ABSTRACT GORDON, BRIANA JULIA. The Sensitivity of Tropical Cyclone Forecast Simulations in the WRF Model to Initial Conditions. (Under the direction of Dr. Gary M. Lackmann). Accurate dynamical forecasts of tropical cyclones are critically important for the prediction of hurricane track and intensity, particularly at times when tropical cyclones (TCs) are threatening to make landfall in populated areas. Over the last two decades, advancements in both conceptual understanding and computing technology have led to increased skill for TC track forecasts; however, intensity forecast skill has seen little improvement over the same time period (Rogers et al. 2006). A major challenge to producing a skillful forecast of TC intensity in a dynamical numerical weather prediction (NWP) model is providing the model with a quality initial condition (IC) (Liu et al. 2000; Hsiao et al. 2010). The aim of this study is to create a more accurate, yet operationally feasible method of producing a TC initial condition in the WRF model. Specifically, this work is done with HURNC, a real-time dynamical hurricane prediction model, which is a version of WRF-ARW that is run twice daily during hurricane season. Due to a lack of available observations over the oceans, where TCs develop and evolve, the GFS 0.5° data, which are used to initialize HURNC, often yields analyzed tropical cyclones with ill-defined vertical structure (Reeder at al. 1991) and underestimated TC intensity (Kurihara et al. 1993). Two 96-hour model simulations are performed for the same initialization time during each of three storms with different initial intensity and locations: Hurricane Ike (2008), Tropical Storm Erika (2009), and Hurricane Earl (2010). To initialize each storm, the control simulation uses the standard GFS 0.5° data and the experimental simulation uses the GFDL bogus vortex (BV) merged with the GFS background. Two additional simulations are performed for Hurricane Earl: 1) an experimental run using a hybrid 3DVAR/EnKF DA (HDA) system, and 2) the same hybrid system with Air Force aircraft reconnaissance environmental dropsonde data assimilated as well (HDA-OBS). It is hypothesized that the hybrid DA ICs will have more forecast skill than will the BV IC, which is artificial in structure and has been shown to exhibit excessive vortex resiliency (Reasor et al. 2004; Mallen et al. 2005) due to a robust vertical eyewall and inner core structure. Furthermore, it is expected that the HDA-OBS IC will have more forecast skill than will the HDA IC, due to the in-situ dropsonde data assimilated into the former model initialization. While the bogus vortex does improve the intensity of the IC for Ike, Erika, and Earl, it does not always improve the forecast, especially at longer lead-times (48 to 96 hours). The BV has a vertically-oriented, strong potential vorticity tower, with a very intense upper-level PV maximum, which persists as a storm feature long after the initial time; this high-altitude PV maximum is not found in the GFS IC. In the additional experiments performed with Earl, the hybrid DA ICs result in large intensity errors at early lead times (0 to 42 hours), during which time both the HDA IC and HDA-OBS IC have greater intensity error than does the BV IC. Also, the HDA- OBS simulation does not display an increase in skill over the HDA simulation, probably due to the lack of near-core data being assimilated in this study as well as to the lack of DA cycling and updating of the model forecast. Conclusions about the effectiveness of the hybrid DA initial condition cannot be drawn based solely on the limited results of this study. However, it is expected that future work using a hybrid DA IC in cycling mode may provide the forecast improvement for the HURNC IC that was sought in this study. The Sensitivity of Tropical Cyclone Forecast Simulations in the WRF Model to Initial Conditions by Briana Julia Gordon A thesis submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the degree of Master of Science Marine, Earth, and Atmospheric Sciences Raleigh, North Carolina 2011 APPROVED BY: _______________________________ ______________________________ Gary M. Lackmann Dr. Lian Xie (Chair of Advisory Committee) ________________________________ Dr. Anantha Aiyyer BIOGRAPHY Briana was born on the upper east side of Manhattan, NY on 4 December 1981, and graduated from Nyack High School in 1999. She first became interested in the weather when Hurricane Gloria came barreling directly through New York City in the Fall of 1985. Her first memory is of unwillingly taking shelter in the hallway of her apartment, where there were no windows. By 1991, Briana was living in Nyack, NY on the Hudson River, and was fascinated by the landfall of Hurricane Bob. Against her mother’s wishes, she put on a poncho and went out on the balcony of her fourth-floor apartment to experience the storm and watch the river overtake her parking lot. Since then, Briana has been known to stand outside in any thunderstorm she can find, and is fascinated by the processes that create these impressive displays of atmospheric science. Briana enjoyed and excelled at earth science, math, and physics in high school, and decided to major in meteorology at Cornell University, where she was fortunate to have the opportunity to do undergraduate research through the Cornell Presidential Research Scholars program. Her undergraduate thesis, conducted under the supervision of her advisor, Dr. Stephen J. Colucci, was titled ―The Evaluation of NCEP Short-Range Ensemble Forecasts for North American Cyclones.‖ Briana completed her B.S. in Atmospheric Science in 2003, graduating magna cum laude with distinction in research. She decided to attend graduate school at the University of Washington, in Seattle, WA, where she obtained her M.Ed. in Curriculum and Instruction, with a concentration in Mathematics Education in 2005 under the advisement of Dr. Elham Kazemi. While working on her degree, Briana worked for The Princeton Review as an academic private ii tutor, teacher, and eventually teacher trainer for SAT, ACT, GRE, and GMAT classes. Upon graduation from the University of Washington, Briana moved back to New York, where she obtained a full time position with The Princeton Review as a book editor for their publishing division. She continued tutoring in her ―spare‖ time as a master tutor for the NYC Tutoring division. By the middle of 2006, Briana was promoted to Senior Editor of the Test Prep and Research & Development books departments. While she loved this work, she missed working with atmospheric science, and decided to go back to graduate school at North Carolina State University in 2008 to pursue her M.S. of Atmospheric Science. She has been fortunate to work under the advisement of Dr. Gary M. Lackmann, who has allowed her the flexibility to design and work on research she finds interesting and challenging. iii ACKNOWLEDGMENTS Support for this research was received from the Renaissance Computing Institute (RENCI) and from the NOAA Collaborative Science, Technology, and Applied Research (CSTAR) program (Grant # NA10NWS4680007) awarded to North Carolina State University. Much of the data used in this study were provided by the National Centers for Environmental Prediction (NCEP) and the Geophysical Fluid Dynamics Lab (GFDL). Thank you as well to the Air Force for the use of their aircraft reconnaissance data and to the National Hurricane Center for their HWND analyses and detailed storm reports. The National Center for Atmospheric Research (NCAR) is also acknowledged for the development and sharing of the WRF model, as is Ryan Torn of the University of Albany for the original WRF to GEMPAK converter. Thank you to Dr. Robert Hart of Florida State University for sharing his GrADS software, which was used to create the merged bogus vortex initial (BV) condition. I am also grateful to RENCI for the use of their computing platforms and to Dr. Brian Etherton for the use of his hybrid DA IC system. I would like to express my deep gratitude to my committee members, Drs. Gary Lackmann, Anantha Aiyyer, and Brian Etherton for their help and guidance through this process, as well as to Dr. Lian Xie for volunteering to be a committee substitute. I am especially grateful to Dr. Brian Etherton for his help troubleshooting cryptic WRF errors and for his patience to help me learn data assimilation. I would particularly like to thank my advisor, Dr. Gary Lackmann, not only for always helping, guiding, and challenging me to be a better scientist, but also for providing support, encouragement, and friendship iv throughout this process. I am lucky to have landed in the Forecast lab, and would like to thank my labmates, Kelly Mahoney, Adam Baker, Kevin Hill, and Megan Gentry for their help, friendship, support, and many moments of stomach-ache-inducing laughter that kept me sane throughout my time in the lab. I would also like to thank all my friends and fellow graduate students from the NCSU MEAS department, who provided help, empathy, and welcome sanity breaks amidst the hard work. There are too many to list here, but I am grateful for your friendship. Lastly, and most importantly, I’d like to thank all of my friends and family, particularly my mom, Anna, my dad, Barry, and my stepmom, Kat, for their tireless support, encouragement, devotion, help, love, acceptance, pride, and laughter over the years. My parents have always raised me to love learning, pursue knowledge, and follow my dreams. I am incredibly fortunate and grateful that they have been supportive of every choice that I have ever made, both personally and professionally. v TABLE OF CONTENTS List of Figures………………………………………………………………………… viii 1.