black lines are contours of hail mass of (1, 3, 6, and 9) × 107 kg. The red vertical lines in the background indicate the times that plowable hail was reported. The blue horizontal lines indicate the heights of the 0 °C, -10 °C, and -25 °C isotherms from the operational soundings listed in Tab. 4.2...... 105

4.12: As in Fig. 4.11, but for the minimum differential reflectivity...... 106

4.13: As in Fig. 4.11, but for the minimum correlation coefficient...... 107

4.14: As in Fig. 4.11, but for the median specific differential phase...... 108

4.15: Accumulated hail depths estimated from the radar data on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. Squares indicate the locations of the plowable hail reports. Inferred areas of accumulating hail that occurred in sparsely populated locations are circled...... 112

4.16: Time series of storm total graupel mass (blue lines), lightning flash rate (black solid lines), and the area of the 40 dBZ-isoecho at the approximate height of the -10 °C isotherm (red lines) for the hailstorms

xv on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. The dashed black lines indicate the times that plowable hail was reported...... 113

xvi List of Abbreviations

AGL Above Ground Level

ARM Atmospheric Radiation Measurement

ARW Advanced Research WRF

CAPE Convective Available Potential Energy

CAPPI Constant-Altitude Plan Position Indicator

CCN Cloud Condensation Nuclei

CoCoRaHS Community Collaborative Rain, Hail, and Snow [Network]

COLMA Colorado Lightning Mapping Array

FIR Finite Impulse Response [Filter]

IHOP International H2O Project

KDEN Denver, Colorado

KFTG Front Range Airport, Colorado

LST Local Standard Time

MSL Mean Sea Level

NCAR National Center for Atmospheric Research

NCL NCAR Command Language

NOA National Observatory of Athens

NOAA National Oceanic and Atmospheric Administration

NOXP NOAA X-band, Dual-Polarized

PARSIVEL Particle Size and Velocity

PIADP Path-Integrated Differential Attenuation

PIAH Path-Integrated Attenuation at Horizontal [Polarization]

xvii PID Particle Identification [Scheme]

PSD Particle Size Distribution

PWAT Column-Integrated Precipitable Water Vapor

RH Relative Humidity

SCWC Self-Consistent With Constraints

SGP Southern Great Plains

SHV Simultaneous Horizontal and Vertical [Polarization]

SNR Signal-to-Noise Ratio

SQI Signal Quality Index

TBS Three-Body Scattering

USD United States Dollars

UTC Coordinated Universal Time

VCP Velocity Coverage Pattern

VHF Very High Frequency

VORTEX2 Second Verification of the Origins of Rotation in Tornadoes Experiment

WRF Weather Research and Forecasting [Model]

WSR-88D Weather Surveillance Radar – 1988 Doppler

xviii 1 INTRODUCTION

The accuracy of short-term weather forecasts (i.e., “nowcasts”) and numerical predictions of thunderstorm severity continues to be limited due to the complex interactions between thunderstorm dynamics, microphysics, and thermodynamics (Grzych et al. 2007; Snook and Xue 2008; Morrison and

Milbrandt 2011; van Weverberg et al. 2012). In an effort to address this problem, advanced measurement networks have recently been established, such as the 2012 Colorado Lightning Mapping Array (COLMA;

Rison et al. 2012) and the 2012 dual-polarization upgrade to the Weather Surveillance Radar-1988

Doppler (WSR-88D) network. At the same time, those who operate numerical weather prediction models must determine how to use recent increases in computer power most efficiently, which has resulted in the use of double-moment microphysics schemes (e.g., Thompson et al. 2004; Milbrandt and Yau 2005a,b;

Thompson et al. 2008; Morrison et al. 2009; Lim and Hong 2010) that predict both hydrometeor mixing ratio and number concentration, along with convection-allowing horizontal grid spacings of 4 km or less.

However, interactions between thunderstorm dynamics, microphysics, and thermodynamics are still not fully understood, especially under a changing climate with different cloud condensation nuclei (CCN) concentration, low-level moisture, and vertical wind shear.

Now that numerical weather models can represent two moments of the particle size distribution

(PSD), would direct measurements of the PSD from ground-based disdrometers, combined with three- dimensional dual-polarization radar data, be useful for improving our understanding of the severe thunderstorm PSD? In addition, can data from the advanced measurement networks, including COLMA and the upgraded WSR-88D, enable forecasters to issue more accurate nowcasts of severe thunderstorms, especially in regards to accurate predictions of hail size and amount? Further, what role do CCN play in modulating the interactions between dynamics, microphysics, and thermodynamics in severe thunderstorms? It could be argued that because severe thunderstorm updraft speeds approach 60 m s-1, the

1 specific characteristics of the CCN distribution are relatively unimportant to the nucleation process and therefore do not affect the microphysics and thermodynamics of the storm.

To answer these questions, I analyze dual-polarization radar observations from the WSR-88D network and a mobile radar, PSD measurements from optical disdrometers, and three-dimensional lightning data from COLMA. I also use the Weather Research and Forecasting (WRF) Model (Skamarock et al. 2008) to determine how changing CCN concentrations affect supercell thunderstorms. My overall research focus is to test new observational methods to study microphysical processes in severe thunderstorms and to improve our understanding of these processes by analyzing disdrometer, dual- polarization radar, lightning, and model data to achieve three primary objectives: 1) to quantify the usefulness of mobile disdrometer observations and assess the uncertainty between the disdrometer measurements and dual-polarization radar observations, 2) to investigate the effects of cloud condensation nuclei (CCN) concentration on the microphysics and thermodynamics of supercell thunderstorms through idealized simulations, and 3) to examine synoptic weather, dual-polarization radar, and lightning data from a series of accumulating hailstorms along the Colorado Front Range to enable forecasters to better understand and predict similar events in the future.

In Chapter 2, collocated optical disdrometer and dual-polarization radar measurements from six severe thunderstorms observed during the second Verification of the Origins of Rotation in Tornadoes

Experiment (VORTEX2; Wurman et al. 2012) are compared to quantify the similarity in the reflectivity

(Z), differential reflectivity (ZDR), and hydrometeor type recorded by the two instruments. The work in

Chapter 2 was published in Monthly Weather Review (Kalina et al. 2014a; “Comparison of Disdrometer and X-Band Mobile Radar Observations in Convective Precipitation”). The objectives of this study were

1) to develop a new particle identification scheme for optical disdrometers in convective weather, 2) to quantify the accuracy of differential-phase based radar attenuation correction schemes in extreme weather conditions such as heavy rain and large hail, and 3) to determine the conditions under which disdrometer and attenuation-corrected radar Z, ZDR, and particle type should not be assimilated into numerical models due to the questionable accuracy of these data. Understanding the errors that are present in microphysical 2 observations of severe thunderstorms is an important first step in using these data for model validation purposes.

In Chapter 3, the microphysical and thermodynamic responses of supercell thunderstorms to changes in CCN concentration are examined through a series of idealized simulations using the WRF

Model. The work in Chapter 3 was published in Journal of the Atmospheric Sciences (Kalina et al. 2014b;

“Aerosol Effects on Idealized Supercell Thunderstorms in Different Environments”). The goals of this research were 1) to quantify the changes in microphysical process rates, hydrometeor mass budgets, cold pool size and strength, and precipitation amount across the observed range of CCN concentrations (100 to

10 000 cm-3) in Earth’s atmosphere, and 2) to evaluate how the magnitudes of these changes are affected by the varying low-level relative humidity and vertical wind shear conditions in which supercell thunderstorms form. Numerical weather prediction models used in operational weather forecasting, such as the Global Forecast System (GFS), North American Mesoscale Forecast System (NAM), and a variety of configurations of the WRF Model, all contain microphysics schemes with prescribed CCN concentrations that are not currently adjusted on a day-to-day or event-to-event basis. As computing resources to run these models continue to increase, however, we may soon assimilate CCN concentration observations if such data are necessary to produce accurate forecasts of tornadogenesis, hail size, and precipitation amount in severe thunderstorms. Chapter 3 therefore explores the sensitivity of numerical forecasts of supercell thunderstorms to CCN concentration to determine if assimilating data on CCN concentration is necessary.

In Chapter 4, the synoptic weather conditions surrounding four extreme hailstorms that produced

15-60 cm of hail accumulations along the Colorado Front Range are examined, in conjunction with the dual-polarization radar and lightning data from each hailstorm. Previous “plowable” hailstorms resulted in numerous motor vehicle accidents along heavily-trafficked roads (including I-70), damaged aircraft at

Denver International Airport, and triggered water rescues as hail melted and combined with heavy rain to flood streets and float vehicles that attempted to traverse the rising water. Due to the high-impact, multi- hazard nature of these events, it is critical that forecasters are able to identify these hailstorms so that they 3 can issue appropriate warnings to the public. The purpose of this work is to analyze the synoptic weather patterns that are conducive to thunderstorms that produce deep hail accumulations and to provide forecasters with a set of dual-polarization radar and lightning signatures that can be used to identify these hailstorms in real time. Towards this goal, I have collaborated with meteorologists from the National

Weather Service Denver/Boulder Forecast Office to ensure that the work in Chapter 4 is of maximal benefit to forecasters. The contents of this chapter will be submitted to Weather and Forecasting (Kalina et al. 2015) in a manuscript titled “An Overview of Colorado Plowable Hailstorms: Synoptic Weather,

Dual-Polarization Radar, and Lightning Data.”

The overall significance of the work discussed here is that it examines new instruments and techniques to analyze microphysical processes in severe thunderstorms. This information is needed to enhance our understanding of convective microphysics and to improve numerical weather predictions and operational nowcasts of severe thunderstorms. This dissertation is an important part of this process, as it provides an assessment of the usefulness of new microphysical data from disdrometers and mobile radars

(Chapter 2), improves our understanding of the interaction between CCN, microphysics, and thermodynamics in supercell thunderstorms (Chapter 3), and demonstrates how state-of-the-art dual- polarization radar and lightning data can be used to identify and predict accumulating hailstorms (Chapter

4).

4 2 COMPARISON OF DISDROMETER AND X-BAND MOBILE RADAR OBSERVATIONS IN CONVECTIVE PRECIPITATION

This chapter is reprinted with permission from:

Kalina, E. A., K. Friedrich, S. Ellis, and D. Burgess, 2014: Comparison of disdrometer and X-band mobile radar observations in convective precipitation. Mon. Wea. Rev.,142, 2414-2435.

2.0 ABSTRACT

Microphysical data from thunderstorms are sparse, yet they are essential to validate microphysical schemes in numerical models. Mobile, dual-polarization X-band radars are capable of providing a wealth of data that include radar reflectivity, drop shape, and hydrometeor type. However, X-band radars suffer from beam attenuation in heavy rainfall and hail, which can be partially corrected with attenuation correction schemes. In this research, we compare surface disdrometer observations to results from a differential phase-based attenuation correction scheme. This scheme is applied to data recorded by the

National Oceanic and Atmospheric Administration (NOAA) X-band dual-Polarized (NOXP) mobile radar, which was deployed during the second Verification of the Origins of Rotation in Tornadoes

EXperiment (VORTEX2). Results are presented from five supercell thunderstorms and one squall line

(183 minutes of data). The median disagreement (radar-disdrometer) in attenuation-corrected reflectivity

(Z) and differential reflectivity (ZDR) is just 1.0 dB and 0.19 dB, respectively. However, two data subsets reveal much larger discrepancies in Z (ZDR): 5.8 dB (1.6 dB) in a hailstorm and -13 dB (-0.61 dB) when the radar signal quality index (SQI) is less than 0.8. The discrepancies are much smaller when disdrometer and S-band WSR-88D Z are compared, with differences of -1.5 dB (hailstorm) and -0.66 dB (NOXP SQI

< 0.8). A comparison of the hydrometeor type retrieved from disdrometer and NOXP radar data is also presented, in which the same class is assigned 63% of the time.

5 2.1 INTRODUCTION

The lack of surface microphysical and in-situ data is a critical obstacle in our attempts to understand and model severe thunderstorms accurately. Microphysical processes (e.g., accretion, collision and coalescence, drop breakup, melting, and evaporation) affect storm behavior and evolution by serving as a crucial link between the storm dynamics and thermodynamics. For example, melting of hail influences the strength and size of the low-level cold pool, which changes the near-surface buoyancy tendency and, as suggested by several recent studies, the tornadogenesis potential (Markowski et al. 2002; Shabbott and

Markowski 2006; Grzych et al. 2007). To collect the surface microphysical data required to understand and quantify these interactions, PARticle SIze and VELocity (PARSIVEL) optical disdrometers were deployed during the second Verification of the Origins of Rotation in Tornadoes EXperiment

(VORTEX2) to obtain particle diameter and fall speed distributions in severe thunderstorms. For the first time, these deployments were coordinated with X-band mobile polarimetric Doppler radars in severe thunderstorms, which provided a three-dimensional dataset of radar reflectivity (Z), differential reflectivity (ZDR), and differential phase (ΨDP) that is needed to characterize microphysical processes throughout thunderstorms.

The VORTEX2 measurements of supercell thunderstorm microphysics with disdrometers and mobile X-band radars are unprecedented, since both sets of instruments were deployed close to the storm and yielded high-resolution information near and at the surface. This dataset provides researchers with a unique opportunity to compare disdrometer data to output from hydrometeor classification schemes that are based on dual-polarization radar observations. However, the measurement accuracy of both instruments is strongly affected by the severe nature of the storms, which contain hail and strong winds.

To combine in-situ microphysical data at the surface with three-dimensional radar imagery, microphysical data need to be quality controlled and rain and hail particles must be discriminated. In addition, attenuation of the X-band radar signal must be corrected using algorithms that may be error-prone, particularly when the radar samples mixed-phase precipitation. A proven algorithm to correct attenuation in hail does not yet exist (Borowska et al. 2011; Ryzhkov et al. 2013a), although recent efforts to develop 6 a scheme valid in melting hail are presented in Ryzhkov et al. (2013a,b). Because supercell thunderstorms often contain large amounts of hail, attenuation correction schemes designed for rain will not always yield accurate results. In this paper, we compare attenuation-corrected radar data and hydrometeor classifications to surface disdrometer measurements in supercell thunderstorms. Can disdrometer data be used to provide guidance on the performance of radar attenuation correction schemes, and, therefore, to provide a measure of radar data quality? To investigate, we first apply a quality control algorithm and a hydrometeor classification scheme for in-situ disdrometer data that uses the particle size and fall speed distributions from the disdrometer to classify particles as rain, small hail (2 mm < d < 5 mm), and large hail (d > 5 mm; note that in this study, “large” is simply relative to the small hail class, and is not meant to be an argument against the typical definition of large hail of d > 20 mm). We then assess the performance of the attenuation correction scheme by comparing disdrometer-derived Z and ZDR to X-band radar Z and ZDR and to S-band radar Z. Comparisons between the disdrometer hydrometeor classification scheme and an existing scheme for X-band radar data are also provided.

A brief review of the different techniques that can be used to correct attenuation is now given.

Several attenuation correction schemes use the propagation differential phase (ΦDP) and specific differential phase profiles (KDP) to estimate the total and specific attenuation, respectively (e.g., Carey et al. 2000; Testud et al. 2000; Bringi et al. 2001; Anagnostou et al. 2006; Steiner et al. 2009). KDP is the range derivative of ΦDP, which must be calculated from the radar-measured total differential phase (ΨDP).

ΨDP is the sum of ΦDP and the backscatter differential phase (δ). ΨDP, ΦDP, δ, and KDP are related via Eq.

(2.1):

2 , (2.1) where r is the radar range. δ is only significant in the Mie scattering regime, which applies to rain drops with d > 2.3 mm at a temperature of 20 °C at X-band (Ryzhkov et al. 2011). Therefore, δ must be estimated before ΦDP and KDP can be used to correct the attenuation. Each attenuation correction scheme differs in the method used to calculate δ. Anagnostou et al. (2006) uses the differential reflectivity (ZDR)

7 in an iterative approach to estimate δ, while Steiner et al. (2009) applies the iterative finite impulse response filter from Hubbert and Bringi (1995) to the measured ΨDP field. Once corrected Z, ZDR, and KDP are obtained, a fuzzy-logic hydrometeor classification scheme (section 2.3.2.2) can be applied to the radar data to determine the dominant hydrometeor type observed in each range gate (Vivekanandan et al. 1999;

Liu and Chandrasekar 2000; Iwanami et al. 2007; Park et al. 2009; Dolan and Rutledge 2009; Snyder et al. 2010).

To our knowledge, this study is one of the first to use disdrometer observations to analyze the performance of attenuation correction and hydrometeor classification schemes for X-band radar measurements in severe thunderstorms. Most previous studies that compared radar and disdrometer data developed empirical relationships between reflectivity and rainfall rate (e.g., Schuur et al. 2001; Ulbrich and Miller 2001; Bringi et al. 2003; Kanofsky and Chilson 2008; Huang et al. 2010). In addition, such studies have primarily been conducted in stratiform precipitation (e.g., Geotis 1978; Goddard et al. 1982;

Thomson and List 1996; Zhang et al. 2011), while only a few comparisons have been performed in severe thunderstorms (Schuur et al. 2001; Thurai et al. 2010, 2011). X-band radar attenuation correction schemes have mostly been evaluated through comparison with S-band radar data, which are less attenuated (e.g.,

Anagnostou et al. 2006; Steiner et al. 2009; Snyder et al. 2010). Here, we propose an alternative method to evaluate attenuation correction schemes with surface disdrometer data. While we do not argue that this approach outperforms comparisons with S-band radar, the question remains whether disdrometers could be used if no nearby S-band radars are present. In addition, because the S-band WSR-88D dual- polarization upgrade was not complete in 2010, mobile radar ZDR measurements corrected for differential attenuation and results from mobile radar hydrometeor classification algorithms cannot be validated with traditional S-band radar comparisons. Therefore, the goal of this study is to determine the relative quality of the VORTEX2 disdrometer and X-band radar data with the intention of improving results from future observational analyses and numerical modeling studies that use these datasets for microphysical retrievals.

8 2.2 CASES, INSTRUMENTS, AND DATA COLLECTION

2.2.1 CASE SELECTION

Coordinated radar and disdrometer data were obtained from ~36 severe thunderstorms during the second year of VORTEX2, a 12-week field campaign conducted in the Great Plains of the United States during May and June of 2009 and 2010 (Wurman et al. 2012). For this study, three criteria were used for case selection: 1) radar data above the disdrometer sites were available for at least five minutes, 2) the disdrometers observed Z > 20 dBZ (approximate threshold between drizzle and light rain; Rinehart 2004) for at least five minutes, and 3) the distance between the radar and the disdrometers was less than 45 km.

Based on these criteria, we consider data from five supercell thunderstorms and one squall line (total of

~183 minutes of data). Table 2.1 provides details on the cases, and Fig. 2.1 shows examples of radar reflectivity from each case at an elevation angle of 1°.

Tab. 2.1: Deployment details for the cases included in this analysis. All of the cases listed are supercell thunderstorms, except for the squall line of 12 June 2010.

Date Times Location Disdrometers NOXP WSR-88D (UTC) Distance (km) Distance (km) (Beam Height, m) (Beam Height, km) 17 May 2218-2333 Artesia, CU01, UF04-07 KFDX NM 16-45 (280-920) 200-207 (4.1-4.3) 19 May 2059-2122 Kingfisher, CU01, UF01, KTLX OK UF05, UF07 29-38 (550-740) 168-173 (3.1-3.3) 2 June 2320-2326 Benkelman, KGLD NE UF03 23 (430) 67 (0.85) 7 June 0005-0143 Mitchell, CU01, UF01, KCYS NE UF05, UF07 30-37 (580-720) 123-138 (2.0-2.3) 9 June 0118-0139 Scottsbluff, CU01, UF01, KCYS NE UF03, UF05-06 24-27 (460-510) 103-107 (1.5-1.6) 12 June 2100-2203 Gruver, CU01, UF01 KAMA TX UF05-06 10-21 (190-390) 111-113 (1.7)

9

Fig. 2.1: Plan position indicators of attenuation-corrected radar reflectivity measured by NOXP at 1° elevation angle at a) 2212 UTC on 17 May 2010, b) 2118 UTC on 19 May 2010, c) 2320 UTC on 2 June 2010, d) 0002 UTC on 8 June 2010, e) 0130 UTC on 10 June 2010, and f) 2136 UTC on 12 June 2010. Disdrometer and radar locations are denoted by open circles and filled squares, respectively. The arrow shows the storm motion direction. 10 2.2.2 DISDROMETER MEASUREMENTS

The OTT PARSIVEL optical disdrometer (Löffler-Mang and Joss 2000) is an integrated laser transmitter-receiver that uses a 180-mm long, 30-mm wide, 1-mm thick light sheet to detect the diameter and fall speed of precipitation particles. More information about the measurement principle can be found in Löffler-Mang and Joss (2000), Löffler-Mang and Blahak (2001), Yuter et al. (2006), and references within. During VORTEX2, two types of disdrometers (Fig. 2.2) were deployed: articulating disdrometers

(denoted as UF01 and UF03), with a measurement volume that was oriented continuously perpendicular to the 10-s running average of the particle trajectory of a 1.2-mm raindrop (Friedrich et al. 2013a), and stationary disdrometers (denoted as CU01, UF04, UF05, UF06, UF07), with a measurement volume that remained fixed and parallel to the ground. For supercell thunderstorms, the disdrometers were deployed in advance of the southern side of the forward-flank downdraft (Fig. 2.3), with the mobile weather radars deployed to the southeast of the thunderstorm. The distance between the disdrometers and the radar ranged from 10 km to 45 km, with a median distance of 20 km. Further details on the deployment strategy are provided in Friedrich et al. (2013a).

2.2.3 RADAR MEASUREMENTS

Radar data were obtained from the National Oceanic and Atmospheric Administration (NOAA)

X-band dual-Polarized (NOXP) mobile radar (Palmer et al. 2009; Burgess et al. 2010). Table 2.2 provides a summary of the radar configuration during VORTEX2. For the cases considered in this research, the size of the radar resolution volume above the disdrometer sites ranged from 74 m x 87 m x 175 m at 10 km range to 74 m x 393 m x 785 m at 45 km range, in range, azimuth, and height, respectively. The radar was electronically leveled and a digital compass was used to record its heading. With respect to the location of the disdrometers, the height of the lowest radar beam ranged from 0.2-1 km AGL, but was mainly below 0.6 km AGL (Tab. 2.1).

11

Fig. 2.2: A photograph of an articulating disdrometer (foreground) and a stationary disdrometer (background) deployed in Artesia, NM on 17 May 2010.

Fig. 2.3: An idealized schematic that shows the disdrometer and radar deployment strategy for supercell thunderstorms. The disdrometers were deployed in a line that was perpendicular to the storm motion vector with an instrument spacing of 0.2-1 km. The radar was deployed ahead of the forward-flank downdraft of the thunderstorm, and was always within 45 km of the disdrometer deployments (ideally within 15 km). The location of the 40-dBZ isoline is shown in black.

12 Tab. 2.2: NOXP radar characteristics for the 2010 VORTEX2 field campaign.

Wavelength 3.21 cm (X-band) Transmission mode Simultaneous transmission and reception (SHV) Beamwidth 1° Range resolution 74 m Azimuthal resolution 0.5° Elevation angles scanned 1° to ≤ 15° (1° increments) Maximum unambiguous range 60.0 km Nyquist velocity 19.9 m/s Pulse repetition frequency 2500 Hz Moments and variables Reflectivity, Doppler velocity, spectrum width, differential reflectivity, differential phase, correlation coefficient

2.3 DATA PROCESSING

2.3.1 DISDROMETER

2.3.1.1 Quality control and hydrometeor classification scheme

A quality control procedure (Fig. 2.4) was applied to the stationary and articulating disdrometer data to address three documented error sources: strong winds, particles that only partially transect the sample volume (i.e., margin fallers), and splashing (e.g., Sevruk 1982; Illingworth and Stevens 1987;

Neŝpor et al. 2000; Schuur et al. 2001; Kruger and Krajewski 2002; Barthazy et al. 2004; Thurai and

Bringi 2005; Yuter et al. 2006; Friedrich et al. 2013b). Friedrich et al. (2013b) analyzed wind-induced errors in PARSIVEL disdrometer data collected in a tropical cyclone and two supercell thunderstorms with wind speeds up to 30 m s-1, using six stationary and two articulating disdrometers. Misclassified particles, with d > 5 mm and unphysically slow fall speeds (v < 1 m s-1), were identified in the stationary

(but not the articulating) disdrometer data at wind speeds as slow as 10 m s-1, and were observed consistently at wind speeds larger than 20 m s-1. To avoid misclassified particles in this analysis, we remove entire time steps in which the stationary disdrometers observed particles with d > 5 mm and slow fall speeds (v < 1 m s-1). Second, margin fallers and splashing raindrops are removed from both types of disdrometer data by eliminating raindrops with a fall speed more than 60% faster or slower than the fall

13 speed-diameter relationship for rain (Gunn and Kinzer 1949; Atlas et al. 1973), based on a study of

PARSIVEL disdrometer accuracy by Jaffrain and Berne (2011).

The decision to classify a particle as rain, small hail, or large hail is based on its diameter and fall speed (Fig. 2.4). While the reader is referred to Friedrich et al. (2013a) for a detailed description of the classification scheme, two notes are merited here. First, we distinguish between small and large hail because small ice particles in thunderstorms are partially melted hailstones that consist of a torus of liquid water that surrounds an ice core (Rasmussen and Heymsfield 1987a; Bringi and Chandrasekar 2001).

Therefore, small and large hail need to be assigned different fractional water contents when the T-matrix method is used to compute the scattering properties of the particle size distribution (see section 2.3.1.2 and Tab. 2.3). Appendix A demonstrates that for the PSDs in this study, which contain relatively few hailstones due to the small sample area of the PARSIVEL disdrometer, the sensitivity of disdrometer Z and ZDR to the small hail fractional water content is less than 0.1 dB. Second, while raindrop fall speed is a well-defined function of diameter (Gunn and Kinzer 1949; Atlas et al. 1973), the various shapes, densities, and water loadings of hail make it possible for an ice particle to have a range of fall speeds for a given diameter, raising the possibility that a hail particle could be erroneously classified as rain. In addition, the disdrometer hydrometeor classification scheme (Fig. 2.4) uses the fall speed curves for dry graupel to define the small hail region, following Friedrich et al. (2013a). However, because the bulk density of small hail in thunderstorms exceeds that of graupel, the graupel fall speed relations are likely less than the fall speeds observed here. Therefore, some of the small hail may be detected in the unclassified region between the small hail and rain classes in Fig. 2.4. Because there is uncertainty over whether particles in this region are rain or small hail, these particles are left unclassified. Appendix B explores the sensitivity of the results to whether particles in the unclassified region are included in the analysis, revealing a mean sensitivity of ~0.2 dB in Z and 0.01 dB in ZDR (relative to Z and ZDR obtained by excluding the particles).

14

Fig. 2.4: Fall speed vs. diameter plot depicting the quality control procedures and the hydrometeor classification scheme applied to the disdrometer data (adapted from Fig. 5 in Friedrich et al. 2013a). The white solid lines are the empirical diameter-fall speed relationships for rain (Atlas et al. 1973), graupel (Locatelli and Hobbs 1974), and hail (Knight and Heymsfield 1983).

Tab. 2.3: Parameters used in the T-matrix program [i.e., canting angle (CA), axis ratio (AR), bulk density (BD), fractional water content (FWC), and temperature (T)]. The mean and standard deviation are denoted by μ and σ, respectively. For comparison to NOXP radar (WSR-88D) data, calculations were performed at a radar frequency of 9.41 GHz (2.895 GHz) and a radar elevation angle of 1° (0.5°).

Hydrometeor Parameters References Rain CA: μ=0°, σ=7.5° Huang et al. 2008 T: 15 °C CA: μ=0°, σ=50° Snyder et al. 2010; Ryzhkov et al. 2011 Small Hail AR: 0.8 Huang et al. 2005 BD: 0.9 g cm-3 Vivekanandan et al. 1993 FWC: 0.5 Huang et al. 2005 T: 0 °C CA: μ=0°, σ=50° Snyder et al. 2010; Ryzhkov et al. 2011 Large Hail AR: 0.8 Knight 1986; Balakrishnan and Zrnic 1990 BD: 0.9 g cm-3 Vivekanandan et al. 1993; Solheim et al. 1999 FWC: 0.2 Aydin et al. 1998 T: 0 °C

15 2.3.1.2 Computation of meteorological variables from disdrometer data

Before calculating Z and ZDR from the disdrometer data, number distributions were accumulated over periods of 60 s to obtain sufficiently large particle samples. The 60-s accumulation time causes higher-frequency variations in the particle size distribution (PSD) to be lost. These variations would likely be retained in the radar data due to the excellent range resolution (74 m), possibly causing the radar and disdrometer measurements to represent different PSDs (this issue is partially addressed by averaging the radar data; see section 2.3.3). Next, the transition (T-) matrix method (Vivekanandan et al. 1991; Bringi and Chandrasekar 2001) was used to compute Z and ZDR for each 60-s time step. Because the scattering properties of rain and ice particles differ, separate rain, small hail, and large hail distributions were input to the T-matrix program for each time step. For rain particles, the drop shape model from Beard and

Chuang (1987) was chosen, which remains accurate at the large drop diameters present in convective weather (i.e., within ±4% of the measured axis ratios at d > 5 mm; Thurai et al. 2009). The raindrops were assumed to have a temperature of 15 °C, a mean canting angle of 0°, and a canting angle standard deviation of 7.5° (Huang et al. 2008). Of these settings, Z and ZDR display the greatest sensitivity to the canting angle standard deviation; however, the change in ZDR is generally less than 0.1 dB across the range of physically reasonable values (1°-10°; Kwiatkowski et al. 1995). Hail particles were assumed to have a canting angle mean and standard deviation of 0° and 50°, respectively (Snyder et al. 2010;

Ryzhkov et al. 2011), and were assigned a temperature of 0 °C. The hailstones were modeled as a uniform mixture of ice and liquid water using the Maxwell-Garnett mixing formula, with ice (liquid water) as the matrix and liquid water (ice) as the inclusions in large (small) hailstones (Bringi and Chandrasekar 2001;

Ryzhkov et al. 2011). Table 2.3 summarizes the parameters used in the T-matrix program for rain, small hail, and large hail.

2.3.2 RADAR DATA PROCESSING

2.3.2.1 Radar attenuation correction scheme

16 The attenuation correction scheme from Steiner et al. (2009) is applied to the radar data. Steiner et al. (2009) evaluated the performance of this scheme with data that were collected in convective storms during the 2002 International H2O Project (IHOP) by the National Observatory of Athens (NOA) X-band radar. For X-band radars with simultaneous horizontal and vertical polarization (SHV) transmit, attenuation correction in heavy rain was found to be most accurate when ZDR was not used to estimate the attenuation or differential attenuation (Steiner et al. 2009). Potential biases in ZDR caused by antenna and depolarization errors as ΨDP increases (Ryzhkov and Zrnic 2007; Hubbert et al. 2010a, 2010b; Zrnic et al.

2010) are thereby avoided.

The Steiner et al. (2009) attenuation correction scheme is modified from the differential phase- based algorithm presented in Anagnostou et al. (2006). The first modification is that δ was removed from

ΨDP with five iterations of the finite impulse response filter from Hubbert and Bringi (1995), rather than using ZDR to estimate δ. The smoothed ΦDP range profile was then used to estimate the path-integrated attenuation at horizontal polarization (Eq. 2.2) and to correct ZH (Eq. 2.3):

Φ Φ 0 , (2.2)

, (2.3) where PIAH is the path-integrated attenuation at horizontal polarization (dB), r is the radar range (km), AH is the specific attenuation at horizontal polarization (dB km-1), γ is an empirical constant equal to 0.3006 in rain (average of γ values in Tab. 3 of Anagnostou et al. 2006), and is the corrected reflectivity at horizontal polarization. Another modification to the Anagnostou et al. (2006) technique was to relate the differential attenuation directly to PIAH following Park et al. (2005) and Gorgucci et al. (2006). Once an initial estimate of PIAH was made, the path-integrated differential attenuation (PIADP) was calculated as a function of PIAH (Eq. 2.4), and ZDR was corrected with Eq. (2.5):

, (2.4)

, (2.5) where is the corrected ZDR and 0.173 in rain.

17 2.3.2.2 Radar hydrometeor classification scheme

Once radar data were corrected for attenuation, the hydrometeor classification scheme from

Snyder et al. (2010) was applied. This fuzzy-logic scheme was originally devised at S-band by Park et al.

(2009) and adapted to X-band by Snyder et al. (2010). The algorithm uses the following dual-polarization radar variables as inputs: ZH, ZDR, 10log(KDP), copolar cross-correlation coefficient at lag zero (ρHV), reflectivity texture, and total differential phase texture. The membership functions used by the scheme are trapezoidal in shape and are derived from T-matrix simulations of observed and idealized PSDs of rain and hail. The membership functions are defined for six hydrometeor classes: ground clutter/anomalous propagation (GC/AP), biological scatterers (BS), big drops (BD), light to moderate rain (RA), heavy rain

(HR), and rain-hail mixture (RH). Only the latter four classes are retained for comparison with the

PARSIVEL disdrometer, since the disdrometer does not detect GC/AP or BS. The output values of the membership functions are weighted according to Park et al. (2009) and then summed for each hydrometeor class. The class with the largest sum is then assigned to the radar range gate.

2.3.3 RADAR-DISDROMETER COMPARISON METHOD

To select data for comparison, radar Z and ZDR were averaged over a 3x3 array of range gates, centered on the gate that contained the disdrometer. Time steps with sharp horizontal reflectivity gradients (> 35 dB km-1) and deployments with ground clutter (Z > 0 dBZ and near-zero Doppler velocity) near the disdrometer sites were excluded from the analysis. Radar and disdrometer data were then paired so that the time difference between the observations did not exceed 30 s.

Despite the exclusion of reflectivity gradients larger than 35 dB km-1 and the averaging of radar range gates, precipitation particle advection and the height difference between the radar beam and the surface are potentially large error sources in this analysis. As an example, consider a raindrop of d = 1 mm, which has a terminal fall speed of v = 4 m s-1. If this drop is at a height of 1 km when the radar observes it, the drop will not reach the ground until 250 s later. Assuming a mean horizontal wind speed of 10 m s-1, the drop will be advected 2.5 km downstream from the point in space where it was observed

18 by the radar, a distance of nearly 34 radar range gates (in the worst case scenario). Thus, the PSDs observed by the radar and disdrometer may be different due to the strong low-level winds that often accompany supercell thunderstorms. A sensitivity test was performed in which the radar data were averaged over various windows, ranging from 3x3 to 11x11 range gates (not shown). Aside from a ~0.25 dB improvement in the ZDR agreement when a 3x3 averaging window was used versus no averaging, the results are not affected by the size of the window. In addition, no correlation was found between the disagreement in the radar and disdrometer data and the radar beam height (not shown). Growth and evaporation of raindrops between the radar beam and the surface, however, are nevertheless error sources.

2.4 RESULTS AND DISCUSSION

2.4.1 RADAR AND DISDROMETER COMPARISON OF Z AND ZDR

2.4.1.1 X-band Radar Z and Zdr

Figure 2.5 presents scatter plots of disdrometer and radar Z for all cases in the analysis (Tab. 2.1 and Fig. 2.1), before (a) and after (b) the attenuation correction scheme was applied. Two data subsets are identified, which will be discussed in sections 2.4.2 and 2.4.3: the 17 May 2010 hailstorm (red plus signs) and data with radar signal quality index (SQI) < 0.8 (blue plus signs). SQI (SIGMET 2009) is related to signal-to-noise ratio (SNR) and spectrum width (W):

exp . (2.6)

177 of the 183 uncorrected radar Z values (97%) are weaker than the corresponding disdrometer observations, with a median difference (radar Z – disdrometer Z) of -17 dB (all data in Fig. 2.5a). When the attenuation correction scheme is applied, the distribution shifts towards larger radar Z (all data in Fig.

2.5b). The median difference (radar Z – disdrometer Z) after attenuation correction is 1.0 dB, and 48% of the radar Z values are weaker than the disdrometer values. These statistics indicate that the attenuation correction scheme has removed the overall negative bias in the radar observations. 25% of the radar Z values are larger than the PARSIVEL disdrometer sampling uncertainty, 30% are smaller, and 45% are within the sampling uncertainty (gray shading in Fig. 2.5), quantified by Jaffrain and Berne (2011). In the 19 Jaffrain and Berne study, two PARSIVEL disdrometers were collocated and sampled ~990 hours of light to moderate rainfall. The standard deviation of the difference in the moments (i.e., the sampling uncertainty) that were derived from each drop size distribution was then calculated (Tabs. B4 and B5 in

Jaffrain and Berne 2011). Since the measurements in this study were taken in severe thunderstorms that often contained heavier rainfall rates, hail, and strong winds, the sampling uncertainty from Jaffrain and

Berne (2011) should be considered a lower bound on the uncertainty in this research.

Scatter plots of ZDR before and after attenuation correction are presented in Fig. 2.5c and Fig.

2.5d, respectively. Differential attenuation is clearly evident in the uncorrected radar ZDR, with 50% of the radar observations having ZDR < 0 dB (Fig. 2.5c). After attenuation correction was applied (Fig. 2.5d), the median difference (radar ZDR – disdrometer ZDR) improves from -2.7 dB to 0.19 dB. 38% of the corrected observations are within the sampling uncertainty of the PARSIVEL disdrometer (gray shaded region in

Fig. 2.5d), compared to 20% prior to correction. Similar to the reflectivity, a large number of points lie outside the sampling uncertainty (gray shaded region in Fig. 2.5d), with 37% (25%) of the radar ZDR values larger (smaller) than the disdrometer sampling uncertainty.

2.4.1.2 S-band radar Z

To provide a benchmark for comparison, disdrometer Z is also compared to Z from the nearest

WSR-88D at 0.5° elevation angle (Fig. 2.6) for all cases in the analysis. In comparing Fig. 2.5b to Fig.

2.6, we note that the median disagreement in radar and disdrometer Z is changed from 1.0 dB to -1.9 dB

(all plus signs) when S-band WSR-88D data are compared to the disdrometer-derived data. However, the median disagreement in Z for two of the data subsets (i.e., 17 May 2010 hailstorm, red plus signs, and

SQI < 0.8, blue plus signs) decreases when WSR-88D data are used, from 5.8 dB to -1.5 dB (hailstorm; red plus signs) and -13 dB to -0.66 dB (SQI < 0.8; blue plus signs). These results imply that the X-band radar attenuation correction scheme partially contributes to the disagreement in these subsets, since S- band radar data compare more favorably with the disdrometer data (this inference, however, is complicated by differences in the X- and S-band radar beam heights; see Tab. 2.1). These details will be

20 explored further in sections 2.4.2-2.4.4, in which three case studies are examined: a supercell with large hail on 17 May 2010, a supercell on 9 June 2010, and a squall line on 12 June 2010.

Fig. 2.5: Comparison of radar and disdrometer observations before (a, c) and after (b, d) attenuation correction for Z (a-b) and ZDR (c-d). The gray shaded region is the sampling uncertainty of the PARSIVEL disdrometer, taken from Jaffrain and Berne (2011). Uncertainties for Z > 50 dBZ and ZDR > 3 dB are outlined in green and were obtained via linear extrapolation. Observations from the hailstorm on 17 May 2010 are plotted in red, while observations with radar SQI < 0.8 are plotted in blue. All other observations are plotted in black. Note that four of the 51 observations from the hailstorm have SQI < 0.8 and are included in the hailstorm subset. The median disagreement (radar – disdrometer) for all data is shown in the upper left, while the bottom right shows the median disagreement for each subset. The number of observations in each plot is 183, consisting of cases described in section 2.2.1 and Tab. 2.1.

21

Fig. 2.6: As in Fig. 2.5b, but for unattenuated S-band WSR-88D Z.

2.4.2 17 MAY 2010: SUPERCELL WITH RADAR Z AND ZDR LARGER THAN DISDROMETER VALUES

On 17 May 2010, a high-precipitation supercell thunderstorm was observed near Artesia, NM.

The disdrometers sampled the forward flank downdraft of the storm (Fig. 2.1a). The authors observed large hail (d ~ 50 mm) between 2220-2232 UTC, which severely damaged the windshields of the deployment vehicles. Figure 2.7 shows a time series plot comparing Z (Fig. 2.7a) and ZDR (Fig. 2.7b) values recorded by the radar and disdrometer CU01 during this event. The total ice volume (assuming spherical particles) and the largest hail size observed by the disdrometer are also plotted with time (Fig.

2.7c and Fig. 2.7d). One might expect that the radar observations would not be heavily attenuated in the portion of the storm nearest the radar and that the attenuation could be accurately corrected. However, the time series data (Fig. 2.7a and Fig. 2.7b) reveal that, on average, corrected Z and ZDR measured by the X- band radar were 8.3 dB and 1.5 dB larger, respectively, than the corresponding measurements from

22 CU01. In fact, during the 27-minute period shown, Z recorded by CU01 never exceeded that of the radar, and the CU01 ZDR was greater than the radar ZDR for only 2 of the 10 time steps. The trend of generally larger attenuation-corrected radar variables relative to all of the disdrometer measurements made on 17

May is also shown in Fig. 2.5b and Fig. 2.5d (red plus signs), a discrepancy that is not reflected in the remainder of the dataset (blue and black plus signs).

We hypothesize that the disagreement evident in the disdrometer and radar observations on 17

May is partly due to large, wet hail causing resonant (Mie) scattering of the radar beam. Due to Mie scattering, strong attenuation at horizontal polarization and differential attenuation have been observed at

C- and X-bands in the presence of hail (e.g., Steiner et al. 2009; Tabary et al. 2009; Snyder et al. 2010;

Borowska et al. 2011). Attenuation correction in the presence of hail is uncertain (Vulpiani et al. 2008;

Borowska et al. 2011; Gu et al. 2011), because the relationship between differential phase and attenuation in a rain/hail mixture has only recently been examined. Ryzhkov et al. (2013a,b) found that the coefficients γ and ε in Eqs. (2.2) and (2.4) differ for rain and melting hail. Therefore, correcting attenuation in mixed-phase precipitation with coefficients meant for rain may lead to large errors (Steiner et al. 2009).

Because it is known that attenuation correction schemes designed for rain may perform poorly in the presence of hail, disdrometers can be used to detect hail on the ground and to flag radar observations beyond the location of the disdrometers as poor quality. Figure 2.8 shows the disdrometer data accumulated over this event as a function of particle fall velocity and diameter, with large hail up to d =

20 mm present (note that the measurement limit of the PARSIVEL disdrometer is d = 26 mm, and undercatchment is likely at these large sizes, as discussed later). Figure 2.7 c shows that between 2218-

2233 UTC, up to 15 cm3 of ice were observed per 1-minute time step by CU01. Although the ice volume decreases to < 5 cm3 min-1 after 2233 UTC, a 5-15 dBZ discrepancy remains in Z, likely because the hail core is located between the radar and disdrometer sites. Range gates located behind the hail core may exhibit erroneous Z values due to the cumulative nature of attenuation correction errors.

23

Fig. 2.7: Time series data recorded by NOXP (solid lines) and disdrometer CU01 (dashed lines) from the supercell thunderstorm with large hail (d ~ 50 mm) observed on 17 May 2010: a) attenuation-corrected radar and disdrometer reflectivity, b) attenuation-corrected radar and disdrometer differential reflectivity, c) disdrometer-observed ice volume, and d) disdrometer-observed maximum hail size. The error bars represent the sampling uncertainty of the PARSIVEL disdrometer.

An important limitation of this analysis is that large hail is sparse (number concentration of 10-2 m-3; Straka 2009) relative to the sample area (54 cm2) of the PARSIVEL disdrometer. Therefore, disdrometer data might not indicate that radar data are suspect if only a few hailstones fall at the disdrometer site. In addition, even in thunderstorms with large amounts of ice, the PARSIVEL

24 disdrometer will underestimate the hailstone concentration due to the small sample area. Undercatchment is especially likely for the largest hailstones, as Fig. 2.7d demonstrates that the largest hailstones detected by disdrometer CU01 during matched observations with the radar are d = 13 mm, despite hailstones of d

= 50 mm being observed. In an attempt to determine whether undercatchment is the primary cause of the disagreement (rather than deficiencies in the attenuation correction scheme), we compare disdrometer Z to that of KFDX, the nearest S-band WSR-88D, at 0.5° elevation angle (Fig. 2.6). For the hailstorm case

(red plus signs in Fig. 2.6), the median disagreement is -1.5 dB when S-band radar is used, but jumps to

5.8 dB for attenuation-corrected X-band radar (Fig. 2.5b). Note, however, that the two radars are sampling at different heights (NOXP at 0.6 km AGL and KFDX at 4.2 km AGL; Tab. 2.1), which prevents us from making a definitive conclusion about the cause of the improved agreement.

Fig. 2.8: Accumulated particle counts recorded by disdrometer CU01 on 17 May 2010, binned by the observed fall speed and diameter. The black lines represent the empirical fall speed-diameter relationships for rain, graupel, and hail that are shown in Fig. 2.4. Hail bins are outlined in red. 25 2.4.3 9 JUNE 2010: SUPERCELL THUNDERSTORM WITH RADAR Z AND ZDR LESS THAN DISDROMETER VALUES

A supercell thunderstorm developed in the late afternoon of 9 June 2010 and moved into the western Nebraska panhandle near Scottsbluff. The core of the thunderstorm passed ~5 km south of the disdrometer deployments, which placed the disdrometers behind the precipitation core relative to the

NOXP radar, which was deployed south of the storm core (Fig. 2.9a). Because the heaviest precipitation passed between the radar and disdrometers, the radar data collected near the disdrometer locations were heavily attenuated. Time series of corrected radar and disdrometer (UF01) data are shown in Fig. 2.10a and Fig. 2.10b. The radar signal quality index (SQI; Eq. 2.6) is also plotted (Fig. 2.9b and Fig. 2.10c).

Between 0130-0140 UTC, the disdrometer recorded Z and ZDR values that are 10-15 dB and 1-3 dB larger, respectively, than the values obtained by the radar after attenuation correction. During this time, SQI ranges from 0.4 to 0.5, which is poor relative to the other observations considered in this analysis (76% have SQI > 0.8). Data with SQI < 0.8 are plotted in blue in Fig. 2.5, and exhibit smaller radar Z relative to the disdrometer, a trend that is not present in the other data subsets (black and red plus signs). We find that the median disagreement in Z (radar Z – disdrometer Z) for SQI < 0.8 is -12 dB, while the median disagreement in Z for SQI > 0.8 is 2.1 dB. These results indicate that it is much more likely for radar Z to be smaller than disdrometer Z when SQI is small and most of the radar signal has been lost due to attenuation. To verify that the disagreement is not solely due to errors in the disdrometer data, we note from Fig. 2.6 (blue plus signs) that the comparison improves markedly for S-band radar data (median disagreement of -0.66 dB versus -13 dB for X-band radar data corrected for attenuation). The radar beam heights are also more similar than in the 17 May hailstorm (0.5 km AGL for NOXP and 1.5 km AGL for

KCYS; Tab. 2.1). This evidence supports our claim that a negative bias exists in the attenuation-corrected

NOXP radar data when SQI < 0.8.

2.4.4 12 JUNE 2010: SQUALL LINE WITH RADAR Z AND ZDR SIMILAR TO DISDROMETER VALUES

26 At approximately 2100 UTC on 12 June 2010, a squall line developed near Gruver, TX. Radar and disdrometer (UF05) time series data are shown in Fig. 2.11. The median attenuation-corrected radar and disdrometer Z and ZDR values are in closer agreement than in previous case studies (5.0 dB and 1.2 dB, respectively). The best agreement (to within 5 dB and 0.5 dB) is found from 2139-2155 UTC, with greater disagreement (> 15 dB and > 1.5 dB) during the first two time steps (i.e., 2130 UTC and 2136

UTC). From 2130-2136 UTC, the radar data (Fig. 2.12) show that a small convective cell (~2 km in north-to-south extent) near the leading edge of the squall line moves over disdrometer UF05. The ice volume recorded by the disdrometer (Fig. 2.11c) is largest during this time period, peaking at 5.6 cm3 min-1 at 2130 UTC, before dropping below 4 cm3 min-1 after 2139 UTC. The time series of maximum hail size (Fig. 2.11d) and accumulated PSD (Fig. 2.13) also depict the hail from the convective cell and show hailstones up to d = 13 mm. From 2130-2136 UTC, hail resulted in larger radar Z and ZDR values relative to the disdrometer data (Fig. 2.11), which is the same result found in section 2.4.2. A radial stripe of suspiciously large radar Z (> 55 dBZ) and ZDR (> 6 dB) values is also present in the radar imagery at

2136 UTC (red circles in Fig. 2.12) in the southwestern portion of the squall line, possibly caused by the presence of hail there.

Following the passage of the convective cell and the hail at ~2137 UTC at the disdrometer site, agreement in Z and ZDR improves and the discrepancies are generally within 5 dB and 0.5 dB, respectively. During the period of agreement (2137-2155 UTC), the radar imagery depicts precipitation that is nearly uniform in time and space near the disdrometer site. Additionally, the observed ice volume remains small (median 0.7 cm3 min-1), suggesting that hail is not biasing the attenuation correction. The large radar SQI (> 0.95, not shown) and the relative absence of hail are the likely reasons for the improved agreement.

27

Fig. 2.9: Plan position indicators of a) attenuation-corrected radar reflectivity and b) signal quality index for the supercell thunderstorm observed by NOXP at 1° elevation angle on 10 June 2010 at 0130 UTC. Black, open circles denote disdrometer locations. The arrow shows the direction of storm motion. The distance between each labeled tick mark is approximately 8 km in X and 11 km in Y. 28

Fig. 2.10: As in Fig. 2.7, but for the supercell thunderstorm observed by disdrometer UF01 on 9 June 2010. The radar signal quality index is shown in c).

29

Fig. 2.11: As in Fig. 2.7, but for the squall line observed by disdrometer UF05 on 12 June 2010.

30

Fig. 2.12: Plan position indicators of attenuation-corrected a) radar reflectivity and b) differential reflectivity for the squall line observed by NOXP at 1° elevation angle on 12 June 2010 at 2136 UTC. The location of disdrometer UF05 is denoted by the black, open circle, and the location of NOXP is annotated. The convective cell is outlined in blue, and an area of large radar reflectivity and differential reflectivity is circled in red. The arrow indicates the storm motion direction. The distance between each labeled tick mark is approximately 18 km in X and 11 km in Y.

Fig. 2.13: As in Fig. 2.8, but for disdrometer UF05 on 12 June 2010. 31 2.4.5 RADAR AND DISDROMETER HYDROMETEOR CLASSIFICATION COMPARISONS

We now compare output from the radar and disdrometer hydrometeor classification schemes. All of the disdrometer and radar data from Fig. 2.5 are included. The disdrometer hydrometeor classes consist of rain (RA), small hail (S. Hail), and large hail (L. Hail), while the radar classes of interest are big drops

(BD), rain (RA), heavy rain (HR), and rain-hail mix (RH). Because the radar scheme does not have firm rain rate or reflectivity thresholds that can be applied to the disdrometer data to discriminate between BD,

RA, and HR, these radar classes are combined into a general rain class. Therefore, the purpose of the comparison is to analyze the agreement between the liquid and ice classes of the two schemes. Due to error sources that include overlapping rain and hail signatures in the radar data, errors in attenuation and differential attenuation correction, and particle advection, we do not expect perfect agreement; however, we can nevertheless use the data to explain under what conditions we expect disagreement and why.

The comparison results are shown in Fig. 2.14. Sectors that represent disagreement (i.e., the disdrometer

(radar) observes hail, but the radar (disdrometer) does not) are separated from the remainder of the chart.

Of the 179 observations, 113 (63%) are in agreement. The most common scenarios are that both schemes detect hail (36% of the time) or that both detect rain (27% of the time). Of the 60 observations that disagree, 13 (7%) of them disagree because the radar scheme identifies hail when the disdrometer does not observe any. One likely explanation for this disagreement is that due to the sparseness of the hail, no hailstones passed through the sample volume of the disdrometer during the same time step as observed by the radar. In addition, it is also possible that the hail melted as it fell from the height of the radar volume to the disdrometer. Since the center of the radar beam was always 0.2-1 km AGL for the observations considered here, this possibility is most likely when small hail was present at the height of the radar beam, which would be more susceptible to complete melting.

Disagreement in the remaining 53 observations (29%) results from the disdrometer detecting hail when the radar does not. It is important to note that the disdrometer ice scheme does not account for the number of hailstones present; if just one hailstone is recorded during a 60-s time step, a classification of hail is still assigned by the disdrometer scheme, even though hail may not be the dominant contributor to 32 the corresponding radar measurements. The median number of hailstones and median hailstone size observed in a 60-s time step by the disdrometer is greater when both the radar and the disdrometer schemes agree that hail is present (median of two hailstones and 8.5 mm when the schemes agree vs. one hailstone and 6.0 mm). The radar classification scheme uses the dual-polarization radar variables to identify the dominant hydrometeor class within the radar volume, but other hydrometeor types may nevertheless be present. Thus, it may be that the radar scheme does not assign the hail class when the radar volume is dominated by rain and there are relatively few, small hailstones present. This is possible because the membership functions for rain and hail in the radar fuzzy logic scheme overlap, but it is only the class with the maximum rule strength (i.e., the maximum weighted sum of the membership function values) that is assigned to the radar range gate. When the disdrometer classifies hail but the radar does not, the median rule strength of the radar hail (rain) class is 0.6 (1.5), compared to 0.5 (1.5) when both instruments classify rain. However, the radar algorithm will not classify hail in either case, because rain is the dominant scatterer in the radar volume.

An additional factor that is likely to cause the PARSIVEL disdrometer to report hail when the radar does not is the possibility that multiple particles may be present within the disdrometer sample volume at the same time. If two 5-mm raindrops were present in the sample volume, for instance, the disdrometer would record a single 10-mm particle, which would be classified as a hailstone (Fig. 2.4). For

-3 -1 a simulated heavy convective rain with intercept parameter N0 = 1400 m mm and rainfall rate R = 300 mm hr-1, Löffler-Mang and Joss (2000) found that the probability of coincidences in the PARSIVEL disdrometer sample volume is ~5%. Therefore, disagreement between the hail detections of the radar and disdrometer schemes is possible due to a combination of particle coincidences in the disdrometer sample volume and the use of the fuzzy logic scheme for classifying the radar data.

33

Fig. 2.14: Pie chart comparing the outputs from the disdrometer and radar hydrometeor classification schemes. The area of each sector in the pie chart is proportional to the percentage of the total number of time steps (179) included in each sector. Each sector is labeled with the class assigned by the disdrometer scheme (i.e., rain, small hail, large hail) in bold, followed by a solidus (/) and the class assigned by the radar scheme (i.e., rain, hail) in italics. The number of time steps in each sector is also listed. Sectors in which the outputs from the two schemes disagree have been separated from the rest of the chart.

2.5 SUMMARY AND CONCLUSIONS

In this paper, we applied an attenuation correction scheme designed for rain (Steiner et al. 2009) to

X-band dual-polarization radar data collected by NOXP in five supercell thunderstorms and one squall line during the VORTEX2 field campaign in 2010. The attenuation-corrected radar Z and ZDR were then compared to those derived from PSDs recorded by PARSIVEL disdrometers using the T-matrix program, which required assumptions to be made about the axis ratio, fractional water content, and canting angle of the observed hailstones (see Tab. 2.3 and appendix A). The Snyder et al. (2010) hydrometeor classification scheme was then applied to the corrected radar data, and the results were compared to the output from a hydrometeor classification scheme that was developed for disdrometers in convective weather.

34 When the disdrometer and attenuation-corrected radar data were compared, it was shown that

45% (38%) of the Z (ZDR) observations agree to within the sampling uncertainty of the PARSIVEL disdrometer (Fig. 2.5b and Fig. 2.5d). A case study analysis of a supercell thunderstorm with large hail demonstrated that the attenuation-corrected X-band radar Z (ZDR) tends to be larger than the values recorded by the disdrometer by 8.3 dB (1.5 dB), respectively (Fig. 2.7a). However, when S-band WSR-

88D and disdrometer Z are compared, the measurements differ by only -1.5 dB (red plus signs in Fig.

2.6). The discrepancy between the X-band radar data and the disdrometer measurements is possibly due to the attenuation correction scheme overcorrecting the radar data within and behind the hail core of the supercell thunderstorm, although undercatchment of large hailstones by the disdrometer may have also contributed to the disagreement. A second case study of a supercell thunderstorm (Fig. 2.10) demonstrated that the disdrometer tends to record larger Z and ZDR (by 13 dB and 0.61 dB, respectively) than the X-band radar when the radar signal quality is poor (SQI < 0.8). For the same data subset, only a

0.66-dB discrepancy in Z exists when the disdrometer and WSR-88D are compared (blue plus signs in

Fig. 2.6). Disagreement between the disdrometer and attenuation-corrected X-band radar data due to poor

SQI is most likely to occur within trailing precipitation that is located behind heavy rainfall (relative to the radar location). A third case study analysis showed that when large hail is not detected, the radar signal quality is good (SQI > 0.8), and the precipitation structure is horizontally homogeneous, Z and ZDR observations from both instruments agree to within 5 dB and 0.5 dB (2137 – 2155 UTC in Fig. 2.11).

When the hydrometeor classification schemes for the radar and the disdrometer are compared

(Fig. 2.14), they agree 63% of the time. Disagreement results when the radar scheme diagnoses hail and the disdrometer scheme does not (7% of the observations) and when the disdrometer observes hail and it is not detected by the radar scheme (29%). Hail may be detected solely by the radar because small hail is present in the radar volume that melts prior to reaching the surface, or because the hail was especially sparse and not detected by the disdrometer. When the opposite situation occurred (i.e., the disdrometer observed hail but the radar scheme did not), it was shown that fewer, smaller hailstones were observed by the disdrometer than when the two instruments both observed hail. In these cases, the non-zero rule 35 strength (i.e., the weighted sum of the membership function values) of the hail class suggests that hail may have been present in the radar volume, but because the radar hydrometeor classification scheme identifies the most dominant hydrometeor in the sampling volume, the hail was not classified.

The data quality analysis presented here may be particularly valuable to those who undertake future

VORTEX2 microphysical process and data assimilation studies. We have shown that attenuation of radar data in severe thunderstorms can be substantial even in the portion of the thunderstorm that is initially penetrated by the radar beam. Further, since the assumptions in the attenuation correction scheme used in this study are not valid in ice, large errors may result in and beyond hail cores. This research may also be helpful to those who use the VORTEX2 disdrometer observations to validate model-predicted surface precipitation types. With the hydrometeor classification scheme in Fig. 2.4, median diameter and number concentration can be derived from the disdrometer data for each hydrometeor class and compared to those from numerical models.

2.6 ACKNOWLEDGMENTS

This research was sponsored by NSF ATM-0910424 and NSF DGE-1144083. Funding for Donald

Burgess came from NOAA/Office of Oceanic and Atmospheric Research under NOAA-University of

Oklahoma Cooperative Agreement #NA11OAR430072. Funding for NOXP data came from NSF ATM-

0802717. We thank Gwo-Jong Huang and Prof. Bringi (Colorado State University) for the T-matrix program, in addition to George Fernandez, Carlos Lopez, and Forrest Masters (University of Florida), who designed the articulating disdrometers and supplied four of the stationary disdrometers. We also thank Stephanie Higgins (University of Colorado) for writing many of the routines that process the disdrometer data. We are grateful to George Fernandez, Stephanie Higgins, Rachel Humphrey, Scott

Landolt, Carlos Lopez, Daniel Nuding, and Cameron Redwine for deploying instruments during

VORTEX2. The helpful feedback from two anonymous reviewers substantially improved an earlier version of this manuscript. Any opinions, findings, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the views of the National Science Foundation.

36 2.7 APPENDIX A: SENSITIVITY TO HAILSTONE CHARACTERISTICS IN THE T-MATRIX PROGRAM

To calculate Z and ZDR from disdrometer data that contain hail, characteristics of the hail must be specified, including axis ratio, fractional water content, and fall behavior. Table 2.3 provides the default values used in this analysis. However, since no measurements of axis ratio, fractional water content, or canting angle were made, the chosen values represent a source of uncertainty. Here we examine the sensitivity of the results to some of the hailstone characteristics in the T-matrix scattering calculations.

To quantify the sensitivity, we select the particle size distribution (PSD) observed by disdrometer

CU01 on 17 May 2010 (Fig. 2.8), since these data are from a hailstorm and should exhibit the greatest sensitivity to the hailstone characteristics. Z and ZDR are calculated for two values of the fractional water content of small hail (0.35 and 0.65), axis ratio of large hail (0.5 and 0.65), and canting angle standard deviation of all hail (35° and 65°). These values, together with those provided in Tab. 2.3, provide a reasonable range over which these characteristics can be expected to vary (Knight 1986; Lesins and List

1986; Ryzhkov et al. 2011). The resulting sensitivities in Z and ZDR for the entire time series recorded by

CU01 are shown in Fig. 2.15, relative to Z and ZDR obtained using the default values of the parameters in

Tab. 2.3. No sensitivity is evident with respect to the small hail fractional water content (Fig. 2.15a). In contrast, up to 2 dB sensitivity in Z and 0.9 dB in ZDR is present when the large hail axis ratio is varied

(Fig. 2.15b), although these sensitivities are only present in ~5 time steps of the disdrometer data. Z and

ZDR display more consistent sensitivity to hail canting angle standard deviation, although this sensitivity remains within 0.2 dB and 0.6 dB, respectively (Fig. 2.15c). Considering only the data included in Fig.

2.5 from CU01 on 17 May (Tab. 2.4), the mean sensitivity in Z (ZDR) to large hail axis ratio and hail canting angle standard deviation, respectively, is just -0.01 dB (0.03 dB) and 0.05 dB (0.1 dB). Similar results (Tab. 2.4) were obtained from the PSD observed by disdrometer UF05 on 12 June 2010 (Fig.

2.13).

In summary, because the PSDs in this study contain little hail relative to rain, Z and ZDR exhibit limited sensitivity to hailstone axis ratio, fractional water content, and canting angle. This result does not 37 mean that Z and ZDR are generally insensitive to hailstone characteristics. Rather, the small sample volume of the PARSIVEL disdrometer caused relatively few hailstones to be observed, which reduced the sensitivity of the results to the hailstone characteristics and meant that rain was the dominant contributor to Z and ZDR.

Tab. 2.4: Mean sensitivity in disdrometer Z and ZDR to small hail fractional water content (FWC), large hail axis ratio (AR), small and large hail canting angle standard deviation (σCA), and the disdrometer hydrometeor classification scheme for two subsets of the data in Fig. 2.5: observations from disdrometer CU01 on 17 May 2010 and observations from disdrometer UF05 on 12 June 2010. All values are relative to those obtained using the default parameters listed in Tab. 2.3 and the disdrometer hydrometeor classification scheme shown in Fig. 2.4.

Sensitivity Test 17 May 2010 (CU01) 12 June 2010 (UF05) ΔZ (dB) ΔZ (dB)

ΔZDR (dB) ΔZDR (dB) FWC = 0.65 0.000471 0.00177 -0.000195 -0.000917 FWC = 0.35 -0.000391 -0.00136 0.000138 0.000702 AR = 0.65 -0.0110 0.0152 0.00954 0.0114 AR = 0.5 -0.00470 0.0399 0.0340 0.0110 0.0412 0.0669 σCA = 35° 0.130 0.190 0.0507 0.0687 σCA = 65° 0.116 0.141 Unclassified particles are rain 0.193 0.0401 0.0116 -0.00168 Unclassified particles are small 0.121 0.0105 hail -0.0133 -0.00687 Small hail particles removed -0.00443 -0.0158 0.00194 0.0103

38

Fig. 2.15: Sensitivity of Z (red lines) and ZDR (blue lines) from disdrometer CU01 on 17 May 2010 to a) fractional water content of small hail, b) axis ratio of large hail, and c) canting angle standard deviation of small and large hail. The sensitivities are relative to the Z and ZDR obtained by using the default values of small hail fractional water content (0.5), large hail axis ratio (0.8), and small and large hail canting angle standard deviation (50°).

39 2.8 APPENDIX B: SENSITIVITY TO DISDROMETER HYDROMETEOR CLASSIFICATION SCHEME

A partially melted hailstone of a given diameter can exhibit a range of fall speeds depending on its fractional water content and bulk density (Rasmussen and Heymsfield 1987a). In the disdrometer hydrometeor classification scheme of Friedrich et al. (2013a), fall speed curves for dry graupel (Locatelli and Hobbs 1974) are used to determine the fall speed thresholds for the small hail region with 2 mm < d <

5 mm (Fig. 2.4). Because small hailstones in supercell thunderstorms have a bulk density greater than that of graupel and are likely embedded in a torus of liquid water, it is reasonable to expect that these particle fall speeds are greater than the graupel curves offered by Locatelli and Hobbs (1974), but somewhat less than the fall speeds for pure rain given by Atlas et al. (1973). In the Friedrich et al. (2013a) classification scheme, such particles are left unclassified (see Fig. 2.4) and are excluded from T-matrix computations of

Z and ZDR, since there is uncertainty about whether they should be modeled as rain or small hail.

To examine the sensitivity of Z and ZDR to these excluded particles, we again use the data recorded by disdrometer CU01 during the 17 May 2010 hailstorm (Fig. 2.8), and compute Z and ZDR for two different tests: one that assumes the previously unclassified particles are rain, and a second that assumes the particles are hail. Figure 2.16 a illustrates the sensitivity in Z (red lines) and ZDR (blue lines) for both the rain (solid lines) and hail (broken lines) tests. Up to 13 dB (1.5 dB) sensitivity in Z (ZDR) is evident for the time step at 2212 UTC. However, in all but one of the other 86 one-minute time steps, negligible sensitivity is present. The reason for the large sensitivity at 2212 UTC is due to small disdrometer reflectivity (23 dBZ, not shown), coupled with a relatively large particle of d = 4.25 mm within the unclassified region. Because of the small total number of particles observed at 2212 UTC, the particle within the unclassified region has marked influence on Z and ZDR. In heavy rainfall typical of the thunderstorms observed in this study, however, Z and ZDR are relatively insensitive to whether particles in the unclassified region are discounted or assumed to be rain or hail. Table 2.4 indicates that for the data included in Fig. 2.5 from CU01 on 17 May, the mean sensitivity to the unclassified region is less than 0.2 dB and 0.02 dB for Z and ZDR, respectively. In fact, when all small hail particles are removed from the 17 40 May PSD, Fig. 2.16b shows that Z and ZDR change by less than 0.2 dB for the entire time series recorded by CU01. For the data included in Fig. 2.5, the mean change in Z (ZDR) is -0.004 dB (0.002 dB); see Tab.

2.4. These tests demonstrate that despite the uncertainty in small hailstone characteristics, the impact on the results is small, likely because the PSDs examined here contain little small hail compared to rain.

Fig. 2.16: As in Fig. 2.15, but for a) the precipitation type of the particles in the unclassified region in Fig. 2.4 and b) the small hail region in Fig. 2.4. The sensitivities are relative to the Z and ZDR obtained by a) excluding the unclassified particles and b) including the small hail particles.

41 3 AEROSOL EFFECTS ON IDEALIZED SUPERCELL THUNDERSTORMS IN DIFFERENT ENVIRONMENTS

This chapter is reprinted with permission from:

Kalina, E. A., K. Friedrich, H. Morrison, and G. H. Bryan, 2014: Aerosol effects on idealized supercell thunderstorms in different environments. J. Atmos. Sci., 71, 4558-4580.

3.0 ABSTRACT

Idealized supercell thunderstorms are simulated with the Weather Research and Forecasting (WRF) model at 15 cloud condensation nuclei (CCN) concentrations (100-10 000 cm-3) using four environmental soundings with different low-level relative humidity (RH) and vertical wind shear values. The Morrison microphysics scheme is used with explicit prediction of cloud droplet number concentration and a variable shape parameter for the raindrop size distribution (results from simulations with a fixed shape parameter are also presented). Changes in the microphysical process rates with CCN concentration are negligible beyond CCN ≈ 3000 cm-3. Changes in cold pool characteristics with CCN concentration are non-monotonic and highly dependent on the environmental conditions. In moist conditions with moderate vertical wind shear, the cold pool area is nearly constant with respect to CCN concentration, while the area is reduced by 84% and 22% in the soundings with dry RH and large vertical wind shear, respectively.

With the exception of the dry RH sounding, domain-averaged precipitation peaks between 500 cm-3 and

5000 cm-3, after which it remains constant or slowly decreases. For the dry RH sounding, the domain- averaged precipitation monotonically decreases with CCN concentration. Accumulated precipitation is enhanced (by up to 25 mm) in the most polluted cases near the updrafts, except for the dry RH sounding.

The different responses for moist and dry soundings are mostly due to increased (decreased) low-level latent cooling from melting hail (evaporating rain) with increasing CCN concentration in the moist soundings. This compensating effect does not exist when the low-level RH is dry.

42 3.1 INTRODUCTION

A variety of convective modes and processes have been shown to exhibit sensitivity to aerosol concentration (e.g., Khain et al. 2005; van den Heever and Cotton 2007; Khain and Lynn 2009; Mansell and Ziegler 2013). However, there is a lack of research that examines how the microphysical processes and resulting thermodynamic structure of thunderstorms vary across the wide range of aerosol concentrations that are possible within the atmosphere. Specifically, it is unclear at what aerosol concentration perturbed microphysical processes become evident, and whether these perturbations continue to grow as aerosol concentration increases or if additional increases have negligible influence above a certain threshold. To understand the response of thunderstorms to a variety of aerosol concentrations, the Weather Research and Forecasting (WRF) model is run at a cloud-resolving horizontal grid spacing of 1 km with the Morrison microphysics scheme, in which the concentration of cloud condensation nuclei (CCN) at 1% supersaturation is varied from 100 cm-3 to 10 000 cm-3. Four different environmental soundings are tested, all of which are supportive of supercell thunderstorms (i.e., thunderstorms with rotating updrafts), which may produce large hail, damaging straight-line winds, flooding, and tornadoes. Mass budgets are analyzed to quantify the effect of increasing CCN concentration on rates of riming, melting, droplet collection, and evaporation. In addition, these perturbed microphysical processes are linked to changes in the low-level cold pool and to differences in accumulated surface precipitation to understand how supercell thunderstorms respond across the wide spectrum of plausible atmospheric aerosol concentrations.

In the atmosphere, aerosol sources include those that are both anthropogenic (e.g., biomass burning, sulfate emissions, motor vehicle exhaust) and natural (e.g., evaporation of sea spray, dust lofting, wildfires) (Penner et al. 1994; Levin and Cotton 2009; Tao et al. 2012). Size radii range from 0.1 μm for

Aitken particles to 100 μm for giant aerosols comprised of sea salt (Rogers and Yau 1989; Levin and

Cotton 2009). In unpolluted conditions over the open ocean, aerosol number concentrations may be as small as 100 cm-3, while environments contaminated by smoke from forest fires may feature concentrations in excess of 10 000 cm-3 (Andreae et al. 2004). Observations from the Southern Great 43 Plains (SGP) site of the Atmospheric Radiation Measurement (ARM) Climate Research Facility in

Lamont, Oklahoma demonstrate that near-surface CCN concentrations on the U.S. Great Plains frequently varied between 1000 cm-3 and 5000 cm-3 on supercell thunderstorm days, and approached 10 000 cm-3 on other days with ordinary convection (Fig. 3.1). Such widely varying CCN number concentration may play an important role in modifying precipitation development in the supercell thunderstorms that commonly affect the Great Plains, since the growth of precipitating liquid hydrometeors starts when water vapor condenses onto CCN to form cloud droplets. The cloud droplet activation rate is affected by the type and size of the aerosols that serve as CCN, and competition for a limited amount of water makes the aerosol concentration critical in determining how large cloud droplets can grow and, therefore, how efficient autoconversion into raindrops will be. Because aerosols can change the rates of cloud microphysical processes, they can also alter the local temperature and moisture profiles by modifying the latent cooling/heating that results from phase changes of water. In this manner, the buoyancy, precipitation efficiency, and the lifetime of the cloud can all be affected by changes in the aerosol properties. For more information about cloud-aerosol interactions, the reader is directed to Levin and Cotton (2009) and Tao et al. (2012).

With ongoing pollution emissions from urban areas and a recent increase in the number of large wildfires in the western United States and elsewhere (Diaz and Swetnam 2013; Luo et al. 2013; Lang et al. 2014), the need to understand how enhanced aerosol concentrations affect a broad range of convective modes has increased. In polluted environments, previous modeling studies that examined the effects of

CCN concentration on thunderstorm characteristics found evidence for delayed onset of precipitation

(Tao et al. 2007; van den Heever and Cotton 2007; Storer et al. 2010; Mansell and Ziegler 2013), suppressed raindrop collision and coalescence (van den Heever and Cotton 2007; Fan et al. 2007; Lerach et al. 2008; Storer et al. 2010), decreased cold pool size (Lerach et al. 2008; Storer et al. 2010; Lerach and

Cotton 2012), and faster updraft speeds due to enhanced latent heat release (Seifert and Beheng 2006; van den Heever et al. 2006; Fan et al. 2007; Ntelekos et al. 2009; Mansell and Ziegler 2013). However, there is disagreement on whether surface precipitation is enhanced or decreased in polluted environments, with 44 sensitivity to the low-level (0-3 km) relative humidity evident (Fan et al. 2007; Khain et al. 2008; Khain and Lynn 2009; Fan et al. 2009). Many recent modeling studies present results from only two or three different CCN concentrations (e.g., Lerach et al. 2008; Khain and Lynn 2009; Lerach and Cotton 2012), leading to uncertainty in how to apply these results to the broad range of observed CCN concentrations. In addition, while some studies have investigated aerosol effects on supercell thunderstorms (e.g., Seifert and Beheng 2006; Lerach et al. 2008; Khain and Lynn 2009; Storer et al. 2010; Lebo and Seinfeld 2011;

Morrison 2012; Lerach and Cotton 2012; Lebo et al. 2012), many studies have focused on non- supercellular storms. Results from studies of aerosol impacts on other convective modes may not be applicable to supercell thunderstorms due to the strong vertical wind shear of the environment, which results in updrafts that are strongly influenced by dynamic perturbation pressure as well as buoyancy

(Rotunno and Klemp 1985). This study is unique because it examines the influence of aerosol concentration on supercell thunderstorms for 15 CCN concentrations that vary from 100 cm-3 to 10 000 cm-3 across different low-level relative humidity and vertical wind shear environments. Thus, we will determine whether aerosol-induced perturbations in the microphysics and thermodynamics increase monotonically or non-monotonically with CCN concentration, whether these perturbations cease to increase beyond a certain limit, and how these changes vary across different environments. While a few other studies (Seifert and Beheng 2006; Fan et al. 2009; Storer et al. 2010; Lebo and Seinfeld 2011;

Lerach and Cotton 2012) have tested the sensitivity of thunderstorms in sheared environments to CCN and relative humidity effects, the objective of this research is to investigate how individual microphysical process rates vary across a spectrum of CCN concentrations, up to a larger value (CCN = 10 000 cm-3) than these previous studies.

In section 3.2, we discuss the WRF model configuration, the four different soundings used to initialize the model, and the Morrison microphysics scheme as it pertains to this study. Section 3.3 presents and compares results for each initial sounding. The research is summarized and concluding remarks are provided in section 3.4.

45

Fig. 3.1: Maximum daily surface CCN (at supersaturation between 0.9 and 1.1%) and condensation nuclei (CN) number concentrations at the DOE-ARM Southern Great Plains (SGP) site from 20 April to 10 June 2011. Days with convective activity (i.e., showers and/or thunderstorms) near the SGP site are indicated in red, and days with supercell thunderstorms are shown in purple. Data were obtained from the DOE- ARM online archive (http://www.arm.gov).

3.2 METHODS

3.2.1 MODEL CONFIGURATION

This study uses version 3.3 of the three-dimensional, non-hydrostatic WRF model (Skamarock et al. 2008) with the Advanced Research WRF (ARW) core. ARW evaluates the equations of motion with a third-order Runge-Kutta integration scheme. An Arakawa C-grid is used to stagger the components of the three-dimensional wind one-half grid length away from the thermodynamic variables (e.g., potential temperature), which are evaluated at the center of each grid cell. The model utilizes terrain-following

(eta) coordinates.

46 The chosen model configuration is similar to the idealized supercell thunderstorm test case that is provided with WRF and to those of Morrison and Milbrandt (2011), Lebo et al. (2012), and Morrison

(2012). Idealized simulations are chosen so that the effect of CCN concentration on the microphysics and thermodynamics of the supercell thunderstorm can be quantified in the absence of secondary feedbacks from radiative, boundary layer, and surface layer processes. The domain has a horizontal grid spacing of 1 km and spans 200 km in both the zonal and meridional directions. The horizontal boundaries are periodic to ensure conservation of total mass within the domain. A time step of 2 s is used, except for the acoustic modes, for which a 0.33-s time step is used. In the vertical, an exponentially stretched grid with 70 levels and a nearly constant spacing of ~300 m is selected. The model top is at z = 24 km, and a Rayleigh damper with a damping coefficient of 0.003 s-1 is used within the upper 5 km to eliminate gravity waves that reflect off the upper boundary.

Horizontal and vertical advection are calculated using monotonic fifth- and third-order schemes, respectively. Turbulent diffusion is computed with a 1.5-order turbulent kinetic energy scheme

(Skamarock et al. 2008). The Morrison scheme, discussed in more detail below, is used to represent microphysical processes. Radiative transfer, surface fluxes, and Coriolis force are neglected for simplicity. Convection is initiated with a warm perturbation in the potential temperature field. The maximum amplitude of the perturbation is 3 K, it is centered at z = 1.5 km, and it is 20 km wide in the horizontal and 3 km in height. The model equations are integrated for two hours, with output written every 10 minutes.

The default sounding (hereafter referred to as def) used to initialize the model (Fig. 3.2) is based on Weisman and Klemp (1982), with the environmental wind making a quarter-circle when plotted on a hodograph (Fig. 3.3). To study the sensitivity of the results to environmental conditions, simulations are also conducted with three additional soundings (Figs. 3.2 -3.3 ; Tab. 3.1), in which the low-level relative humidity and vertical wind shear of the def sounding are modified. These soundings are called low relative humidity (loRH), high relative humidity (hiRH), and high vertical wind shear (hiWS). In the loRH case, the mean surface-800 mb relative humidity from the default input sounding is reduced from 47

Fig. 3.2: Skew-T log-P diagram with the soundings used to initialize the WRF model, including the default (def) sounding and the soundings used for the sensitivity tests: low relative humidity (loRH; dashed line), high relative humidity (hiRH; dotted line), and high vertical wind shear (hiWS; rightmost wind barbs). The solid red line is the temperature profile, while the dewpoint temperature profiles are shown in blue. The wind speed and direction are represented by two sets of wind barbs on the right side of the diagram: one set for the hiWS sensitivity test and one set for all other simulations (def).

48

Fig. 3.3: Hodograph of the wind profile used to initialize the WRF model in the high wind shear case (red line; hiWS) and in all other cases (blue line; def). Each filled circle represents an individual wind vector from the skew-T log-P diagram in Fig. 3.2. The numbers and tick marks along the red and blue lines indicate the height above the surface (in km), and the numbers along the concentric circles indicate the wind speed (in m s-1).

49 ~80% to ~61% (Fig. 3.2 and Tab. 3.1). The result is an “inverted-V” sounding (Brady and Szoke 1989;

Johns and Doswell 1992) that is typical of the western U.S. Great Plains in the summer months and that is conducive to high-based, low-precipitation supercell thunderstorms (e.g., Bluestein and Parks 1983; Grant and van den Heever 2014). In the hiRH case, the mean surface-800 mb relative humidity is increased to

~91%, yielding a surface-based Convective Available Potential Energy (CAPE) of 5138 J kg-1 (compared to 2745 J kg-1 in the def sounding; Tab. 3.1). Finally, in the hiWS case, the wind speed is increased by ~5 m s-1 per km until it reaches a maximum value of ~25 m s-1 at z = 4.25 km, above which the wind speed is constant. As a result, the hiWS sounding has a 0-3 km bulk shear of ~16 m s-1 compared to ~8 m s-1 in the def sounding (Fig. 3.3).

Previous modeling studies (e.g., Weisman and Klemp 1982, 1984) have suggested that the convective mode depends on the convective bulk Richardson number (R), given by

, (1) . where and are the difference between the density-weighted mean wind calculated over the 0-6 km layer and a 500-m deep surface layer. Weisman and Klemp (1984) found that 15 < R < 45 strongly favors supercell thunderstorms, with R = 18 producing the fastest updrafts relative to the CAPE. Table 3.1 shows that R ranges from 12.4 (loRH) to 65.8 (hiRH) for the soundings here. While these values lie outside of the range Weisman and Klemp (1984) suggested that most favors supercell thunderstorms, we nevertheless observe discrete cell development, storm splitting, and substantial updraft rotation (see section 3.3.1) in all of the simulations.

Tab. 3.1: Relative humidity, Convective Available Potential Energy (CAPE), and Richardson number for the soundings used to initialize the WRF model in the default (def), low relative humidity (loRH), high relative humidity (hiRH), and high vertical wind shear (hiWS) cases.

Case Mean Surface-800 mb Surface-based Bulk Richardson Relative Humidity (%) CAPE (J kg-1) Number def 79.9 2745 35.2 loRH 61.3 967 12.4 hiRH 90.9 5138 65.8 hiWS 79.9 2745 16.8

50 3.2.2 MICROPHYSICS SCHEME

The Morrison double-moment microphysics scheme (Morrison and Pinto 2005; Morrison et al.

2005; Morrison et al. 2009) is selected to model microphysical processes. The scheme considers five hydrometeor species, all of which are assumed to consist of spherical particles: cloud droplets, cloud ice, rain, snow, and a rimed ice category. The latter uses bulk density and fall speed characteristics that are typical of either graupel or hail (user selectable). In this research, hail is chosen as the rimed ice category, which may be more suitable than graupel for studies of continental deep convection (McCumber et al.

1991; Bryan and Morrison 2012). The hydrometeor size distributions N(D) are represented by gamma functions of the form:

(3.2) where D is the particle diameter, N0 is the intercept parameter, μ is the shape parameter, and λ is the slope parameter. N0 and λ are predicted for each hydrometeor species via Eqs. (3.3) and (3.4):

(3.3)

(3.4) where q is the hydrometeor mass mixing ratio, Γ is the Euler gamma function and c is a parameter in the power law that relates the diameter and mass (m) of the hydrometeors:

(3.5)

Since all particles are assumed to be spherical, c = π/6 × ρ, where ρ is the bulk density of the hydrometeor class, given by Tab. 4 in Morrison and Milbrandt (2011).

For cloud ice, snow, and hail, μ is set to zero. For cloud droplets, the value of μ is a function of the predicted cloud droplet number concentration according to Martin et al. (1994) and varies from 2 to

10. For the simulations presented in sections 3.3.1-3.3.3, a variable shape parameter for the raindrop size distribution is used. Specifically, μ is diagnosed with the shape-slope relation from Cao et al. (2008), allowing μ to vary with λ:

51 0.0201 0.902 1.718 (3.6)

This relationship was derived from data collected in central Oklahoma by three two-dimensional video disdrometers from May 2005 to May 2007. In total, 14 200 1-min drop spectra were collected in all seasons in both stratiform and convective rain events that varied in rainfall rate from 0.1 mm hr-1 to 100 mm hr-1 (Cao et al. 2008). While a few other shape-slope relations exist in the literature for more tropical climates (e.g., Florida, Zhang et al. 2001; India, Rao et al. 2006), they are not considered here because shape-slope relations exhibit regional dependence (Cao et al. 2008) and supercell thunderstorms are primarily mid-latitude phenomena.

The use of a variable μ scheme for rain helps address the problem of excessive size sorting that has been noted in double-moment bulk microphysics schemes with fixed μ (Wacker and Seifert 2001;

Milbrandt and Yau 2005a; Milbrandt and McTaggart-Cowan 2010; Kumjian and Ryzhkov 2012). To determine μ within the microphysics scheme, an initial guess of μ = 0 is made and Eqs. (3.4) and (3.6) are iterated until λ converges to within 0.1%. This formula is not extrapolated to values of λ larger than the

Cao et al. (2008) data range (20 mm-1), giving a maximum μ of approximately 8.28. The minimum allowed μ for rain is 0. In section 3.3.4, results from the variable μ scheme are compared to results from a set of μ = 0 simulations with the default sounding. We note that comparing simulations in which the hail shape parameter is fixed versus allowed to vary might also be interesting, although this topic is left unexplored here due to limited observations of the shape-slope relation for hail.

In this research, the CCN spectrum is represented by a power law relationship (Pruppacher and

Klett 1997):

(3.7)

-3 where NCCN is the number concentration of activated cloud condensation nuclei (cm ), S is the supersaturation ratio in percent, C is the CCN concentration (cm-3) at S = 1%, and k is a unitless constant.

As S increases, more CCN activate. Tables 9.1 and 9.2 in Pruppacher and Klett (1997) demonstrate that while C is observed to be larger in continental vs. maritime air masses, there is no clear dependence of k on the type of air mass or on the value of C. Therefore, in this study, we assume that k is equal to 0.7, 52 which is the average of the values in Pruppacher and Klett (1997). For simplicity, the initial vertical distribution of CCN is assumed to be constant with height and CCN transport and sources and sinks

(other than scavenging from cloud droplet nucleation) are not considered. The total number concentration of CCN plus cloud droplets within a grid cell remains constant throughout the simulation so that the CCN concentration reverts to the background value upon cloud evaporation.

Throughout the activation process, mixing is treated as homogeneous; that is, entrainment of subsaturated air causes the mean cloud droplet radius to decrease while the droplet concentration remains constant. At cloud base, droplet activation is performed using the approach from Rogers and Yau (1989), which depends only on updraft speed and assumes that no initial cloud water exists:

. 0.880.07 , (3.8) where wef is the sum of the resolved and subgrid vertical velocity. At cloud base, this relation was shown by Dearden (2009) to produce cloud droplet concentrations to within 10-20% of those obtained when supersaturation is resolved explicitly (requiring a time step of 0.1s or shorter). However, within the cloud, Eq. (3.8) overestimates the number of activated cloud droplets because it fails to account for the reduction in supersaturation due to the condensational growth of existing cloud droplets (Dearden 2009).

A different approach is therefore used in the cloud interior, in which the supersaturation is assumed to be in quasi-equilibrium; that is, the production of supersaturation from upward motion is assumed to be balanced by the loss from depositional growth of existing hydrometeors (Eq. 3.9; adapted from Morrison et al. 2005).

(3.9)

In the above equation, qv is the water vapor mixing ratio, qsw is the liquid water saturation mixing ratio, qsi is the ice saturation mixing ratio, T is the temperature, g is the gravitational acceleration, cp is the specific heat of air at constant pressure, Q1 = 1 + (dqsw/dT)(Ls/cp), Q2 = 1 + (dqsi/dT)(Ls/cp), Ls is the latent heat of

53 sublimation, and τc, τr, τi, τs, and τh are the saturation relaxation time scales of cloud droplets, rain, ice crystals, snow, and hail, respectively. The supersaturation obtained from Eq. (3.9) is then substituted into

Eq. (3.7) to obtain the predicted number of activated cloud droplets in the cloud interior. However, if the predicted number of activated cloud droplets is less than the number that already exists, no new CCN are activated. In addition, the number of activated cloud droplets is never allowed to exceed that predicted by

Eq. (3.8).

Ice crystals are produced through primary nucleation, the Hallett-Mossop process (i.e., rime splintering), and through freezing of cloud droplets (Morrison and Milbrandt 2011). Autoconversion of ice crystals to snow is based on the approach of Harrington et al. (1995) assuming a threshold size of 125

μm. Snow can then be rimed to form hail particles. During melting, snow and hail particles are assumed to have constant mean mass diameter, and the total number concentration of rain, snow, and hail particles is conserved. All ice crystals are melted into rain when the ambient air temperature exceeds 0 °C. Liquid water particles that are shed from hailstones are assumed to have a mean mass diameter of 1 mm based on laboratory (Carras and Macklin 1973; Lesins et al. 1980; Rasmussen et al. 1984) and field (Rasmussen and Heymsfield 1987b) experiments. These studies consistently showed that water drops shed from hailstones undergoing wet growth have diameters of 0.5-2 mm, with a mode of 1 mm.

Raindrop breakup is treated implicitly by limiting the bulk collection efficiency of raindrops (Ec) at mean mass diameters (Dmr) that exceed a specified threshold, Dth (Verlinde and Cotton 1993; Morrison et al.

2012). When Dmr is less than Dth, the collection efficiency is assumed to be one. Once Dmr exceeds Dth, the collection efficiency is reduced according to

2exp2300 , Dmr > Dth (3.10)

Here, Dth is set to a constant value of 300 μm, which is the default value in the public release of the

Morrison scheme in WRF version 3.3. The size and strength of the low-level cold pool are sensitive to the choice of Dth, since larger values of Dth will increase Dmr and thereby reduce the evaporation rate (see

Morrison and Milbrandt 2011 and Morrison et al. 2012). Further discussion of how the choice of Dth influences the evaporation rate and therefore the cold pool, however, is beyond the scope of this paper. 54 Simulations are conducted for 15 different values of C that range between 100 cm-3 and 10 000 cm-3: 100,

250, 500, 750, 1000, 1500, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, and 10 000 cm-3. Note that

10 000 cm-3 is a larger CCN concentration than has been used in previous numerical studies of CCN concentration and supercell thunderstorms (e.g., 1500 cm-3 in Khain and Lynn 2009; 2000 cm-3 in Lerach and Cotton 2012). However, observations from the Department of Energy’s Atmospheric Radiation

Measurement (ARM) site in the Southern Great Plains (SGP; Fig. 3.1) show that CCN concentrations can reach at least 6000 cm-3 on supercell thunderstorm days (12 May) and can approach 10 000 cm-3 on days with ordinary convection (18 May).

3.3 RESULTS

3.3.1 CCN EFFECTS ON HYDROMETEOR CHARACTERISTICS AND MICROPHYSICAL PROCESSES

Reflectivity that exceeds 0 dBZ first appears at t ≈ 20 min in each of the simulations. A hook echo becomes evident in radar reflectivity at t ≈ 30 min (Fig. 3.4a). By t ≈ 60 min, the supercell thunderstorm has clearly split into left- and right-moving updrafts, and the hook echo associated with the right mover has become more distinct (Fig. 3.4b). The two cells move at ~90° from each other (Fig. 3.4c- d), and by t = 120 min, the area with reflectivity greater than 0 dBZ is ~8200 km2 (21% of the domain).

The above results are similar for both the CCN = 100 cm-3 and CCN = 10 000 cm-3 runs. Note that 2-5 km updraft helicities (Kain et al. 2008) of 1000-2700 m-2 s-2 are present throughout the simulations of def, hiRH, and hiWS, while values in excess of 300 m-2 s-2 are simulated for the loRH case (Fig. 3.5), justifying our classification of these thunderstorms as supercells.

The next several analyses compare only the cleanest (CCN = 100 cm-3) and dirtiest (CCN = 10

000 cm-3) simulations for each of the four soundings, considering differences in the mean diameters (Fig.

3.6), number concentrations (Fig. 3.7), and mass mixing ratios (Fig. 3.8) of cloud droplets, rain, and hail.

Each of these figures consists of conditional (i.e., only non-zero values), domain-averaged vertical profiles at t = 120 min. Due to more CCN, the dirtiest simulation (independent of the initial sounding) has

55

Fig. 3.4: Horizontal cross-sections of simulated radar reflectivity (assuming a 10-cm wavelength) at z = 1 km AGL at a) t = 30 min, b) t = 60 min, c) t = 90 min, and d) t = 120 min using the default (def) sounding and a CCN concentration of 10 000 cm-3.

56

Fig. 3.5: Updraft helicity (integrated over 2-5 km AGL) from t = 10 min to t = 120 min at ten minute intervals for the CCN = 10 000 cm-3 runs of a) def, b) hiRH, c) loRH, and d) hiWS.

57

Fig. 3.6: Conditional, domain-averaged vertical profiles of hydrometeor mean mass diameter at t = 120 min for cloud droplets (green lines), rain (blue lines), and hail (purple lines) for a) def, b) hiRH, c) loRH, and d) hiWS soundings. Results from the cleanest (CCN = 100 cm-3; solid lines) and dirtiest (CCN = 10 000 cm-3; dashed lines) simulations are shown.

58

Fig. 3.7: As in Fig. 3.6, but for hydrometeor number concentration.

59

Fig. 3.8: As in Fig. 3.6, but for hydrometeor mass mixing ratio.

60 smaller (5 μm vs. 17 μm; Fig. 3.6), more numerous (~4000 cm-3 vs. ~100 cm-3; Fig. 3.7) cloud droplets than the cleanest simulation. This result is known as the aerosol indirect effect, first noted by Twomey

(1974) and later simulated by Khain et al. (2005), Fan et al. (2007), Tao et al. (2007), Storer et al. (2010), and many other recent studies across a wide variety of convective modes. Because the mean cloud droplet size is smaller in the dirtiest run, fewer cloud droplets make the transition into rain and hail particles, with a 97% (85%) smaller number concentration of rain (hail) at 5 km (10 km) compared to the cleanest simulation of the def sounding (Fig. 3.7a; note that a logarithmic scale is used). Similar trends are seen in the number concentration profiles that result from the other soundings (Fig. 3.7b-d). Although there are fewer raindrops and hailstones in the dirtiest simulation, the rain and hail particles that are present have three times the cloud water mass available for collection (0.9 g kg-1 vs. 0.3 g kg-1 at 9 km; Fig. 3.8a), which leads to mean mass diameters that are 30% larger for rain and 3% larger for hail near the surface in the dirtiest run with the def sounding (Fig. 3.6a; Tab. 3.2). Similar results (fewer, but larger rain and hail particles in polluted conditions) were also simulated by Storer et al. (2010), Lerach and Cotton (2012), and others. For the other soundings, the only notable departure from this pattern is with the loRH sounding, in which the mean mass diameter of rain is about equal in the cleanest and dirtiest runs near the surface (Fig. 3.6c; Tab. 3.2), despite being larger in the dirtiest run from 2-8 km (as in the other soundings).

The differences in the number concentration and mean diameter of the hydrometeors in the cleanest and dirtiest runs lead to important changes in the microphysical process rates. These changes can be analyzed by calculating the liquid (Fig. 3.9) and ice (Fig. 3.10) mass budgets, similar to the approach used by Khain et al. (2011) and others. Table 3.2 summarizes the differences in the vertically-integrated microphysical process rates between the cleanest and dirtiest runs for each of the four soundings.

Collection of cloud droplets by rain generally proceeds at a slower rate between 2-9 km in the dirtiest runs

(Fig. 3.9) at both t = 80 min (thin lines) and t = 120 min (thick lines) due to the smaller rain mass mixing ratio (Fig. 3.8). However, because there are more cloud droplets to collect per raindrop, the resulting raindrops have a mean mass diameter that is up to 0.4 mm larger compared to the cleanest runs from 0-4 61 Tab. 3.2: The percent change in microphysical and thermodynamic quantities between the cleanest (CCN = 100 cm-3) and dirtiest (CCN = 10 000 cm-3) simulations (DIRTIEST – CLEANEST) at t = 120 min for all cases: default (def), high relative humidity (hiRH), low relative humidity (loRH), high vertical wind shear (hiWS), and the default sounding with μ for rain set to zero (zero μ or ZMU). Cold pool characteristics, precipitation, and hydrometeor diameters are calculated at the lowest model level (z = 170 m).

def hiRH loRH hiWS ZMU Cold Pool Size -3.3 2.4 -83.8 -21.8 -13.0 Mean Cold Pool θ’ -6.3 -9.6 -14.5 -19.1 -10.2 Mean Total Precipitation 7.9 12.2 -50.1 3.2 -6.4 Mean Rain Diameter 29.7 23.6 0.37 35.3 15.5 Mean Hail Diameter 2.9 2.0 2.1 0.52 1.0 Vert. Int. Melting -10.4 -2.3 -59.3 -34.8 -20.1 Vert. Int. Riming w/Cloud 14.6 26.3 -36.7 2.2 3.0 Vert. Int. Riming w/Rain -78.0 -57.1 -94.8 -72.9 -72.0 Vert. Int. Total Riming -23.4 -2.6 -54.6 -29.7 -20.2 Vert. Int. Evaporation -43.7 -33.5 -62.6 -45.0 -33.2 Vert. Int. Collection -37.3 -9.3 -74.9 -47.6 -30.9

km (Fig. 3.6). Due to the larger raindrop size in conjunction with the reduced rain number concentration, rain evaporation at these heights is reduced in the dirtiest runs. An exception is the loRH case (Fig. 3.9c), in which the drier boundary layer is associated with a much higher cloud base (~1.8 km versus ~0.5 km in the other soundings) in both the cleanest and dirtiest runs (Fig. 3.7c), thereby preventing cloud droplet collection by rain below z = 2 km and leading to rain mean mass diameters that are about the same near the surface in both the cleanest and dirtiest runs (Fig. 3.6c; Tab. 3.2). In regards to ice processes, riming of hailstones with cloud droplets (d < 0.1 mm) is increased between 6-11 km with more CCN (Fig. 3.10) in def and hiRH. This trend is not as apparent in hiWS and is reversed in loRH, likely because both hiWS and loRH feature reduced differences in cloud droplet mass between the cleanest and dirtiest runs near 5 km (Fig. 3.8c-d). On the other hand, riming of hail with raindrops is decreased by up to 2×10-4 g kg-1 s-1 in the dirtiest simulation of all of the soundings due to the reduced rain number concentration. Considering the melting rate, the updraft-influenced (w > 10 m s-1) freezing level is located between 4.0-4.4 km in all of the soundings (not shown). The peak melting rate is located at 2.75-3.25 km, although the peak rate is up to 500 m closer to the surface in the dirtiest runs, likely because the larger diameter hailstones (Fig.

62 3.6) have faster fall speeds and, therefore, get closer to the surface before melting substantially. We also note an important difference between the def/hiRH versus the loRH/hiWS cases in Fig. 3.10: namely, more melting occurs in the lowest ~2 km in the dirtiest (relative to the cleanest) runs of def and hiRH at both t = 80 min and t = 120 min, while the opposite trend is apparent in loRH and hiWS. The increased melting at larger CCN concentrations in def and hiRH may be due to the dirtiest run having larger hail number concentrations than the cleanest run from 0-2 km (Fig. 3.7a-b), a trend that does not occur in loRH (Fig. 3.7c) or hiWS (Fig. 3.7d). The additional latent cooling from melting hail in the dirtiest runs of def and hiRH compensates for the decreased evaporative cooling in these runs, thereby impacting the cold pool characteristics in def and hiRH, which will be discussed later.

Now considering all of the CCN values (Fig. 3.11), it is apparent that it is not necessary for the

CCN concentration to be increased to 10 000 cm-3 for the microphysical processes to be substantially perturbed. In fact, for each of the soundings, perturbations in the rates of vertically-integrated melting, evaporation, and riming of hailstones generally saturate by CCN = 3000 cm-3, after which these rates exhibit little change. The microphysical processes that directly involve cloud droplets (i.e., collection of cloud droplets by rain and riming of hailstones by cloud droplets) are most sensitive to further increases in CCN concentration above 2000-3000 cm-3, although the rate of change in these processes is generally smaller above CCN = 3000 cm-3 in each of the soundings. While Fig. 3.11 shows results for t = 120 min, similar results are obtained for each of the model outputs from t = 60 min to t = 120 min. These results are also similar to Khain et al. (2011), who used a two-dimensional spectral bin model to conclude that the microphysical processes in a multicell thunderstorm were less sensitive to further increases in CCN concentration above 3000 cm-3. Together, these results suggest that extreme concentrations of CCN, such as those observed downwind of forest fires or highly polluted urban areas, may not be necessary to perturb the microphysical processes substantially. Further, Fig. 3.11 demonstrates that after the CCN concentration exceeds 3000 cm-3, additional increases in CCN concentration have progressively less impact on the microphysical processes. An important caveat to this result, however, is that the Morrison microphysics scheme does not include the effects of wet scavenging on the CCN concentration. If the 63 number of CCN available to perturb the microphysical processes is limited by wet scavenging, then larger

CCN concentrations might be needed to achieve the maximum perturbation in those processes.

Fig. 3.9: Domain-averaged vertical profiles of the rate of cloud droplet collection by rain (green lines) and rain evaporation rate (blue lines) at t = 80 min (thin lines) and t = 120 min (thick lines) for a) def, b) hiRH, c) loRH, and d) hiWS soundings. Solid (dashed) lines represent profiles from the cleanest (dirtiest) simulation.

64

Fig. 3.10: As in Fig. 3.9, but for the rate of riming hailstones with cloud droplets (green lines), rate of riming hailstones with rain (blue lines), and the melting rate of hail (purple lines).

65

Fig. 3.11: Vertically-integrated, horizontally averaged microphysical process rates versus CCN concentration at t = 120 min for a) def, b) hiRH, c) loRH, and d) hiWS soundings.

66 3.3.2 CCN EFFECTS ON COLD POOL SIZE AND STRENGTH

Since the size and strength of the low-level cold pool is determined by the amount of latent cooling caused by evaporating rain and melting hail (e.g., Gilmore and Wicker 1998; Tao et al. 2007;

James and Markowski 2010), it is reasonable to expect that changes in the rates of these processes will influence the cold pool. Here, we define the cold pool to be the area at the lowest model level (z = 170 m) that has a perturbation potential temperature colder than -2 K, consistent with recent studies by Morrison and Milbrandt (2011) and Morrison (2012). Figure 3.12 depicts the variation in the size and the mean perturbation potential temperature (θ’) of the cold pool with CCN concentration. It is evident from this figure that the relationship between the cold pool characteristics and the CCN concentration is not monotonic, in contrast to what has been reported or suggested in several other studies of supercell thunderstorms (e.g., Lerach et al. 2008; Storer et al. 2010; Lerach and Cotton 2012). In addition, the response of the cold pool to CCN concentration is highly dependent on the initial sounding used: In def

(Fig. 3.12a) and hiRH (Fig. 3.12b), the cold pool size at t = 120 min remains nearly constant at 6400 km2 regardless of the CCN concentration, while the mean cold pool temperature increases by ~0.4 K from

CCN = 100 cm-3 to CCN = 4000 cm-3 before remaining nearly constant at larger CCN concentrations. In contrast, loRH (Fig. 3.12c) exhibits a rapid decrease in cold pool size at t = 120 min, from 1200 km2

(CCN = 100 cm-3) to 400 km2 (CCN = 3000 cm-3), followed by a much more gradual decrease from 400 km2 to 250 km2 between CCN = 3000 cm-3 and CCN = 10 000 cm-3. The response of the cold pool in hiWS (Fig. 3.12d) is similar to that of loRH, with a rapid decrease in cold pool size from 7700 km2 (CCN

= 100 cm-3) to 6000 km2 (CCN = 3000 cm-3), followed by nearly constant size between CCN = 3000 cm-3 and CCN = 10 000 cm-3. These trends in the cold pool response to CCN concentration are summarized in

Tab. 3.2.

To determine why the cold pool response to increasing CCN concentration differs so greatly between def/hiRH (little change in size) and loRH/hiWS (rapid decrease in cold pool size between CCN =

100 cm-3 and CCN = 3000 cm-3), we examine the ice mass budget (Fig. 4.10). In def/hiRH, the rate of melting hail from 0-2 km in the dirtiest run actually exceeds that of the cleanest run, consistent with larger 67

Fig. 3.12: Total area (solid lines) and mean perturbation potential temperature (dashed lines) of the cold pool at the lowest model level (z = 170 m) at t = 100 min (blue lines) and t = 120 min (red lines) versus CCN concentration for a) def, b) hiRH, c) loRH, and d) hiWS soundings.

68 hail number concentrations (Fig. 3.7a-b) and more hail mass (Fig. 3.8a-b) in this layer in the dirtiest run.

The increased latent cooling in def/hiRH from melting hail offsets the reduced latent cooling due to less rain evaporation (Fig. 3.9a-b), resulting in little change in the size of the cold pool as CCN concentration increases (Fig. 3.12a-b). These trends are not evident in loRH/hiWS, which have much smaller melting rates from 0-2 km in the dirtiest run, approximately equal hail number concentration and mass, and cold pools that shrink in size between CCN = 100 cm-3 and CCN = 3000 cm-3.

3.3.3 CCN EFFECTS ON SURFACE PRECIPITATION

Similar to that of the cold pool characteristics, the response of the domain-averaged precipitation to increases in CCN concentration (Fig. 3.13) is dependent on environmental conditions. For def (Fig.

3.13a), the domain-averaged precipitation increases by ~0.1 mm as the CCN concentration is increased from CCN = 100 cm-3 to CCN = 1000 cm-3, rises much more slowly from CCN = 1000 cm-3 to its peak at

CCN = 5000 cm-3, and then slowly declines by ~0.03 mm between CCN = 5000 cm-3 and CCN = 10 000 cm-3. The pattern for hiRH (Fig. 3.13b) is similar, but the initial increase in precipitation between CCN =

100 cm-3 and CCN = 1000 cm-3 is larger than in def (0.2 mm versus 0.1 mm), and the precipitation reaches a peak value between CCN = 2000 cm-3 and CCN = 4000 cm-3 and then remains roughly constant through CCN = 10 000 cm-3, rather than slowly decreasing as in def. In contrast, increasing CCN concentration causes domain-averaged precipitation to decline nearly monotonically in loRH (Fig. 3.13c), with the most rapid decrease in precipitation from CCN = 100 cm-3 to CCN = 3000 cm-3. In addition, while the absolute change in domain-averaged precipitation in loRH is similar to that of the other soundings (~0.1 mm at t = 120 min), the relative change is much larger in loRH (50% vs. ~10%; Tab. 3.2) due to the small total precipitation (~0.16 mm vs. 1.3-2.0 mm). Our finding that supercell thunderstorm precipitation increases (in a domain-averaged sense) for relatively humid low-level conditions (def/hiRH), but decreases for relatively dry low-level conditions (loRH) supports the results of Khain et al. (2008), who found that high (low) relative humidity causes net total condensate gain (loss) due to enhanced CCN concentrations across a variety of convective modes. Nevertheless, our results show that the largest

69

Fig. 3.13: Domain-averaged, accumulated surface precipitation at t = 90 min (blue line), t = 100 min (green line), t = 110 min (yellow line), and t = 120 min (red line) versus CCN concentration for a) def, b) hiRH, c) loRH, and d) hiWS soundings.

70 changes in precipitation for each of the four soundings are achieved between CCN = 100 cm-3 and CCN =

3000 cm-3 (Fig. 3.13), with much smaller changes at CCN concentrations larger than 3000 cm-3. This pattern mirrors that of the vertically-integrated microphysical process rates shown in Fig. 3.11, which also change much more gradually once the CCN concentration exceeds 3000 cm-3. Similar trends were also found in Khain et al. (2011), in which precipitation in a multicell storm exhibited little sensitivity to additional increases in CCN concentration above 3000 cm-3.

The spatial distribution of the accumulated precipitation (Fig. 3.14) reveals a few important differences between the cleanest and dirtiest runs for each sounding. First, the most polluted runs of def

(Fig. 3.14a) and hiRH (Fig. 3.14b) have up to 25 mm more precipitation along (and to the immediate left) of the tracks of both the left- and right-moving updrafts. This trend is also partly reflected in hiWS, although the enhancement in precipitation along the updraft tracks in the most polluted run is up to 18 mm and is less spatially uniform than in def and hiRH. Note that the precipitation reduction along the track of the right-moving hiRH storm from t = 40-60 min is due to secondary convective development that follows the main storm. The pattern is completely different in loRH, however, with decreases in accumulated precipitation of up to 18 mm along and to the right of the updraft tracks in the most polluted case. To evaluate whether these changes in precipitation are merely due to shifts in the updraft tracks, purple (black) contours of updraft velocity in the cleanest (dirtiest) simulations are included in Fig. 3.14 to indicate the approximate updraft tracks. In def, hiRH, and hiWS, the left-moving updraft moves farther to the left in the most polluted run, especially after t = 60 min in hiRH and t = 80 min in def and hiWS.

This shift in the track of the left-moving updraft is therefore at least partly responsible for the precipitation enhancement along and to the left of the left-moving updraft in the dirtiest runs of def, hiRH, and hiWS, and demonstrates that the CCN concentration can indirectly influence the path of the supercell thunderstorm, likely by changing the characteristics of the low-level cold pool. However, the enhanced precipitation along and to the left of the right-moving updraft cannot be attributed to a track shift, as the path of the right-moving updraft either remains the same (hiWS) or shifts slightly to the right (def/hiRH) in the most polluted runs. It is possible that the 30% (rain) and 3% (hail) increases in the mass mean 71

Fig. 3.14: Difference in accumulated surface precipitation between the dirtiest (CCN = 10 000 cm-3) and cleanest (CCN = 100 cm-3) simulations at t = 120 min (color fill) for a) def, b) hiRH, c) loRH, and d) hiWS soundings. The purple and black contours indicate the maximum updraft speeds that were simulated at z = 5 km for the duration of the cleanest and dirtiest simulations, respectively. These contours range from 10 m s-1 to 30 m s-1 at an interval of 10 m s-1. The approximate locations of the main left- and right-moving updrafts at several times during the simulations are also indicated.

72 diameter between the cleanest and dirtiest runs explain this shift in precipitation associated with the right- moving updraft. The larger rain and hail particles in the dirtiest run are not advected as far from the updraft because these heavier particles have faster fall speeds and are not carried aloft where strong horizontal winds are present, according to size sorting theory (Browning and Donaldson 1963; Browning

1964; Hall et al. 1984; Ryzhkov et al. 2005; Tessendorf et al. 2005; Kumjian and Ryzhkov 2012).

3.3.4 COMPARISON TO SIMULATIONS WITH RAIN μ SET TO ZERO

The results presented above are from simulations in which the shape parameter μ of the raindrop size distribution (see Eq. 3.2) is allowed to vary (we refer to these simulations as varying μ or VMU hereafter). However, to our knowledge, the sensitivity of CCN effects to the shape parameter for rain has not yet been explored in the literature. To determine how the choice of the shape parameter affects the results, we now compare results from VMU to simulations in which μ for rain is set to zero (zero μ or

ZMU) and the default sounding (Fig. 3.2) is used.

Figure 3.15 shows the relationship between CCN concentration and the vertically-integrated microphysical process rates (Fig. 3.15a), cold pool characteristics (Fig. 3.15b), domain-averaged surface precipitation (Fig. 3.15c), and the spatial distribution of precipitation (Fig. 3.15d). Tab. 3.2 summarizes the changes in these quantities for ZMU between the cleanest and dirtiest runs. The vertically integrated changes in the microphysical process rates in VMU (Fig. 3.11a) and ZMU (Fig. 3.15a) are similar (Tab.

3.2). However, while the cold pool area changes little with CCN concentration in VMU (Fig. 3.12a), the area decreases by 13% from the cleanest to the dirtiest run in ZMU (Fig. 3.15b; Tab. 3.2). There are also differences in the precipitation response to the CCN concentration. Domain-averaged precipitation does not peak until CCN = 5000 cm-3 in VMU (Fig. 3.13a), but peaks at CCN = 250 cm-3 and declines slowly thereafter in ZMU (Fig. 3.15c). The precipitation enhancement near and to the left of the updraft tracks in

ZMU is also less apparent (Fig. 3.15d vs. Fig. 3.14a).

When the liquid and ice mass budgets for VMU (Fig. 3.9a and Fig. 3.10a) and ZMU (Fig. 3.16a and Fig. 3.16b) are compared, it is clear that the cold pool is more sensitive to CCN concentration in

73

Fig. 3.15: As in a) Fig. 3.11, b) Fig. 3.12, c) Fig. 3.13, and d) Fig. 3.14, but for simulations with the default sounding and the shape parameter μ in the raindrop size distribution set to zero.

74 ZMU because a larger reduction in the rain evaporation rate occurs in ZMU below 1 km as the CCN concentration increases (Fig. 3.16a). In addition, the larger hail melting rates in the z = 0-2 km layer in the dirtiest run of VMU (Fig. 3.10a) are not present in ZMU (Fig. 3.16b). Taken together, these two changes indicate that low-level latent cooling decreases more with CCN concentration in ZMU, which explains the greater reduction in cold pool size in those runs. It is likely that the differences in the evaporation and melting rates in VMU and ZMU are related to differences in the amount of raindrop size sorting in the two sets of simulations. Microphysics schemes that assume a value of μ = 0 for rain are known to produce more aggressive raindrop size sorting, with larger (smaller) values of raindrop mean mass diameter near the surface (aloft), than schemes that have μ > 0 (Milbrandt and Yau 2005a). Our results confirm this trend. At t = 120 min in the CCN = 100 cm-3 run of ZMU, the raindrop mean mass diameter (not shown) is 14% larger near the surface and 48% smaller at z = 6 km than in VMU.

Fig. 3.16: As in a) Fig. 3.9 and b) Fig. 3.10 at t = 120 min, but for simulations with the default sounding and the shape parameter μ in the raindrop size distribution set to zero.

75 3.4 SUMMARY AND CONCLUSIONS

In this study, the Weather Research and Forecasting (WRF) model was run at a horizontal grid spacing of

1 km to investigate how cloud condensation nuclei (CCN) concentration affects the microphysical process rates, particle diameters, cold pool size and strength, and accumulated precipitation in idealized supercell thunderstorms. The Morrison microphysics scheme was modified to include explicit prediction of cloud droplet concentration and a variable shape parameter for the raindrop size distribution. The results were examined across 15 different CCN concentrations between 100 cm-3 and 10 000 cm-3 and for four environmental soundings with different values of low-level relative humidity and vertical wind shear.

While a few other studies (Seifert and Beheng 2006; Fan et al. 2009; Storer et al. 2010; Lebo and Seinfeld

2011; Lerach and Cotton 2012) have tested the sensitivity of thunderstorms in sheared environments to

CCN and relative humidity effects, our study has shown how individual microphysical process rates vary across a spectrum of CCN concentrations, up to a larger value (CCN = 10 000 cm-3) than these previous studies. In addition, we have compared results from simulations with a fixed versus variable shape parameter for the raindrop size distribution, which can qualitatively change some results. Relative to the cleanest simulation (CCN = 100 cm-3), the dirtiest simulation (CCN = 10 000 cm-3) of each sounding features larger cloud droplet number concentrations (~4000 cm-3 vs. ~100 cm-3; Fig. 3.7) and smaller concentrations of rain (e.g., 10-4 vs. 10-2 cm-3 at z = 5 km) and hail (e.g., 10-3 cm-3 vs. 10-2 cm-3 at z = 10 km) from 2-10 km. Fewer cloud droplets undergo autoconversion in the dirtiest simulation because the cloud droplets are smaller (5 μm vs. 17 μm; Fig. 3.6) due to increased competition for liquid water.

However, the rain and hail particles in the dirtiest runs are up to 30% and 3% larger near the surface (Fig.

3.6; Tab. 3.2), respectively, than in the cleanest runs as they collect the enhanced bulk cloud droplet mass. Due to the reduced concentration and larger mean size of rain and hail particles in the most polluted runs, the mass budget analyses (Figs. 3.9-3.10) and Tab. 3.2 reveal reduced rates of evaporation for all soundings, reduced rates of melting at all heights in loRH/hiWS, and reduced rates of melting between

2.25-4.25 km in def/hiRH with increased melting below 2.25 km due to larger hail number concentrations in that layer (Fig. 3.7a-b). Collection of cloud droplets by rain and riming of hail with rain were reduced 76 in the dirtiest run of each sounding (Tab. 3.2) due to smaller rain and hail number concentrations. When all of the different CCN concentrations were examined (100-10 000 cm-3), it was found that the perturbation in these microphysical process rates, particularly with respect to evaporation, melting, and riming of hail with rain, saturates by CCN ≈ 3000 cm-3 (Fig. 3.11), independent of the initial sounding.

This result indicates that extreme CCN concentrations of 5000-10 000 cm-3 are not necessary to alter the microphysical processes of supercell thunderstorms substantially.

The response of the area and temperature of the low-level cold pool to CCN concentration is found to be non-monotonic and greatly dependent on the environmental conditions (Fig. 3.12). For both def and hiRH, the cold pool size shows almost no dependence on CCN concentration due to compensating changes in latent cooling from melting hail versus evaporating rain. Changes in the mean cold pool temperature are also small, with the mean perturbation potential temperature initially increasing by ~0.5 K, reaching a peak at CCN ≈ 3000 cm-3, and then decreasing by ~0.15 K at larger CCN concentrations. The cold pool response to CCN concentration is limited in def and hiRH because while low-level evaporative cooling decreases with CCN concentration in these simulations, low-level latent cooling from melting hail actually increases by up to ~50% (Fig. 3.10) due to larger near-surface hail mixing ratios (Fig. 3.8) in the dirtiest runs. In contrast, cold pool size decreases dramatically in loRH from 1200 km2 (for CCN = 100 cm-3) to 200 km2 (for CCN = 10 000 cm-3), while the mean temperature increases by 0.45 K. The trends in the cold pool characteristics of hiWS resemble those of loRH.

However, the cold pool size (temperature) does not decrease (increase) after CCN = 3000 cm-3 in hiWS, whereas the rates of areal decrease and warming merely slow in loRH after CCN = 3000 cm-3.

When the relationship between domain-averaged precipitation and CCN concentration was examined (Fig. 3.13; Tab. 3.2), it was found that larger CCN concentrations produced more precipitation in def, hiRH, and hiWS, up to a peak accumulation at CCN = 5000 cm-3 in def, CCN = 4000 cm-3 in hiRH, and CCN = 500 cm-3 in hiWS. At larger CCN concentrations, accumulations either remained the same (hiRH) or decreased slightly (def/hiWS). In contrast, domain-averaged precipitation in loRH decreased almost monotonically from 0.16 mm (CCN = 100 cm-3) to 0.08 mm (CCN = 10 000 cm-3). 77 Maps of the precipitation difference between the cleanest and dirtiest simulations for each sounding (Fig.

3.14) demonstrate that much of the precipitation enhancement at large CCN concentrations in def, hiRH, and hiWS occurs near and to the left of the tracks of the left- and right-moving updrafts.

Lastly, results from simulations with the default sounding and a variable shape parameter μ of the raindrop size distribution were compared to simulations in which μ was set to zero. The size of the cold pool decreased by ~13% as the CCN concentration increased in the μ = 0 runs (Fig. 3.15b; Tab. 3.2), while little change occurred when μ was allowed to vary (Fig. 3.12a). Domain-averaged precipitation also slowly decreased with CCN concentration when μ = 0 (Fig. 3.15c), despite having increased between

CCN = 100 cm-3 and CCN = 5000 cm-3 in the variable μ runs (Fig. 3.13a). The differences between the two sets of simulations were caused by a larger reduction in low-level latent cooling as CCN concentration increased in the μ = 0 simulations (Fig. 3.16).

The results herein highlight the complex interactions between microphysical processes, precipitation, and thermodynamics in supercell thunderstorms and the sensitivity that these processes display towards pollutant concentration. While changes in the individual microphysical process rates may be fairly large and monotonic, the impacts on the cold pool characteristics and the accumulated precipitation are generally smaller (in a relative sense) and non-monotonic due to compensating changes in the microphysical processes. Low-precipitation supercell thunderstorms, however, may be an exception to this statement. Here, an 84% reduction in the cold pool area and a 50% decrease in the domain- averaged precipitation occurred in polluted conditions with dry low-level relative humidity (Tab. 3.2).

This result indicates that the response of supercell thunderstorms to CCN concentration is highly dependent on the environmental conditions, even in an idealized modeling framework in which the secondary feedback between the initial conditions and physical processes such as radiative transfer and surface fluxes are neglected. Since differences in the relative humidity and vertical wind shear can change the cold pool and precipitation responses to CCN concentration, future studies that examine observational evidence to validate the trends seen in numerical models will likely need to stratify results by

78 environmental conditions. While unexplored in this paper, it also should be noted that aerosols might further perturb convective processes in their role as ice nuclei, a topic that requires additional research.

3.5 ACKNOWLEDGEMENTS

We thank Dr. David Dowell (NOAA) for his helpful feedback. We also thank Dr. Amy Solomon

(NOAA) for modifying the public release of the Morrison microphysics scheme in version 3.1 of the

WRF model to include explicit prediction of cloud droplet concentration. Feedback from three anonymous reviewers substantially improved this manuscript. This material is based upon work supported by the National Science Foundation, in part by a Graduate Research Fellowship (DGE-1144083) and in part by NSF ATM 0910424. Hugh Morrison was partially supported by U.S. DOE ASR DE-SC0008648.

We would like to acknowledge high-performance computing support from Yellowstone

(ark:/85065/d7wd3xhc) provided by NCAR's Computational and Information Systems Laboratory, sponsored by the National Science Foundation. Figures were made with the National Center for

Atmospheric Research (NCAR) Command Language (NCL), version 6.0.0 (Brown et al. 2012). Any opinions, findings, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the views of the National Science Foundation.

79 4 AN OVERVIEW OF COLORADO PLOWABLE HAILSTORMS: SYNOPTIC WEATHER, DUAL-POLARIZATION RADAR, AND LIGHTNING DATA

This chapter is in preparation:

Kalina, E. A., K. Friedrich, B. C. Motta, W. Deierling, G. T. Stano, and N. N. Rydell, 2015: An overview of Colorado plowable hailstorms: Synoptic weather, dual-polarization radar, and lightning data. Wea. Forecasting, in preparation.

4.0 ABSTRACT

Synoptic weather, S-band dual-polarization radar, and total lightning observations are analyzed from four thunderstorms that caused “plowable” hail accumulations of 15-60 cm in localized areas of the Colorado

Front Range. Results indicate that moist, relatively slow (5-15 m s-1) southwesterly to westerly flow at

500 hPa and postfrontal low-level upslope flow, with 2-m dewpoint temperatures of 11-19 °C at 1200

LST, were present on each plowable hail day. This pattern resulted in column-integrated precipitable water values that were 142-212% of the monthly climatological means and freezing level heights that were, counterintuitively, 200-700 m higher than average.

Radar data indicate that 1-3 maxima in reflectivity (Z; 68-75 dBZ) and 50-dBZ echo top height

(11-15 km MSL) occurred over the lifetime of each hailstorm. These maxima, which imply an enhancement in updraft strength, resulted in increased graupel and hail production and accumulating hail at the surface within 30 min of the highest echo tops. The hail core had Z~70 dBZ, differential reflectivity

(ZDR) of 0 to -4 dB, and correlation coefficient (ρHV) of 0.90-0.95. Time-height plots reveal that these minima in ZDR and ρHV gradually descended to the surface after originating at heights of 6-10 km MSL

~15-30 min prior to accumulating hailfall. Hail accumulations estimated from the radar data pinpoint the times and locations of plowable hail, with depths greater than 5 cm collocated with the plowable hail

80 reports. Each plowable hail event was accompanied by lightning flash rates that were near the maximum observed thus far within the thunderstorm.

4.1 INTRODUCTION

Thunderstorms that result in deep hail accumulations pose a substantial risk to life and property.

Numerous such hailstorms have previously resulted in motor vehicle accidents, road closures, airport delays, urban and river flooding, and water rescues (Chappell and Rodgers 1988; Grahame et al. 2009;

Schlatter and Doesken 2010). Damage from one hailstorm, which caused 25 cm of hail accumulation in a small town in southwestern England, was estimated to cost 1M British Pounds (~1.8M USD in 2015 dollars; Grahame et al. 2009). A number of similar events have occurred in and near the Denver, CO metro area (Tab. 4.1; Knight et al. 2008; Schlatter et al. 2008; Schlatter and Doesken 2010), impacting thousands of people. Following these hailstorms, some roads remained impassable until snowplows and bulldozers were used to clear them (Fig. 4.1), leading these events to be referred to as “plowable” hailstorms. Despite the extreme nature of these storms, some of the events, such as the 9 September 2013 hailstorm in Lakewood, CO (Tab. 4.1), did not even merit a severe thunderstorm warning, since the maximum hailstone diameter (~13 mm) was much smaller than the warning criteria of 25.4 mm.

Examples of similar events exist in the literature, and were reported to consist either of small hail (d < 10 mm; Grahame et al. 2009) or of a mixture of low-density small and large hailstones (Knight et al. 2008;

Schlatter et al. 2008). However, based on public reports from the Community Collaborative Rain, Hail, and Snow (CoCoRaHS) network1 and the Storm Events Database2, two of the events in Tab. 4.1 were accompanied by large hail of up to 44.5 mm that did considerable damage to structures (3 August 2013 and 21 May 2014), so not all deep hail accumulations consist entirely of small or low density hailstones.

1 http://www.cocorahs.org/

2 http://www.ncdc.noaa.gov/stormevents/

81 In addition, severe wind gusts that exceeded 25 m s-1 and tornadoes accompanied the 3 August and 21

May hailstorms (Tab. 4.1).

Tab. 4.1: Characteristics of Colorado plowable hailstorms in 2013-2014 derived from the Community Collaborative Rain, Hail, and Snow (CoCoRaHS) network and NOAA’s Storm Events Database. Hail times and locations correspond to the plowable hail reports, and severe weather (other than large hail) includes any tornadoes or wind gusts greater than 25 m s-1.

Analysis times Hail time (UTC) Max hail diameter (mm) Severe weather (UTC) Location 3 August 2013 2216 44.5 3 EF0 tornadoes 1842-2356 Windsor 25.7 m s-1 wind gust 22-23 August 2013 2339 44.5 None 2138-0021 Ken Caryl 9 September 2013 2100 12.7 None 1829-2258 Lakewood 21 May 2014 2030 25.4 5 EF0 tornadoes 1725-2247 Green Valley Ranch 30.9 m s-1 wind gust

Fig. 4.1: Hail being plowed in Lakewood, CO after the 9 Sept 2013 hailstorm. Reprinted with permission from http://www.thedenverchannel.com/news/hail-rain-pours-in-lakewood-wheat-ridge. Photo credit: 7NEWS Reporter Marshall Zelinger. 82 While plowable hailstorms are not always associated with severe weather and do not occur as frequently as other hailstorms, forecasters are nevertheless required to predict them accurately due to the considerable threats they pose to people, transportation, and infrastructure. However, knowledge of synoptic weather conditions and operational radar features associated with thunderstorms that produce deep hail accumulations is limited. The only two case studies of such storms in the refereed literature consist of single-polarization radar data (Knight et al. 2008; Schlatter et al. 2008; Grahame et al. 2009).

Therefore, dual-polarization radar characteristics of plowable hailstorms, available to forecasters since the

2012 upgrade to the Weather Surveillance Radar – 1988 Doppler (WSR-88D) network, are unexplored.

What are the typical radar features or products from the S-band operational radar network that forecasters can use in real-time to infer that plowable hail is imminent or occurring? Further, are there other state-of- the-art instruments, such as three-dimensional total lightning detection networks, that can provide additional insight into the microphysical processes that contribute to plowable hail? To address these questions, this research examines the synoptic weather conditions and the radar and lightning characteristics of four plowable hailstorms that occurred along the Colorado Front Range between August

2013 and May 2014. To our knowledge, this is the first study to present such an analysis on plowable hailstorms.

The S-band dual-polarization radar characteristics of severe thunderstorms with large hail have been well documented. Since radar reflectivity (Z) is proportional to the sixth power of the particle diameter, reflectivity is often used to identify hailstorms. Typically, hailstorms have Z > 60 dBZ at S

Band (Kumjian and Ryzhkov 2008; Snyder et al. 2010). Storms with giant hail (d > 50.8 mm) have Z in excess of 65-70 dBZ (Ryzhkov et al. 2010). For a given hailstone size, Z is larger for hailstones with greater fractional water content, since the liquid water coating that develops on hailstones undergoing wet growth is highly reflective (Snyder et al. 2010). Therefore, wet hail and giant hail may be associated with similar Z values, requiring the use of differential reflectivity (ZDR) to distinguish between the two. ZDR is the logarithmic ratio of the reflectivity in the horizontal to the reflectivity in the vertical. Giant hail typically has ZDR < -0.5 dB (Ryzhkov et al. 2010), large hail (25.4 mm < d < 50.8 mm) has ZDR near 0 dB 83 (Balakrishnan and Zrnic 1990; Kumjian and Ryzhkov 2008; Snyder et al. 2010; Kennedy et al. 2014) due to the tumbling nature of hailstones (Lesins and List 1986; Herzegh and Jameson 1992), and small (d <

25.4 mm), wet hail has ZDR > 0 dB, sometimes exceeding 4 dB (Ryzhkov et al. 2013a), due to the torus of liquid water that envelopes the melting hailstone (Rasmussen and Heymsfield 1987a). A third radar variable, the copolar cross-correlation coefficient (ρHV), also responds strongly to hail. ρHV ranges from zero to one and quantifies the degree of similarity in the shape and orientation of particles within the radar volume. In rain, ρHV normally exceeds 0.97, but in hail, ρHV can range from 0.8 to 0.95 due to the diversity of shapes and orientations typical of hailstones (Ryzhkov et al. 2013b). The largest reductions in ρHV occur when large hail is mixed with rain in the radar volume (Balakrishnan and Zrnic 1990).

In addition to the changes in the radar variables that indicate hail is occurring, a number of studies have also shown that in some thunderstorms, increases in lightning flash rate precede severe weather events, including hailfall, by 5-20 minutes (e.g., Williams et al. 1999; Goodman et al. 2005; Wiens et al.

2005; Schultz et al. 2009; Darden et al. 2010; Rudlosky and Fuelberg 2013). Lightning flash rate has been found to be strongly correlated with updraft strength, updraft volume, and graupel mass (Wiens 2005;

Wiens et al. 2005; Tessendorf et al. 2007; Deierling and Petersen 2008; Deierling et al. 2008). Thus, lightning data can help forecasters to assess thunderstorm intensity and to determine whether a storm is in the developing, mature, or weakening phases of its lifecycle (Darden et al. 2010; Rudlosky and Fuelberg

2013). The lightning characteristics of plowable hailstorms, however, have yet to be investigated. Do increases in lightning flash rate precede the occurrence of accumulating hail, even in cases when the hailstones are too small to be classified as severe? This study will examine three-dimensional total lightning data from the four previously mentioned plowable hailstorms to determine if total lightning information can aid forecasters in predicting similar events in the future.

An overview of the four hailstorms analyzed in this research is provided in section 4.2, which also includes a discussion of the radar and lightning data and the analysis methods. An examination of the synoptic weather, dual-polarization radar, and total lightning characteristics of the four cases is presented in section 4.3. section 4.4 summarizes the results and provides concluding remarks. 84

4.2 DATA AND METHODS

4.2.1 OVERVIEW OF CASES

The hailstorms discussed in this research all occurred in August and September 2013 and May

2014 along the Colorado Front Range (Fig. 4.2) and produced hail accumulations of at least 15 cm within

30 min. The location and time of the plowable hail reports, in addition to details about the characteristics of these storms, are provided in Tab. 4.1. With the exception of the 9 September case, these hailstorms were considered severe thunderstorms, as two of the storms produced severe wind gusts and multiple EF0 tornadoes (3 August and 21 May) and all but the 9 September case produced severe hail (Tab. 4.1). The maximum diameter of the hailstones in each case ranged from 12.7 mm to 44.5 mm during the hail accumulation. The location, time, and maximum diameter of the hailstones are based on data from the

CoCoRaHS network and the Storm Events Database. Note that this study focuses on four examples of plowable hailstorms that occurred in 2013 and 2014, since three-dimensional lightning data were not readily available prior to 2013. From 2013-2014, we are aware of a total of nine plowable hailstorms that occurred in Colorado.

4.2.2 RADAR DATA AND OPERATIONAL SOUNDINGS

For each of the thunderstorms in Tab. 4.1, dual-polarization radar data were obtained from the

Weather Surveillance Radar – 1988 Doppler (WSR-88D) located at Front Range Airport (KFTG; 1.68 km

MSL) in Colorado. The radar was operated in Velocity Coverage Pattern (VCP) 212, and scanned 14 elevation angles from 0.5° to 19.5°. In all cases, Z > 0 dBZ was first observed to the west of KFTG at a distance of 111-152 km from the radar site (Fig. 4.2) and then Z gradually increased as the storms approached the radar. The minimum distances between the storms and the radar ranged from 9-44 km. At the time of the plowable hail reports, the distances from the radar ranged from 18 km (21 May) to 78 km

(3 August), which caused the height of the lowest radar beam (0.5° elevation angle) to range from 0.2 to

1.3 km AGL (Fig. 4.2b). The data analysis period (Tab. 4.1) for each storm began when Z > 0 dBZ in the 85 eventual hailstorm first appeared in the radar volume. Analysis continued until the convective core of the hailstorm (defined here as Z > 30 dBZ) merged with other convective cores and became indistinguishable in subsequent radar data. This occurred as early (long) as 39 (138) min after the time that plowable hail was reported.

All radar volumes during the periods of analysis were manually edited with the Solo II radar software3 from the National Center for Atmospheric Research (NCAR) to remove echoes unrelated to the plowable hailstorms, including echoes from non-meteorological targets such as ground clutter and precipitation in the vicinity of the hailstorms but unrelated to them. Fields of Z, Doppler velocity (Vr), spectrum width (W), and ZDR were considered for this purpose. Each elevation angle scan in the radar volume was examined individually, since thunderstorm echoes were often tilted with height. Following the results in Giuli et al. (1991) and Park et al. (2009), radar gates that contained ground clutter were

-1 -1 visually identified by nearly constant Z over time, Vr near 0 m s , and W > 8 m s . The latter criterion was used to identify the boundary between radar gates that contained pure weather echoes and those that contained a mixture of weather and clutter echoes. Other non-meteorological echoes, which consisted mainly of biological scatterers, were visually identified by radar gates that had Z < 25 dBZ and spatially inhomogeneous ZDR > 4 dB (Park et al. 2009). When showers and thunderstorms other than the hailstorm occurred in the radar volume, these echoes were removed unless the convective core (Z > 30 dBZ) of the shower or thunderstorm merged with the convective core of the hailstorm at the lowest elevation angle

(0.5°). Areas of precipitation consisting entirely of Z < 30 dBZ that were not contiguous with the hailstorm at 0.5° elevation angle were removed.

3 https://www.eol.ucar.edu/software/solo-ii

86

Fig. 4.2: Maps showing the locations of hail reports (diamonds), the KFTG radar (cross), COLMA stations (squares), the center of COLMA (plus sign), and the approximate storm tracks (lines) relative to a) the elevation of the topography (km MSL) and b) the height of the center of the lowest radar beam (km AGL). Dashed lines indicate areas of beam blockage along the storm tracks. The numbers indicate a) the start and end times (UTC) of the analysis periods for each case and b) the distances (km) from the plowable hail reports to the KFTG radar (cross) and to the COLMA center (plus sign), respectively. 87 After editing the radar data, NCAR’s Radx C++ software package4 was used to calculate the specific differential phase (KDP) from the total differential phase (ΦDP) measured by the radar. To calculate KDP, a finite impulse response (FIR) filter with a length of 10 range gates was iteratively applied to ΦDP four times to smooth it. KDP was then calculated from the smoothed ΦDP over nine range gates, centered on the gate of interest. Next, the NCAR particle identification scheme (PID; Vivekanandan et al.

1999) was applied to the data. The PID is a fuzzy logic algorithm that uses trapezoidal membership functions for seven input variables and 14 particle classes to estimate the most dominant contributor to the radar signal in a given range gate. The PID uses Z, ZDR, KDP, ρHV, standard deviation of ZDR, standard deviation of ΦDP, and air temperature as input variables, with the standard deviations calculated over nine range gates. Air temperature data were obtained from the 0000 UTC atmospheric soundings at Denver during the evenings of the plowable hail cases, except for the 21 May case, when an 1800 UTC sounding was available. For each of the aforementioned input variables, the PID assigns a value between zero and one to each range gate for each of the following particle classes: cloud droplets, drizzle, light rain, moderate rain, heavy rain, rain-hail mix, hail, graupel-small hail mix, graupel-rain mix, dry snow, wet snow, ice crystals, irregular ice crystals, and supercooled liquid droplets. The seven values belonging to a given particle class are then summed, and the class with the largest sum is assigned to the radar gate.

After the PID was assigned, the Radx software package was used to regrid the polar coordinate radar data to a Cartesian coordinate system using an 8-point linear interpolation scheme. The azimuthal equidistant map projection was selected for the Cartesian grid, which spanned 400 km by 400 km in the horizontal and 15 km MSL in the vertical, with the KFTG radar located at the grid center. Each grid cell had horizontal and vertical dimensions of 0.5 km. To interpolate the radar variables to a given grid cell, a minimum of five valid data points (out of a possible eight) needed to be present. Since the PID field is a

4 http://www.ral.ucar.edu/projects/titan/docs/radial_formats/radx.html

88 discrete field, it was not interpolated and was instead assigned to each grid cell using the nearest neighbor approximation.

Graupel (Eq. 4.1) and hail (Eq. 4.2) mass concentrations were then estimated from the radar reflectivity using the relations from Heymsfield and Miller (1988):

0.5 M g  0.0052Z (4.1)

0.71 M h  0.000044Z (4.2)

-3 Here, Mg and Mh are the mass concentrations of graupel and hail, respectively, in g m , and Z is the radar reflectivity in mm6 m-3. The graupel relation was applied to all of the Cartesian radar grid cells that were classified by the NCAR PID as graupel/small hail or graupel/small hail/rain mix, while the hail relation was applied to all of the grid cells classified as hail or hail/rain mix, as in Deierling et al. (2008). Eqs. (1) and (2) were derived from in situ aircraft measurements of ice particle size spectra (0.0125 < d < 40 mm) in the updrafts of a single cell thunderstorm (Eq. 4.1) and a supercell thunderstorm (Eq. 4.2). These thunderstorms occurred in eastern Montana, an environment that is geographically and climatologically similar to eastern Colorado. These Z-M relationships have been applied to a variety of single cell, multicell, and supercell thunderstorms across the U.S. (Deierling et al. 2008). While it is acknowledged that large absolute errors in ice mass estimates from these relations likely exist due to the limited number of thunderstorms sampled and due to the inherent difficulties in making accurate airborne measurements of particle size distributions, the focus of our study is not on the absolute values of the ice masses but on the relative changes in these masses over the lifetimes of the hailstorms.

4.2.3 LIGHTNING DATA

Three-dimensional lightning data were obtained from the Colorado Lightning Mapping Array

(COLMA) in northern Colorado (Fig. 4.2; Rison et al. 2012). COLMA was installed in the spring of 2012 and is owned and operated by the New Mexico Institute of Mining and Technology. The array consists of

16 stations, each equipped with a receiving antenna that is sensitive to Very High Frequency (VHF) radiation of ~60 MHz in frequency, at which lightning discharges emit strongly. The location (x, y, z) and 89 time t of a VHF source is determined from time-of-arrival (ti) information recorded by Global Positioning

System (GPS) receivers at multiple COLMA stations (Eq. 4.3).

(x  x )2  (y  y )2  (z  z )2 t  t  i i i (4.3) i c

Above, the location of the receiving station is (xi, yi, zi) and c is the propagation speed of the VHF radiation. If ti is measured by at least four stations, the four unknowns x, y, z, and t can be determined from Eq. (4.3). The errors in the radial and vertical positions of VHF sources are proportional to (r / D)2 and (z / D)2 , respectively, where r is the radial distance from the array center to the lightning source, z is the altitude of the source, and D is the diameter of the lightning mapping array (~100 km here). COLMA is capable of detecting lightning sources up to 350 km away from the array center (Rison et al. 2012). For the hailstorms examined here, the maximum distance from the array center (Fig. 4.2b) generally occurred at the beginning of each storm’s development, but was much less than the detection limit of 350 km. At the time of the plowable hail reports, the distance of the storms from the array center ranged from 46 km to 131 km (Fig. 4.2b).

The individual VHF sources detected by COLMA were processed with the flash creation algorithm described by McCaul et al. (2005, 2009) to filter out noise sources such as airplane tracks and to combine the remaining VHF sources into lightning flashes. VHF sources were assumed to be part of the same lightning flash if they satisfied certain temporal and spatial criteria. First, the sources must have occurred within 0.3 seconds of each other to be grouped into the same flash. Next, the radial distance

2 between successive sources must not have exceededr /1000 . For example, the maximum allowable radial distance between sources at 200 km range was 40 km (McCaul et al. 2009). This criterion reflects the dependence of the error in the radial position of a source on its radial distance from the array center.

Additionally, sources were not allowed to be more than 0.05 rad (~2.9°) apart in azimuth (the maximum expected azimuth error) to be grouped into the same flash. To prevent noise sources from bridging the time and/or distance between two separate flashes, sources with arrival times that had reduced chi-square

90 goodness of fit values (described in Thomas et al. 2004; their Eq. A2) of more than 2.0 were not grouped into flashes. In addition, flashes with fewer than ten sources were eliminated from the data, as in Wiens et al. (2005) and Tessendorf et al. (2007). Figure 4 in Wiens et al. (2005) demonstrates that, compared to the

10-source threshold, the trends in lightning flash rate are unchanged when thresholds of 50 and 100 sources are used to filter the flash data.

After lightning flashes were created, the sources from each flash were gridded into a Cartesian volume that was identical to the one used for the radar data (section 4.2.2). To exclude lightning flashes from all thunderstorms other than the plowable hailstorm, the initial source of each lightning flash was checked to determine if it was located within a vertical column of the Cartesian radar data that had Z ≥ 0 dBZ somewhere within the column (after other precipitation and non-meteorological echoes were removed from the radar data). Flashes with initial sources in regions of Z < 0 dBZ were excluded. One- minute flash rates and individual flash areas were then calculated from the remaining flashes. For each lightning flash, the flash area was estimated by first counting the number of grid cells that contained at least one lightning source from the flash under consideration and then by multiplying the total count by the area of one grid cell (0.25 km2).

4.3 RESULTS AND DISCUSSION

4.3.1 METEOROLOGICAL CONDITIONS

This section examines the synoptic weather conditions that favored the occurrence of plowable hail in the four thunderstorms listed in Tab. 4.1. Figure 4.3 shows the 500 hPa height, air temperature, dewpoint temperature, and wind vectors as measured by rawinsondes at 1200 UTC on the morning of each hailstorm. On 3 August (Fig. 4.3a) and 22 August (Fig. 4.3b), a ridge axis is aligned north-south across central Colorado with lower heights to the west across Utah and Nevada. An upper-level trough and closed upper-level low are seen approaching Colorado on 9 September (Fig. 4.3c) and 21 May (Fig.

4.3d), respectively. These weather features resulted in 500 hPa winds from the southwest or west at 5-15

91 m s-1 at Denver, CO (KDEN) on each of the four days. The light to moderate southwesterly flow transported a rich plume of mid-level subtropical moisture northward from the eastern Pacific, as evidenced by 500 hPa dewpoint depressions of ≤ 7 °C (except on 3 August; Fig. 4.3a). The weak steering winds at 500 hPa, in tandem with the subtropical moisture, allowed the resultant thunderstorms to move slowly and to produce heavy precipitation rates. These factors favored accumulating hail at the surface.

In addition to the similarities in the 500 hPa pattern, the synoptic weather features near the surface were also similar for the plowable hail cases examined here. Figure 4.4 demonstrates that all of the cases occurred in low-level easterly upslope flow behind a cold front that moved through eastern

Colorado earlier in the day. The upslope flow acted to moisten the low-level air mass, resulting in dewpoint temperatures at KDEN that ranged from 11 °C (9 September; Fig. 4c) to 19 °C (3 August, Fig.

4.4a) at 18 UTC (12 LST). The combination of the low-level moisture and relatively warm near-surface air temperatures, which ranged from 20-28°C (Fig. 4.4), resulted in surface-based Convective Available

Potential Energy (CAPE) values from 1022-2568 J kg-1 at KDEN during the afternoons of the plowable hailstorms (Tab. 4.2). In addition, the low-level easterly upslope flow that gradually veered and strengthened to mid-level westerly flow of 10-25 m s-1 (Fig. 4.5) contributed to 0-6 km AGL bulk wind shear values of ~18 m s-1 during each event (Tab. 4.2). The presence of vertical wind shear may have acted to increase hail production in the thunderstorms, since wind shear has been shown to prolong the residence time of hailstones within the thunderstorm updraft (e.g., Dessens 1960, Das 1962, Longley and

Thompson 1965, Berthet et al. 2013).

Tab. 4.2: Surface-based Convective Available Potential Energy (SBCAPE), 0-6 km AGL bulk shear, and total precipitable water vapor (PWAT) derived from Denver rawinsonde soundings (Fig. 4.5) for each of the cases listed in Tab. 4.1.

Time SBCAPE (J kg-1) Bulk shear (m s-1) PWAT (mm) 4 August 2013 0 UTC 1022 18.4 28.5 23 August 2013 0 UTC 2568 17.6 32.6 10 September 2013 0 UTC 1342 18.6 30.8 21 May 2014 18 UTC 1740 18.7 15.8

92

Fig. 4.3: Observations at the 500-hPa pressure level at 1200 UTC: Air temperature (°C, red numbers), dewpoint temperature (°C, green numbers), geopotential height (dm, purple numbers), and wind barbs (knots, blue) on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. Temperature (dashed thin red lines) and height (black lines) are contoured at intervals of 2 °C and 6 dm, respectively. Dashed thick red lines denote the positions of trough axes.

93

Fig. 4.4: Surface observations at 1800 UTC: Air temperature (°F, red numbers), dewpoint temperature (°F, green numbers), mean sea level pressure (hPa, large tan numbers), mean sea level pressure change relative to three hours earlier (10×hPa, small tan numbers), and wind barbs (knots, blue) on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. Mean sea level pressure (brown lines) is contoured at intervals of 4 hPa. Frontal boundaries, trough axes, dry lines, and high- and low-pressure systems are denoted by their standard symbols at the surface.

94

Fig. 4.5: Skew-T log-P diagram with air temperature (solid lines), dewpoint temperature (dotted lines), and wind velocity (barbs) at KDEN on a) 0000 UTC 4 Aug 2013 (black), b) 0000 UTC 23 Aug 2013 (blue), c) 0000 UTC 10 Sept 2013 (green), and d) 1800 UTC 21 May 2014 (red).

95 Each of the four hailstorms occurred on days with large amounts of atmospheric moisture, based on the column-integrated precipitable water vapor (PWAT) values calculated from the rawinsonde soundings at KDEN on the mornings (1200 UTC) and evenings (0000 UTC) of the hailstorms. Figure

4.6a demonstrates that the PWAT ranged from 19-33 mm for this set of eight soundings. To put into perspective how anomalous these values were, Fig. 4.6a compares the measured PWAT to the monthly climatological mean PWAT values5, calculated from all KDEN rawinsondes from 1948-2014. Maximum

PWAT values on the four plowable hail days ranged from 142-212% of the monthly climatological means

(Fig. 4.6a). In fact, both of the soundings on 22 August and the evening sounding on 9 September had

PWAT values that were more than two standard deviations greater than average. The anomalously large atmospheric moisture values are further highlighted by the 9 September plowable hail event, which marked the beginning of the Great Colorado Flood (9-16 September 2013) that resulted from over 400 mm of rainfall in localized areas along the Colorado Front Range (Friedrich et al. 2015a; Friedrich et al.

2015b; Gochis et al. 2015). These events suggest that, at least in eastern Colorado, plowable hailstorms may be more likely to occur on days in which forecasters also expect a risk of flash flooding from slow- moving thunderstorms, since the synoptic weather conditions are similar between the two types of events.

In addition to using PWAT as an indicator of the risk for plowable hail, it also worth investigating whether anomalously low freezing level heights were present on the plowable hail days. The freezing level height was calculated from the rawinsonde soundings at KDEN on the mornings and evenings of the plowable hailstorms and compared to the 1970-2000 monthly mean freezing level heights6, as determined from the National Centers for Environmental Prediction/NCAR reanalysis of air temperature (Kalnay et al. 1996). This comparison reveals that the height of the freezing level was 200-700 m higher than average on the plowable hail days (Fig. 4.6b). While low freezing level heights are frequently associated

5 http://www.crh.noaa.gov/unr/include/pw.php?sid=dnr

6 http://www.wrcc.dri.edu/cwd/products/

96 with thunderstorms that produce large hail (e.g., Pappas 1962; Xie et al. 2010), the Clausius-Clapeyron relation suggests that the freezing level is likely to be higher than average when anomalously large atmospheric moisture values are present, as in these cases. Therefore, the freezing level height may not be useful in identifying days in which plowable hailstorms may occur in eastern Colorado.

Fig. 4.6: Bar plots of a) column-integrated precipitable water vapor and b) freezing level height from KDEN rawinsondes at 1200 UTC on the morning of the plowable hailstorm (blue) and at 0000 UTC on the evening of the plowable hailstorm (red). The monthly mean climatological values of precipitable water and freezing level height are shown in green.

97 4.3.2 RADAR ANALYSIS

In this section, the dual-polarization radar variables from the four plowable hailstorms are first examined near the time of the hail reports. Later, the temporal evolution of the vertical profiles of the radar variables is examined to determine the radar characteristics that suggest hail may be accumulating at the surface. Finally, the total hail accumulation is estimated from the radar data to determine if forecasters can use this method to pinpoint areas of deep hail accumulation, thereby distinguishing thunderstorms with large, non-accumulating hailstones from those that produce deep hail.

4.3.2.1 Near-surface radar features during hail accumulation

Figure 4.7 presents constant altitude plan position indicators (CAPPIs) of Z at the radar volume scan closest in time to the plowable hail reports. The heights of the CAPPIs, which range from 2.5-3.5 km

MSL, represent the lowest heights at which radar data were available. A prominent hook echo is evident

15 km west of the radar in the 21 May supercell thunderstorm (Fig. 4.7d), which produced five separate tornadoes during its lifetime. All of the hailstorms had maximum Z values that exceeded 65 dBZ, and the storms on 3 August (Fig. 4.7a) and 9 September (Fig. 4.7c) had Z > 70 dBZ. As discussed in section 4.1,

Z > 65 dBZ can either indicate the presence of giant hail (d > 50 mm) or the presence of small, partially melted hailstones with reflective water coatings. Since hail up to d = 45 mm was reported on 3 August and 22 August, the large Z in these storms may have been due to giant hail if the largest hailstones were not measured. In contrast, the small, melting hailstones on 9 September (d ≤ 13 mm) likely contributed to

Z in excess of 70 dBZ.

CAPPIs of ZDR, ρhv, and KDP are presented in Figs. 4.8-4.10. Within the maximum reflectivity regions of the storms (black contour lines; Z > 50 dBZ), the CAPPIs depict minimum ZDR and ρhv values of 0 to -3 dB (Fig. 4.8) and 0.90-0.95 (Fig. 4.9), respectively, in each storm. The smallest ZDR and ρhv are found in the 3 August and 22 August storms, which contained the largest hail, a result that is expected

(Ryzhkov et al. 2010). KDP values vary substantially between the four cases (Fig. 4.10). In the 3 August

-1 (Fig. 4.10a) and 21 May (Fig. 4.10d) thunderstorms, KDP values range from 0-2 deg km in the hail cores.

98 In the hailstorms on 22 August (Fig. 4.10: As in Fig. 4.7, but for specific differential phase.) and 9

-1 September (Fig. 4.10c), however, KDP ranges from 4-6 deg km in the core. Since KDP is zero for spherical hailstones and is proportional to the rain rate, larger values of KDP indicate rain mixed with hail and/or a large number of water-coated hailstones. The hailstorm on 9 September indeed produced flooding rainfall along with a large quantity of partially melted hailstones, which is reflected by the large

KDP values on this date.

In the CAPPIs from 22 August and 21 May, a ribbon of unusually small ZDR (0 to -3 dB) stretches radially behind the hail core of the storm (Figs. 4.8b and 4.8d). Although not shown in the CAPPI (Fig.

4.8a), radar data from the 3 August storm also occasionally contained this ribbon of small ZDR. This feature is evidence of three-body scattering (TBS; Zrnić 1987; Hubbert and Bringi 2000; Kumjian et al.

2010), which occurs when energy from the radar beam is scattered by hail to the ground, which in turn scatters the energy back to the hail and finally to the radar. Kumjian et al. (2010) suggested that TBS at S

Band is indicative of hailstones with 20 mm < d < 50 mm, since this signature is not seen in storms that contain mostly small hail (d < 20 mm) or predicted from scattering calculations with exclusively giant hail (d > 50 mm). The absence of TBS in the radar data from the 9 September storm, which contained small hail (d ≤ 13 mm), corroborates this information. Since a large amount of hail with 20 mm < d < 50 mm is likely to produce the strongest TBS signature, storms that exhibit these signatures on days in which the synoptic environment favors accumulating hail may contain severe hail and have the potential to cause deep hail accumulations.

99

Fig. 4.7: Constant altitude plan position indicators of reflectivity at a) 2216 UTC 3 Aug 2013 at z = 3.5 km MSL, b) 2344 UTC 22 Aug 2013 at z = 3 km MSL, c) 2107 UTC 9 Sept 2013 at z = 2.5 km MSL, and d) 2028 UTC 21 May 2014 at z = 2.5 km MSL. The black lines are contours of reflectivity from 50 dBZ to 70 dBZ at intervals of 5 dBZ. The white plus signs indicate the locations of the plowable hail reports.

100

Fig. 4.8: As in Fig. 4.7, but for differential reflectivity.

101

Fig. 4.9: As in Fig. 4.7, but for correlation coefficient.

102

Fig. 4.10: As in Fig. 4.7, but for specific differential phase.

103 4.3.2.2 Time-height evolution of radar features

The dual-polarization radar variables from the four hailstorms are also shown as a function of height and time in Figs. 4.11-4.14. Maximum Z values from each height and radar scan are shown in Fig.

4.11, while minimum ZDR and ρhv and median KDP values (taken from grid cells with Z ≥ 50 dBZ) are shown in Figs. 4.12-4.14, respectively. The 50-dBZ threshold is used to restrict the radar data to times and locations in the storm that were likely involved in hail production and to highlight the time evolution of the 50-dBZ echo top. The total graupel mass (gray lines) and hail mass (black lines) at each time and height throughout the entire storm are also shown.

From Fig. 4.11, each of the hailstorms consisted of 1-3 maxima in 50-dBZ echo top height and Z over time, separated by intervals when the 50-dBZ echo top height was lower (6-10 km versus 11-15 km) and Z was relatively weaker (60-65 dBZ versus 68-74 dBZ). These maxima occurred every one to two hours and are suggestive of storm updraft speeds that pulsed in intensity over time. Periods of intense graupel production (>5×107 kg; gray lines in Fig. 4.11) preceded the maxima in reflectivity by ~30 min, while the greatest hail production (>107 kg; black lines in Fig. 4.11) was generally coincident with the largest reflectivity values. In regards to the occurrence of plowable hail, all of the reports (denoted by red vertical lines in Fig. 4.11) were associated with maximum column Z > 70 dBZ and 50-dBZ echo top heights that were near the maximum observed over the analysis period. These features were either coincident with the hail report or occurred up to 30 min prior to it (e.g., 21 May 2014, Fig. 4.11d). In the

30 min that followed the plowable hail reports, Z weakened considerably by >10 dB in each of the four cases, possibly initiated by loading of the updraft with large quantities of hail.

The time-height evolution of the dual-polarization radar variables will now be examined.

Typically, large, dry hailstones have ZDR around 0 dB due to their near spherical shape and have small ρhv

(relative to rain) of 0.8-0.95. However, large hail can also be associated with unusually negative ZDR values of -5 dB or less due to TBS (Hubbert and Bringi 2000), as discussed in section 4.3.2.1. As hailstones descend below the melting layer, the smaller hailstones (d < 20 mm) develop a liquid water

104 coating, which results in positive ZDR below the melting layer in hail shafts that consist primarily of small hailstones.

Fig. 4.11: Time-height plots of the maximum reflectivity for Z ≥ 50 dBZ for the hailstorms on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. Gray lines are contours of graupel mass of (1, 5, 10, 15, and 20) × 107 kg. Black lines are contours of hail mass of (1, 3, 6, and 9) × 107 kg. The red vertical lines in the background indicate the times that plowable hail was reported. The blue horizontal lines indicate the heights of the 0 °C, -10 °C, and -25 °C isotherms from the operational soundings listed in Tab. 4.2.

105

Fig. 4.12: As in Fig. 4.11, but for the minimum differential reflectivity.

106

Fig. 4.13: As in Fig. 4.11, but for the minimum correlation coefficient.

107

Fig. 4.14: As in Fig. 4.11, but for the median specific differential phase.

108 Time-height plots of ZDR, ρhv, and KDP are shown in Fig. 4.12, Fig. 4.13, and Fig. 4.14, respectively. Overall, the radar variables from the 22 August case (Figs. 4.12b-4.14b) most clearly distinguish between the period of hail accumulation and the remainder of the storm. The plowable hail report at 2339 UTC was coincident with minimum ZDR and ρhv values of -4 dB and 0.5 at z = 2.5 km, respectively. These small near-surface values were associated with a column of negative ZDR (0 to -2 dB) and small ρhv (0.75-0.95) that extended to z = 9 km, but was most evident at z < 4 km where the largest hailstones were likely located. After peaking throughout the column at ~2320 UTC, median KDP values decreased slightly as hail production increased (Fig. 4.14b). Out of the four events, the largest hailstones occurred on 3 August and 22 August (44.5 mm; Tab. 4.1), which may explain why the signals in the radar variables were so pronounced during the 22 August case.

In the other cases, similar signals can be seen in ZDR, ρhv, and KDP, although the variables generally discriminate less strongly between the times of plowable hail and the remainder of the storm than in the 22 August case. On 3 August, there were a number of times when minimum ZDR < -2.5 dB

(Fig. 4.12a) and ρhv < 0.75 (Fig. 4.13a) overlapped, although this result is not surprising since this storm produced large hail for much of its lifetime, unlike the 22 August storm. However, these values did not extend much below z = 5 km until 2200 UTC (16 min prior to the plowable hail report), when a column of negative ZDR and small ρhv gradually descended towards the surface with time, reaching the lowest height sampled by the radar at 2215 UTC. Similar to the 22 August case, a peak in KDP occurred ~15 min prior to the plowable hail report. KDP then decreased as hail production maximized (Fig. 4.14a). A similar sequence of events occurred on 21 May (Figs. 4.12d-4.14d), when a pocket of ZDR ~ -2 dB and ρhv ~ 0.7 formed near z = 5 km at 1935 UTC and reached the lowest radar level (z = 2 km) at 2015 UTC (15 min prior to the plowable hail report). The peak in KDP prior to the occurrence of plowable hail, however, is absent on this day.

Finally, the time-height cross sections of ZDR and ρhv demonstrate the least utility for distinguishing the occurrence of plowable hail on 9 September (Figs. 4.12c and 4.13c), and we note that this storm produced the smallest hailstones (d = 12.7 mm; Tab. 4.1) out of the four cases. The smallest 109 values of ZDR and ρhv below z = 3 km occurred at 2030 UTC, 30 min before the plowable hail report when the storm was still over the Foothills west of Ken Caryl, CO. By the time hail began to accumulate at the surface at 2100 UTC, minimum ZDR (ρhv) values had increased by about 1 dB (0.3) at z = 2.5 km, possibly

-1 due to the presence of water-coated hailstones (inferred from KDP > 3 deg km ; Figs. 4.10c and 4.14c).

Nevertheless, the peak in Z of ~73 dBZ at 2100 UTC (Fig. 4.7c) suggests that the most extensive hailfall was occurring at this time. The CAPPIs (Figs. 4.8c and 4.9c), which depict minima in ZDR and ρhv collocated with Z > 70 dBZ at 2107 UTC, also support this conclusion.

4.3.2.3 Estimating hail accumulation from radar data

The radar characteristics discussed thus far are not exclusive to thunderstorms that produce deep hail accumulations. Z > 70 dBZ, descending columns of ZDR near 0 dB and ρhv < 0.95 in the thunderstorm core, and TBS signatures have all been observed in non-accumulating hailstorms (e.g., Hubbert and

Bringi 2000; Kumjian and Ryzhkov 2008; Ryzhkov et al. 2010). To identify the occurrence of plowable hail in real time, we propose that forecasters estimate the hail accumulation (hAcc) from the radar data using Eq. (4.4).

v tcurrent hAcc   tt M h,t (4.4)  h tt0

-3 Mh is the hail mass concentration (kg m ) at the lowest radar level, determined from Eq. (4.2) and the method discussed in section 4.2.2, Δt is the time (s) between successive radar scans, v is the hail fall speed

-1 -3 (cm s ), ρh is the hail bulk density (kg m ), and η is the fractional space occupied by ice (rather than air) once the hailstones accumulate on the ground. We have assumed v = 1500 cm s-1 (appropriate for a d = 2

-3 cm hailstone; Pruppacher and Klett 1997), ρh = 900 kg m , and η = 0.64, the closest possible random packing of monodisperse spheres (Scott and Kilgour 1969). For each radar grid cell, the product tM h can be computed for all radar scans since the formation of the hailstorm (t0) through the current time

(tcurrent) and then summed to map the storm total hail accumulation.

110 The results of applying this procedure to the radar scans within the analysis periods indicated in

Tab. 4.1 are shown in Fig. 4.15. In all four cases, the plowable hail reports (black squares in Fig. 4.15) are collocated with hAcc > 5 cm, whereas the remainder of each hail swath mostly contains hAcc < 1.5 cm.

Two exceptions (circled areas) occur on 3 August (Fig. 4.15a) and 21 May (Fig. 4.15d), when hAcc near

10 cm is noted well to the northwest and east of the hail reports, respectively. The area on 3 August is located in an unpopulated region of the Foothills to the south of the Wyoming border (Fig. 4.2), and thus it cannot be determined whether the estimated hail accumulations actually occurred. While the circled area in Fig. 4.15d is also sparsely populated, storm chasers reported and photographed hail accumulations of at least 10 cm in this area in the wake of the 21 May hailstorm. These results indicate that the above technique is capable of distinguishing between times and locations in which accumulating hail does and does not occur.

4.3.3 LIGHTNING AND ICE MASS ANALYSIS

Here we investigate whether three-dimensional total lightning data, when examined in conjunction with the dual-polarization radar data, offer any additional insight into the likelihood of plowable hail. Time series of the lightning flash rate and the storm total graupel mass are shown in Fig.

4.16. Lightning flash rates at the time of the plowable hail reports (denoted by dashed black vertical lines) ranged from 25 min-1 in the 9 September storm (Fig. 4.16c) to 220 min-1 on 3 August (Fig. 4.16a), the latter of which was closest to the lightning mapping array (Fig. 4.2). Because the detection efficiency of the array is inversely proportional to the square of the distance from its center, the focus here is not on the absolute magnitudes of the flash rates but on how the lightning activity evolved over time in relation to the graupel content and the hail reports. In three of the four cases (the exception being 9 September), plowable hail occurred as the lightning flash rate was increasing, in some cases dramatically (e.g., flash rates tripled in the 30 min prior to the 22 August plowable hail report; Fig. 4.16b). Further, in all of the storms, the flash rates at the time of the plowable hail report were at or near the largest observed in the storm at the time of the report.

111

Fig. 4.15: Accumulated hail depths estimated from the radar data on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. Squares indicate the locations of the plowable hail reports. Inferred areas of accumulating hail that occurred in sparsely populated locations are circled.

112 The time series plots in Fig. 4.16 demonstrate that the increases in lightning flash rate that occurred during or just prior to accumulating hailfall were also accompanied by increases in the storm total graupel mass (with the period from 2000-2045 UTC on 21 May being an exception; Fig. 4.16d).

Overall, it is evident that the flash rates are well correlated with the total graupel mass. This correlation ranges from 0.77 to 0.83 over the lifetime of each storm (not shown). The graupel mass is also well correlated with other lightning characteristics, such as the maximum observed flash area (not shown).

This correlation ranges from 0.64 to 0.74, although it is acknowledged that the growth in the size of the hailstorm over time is a confounding variable that increases both the flash area and the total graupel mass.

Fig. 4.16: Time series of storm total graupel mass (blue lines), lightning flash rate (black solid lines), and the area of the 40 dBZ-isoecho at the approximate height of the -10 °C isotherm (red lines) for the hailstorms on a) 3 Aug 2013, b) 22 Aug 2013, c) 9 Sept 2013, and d) 21 May 2014. The dashed black lines indicate the times that plowable hail was reported.

113 For storm electrification to occur, graupel must be present in conjunction with ice crystals and supercooled liquid water, since rebounding collisions between ice crystals and graupel act to generate charge within the cloud (e.g., Williams et al. 1991; Saunders 1993; Ziegler and MacGorman 1994). The time-height plots of reflectivity and graupel and hail mass (Fig. 4.11) demonstrate that graupel production dramatically increases when maxima in echo top height and reflectivity occur. These maxima are suggestive of intense updrafts that are supportive of both graupel and hail formation and that cause increases in lightning flash rate due to the additional graupel mass. While forecasters may not be able to calculate the total graupel mass easily, Fig. 4.16 demonstrates that another quantity, the area of the 40- dBZ isoecho at the approximate height of the -10 °C isotherm (determined from atmospheric soundings;

Tab. 4.2), closely tracks the time series of storm total graupel mass and can be used as a proxy. It is not surprising that this quantity closely mirrors the trend in graupel mass, since Z > 40 dBZ at -10 °C likely requires the existence of graupel and/or hail at this height. The presence of 40-dBZ reflectivity at the height of the -10 °C isotherm has also been successfully used to predict the onset of lightning (e.g., Dye et al. 1989; Gremillion and Orville 1999; Vincent et al. 2003), a further indication of its relationship to graupel mass.

4.4 SUMMARY AND CONCLUSIONS

In this paper, we examined dual-polarization radar (Z, ZDR, ρHV, KDP, graupel and hail mass) and total lightning observations from four hailstorms that resulted in hail accumulations of 15 to 60 cm along the

Colorado Front Range in 2013 and 2014 (Tab. 4.1; Fig. 4.2). The lightning and radar data were interpolated to a Cartesian grid with 0.5-km spacing, and relationships between the radar and lightning variables and hail production were investigated that may assist forecasters in recognizing and predicting future plowable hailstorms. The synoptic weather conditions that favor the development of these storms were also examined.

114 When analyzing the synoptic weather patterns prior to the occurrence of the hailstorms, it was found that moist southwesterly to westerly flow existed at 500 hPa at relatively slow speeds (5-15 m s-1;

Fig. 4.3). Post-frontal, upslope low-level flow was also present that contributed to moist surface dewpoint temperatures of 11-19 °C (Fig. 4.4). The upper-air pattern and upslope low-level flow resulted in ample atmospheric moisture and relatively light steering winds that favored deep hail accumulations through a combination of slow storm motions and heavy precipitation rates. Column-integrated precipitable water vapor observations from rawinsondes at Denver, CO revealed values that were 142-212 percent of the mean monthly values on the plowable hail days (Tab. 4.2; Fig. 4.6a). Counterintuitively, the freezing level was 200-700 m higher than the monthly climatological means (Fig. 4.6b), indicative of the relatively warm atmospheric column that contributed to the large moisture content. Therefore, forecasters in eastern

Colorado may not be able to rely on the freezing level height to assess the likelihood of these extreme hailstorms.

When dual-polarization radar data from the hailstorms were examined, it was found that all of the storms had maximum column reflectivity in excess of 70 dBZ and 50-dBZ echo tops of 11-15 km MSL within 30 min prior to the plowable hail reports (Figs. 4.7 and 4.11). Increases in the intensity of the thunderstorm updraft (inferred from increases in Z and echo top height) first resulted in greater graupel production (5-20×107 kg) and then increased hail production (1-9×107 kg), generally within 30 min of updraft intensification (Fig. 4.11). The largest hail masses were coincident with maxima in Z and the plowable hail reports. All of the hailstorms also exhibited small values of ZDR and ρHV of 0 to -4 dB and

0.90-0.95, respectively, within the hail core (Figs. 4.8-4.9, 4.12-4.13), indicative of nearly spherical frozen hydrometeors likely mixed with rain. These reductions in ZDR and ρHV began within the midlevels of the radar volume scan (6-10 km MSL) and then generally descended to the surface within the next 15-

30 min. The hailstorms with the smallest values of ZDR and ρHV (i.e., 3 August and 22 August) also produced the largest hailstones during the plowable hail events (Tab. 4.1). Reductions in ZDR and ρHV were much more limited in area and magnitude in the 9 September hailstorm, which produced maximum hail

-1 diameters of only 12.7 mm. This storm also exhibited the largest KDP values (4-6 deg km ; Figs. 4.10c 115 and 4.14c) at the time of hail accumulation, a further indication of the presence of small, partially melted hailstones accompanied by heavy rain. While the radar features discussed above are not exclusive to plowable hailstorms, Fig. 4.15 demonstrates that the time and location of plowable hail can be pinpointed by accumulating the hail mass derived from successive radar volume scans. Areas of plowable hail have hail depths in excess of 5 cm, while locations that received non-plowable amounts of hail have hail depths mostly less than 1.5 cm.

Last, total lightning and dual-polarization radar data were examined together to determine if there were any relationships between these data that were associated with the plowable hail reports. It was found that in each storm, the lightning flash rate was near the maximum observed thus far within the storm at the time of the plowable hail reports (Fig. 4.16). The lightning flash rate and maximum flash area were well correlated (r~0.8 and 0.7, respectively) with the storm total graupel mass (Fig. 4.16) calculated from the radar data, which generally increased in the 30 min prior to the plowable hail report and provided the necessary collisions between graupel and ice crystals to electrify the hailstorm further.

It is emphasized that the relationships between the synoptic weather, radar, and lightning variables analyzed here are based on four plowable hailstorms. Future work should focus on additional analyses using a larger sample set of hailstorms so that statistical relationships can be established.

Nevertheless, forecasters can use the results from this initial study to detect similar synoptic weather patterns that may be conducive to plowable hailstorms. Once it is known that the weather pattern favors storms with accumulating hail, the dual-polarization radar and total lightning data can be used in conjunction with each other to determine the likelihood that a particular storm will result in substantial hail accumulations.

4.5 ACKNOWLEDGEMENTS

We thank the employees of the National Weather Service Forecast Office in Boulder, CO for providing valuable feedback on this study. We also thank Mike Dixon (NCAR) for his assistance in using the Radx

C++ software package for radar data processing, and Scott Ellis (NCAR) for his help in using Solo II to

116 view and edit the radar data. This material is based upon work supported by the National Science

Foundation Graduate Research Fellowship under DGE-1144083.

117 5 OVERALL CONCLUSION

5.1 SUMMARY OF MAJOR FINDINGS

The work presented in this dissertation includes the first studies 1) to analyze the usefulness and accuracy of the disdrometer and X-band dual-polarization radar observations collected during VORTEX2 (Chapter

2; Kalina et al. 2014a), 2) to quantify the changes in idealized supercell thunderstorm microphysical processes, cold pool area and temperature, and precipitation amount across the entire range of CCN concentrations (100 to 10 000 cm-3) observed in the Earth’s atmosphere (Chapter 3; Kalina et al. 2014b), and 3) to investigate the synoptic weather conditions as well as the dual-polarization radar and lightning characteristics associated with “plowable” hailstorms in Colorado (Chapter 4; Kalina et al. 2015).

In Chapter 2, dual-polarization radar data (Z, ZDR, and particle type) from the mobile NOXP radar and the operational WSR-88D network were compared to optical disdrometer data. These measurements were collected in five supercell thunderstorms and one squall line during VORTEX2. The uncertainty between the disdrometer and dual-polarization radar measurements was quantified to determine the usefulness of the observations for purposes of data assimilation, model validation, and microphysical analyses. The results demonstrated that the attenuation-corrected NOXP radar and disdrometer data agreed, on average, to within 1 dB (Z) and 0.2 dB (ZDR). However, greater disagreement occurred in large hail (6-dB and 1.6-dB discrepancies in Z and ZDR, respectively) and when heavy rain decreased the NOXP radar signal quality index to less than 0.8 (13-dB and 0.6-dB discrepancies in Z and ZDR, respectively).

When the disdrometer and WSR-88D radar data were compared, the agreement in Z was much improved to 1.5 dB in large hail and 0.7 dB in heavy rain. These results suggest that the attenuation correction scheme (Steiner et al. 2009) that was applied to the mobile NOXP radar data introduced substantial error to the data when large hail or poor radar signal quality were present. Therefore, X-band radar data from range gates that contain these conditions may not be appropriate for quantitative precipitation estimation or for assimilation into numerical weather prediction models. Using a new particle identification scheme

118 for optical disdrometer observations in convective weather (Friedrich et al. 2013; Kalina et al. 2014), it was determined that the particle types observed by the radar and the disdrometers were the same 63% of the time. Discrepancies were caused by hail melting into rain between the height of the radar beam and the surface disdrometer location and by differences in the sampling techniques of the two instruments.

Overall, the measured PSD from the disdrometers have similar accuracy to the radar data, and therefore can be used for data assimilation, model validation, and microphysical analyses.

Chapter 3 explored the effect of CCN concentration on idealized supercell thunderstorm microphysical, thermodynamic, and precipitation processes for varying low-level relative humidity and vertical wind shear conditions in the WRF Model. Fifteen different CCN concentrations between 100 cm-3 and 10 000 cm-3 were tested with three different values of the low-level relative humidity (61%,

80%, and 91%) and two different values of the bulk vertical wind shear (8 m s-1 and 16 m s-1). The results from this study indicate that, relative to a clean atmosphere, changes in supercell thunderstorm microphysical process rates (i.e., accretion, collision and coalescence, melting, and evaporation) generally peak at CCN ≈ 3000 cm-3 before saturating. Therefore, substantial changes in supercell thunderstorm microphysics can occur even under relatively moderate pollutant concentrations. The simulations also revealed that the response of the area and temperature of the near-surface cold pool (i.e., the air cooled by evaporating rain and melting hail) is highly dependent on the environmental conditions. In moist conditions with moderate vertical wind shear, the cold pool area was found to be nearly constant with respect to CCN concentration, while the area was reduced by 84% and 22% in the soundings with dry relative humidity (i.e., 61%) and large vertical wind shear (i.e., 16 m s-1), respectively. Finally, I determined that CCN concentration affects the spatial distribution of precipitation in supercell thunderstorms, with up to 25 mm more rainfall near the thunderstorm updrafts in the most polluted cases when the relative humidity was at least 80%. In summary, although the atmospheric dynamics exert a first-order effect on supercell thunderstorm evolution, aerosol effects are nonetheless an important second-order consideration. As mesoscale model physics continues to improve, the above results suggest

119 that it may indeed be important to assimilate observations of CCN concentration to predict supercell thunderstorm characteristics accurately.

In Chapter 4, the synoptic weather conditions conducive to the formation of thunderstorms that result in hail accumulations greater than 15 cm (also referred to as “plowable” hailstorms) were analyzed.

In addition, dual-polarization radar and lightning data from the operational radar and lightning networks were examined to determine the predictability of similar extreme events in the future. On days with plowable hailstorms, southwesterly to westerly 500 hPa winds of 5-15 m s-1 were observed, which advected moist air from the eastern Pacific over Colorado and resulted in slow storm motions. Column- integrated precipitable water vapor values ranged from 142-212% of average, which were further enhanced by cold-frontal passages that resulted in moist easterly low-level upslope flow prior to convective initiation.

Time-height plots of differential reflectivity (ZDR) and correlation coefficient (ρhv) from the

Weather Surveillance Radar-1988 Doppler (WSR-88D) at Front Range Airport (KFTG) showed notable decreases in ZDR (from 1-3 dB to less than 0 dB) and ρhv (from near 1.0 to 0.90-0.95) in the low-levels (0-

3 km AGL) prior to the accumulation of hail at the surface, with 15-30 min of lead time in some cases. It was also shown that the time and location of plowable hail can be pinpointed in real time by accumulating the hail mass derived from the operational radar data. This new technique provides an estimate of the hail depth on the ground. Depths greater than 5 cm were associated with locations that received plowable hail, which were distinguished from much larger areas that received hail that did not accumulate (associated with depths less than 1.5 cm). The plowable hail reports were also accompanied by an increase in lightning flash rate, which suggests that enhanced lightning activity may provide forecasters with an additional indication that intense hailfall is imminent. This relationship likely exists due to the increase in graupel mass that occurred 30-60 min prior to accumulating hail, as graupel mass was found to be well- correlated with lightning flash rate (r~0.8) and maximum flash area (r~0.7) and provided the necessary rebounding collisions with ice crystals to result in cloud electrification. Overall, the results in Chapter 4

120 demonstrate how advanced operational measurement networks, such as WSR-88D and COLMA, can improve the ability of forecasters to nowcast extreme weather events such as deep hail.

5.2 OUTLOOK

Several opportunities exist to extend the work presented here. While the application of attenuation correction schemes to mobile C- and X-band radar data is a common approach and is often assumed to result in correct (or at least approximately correct) Z and ZDR, the results of attenuation correction schemes should always be validated against other datasets when available. Chapter 2 offers a systematic validation of the Steiner et al. (2009) scheme against disdrometer and S-band radar data, but there are many other attenuation correction schemes currently in use (e.g., Carey et al. 2000; Testud et al. 2000;

Bringi et al. 2001; Anagnostou et al. 2006) and these should be validated as well. Currently, it is unknown whether these schemes also perform poorly in areas of large hail and limited radar signal quality as the

Steiner et al. (2009) scheme does. Unless a particular scheme is tested, hydrometeor classification and quantitative precipitation estimation from attenuation-corrected Z and ZDR have unknown accuracy, and assimilating these attenuation-corrected data into numerical models may indeed degrade their performance.

In recent years (2010-present), a substantial increase in our understanding of how CCN concentration affects thunderstorm microphysical and thermodynamic processes has been achieved.

However, a similar subject that has received comparatively little attention is the influence that ice nuclei

(IN) concentration (and composition) might have on these storms. IN concentrations at the height of the -

20 °C isotherm typically range from 0.001 cm-3 to 0.01 cm-3 (Pruppacher and Klett 1997), which means that IN are at least ten thousand times sparser than CCN at this height. Nevertheless, IN are the building blocks for graupel and hail particles, which can substantially alter the latent heating profile within a thunderstorm through microphysical processes such as riming and melting. More research is needed to understand how the properties of IN can influence thunderstorm evolution and behavior.

121 Finally, the analysis of synoptic weather, radar, and lightning trends in Colorado thunderstorms that produced deep (greater than 15 cm) hail accumulations is based on four cases. Additional work should focus on analyzing other plowable hailstorms to determine whether the synoptic weather pattern of slow, moist westerly mid-level flow and post-frontal low-level upslope flow is consistent across a larger number of cases. Radar and lightning data from a larger sample set would also allow thresholds for Z,

ZDR, ρhv, hail mass, and lightning flash rate to be determined that could be used to distinguish plowable hailstorms from other hail-producing thunderstorms. Finally, plowable hailstorms are not exclusive to

Colorado or even to the United States (Grahame et al. 2009; Schlatter and Doesken 2010), and an examination of the synoptic weather patterns that produce these storms in regions that are geographically and climatologically dissimilar to the U.S. Intermountain West should be undertaken.

122 REFERENCES

Anagnostou, M. N., E. N. Anagnostou, and J. Vivekanandan, 2006: Correction for rain path specific and differential attenuation of X-band dual-polarization observations. IEEE Geosci. Remote Sens., 44, 2470–2480.

Andreae, M. O., D. Rosenfeld, P. Artaxo, A. A. Costa, G. P. Frank, K. M. Longlo, and M. A. F. Silva- Dias, 2004: Smoking rain clouds over the Amazon. Science, 303, 1337–1342.

Atlas, D., R. C. Srivastava, and R. S. Sekhon, 1973: Doppler radar characteristics of precipitation at vertical incidence. Rev. Geophys. Space Phys., 11, 1–35.

Aydin, K., S. H. Park, and T. M. Walsh, 1998: Bistatic dual-polarization scattering from rain and hail at S- and C-band frequencies. J. Atmos. Oceanic Technol., 15, 1110–1121.

Balakrishnan, N., and D. S. Zrnić, 1990: Use of polarization to characterize precipitation and discriminate large hail. J. Atmos. Sci., 47, 1525–1540.

Barthazy, E., S. Göke, R. Schefold, and D. Högl, 2004: An optical array instrument for shape and fall velocity measurements of hydrometeors. J. Atmos. Oceanic Technol., 21, 1400–1416.

Beard, K. V., and C. H. Chuang, 1987: A new model for the equilibrium shape of raindrops. J. Atmos. Sci., 44, 1509–1524.

Berthet, C., E. Wesolek, J. Dessens, and J. L. Sanchez, 2013: Extreme hail day climatology in southwestern France. Atmos. Res., 123, 139–150.

Bluestein, H. B., and C. R. Parks, 1983: A synoptic and photographic climatology of low-precipitation severe thunderstorms in the southern Plains. Mon. Wea. Rev., 111, 2034–2046.

Borowska, L., A. Ryzhkov, D. Zrnic, C. Simmer, and R. Palmer, 2011: Attenuation and differential attenuation of 5-cm-wavelength radiation in melting hail. J. Appl. Meteor. Climatol., 50, 59–76.

Brady, R. H., and E. J. Szoke, 1989: A case study of nonmesocyclone tornado development in northeast Colorado: Similarities to waterspout formation. Mon. Wea. Rev., 117, 843–856.

Bringi, V. N., and V. Chandrasekar, 2001: Polarimetric Doppler Weather Radar: Principles and Applications. Cambridge University Press, 662 pp.

Bringi, V. N., T. D. Keenan, and V. Chandrasekar, 2001: Correcting C-band radar reflectivity and differential reflectivity data for rain attenuation: A self-consistent method with constraints. IEEE Trans. Geosci. Remote Sens., 39, 1906–1915.

Bringi, V. N., V. Chandrasekar, J. Hubbert, E. Gorgucci, W. L. Randeu, and M. Schoenhuber, 2003: Raindrop size distribution in different climatic regimes from disdrometer and dual-polarized radar analysis. J. Atmos. Sci., 60, 354–365.

Brown, D., R. Brownrigg, M. Haley, and W. Huang, 2012: The NCAR Command Language (NCL) (version 6.0.0). UCAR/NCAR Computational and Information Systems Laboratory, Boulder, CO. [Available online at http://dx.doi.org/10.5065/D6WD3XH5.]

123

Browning, K. A., 1964: Airflow and precipitation trajectories within severe local storms which travel to the right of the winds. J. Atmos. Sci., 21, 634–639.

Browning, K. A., and R. J. Donaldson, 1963: Airflow and structure of a tornadic thunderstorm. J. Atmos. Sci., 20, 533–545.

Bryan, G. H., and H. Morrison, 2012: Sensitivity of a simulated squall line to horizontal resolution and parameterization of microphysics. Mon. Wea. Rev., 140, 202–225.

Burgess, D. W., E. R. Mansell, C M. Schwarz, and B. J. Allen, 2010: Tornado and tornadogenesis events seen by the NOXP, X-Band, dual-polarization radar during VORTEX2 2010. Preprints, 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., 5.2. [Available online at http://ams.confex.com/ams/25SLS/techprogram/paper_176164.htm.]

Cao, Q., G. Zhang, E. Brandes, T. Schuur, A. Ryzhkov, and K. Ikeda, 2008: Analysis of video disdrometer and polarimetric radar data to characterize rain microphysics in Oklahoma. J. Appl. Meteor. Climatol., 47, 2238–2255.

Carey, L. D., S. A. Rutledge, D. A. Ahijevych, and T. D. Keenan, 2000: Correcting propagation effects in C-band polarimetric radar observations of tropical convection using differential propagation phase. J. Appl. Meteor., 39, 1405–1433.

Carras, J. N., and W. C. Macklin, 1973: The shedding of accreted water during hailstone growth. Quart. J. Roy. Meteor. Soc., 99, 639–648.

Chappell, C. F., and D. M. Rodgers, 1988: Meteorological analysis of the Cheyenne, Wyoming, flash flood and hailstorm of 1 August 1985. NOAA Tech. Report ERL 435-FSL 1, 50 pp.

Darden, C. B., D. J. Nadler, B. C. Carcione, R. J. Blakeslee, G. T. Stano, and D. E. Buechler, 2010: Utilizing total lightning information to diagnose convective trends. Bull. Amer. Meteor. Soc., 91, 167–175.

Das, P., 1962: Influence of the wind shear on the growth of hail. J. Atmos. Sci., 19, 407–414.

Dearden, C., 2009: Investigating the simulation of cloud microphysical processes in numerical models using a one-dimensional dynamical framework. Atmos. Sci. Lett., 10, 207–214.

Deierling, W., and W. A. Petersen, 2008: Total lightning activity as an indicator of updraft characteristics. J. Geophys. Res., 113, D16210, doi:10.1029/2007JD009598.

Deierling, W., W. A. Petersen, J. Latham, S. Ellis, and H. J. Christian, 2008: The relationship between lightning activity and ice fluxes in thunderstorms. J. Geophys. Res., 113, D15210, doi:10.1029/2007JD009700.

Dessens, H., 1960: Severe hailstorms are associated with very strong winds between 6,000 and 12,000 meters. Physics of Precipitation, Geophys. Monogr., No. 5, Amer. Geophys. Union, 333–338.

Diaz, H. F., and T. W. Swetnam, 2013: The wildfires of 1910: Climatology of an extreme early twentieth- century event and comparison with more recent extremes. Bull. Amer. Meteor. Soc., 94, 1361–1370.

124

Dolan, B., and S. A. Rutledge, 2009: A theory-based hydrometeor identification algorithm for X-band polarimetric radars. J. Oceanic Atmos. Technol., 46, 2071–2088.

Dye, J. E., W. P. Winn, J. J. Jones, and D. W. Breed, 1989: The electrification of New Mexico thunderstorms: 1. Relationship between precipitation development and the onset of electrification. J. Geophys. Res., 94, 8643–8656.

Fan, J., R. Zhang, G. Li, and W.-K. Tao, 2007: Effects of aerosols and relative humidity on cumulus clouds. J. Geophys. Res., 112, D14204, doi:10.1029/2006JD008136.

Fan, J., and Coauthors, 2009: Dominant role by vertical wind shear in regulating aerosol effects on deep convective clouds. J. Geophys. Res., 114, D22206, doi:10.1029/2009JD012352.

Friedrich, K., E. A. Kalina, F. J. Masters, and C. R. Lopez, 2013a: Drop-size distributions in thunderstorms measured by optical disdrometers during VORTEX2. Mon. Wea. Rev., 141, 1182– 1203.

Friedrich, K., S. A. Higgins, F. J. Masters, and C. R. Lopez, 2013b: Articulating and stationary PARSIVEL disdrometer measurements in conditions with strong winds and heavy rainfall. J. Atmos. Oceanic Technol., 30, 2063–2080.

Friedrich, K., E. A. Kalina, J. D. Aikins, D. J. Gochis, and R. Rasmussen, 2015a: Precipitation and cloud structures of intense rain during the 2013 Great Colorado Flood. J. Hydrometeor., in press.

Friedrich, K., E. A. Kalina, J. D. Aikins, D. J. Gochis, P. Kucera, K. Ikeda, and M. Steiner, 2015b: Raindrop size distribution and rain characteristics during the 2013 Great Colorado Flood. J. Hydrometeor., in review.

Geotis, S. G., 1978: Comparison of reflectivity measurements by radar and by disdrometer. J. Appl. Meteor., 17, 1403–1405.

Gilmore, M. S., and L. J. Wicker, 1998: The influence of midtropospheric dryness on supercell morphology and evolution. Mon. Wea. Rev., 126, 943–958.

Giuli, D., M. Gherardelli, A. Freni, T. A. Seliga, and K. Aydin, 1991: Rainfall and clutter discrimination by means of dual-linear polarization radar measurements. J. Atmos. Oceanic Technol., 8, 777–789.

Gochis, D. J., and Coauthors, 2015: The Great Colorado Flood of September 2013. Bull. Amer. Meteor. Soc., in press.

Goddard, J. W. F., S. M. Cherry, and V. N. Bringi, 1982: Comparison of dual-polarization radar measurements of rain with ground-based disdrometer measurements. J. Appl. Meteor., 21, 252–256.

Goodman, S. J., and Coauthors, 2005: The North Alabama Lightning Mapping Array: Recent severe storm observations and future prospects. Atmos. Res., 76, 423–437.

Grahame, N., B. Riddaway, A. Eadie, B. Hall, and E. McCallum, 2009: Exceptional hailstorm hits Ottery St Mary on 30 October 2008. Wea., 64, 255–263.

125 Grant, L. D., and S. C. van den Heever, 2014: Microphysical and dynamical characteristics of low- precipitation and classic supercells. J. Atmos. Sci., 71, 2604–2624.

Gremillion, M. S., and R. E. Orville, 1999: Thunderstorm characteristics of cloud-to-ground lightning at the Kennedy Space Center, Florida: A study of lightning initiation signatures as indicated by the WSR-88D. Wea. Forecasting, 14, 640–649.

Grzych, M. L., B. D. Lee, and C. A. Finley, 2007: Thermodynamic analysis of supercell rear-flank downdrafts from Project ANSWERS. Mon. Wea. Rev., 135, 240–246.

Gu, J.-Y., A. Ryzhkov, P. Zhang, P. Neilley, M. Knight, B. Wolf, and D.-I. Lee, 2011: Polarimetric attenuation correction in heavy rain at C band. J. Appl. Meteor. Climatol., 50, 39–58.

Gunn, R., and G. D. Kinzer, 1949: The terminal velocity of fall for water droplets in stagnant air. J. Meteor., 6, 243–248.

Hall, M. P. M., J. W. F. Goddard, and S. M. Cherry, 1984: Identification of hydrometeors and other targets by dual-polarization radar. Radio Sci., 19, 132–140.

Harrington, J. Y., M. P. Meyers, R. L. Walko, and W. R. Cotton, 1995: Parameterization of ice crystal conversion processes due to vapor deposition for mesoscale models using double-moment basis functions. Part I: Basic formulation and parcel model test. J. Atmos. Sci., 52, 4344–4366.

Herzegh, P. H., and A. R. Jameson, 1992: Observing precipitation through dual-polarization radar measurements. Bull. Amer. Meteor. Soc., 73, 1365–1374.

Heymsfield, A. J., and K. M. Miller, 1988: Water vapor and ice mass transported into the anvils of CCOPE thunderstorms: Comparison with storm influx and rainout. J. Atmos. Sci., 45, 3501–3514.

Huang, G. J., V. N. Bringi, S. van den Heever, and W. Cotton, 2005: Polarimetric radar signatures from RAMS microphysics. Preprints, 32nd Int. Conf. on Radar Meteorology, Albuquerque, NM, Amer. Meteor. Soc., P11R.6. [Available online at http://ams.confex.com/ams/pdfpapers/96261.pdf.]

Huang, G. J., V. N. Bringi, and M. Thurai, 2008: Orientation angle distributions after an 80-m fall using a 2D video disdrometer. J. Atmos. Oceanic Technol., 25, 1717–1723.

Huang, G. J., V. N. Bringi, R. Cifelli, D. Hudak, and W. A. Petersen, 2010: A methodology to derive radar reflectivity–liquid equivalent snow rate relations using C-band radar and a 2D video disdrometer. J. Atmos. Oceanic Technol., 27, 637–651.

Hubbert, J. V., and V. N. Bringi, 1995: An iterative filtering technique for the analysis of coplanar differential phase and dual-frequency radar measurements. J. Atmos. Oceanic Technol., 12, 643–648.

Hubbert, J. C., and V. N. Bringi, 2000: The effects of three-body scattering on differential reflectivity signatures. J. Atmos. Oceanic Technol., 17, 51–61.

Hubbert, J. C., S. M. Ellis, M. Dixon, and G. Meymaris, 2010a: Modeling, error analysis, and evaluation of dual-polarization variables obtained from simultaneous horizontal and vertical polarization transmit radar. Part I: Modeling and antenna errors. J. Atmos. Oceanic Technol., 27, 1583–1598.

126 Hubbert, J. C., S. M. Ellis, M. Dixon, and G. Meymaris, 2010b: Modeling, error analysis and evaluation of dual-polarization variables obtained from simultaneous horizontal and vertical polarization transmit radar. Part II: Experimental data. J. Atmos. Oceanic Technol., 27, 1599–1607.

Illingworth, A. J., and C. J. Stevens, 1987: An optical disdrometer for the measurement of raindrop size spectra in windy conditions. J. Atmos. Oceanic Technol., 4, 411–421.

Iwanami, K., K. Kusunoki, N. Orikasa, M. Maki, R. Misumi, and M. Murakami, 2007: Hydrometeor type classification in winter clouds using X-band polarimetric radar measurements–Comparison of X-band polarimetric radar data with in-situ measurements by HYVIS. Preprints, 33rd Conf. on Radar Meteorology, Cairns, QLD, Australia, Amer. Meteor. Soc., P10.11. [Available online at http://ams.confex.com/ams/pdfpapers/123072.pdf.]

Jaffrain, J., and A. Berne, 2011: Experimental quantification of the sampling uncertainty associated with measurements from PARSIVEL disdrometers. J. Hydrometeor., 12, 352–370.

James, R. P., and P. M. Markowski, 2010: A numerical investigation of the effects of dry air aloft on deep convection. Mon. Wea. Rev., 138, 140–161.

Johns, R. H., and C. A. Doswell, 1992: Severe local storms forecasting. Wea. Forecasting, 7, 588–612.

Kain, J. S., and Coauthors, 2008: Some practical considerations regarding horizontal resolution in the first generation of operational convection-allowing NWP. Wea. Forecasting, 23, 931–952.

Kalina, E. A., K. Friedrich, S. M. Ellis, and D. W. Burgess, 2014a: Comparison of disdrometer and X- band mobile radar observations in convective precipitation. Mon. Wea. Rev., 142, 2414–2435.

Kalina, E. A., K. Friedrich, H. Morrison, and G. H. Bryan, 2014b: Aerosol effects on idealized supercell thunderstorms in different environments. J. Atmos. Sci., 71, 4558–4580.

Kalina, E. A., K. Friedrich, B. C. Motta, W. Deierling, G. T. Stano, and N. N. Rydell, 2015: An overview of Colorado plowable hailstorms: Synoptic weather, dual-polarization radar, and lightning data. Wea. Forecasting, in preparation.

Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.

Kanofsky, L., and P. B. Chilson, 2008: An analysis of errors in drop size distribution retrievals and rain bulk parameters with a UHF wind profiling radar and a two-dimensional video disdrometer. J. Atmos. Oceanic Technol., 25, 2282–2292.

Kennedy, P. C., S. A. Rutledge, B. Dolan, and E. Thaler, 2014: Observations of the 14 July 2011 Fort Collins hailstorm: Implications for WSR-88D-based hail detection and warnings. Wea. Forecasting, 29, 623–638.

Khain, A. P., and B. Lynn, 2009: Simulation of a supercell storm in clean and dirty atmosphere using weather research and forecast model with spectral bin microphysics. J. Geophys. Res., 114, D19209.

Khain, A. P., D. Rosenfeld, and A. Pokrovsky, 2005: Aerosol impact on the dynamics and microphysics of deep convective clouds. Quart. J. Roy. Meteor. Soc., 131, 2639–2663.

127 Khain, A. P., N. BenMoshe, and A. Pokrovsky, 2008: Factors determining the impact of aerosols on surface precipitation from clouds: An attempt at classification. J. Atmos. Sci., 65, 1721–1748.

Khain, A. P., D. Rosenfeld, A. Pokrovsky, U. Blahak, and A. V. Ryzhkov, 2011: The role of CCN in precipitation and hail in a mid-latitude storm as seen in simulations using a spectral (bin) microphysics model in a 2D dynamic frame. Atmos. Res., 99, 129–146.

Knight, N. C., 1986: Hailstone shape factor and its relationship to radar interpretation of hail. J. Climate Appl. Meteor., 25, 1956–1958.

Knight, N. C., and A. J. Heymsfield, 1983: Measurement and interpretation of hailstone density and terminal velocity. J. Atmos. Sci., 40, 1510–1516.

Knight, C. A., P. T. Schlatter, and T. W. Schlatter, 2008: An unusual hailstorm on 24 June 2006 in Boulder, Colorado. Part II: Low-density growth of hail. Mon. Wea. Rev., 136, 2833–2848.

Kruger, A., and W. F. Krajewski, 2002: Two-dimensional video disdrometer: A description. J. Atmos. Oceanic Technol., 19, 602–617.

Kumjian, M. R., and A. V. Ryzhkov, 2008: Polarimetric signatures in supercell thunderstorms. J. Appl. Meteor. Climatol., 47, 1940–1961.

Kumjian, M. R., and A. V. Ryzhkov, 2012: The impact of size sorting on the polarimetric radar variables. J. Atmos. Sci., 69, 2042–2060.

Kumjian, M., J. Picca, S. Ganson, A. Ryzhkov, and D. Zrnić, 2010: Three body scattering signatures in polarimetric radar data. NOAA/NSSL Tech. Report, 12 pp. [Available online at http://www.nssl.noaa.gov/publications/wsr88d_reports/FINAL_TBSS.doc.]

Kwiatkowski, J. M., A. B. Kostinski, and A. R. Jameson, 1995: The use of optimal polarizations for studying the microphysics of precipitation: Nonattenuating wavelengths. J. Atmos. Oceanic Technol., 12, 96–114.

Lang, T. J., S. A. Rutledge, B. Dolan, P. Krehbiel, W. Rison, and D. T. Lindsey, 2014: Lightning in wildfire smoke plumes observed in Colorado during summer 2012. Mon. Wea. Rev., 142, 489–507.

Lebo, Z. J., and J. H. Seinfeld, 2011: Theoretical basis for convective invigoration due to increased aerosol concentration. Atmos. Chem. Phys., 11, 5407–5429.

Lebo, Z. J., H. Morrison, and J. H. Seinfeld, 2012: Are simulated aerosol-induced effects on deep convective clouds strongly dependent on saturation adjustment? Atmos. Chem. Phys., 12, 9941–9964.

Lerach, D. G., and W. R. Cotton, 2012: Comparing aerosol and low-level moisture influences on supercell tornadogenesis: Three-dimensional idealized simulations. J. Atmos. Sci., 69, 969–987.

Lerach, D. G., B. J. Gaudet, and W. R. Cotton, 2008: Idealized simulations of aerosol influences on tornadogenesis. Geophys. Res. Lett., 35, L23806. doi:10.1029/2008GL035617.

Lesins, G. B., and R. List, 1986: Sponginess and drop shedding of gyrating hailstones in a pressure- controlled icing wind tunnel. J. Atmos. Sci., 43, 2813–2825.

128 Lesins, G. B., R. List, and P. I. Joe, 1980: Ice accretions. Part I: Testing of new atmospheric icing concepts. J. Rech. Atmos., 14, 347–356.

Levin, Z., and W. R. Cotton, Eds., 2009: Aerosol Pollution Impact on Precipitation: A Scientific Review. Springer, 386 pp.

Lim, K.-S. S., and S.-Y. Hong, 2010: Development of an effective double-moment cloud microphysics scheme with prognostic cloud condensation nuclei (CCN) for weather and climate models. Mon. Wea. Rev., 138, 1587–1612.

Liu, H., and V. Chandrasekar, 2000: Classification of hydrometeors based on polarimetric radar measurements: Development of fuzzy logic and neuro-fuzzy systems, and in situ verification. J. Atmos. Oceanic Technol., 17, 140–164.

Locatelli, J. D., and P. Hobbs, 1974: Fall speed and masses of solid precipitation particles. J. Geophys. Res., 79, 2185–2197.

Löffler-Mang, M., and J. Joss, 2000: An optical disdrometer for measuring size and velocity of hydrometeors. J. Atmos. Oceanic Technol., 17, 130–139.

Löffler-Mang, M., and U. Blahak, 2001: Estimation of the equivalent radar reflectivity factor from measured snow size spectra. J. Appl. Meteor., 40, 843–849.

Longley, R. W., and C. E. Thompson, 1965: A study of causes of hail. J. Appl. Meteor., 4, 69–82.

Luo, L., Y. Tang, S. Zhong, X. Bian, and W. E. Heilman, 2013: Will future climate favor more erratic wildfires in the western United States? J. Appl. Meteor. Climatol., 52, 2410–2417.

Mansell, E. R., and C. L. Ziegler, 2013: Aerosol effects on simulated storm electrification and precipitation in a two-moment bulk microphysics model. J. Atmos. Sci., 70, 2032–2050.

Markowski, P. M., J. M. Straka, and E. N. Rasmussen, 2002: Direct surface thermodynamic observations within the rear-flank downdrafts of nontornadic and tornadic supercells. Mon. Wea. Rev., 130, 1692– 1721.

Martin, G. M., D. W. Johnson, and A. Spice, 1994: The measurement and parameterization of effective radius of droplets in warm stratocumulus clouds. J. Atmos. Sci., 51, 1823–1842.

McCaul, E. W., Jr., J. C. Bailey, J. Hall, S. J. Goodman, R. J. Blakeslee, and D. E. Buechler, 2005: A flash clustering algorithm for North Alabama Lightning Mapping Array data. Conf. on Meteorological Applications of Lightning Data, San Diego, CA, Amer. Meteor. Soc., 5.3. [Available online at ams.confex.com/ams/Annual2005/techprogram/paper_84373.htm.]

McCaul, E. W., Jr., S. J. Goodman, K. M. LaCasse, and D. J. Cecil, 2009: Forecasting lightning threat using cloud-resolving model simulations. Wea. Forecasting, 24, 709–729.

McCumber, M., W. K. Tao, J. Simpson, R. Penc, and S. T. Soong, 1991: Comparison of ice-phase microphysical parameterization schemes using numerical simulations of tropical convection. J. Appl. Meteor., 30, 985–1004.

129 Milbrandt, J. A., and M. K. Yau, 2005a: A multimoment bulk microphysics parameterization. Part I: Analysis of the role of the spectral shape parameter. J. Atmos. Sci., 62, 3051–3064.

Milbrandt, J. A., and M. K. Yau, 2005b: A multimoment bulk microphysics parameterization. Part II: A proposed three-moment closure and scheme description. J. Atmos. Sci., 62, 3065–3081.

Milbrandt, J. A., and R. McTaggart-Cowan, 2010: Sedimentation-induced errors in bulk microphysics schemes. J. Atmos. Sci., 67, 3931–3948.

Morrison, H., 2012: On the robustness of aerosol effects on an idealized supercell storm simulated with a cloud system-resolving model. Atmos. Chem. Phys., 12, 7689–7705.

Morrison, H., and J. O. Pinto, 2005: Mesoscale modeling of springtime arctic mixed-phase stratiform clouds using a new two-moment bulk microphysics scheme. J. Atmos. Sci., 62, 3683–3704.

Morrison, H., and J. A. Milbrandt, 2011: Comparison of two-moment bulk microphysics schemes in idealized supercell thunderstorm simulations. Mon. Wea. Rev., 139, 1103–1130.

Morrison, H., J. A. Curry, and V. I. Khvorostyanov, 2005: A new double-moment microphysics parameterization for application in cloud and climate models. Part I: Description. J. Atmos. Sci., 62, 1665–1677.

Morrison, H., G. Thompson, and V. Tatarskii, 2009: Impact of cloud microphysics on the development of trailing stratiform precipitation in a simulated squall line: Comparison of one- and two-moment schemes. Mon. Wea. Rev., 137, 991–1007.

Morrison, H., S. Tessendorf, K. Ikeda, and G. Thompson, 2012: Sensitivity of a simulated midlatitude squall line to parameterization of raindrop breakup. Mon. Wea. Rev., 140, 2437–2460.

Neŝpor, V., W. F. Krajewski, and A. Kruger, 2000: Wind-induced error of raindrop size distribution measurement using a two-dimensional video disdrometer. J. Atmos. Oceanic Technol., 17, 1483– 1492.

Ntelekos, A. A., J. A. Smith, L. Donner, J. D. Fast, W. I. Gustafson Jr., E. G. Chapman, and W. F. Krajewski, 2009: The effects of aerosols on intense convective precipitation in the northeastern United States. Quart. J. Roy. Meteor. Soc., 135, 1367–1391.

Palmer, R., and Coauthors, 2009: Weather radar education at the University of Oklahoma--An integrated interdisciplinary approach. Bull. Amer. Meteor. Soc., 90, 1277–1282.

Pappas, J. J., 1962: A simple yes-no hail forecasting technique. J. Appl. Meteor., 1, 353–354.

Park, H., A. V. Ryzhkov, D. S. Zrnic, and K. Kim, 2009: The hydrometeor classification algorithm for the polarimetric WSR-88D: Description and application to an MCS. Wea. Forecasting, 24, 730–748.

Park, S. G., V. N. Bringi, V. Chandrasekar, M. Maki, and K. Iwanami, 2005: Correction of radar reflectivity and differential reflectivity for rain attenuation at X-band. Part I: Theoretical and empirical basis. J. Atmos. Oceanic Technol., 22, 1621–1632.

Penner, J. E., and Coauthors, 1994: Quantifying and minimizing uncertainty of climate forcing by anthropogenic aerosols. Bull. Amer. Meteor. Soc., 75, 375–400. 130

Pruppacher, H. R., and J. D. Klett, 1997: Microphysics of Clouds and Precipitation. 2nd ed. Kluwer Academic, 954 pp.

Rao, T. N., N. V. P. Kirankumar, B. Radhakrishna, and D. N. Rao, 2006: On the variability of the shape- slope parameter relations of the gamma raindrop size distribution model. Geophys. Res. Lett., 33, L22809.

Rasmussen, R. M., and A. J. Heymsfield, 1987a: Melting and shedding of graupel and hail. Part I: Model physics. J. Atmos. Sci., 44, 2754–2763.

Rasmussen, R. M., and A. J. Heymsfield, 1987b: Melting and shedding of graupel and hail. Part III: Investigation of the role of shed drops as hail embryos in the 1 August CCOPE severe storm. J. Atmos. Sci., 44, 2783–2803.

Rasmussen, R. M., V. Levizzani, and H. R. Pruppacher, 1984: A wind tunnel and theoretical study on the melting behavior of atmospheric ice particles: III. Experiment and theory for spherical ice particles of radius > 500 μm. J. Atmos. Sci., 41, 381–388.

Rinehart, R. E., 2004: Radar for Meteorologists. Rinehart Publications, 482 pp.

Rison, W., P. R. Krehbiel, R. J. Thomas, D. Rodeheffer, and B. Fuchs, 2012: The Colorado Lightning Mapping Array. 2012 Fall Meeting, San Francisco, CA, Amer. Geophys. Union, AE23B-0319. [Available online at http://fallmeeting.agu.org/2012/eposters/eposter/ae23b-0319.]

Rogers, R. R., and M. K. Yau, 1989: A Short Course in Cloud Physics. Pergamon Press, 209 pp.

Rotunno, R., and J. B. Klemp, 1985: On the rotation and propagation of simulated supercell thunderstorms. J. Atmos. Sci., 42, 271–292.

Rudlosky, S. D., and H. E. Fuelberg, 2013: Documenting storm severity in the mid-Atlantic region using lightning and radar information. Mon. Wea. Rev., 141, 3186–3202.

Ryzhkov, A. V., and D. S. Zrnic, 2007: Depolarization in ice crystals and its effect on radar polarimetric measurements. J. Atmos. Oceanic Technol., 24, 1256–1267.

Ryzhkov, A. V., T. J. Schuur, D. W. Burgess, and D. S. Zrnic, 2005: Polarimetric tornado detection. J. Appl. Meteor., 44, 557–570.

Ryzhkov, A., D. Zrnic, J. Krause, M. Kumjian, and S. Ganson, 2010: Discrimination between large and small hail: Final report. NOAA/NSSL Tech. Report, 18 pp. [Available online at https://www.nssl.noaa.gov/publications/wsr88d_reports/FINAL_HailSize.doc.]

Ryzhkov, A. V., M. Pinsky, A. Pokrovsky, and A. Khain, 2011: Polarimetric radar observation operator for a cloud model with spectral microphysics. J. Appl. Meteor. and Climatol., 50, 873–894.

Ryzhkov, A. V., M. R. Kumjian, S. M. Ganson, and A. P. Khain, 2013a: Polarimetric radar characteristics of melting hail. Part I: Theoretical simulations using spectral microphysical modeling. J. Appl. Meteor. Climatol., 52, 2849–2870.

131 Ryzhkov, A. V., M. R. Kumjian, S. M. Ganson, and P. Zhang, 2013b: Polarimetric radar characteristics of melting hail. Part II: Practical implications. J. Appl. Meteor. Climatol., 52, 2871–2886.

Saunders, C. P. R., 1993: A review of thunderstorm electrification processes. J. Appl. Meteor., 32, 642– 655

Schlatter, T. W., and N. Doesken, 2010: Deep hail: Tracking an elusive phenomenon. Weatherwise, 63, 35–41.

Schlatter, P. T., T. W. Schlatter, and C. A. Knight, 2008: An unusual hailstorm on 24 June 2006 in Boulder, Colorado. Part I: Mesoscale setting and radar features. Mon. Wea. Rev., 136, 2813–2832.

Schultz, C. J., W. A. Petersen, and L. D. Carey, 2009: Preliminary development and evaluation of lightning jump algorithms for the real-time detection of severe weather. J. Appl. Meteor. Climatol., 48, 2543–2563.

Schuur, T. J., A. V. Ryzhkov, D. S. Zrnić, and M. Schönhuber, 2001: Drop size distributions measured by a 2D video disdrometer: Comparison with dual-polarization radar data. J. Appl. Meteor., 40, 1019– 1034.

Scott, G. D., and D. M. Kilgour, 1969: The density of random close packing of spheres. J. Phys. D: Appl. Phys., 2, 863–866.

Seifert, A., and K. D. Beheng, 2006: A two-moment cloud microphysics parameterization for mixed- phase clouds: Part 2: Maritime vs. continental deep convective storms. Meteor. Atmos. Phys., 92, 67– 82.

Sevruk, B., 1982: Methods of correction for systematic error in point precipitation measurement for operational use. Operational Hydrology Rep. 21 Publ. 589, 91 pp. [Available from WMO, Case Postale 2300, CH-1211 Geneva, Switzerland.]

Shabbott, C. J., and P. M. Markowski, 2006: Surface in situ observations within the outflow of forward- flank down drafts of supercell thunderstorms. Mon. Wea. Rev., 134, 1422–1441.

SIGMET, 2009: RVP8 digital IF receiver/Doppler signal processor user's manual. SIGMET, 451 pp. [Available online at ftp://ftp.sigmet.com/outgoing/manuals/RVP8_Users_Manual.pdf.]

Skamarock, W. C., and Coauthors, 2008: A description of the advanced research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp. [Available online at http://www.mmm.ucar.edu/wrf/users/docs/arw_v3.pdf.]

Snook, N., and M. Xue, 2008: Effects of microphysical drop size distribution on tornadogenesis in supercell thunderstorms. Geophys. Res. Lett., 35, L24803.

Snyder, J. C., H. B. Bluestein, G. Zhang, and S. J. Frasier, 2010: Attenuation correction and hydrometeor classification of high-resolution, X-band, dual-polarized mobile radar measurements in severe convective storms. J. Atmos. Oceanic Technol., 27, 1979–2001.

Solheim, F. S., R. Vivekanandan, R. H. Ware, and C. Rocken, 1999: Propagation delays induced in GPS signals by dry air, water vapor, hydrometeors and other atmospheric particulates. J. Geophys. Res., 104, 9663–9670. 132

Steiner, M., G. Lee, S. M. Ellis, and J. Vivekanandan, 2009: Quantitative precipitation estimation and hydrometeor identification using dual-polarization radar: Phase II. NCAR Tech. Rep., 74 pp. [Available online at ftp://ftp.eol.ucar.edu/pub/temp/users/sellis/Steiner_etal_2009/.]

Storer, R. L., S. van den Heever, and G. L. Stephens, 2010: Modeling aerosol impacts on convective storms in different environments. J. Atmos. Sci., 67, 3904–3915.

Straka, J. M., 2009: Cloud and Precipitation Microphysics: Principles and Parameterizations. Cambridge University Press, 406 pp.

Tabary, P., G. Vulpiani, J. J. Gourley, A. J. Illingworth, and O. Bousquet, 2009: Unusually high differential attenuation at C-band: Results from a two-year analysis of the French Trappes polarimetric radar data. J. Appl. Meteor. and Climat., 48, 2037–2053.

Tao, W.-K., X. Li, A. Khain, T. Matsui, S. Lang, and J. Simpson, 2007: Role of atmospheric aerosol concentration on deep convective precipitation: Cloud-resolving model simulations. J. Geophys. Res., 112, D24S18.

Tao, W.-K., J.-P. Chen, Z. Li, C. Wang, and C. Zhang, 2012: Impact of aerosols on convective clouds and precipitation. Rev. Geophys., 50, RG2001.

Tessendorf, S. A., L. J. Miller, K. C. Wiens, and S. A. Rutledge, 2005: The 29 June 2000 supercell observed during STEPS. Part I: Kinematics and microphysics. J. Atmos. Sci., 62, 4127–4150.

Tessendorf, S. A., K. C. Wiens, and S. A. Rutledge, 2007: Radar and lightning observations of the 3 June 2000 electrically inverted storm from STEPS. Mon. Wea. Rev., 135, 3665–3681.

Testud, J., E. Le Bouar, E. Obligis, and M. Ali-Mehenni, 2000: The rain profiling algorithm applied to polarimetric weather radar. J. Atmos. Oceanic Technol., 17, 332–356.

Thomas, R. J., P. R. Krehbiel, W. Rison, S. J. Hunyady, W. P. Winn, T. Hamlin, and J. Harlin, 2004: Accuracy of the Lightning Mapping Array. J. Geophys. Res. Atmos., 109, D14207.

Thomson, A. D., and R. List, 1996: Raindrop spectra and updraft determination by combining Doppler radar and disdrometer. J. Atmos. Oceanic Technol., 13, 465–476.

Thompson, G., R. M. Rasmussen, and K. Manning, 2004: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part I: Description and sensitivity analysis. Mon. Wea. Rev., 132, 519–542.

Thompson, G., P. R. Field, R. M. Rasmussen, and W. D. Hall, 2008: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part II: Implementation of a new snow parameterization. Mon. Wea. Rev., 136, 5095–5115.

Thurai, M., and V. N. Bringi, 2005: Drop axis ratios from a 2D video disdrometer. J. Atmos. Oceanic Technol., 22, 966–978.

Thurai, M., M. Szakall, V. N. Bringi, K. V. Beard, S. K. Mitra, and S. Borrmann, 2009: Drop shapes and axis ratio distributions: Comparisons between 2D video disdrometer and wind tunnel measurements. J. Atmos. Oceanic Technol., 26, 1427–1432. 133

Thurai, M., W. A. Petersen, and L. D. Carey, 2010: DSD characteristics of a cool-season tornadic storm using C-band polarimetric radar and two 2D-video disdrometers. Proc. Sixth European Conf. on Radar Meteorology and Hydrology, Sibiu, Romania, European Meteorological Society, 6 pp. [Available online at http://www.erad2010.org/pdf/oral/tuesday/radpol1/5_ERAD2010_0101.pdf.]

Thurai, M., W. A. Petersen, A. Tokay, C. Schultz, and P. Gatlin, 2011: Drop size distribution comparisons between PARSIVEL and 2-D video disdrometers. Adv. Geosci., 30, 3–9.

Twomey, S., 1974: Pollution and the planetary albedo. Atmos. Environ., 8, 1251–1256.

Ulbrich, C. W., and N. E. Miller, 2001: Experimental test of the effects of Z–R law variations on comparison of WSR-88D rainfall amounts with surface rain gauge and disdrometer data. Wea. Forecasting, 16, 369–375. van den Heever, S. C., and W. R. Cotton, 2007: Urban aerosol impacts on downwind convective storms. J. Appl. Meteor. Climatol., 46, 828–850. van den Heever, S. C., G. G. Carrió, W. R. Cotton, P. J. DeMott, and A. J. Prenni, 2006: Impacts of nucleating aerosol on Florida storms. Part I: Mesoscale simulations. J. Atmos. Sci., 63, 1752–1775. van Weverberg, K., A. M. Vogelmann, H. Morrison, and J. A. Milbrandt, 2012: Sensitivity of idealized squall line simulations to the level of complexity used in two-moment bulk microphysics schemes. Mon. Wea. Rev., 140, 1883–1907.

Verlinde, H., and W. R. Cotton, 1993: Fitting microphysical observations of nonsteady convective clouds to a numerical model: An application of the adjoint technique of data assimilation to a kinematic model. Mon. Wea. Rev., 121, 2776–2793.

Vincent, B. R., L. D. Carey, D. Schneider, K. Keeter, and R. Gonski, 2003: Using WSR-88D reflectivity data for the prediction of cloud-to-ground lightning: A central North Carolina study. Nat. Wea. Dig., 27, 35–44.

Vivekanandan, J., W. M. Adams, and V. N. Bringi, 1991: Rigorous approach to polarimetric radar modeling of hydrometeor distributions. J. Appl. Meteor., 30, 1053–1063.

Vivekanandan, J., R. Raghavan, and V. N. Bringi, 1993: Polarimetric radar modeling of mixtures of precipitation particles. IEEE Trans. Geosci. Remote Sens., 31, 1017–1030.

Vivekanandan, J., S. M. Ellis, R. Oye, D. S. Zrnic, A. V. Ryzhkov, and J. Straka, 1999: Cloud Microphysics Retrieval Using S-band Dual-Polarization Radar Measurements. Bull. Amer. Meteor. Soc., 80, 381–388.

Vulpiani, G., P. Tabary, J. Parent du Chatelet, and F. Marzano, 2008: Comparison of advanced radar polarimetric techniques for operational attenuation correction at C band. J. Atmos. Oceanic Technol., 25, 1118–1135.

Wacker, U., and A. Seifert, 2001: Evolution of rain water profiles resulting from pure sedimentation: Spectral vs. parameterized description. Atmos. Res., 58, 19–39.

134 Weisman, M. L., and J. B. Klemp, 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110, 504–520.

Weisman, M. L., and J. B. Klemp, 1984: The structure and classification of numerically simulated convective storms in directionally varying wind shears. Mon. Wea. Rev., 112, 2479–2498.

Wiens, K. C., 2005: Kinematic, microphysical, and electrical structure and evolution of thunderstorms during the Severe Thunderstorm Electrification and Precipitation Study (STEPS). Ph.D. thesis, Colorado State University, 295 pp.

Wiens, K. C., S. A. Rutledge, and S. A. Tessendorf, 2005: The 29 June 2000 supercell observed during steps. Part II: Lightning and charge structure. J. Atmos. Sci., 62, 4151–4177.

Williams, E. R., R. Zhang, and J. Rydock, 1991: Mixed-phase microphysics and cloud electrification. J. Atmos. Sci., 48, 2195–2203.

Williams, E. R., and Coauthors, 1999: The behavior of total lightning activity in severe Florida thunderstorms. Atmos. Res., 51, 245–264.

Wurman, J., D. Dowell, Y. Richardson, P. Markowski, E. Rasmussen, D. Burgess, L. Wicker, and H. Bluestein, 2012: The Second Verification of the Origins of Rotation in Tornadoes Experiment: VORTEX2. Bull. Amer. Meteor. Soc., 93, 1147–1170.

Xie, B., Q. Zhang, and Y. Wang, 2010: Observed characteristics of hail size in four regions in China during 1980–2005. J. Climate, 23, 4973–4982.

Yuter, S. E., D. E. Kingsmill, L. B. Nance, and M. Loffler-Mang, 2006: Observations of precipitation size and fall speed characteristics within coexisting rain and wet snow. J. Appl. Meteor., 45, 1450–1464.

Zhang, G., J. Vivekanandan, and E. Brandes, 2001: A method for estimating rain rate and drop size distribution from polarimetric radar measurements. IEEE Trans. Geosci. Remote Sens., 39, 830–840.

Zhang, G., S. Luchs, A. V. Ryzhkov, M. Xue, L. Ryzhkova, and Q. Cao, 2011: Winter precipitation microphysics characterized by polarimetric radar and video disdrometer observations in central Oklahoma. J. Appl. Meteor. Climatol., 50, 1558–1570.

Ziegler, C. L., and D. R. MacGorman, 1994: Observed lightning morphology relative to modeled space charge and electric field distributions in a tornadic storm. J. Atmos. Sci., 51, 833–851.

Zrnić, D. S., 1987: Three-body scattering produces precipitation signature of special diagnostic value. Radio Sci., 22, 76–86.

Zrnić, D. S., R. J. Doviak, G. Zhang, and A. V. Ryzhkov, 2010: Bias in differential reflectivity due to cross-coupling through the radiation patterns of polarimetric weather radars. J. Atmos. Oceanic Technol., 27, 1624–1637.

135