Understanding and Predicting the Impact of Outflow on Tropical Cyclone Intensification and Structure (“TCI-14” & “TCI-15”) An FY14-18 ONR Departmental Research Initiative Ronald J. Ferek and CDR Joel Feldmeier, 322MM 3 March 2015

Recent ONR TC Field Campaigns NRL P-3 with ELDORA radar & wind LIDAR NRL P-3 with ELDORA radar P-3 with ELDORA radar (TCS08) For more than 2 decades, ONR 6.1 investments have driven the nation’s basic research agenda on TC science. At recent peer review: John Snow, OU Dean and chair of NOAA’s SAB called ONR ‘the nation's main basic research program for tropical cyclones.” TCM: fundamental progress led to large reductions in track errors CBLAST: put intensity in reach for the first time (first realistic pressure/wind relationship in models); lead NOAA to establish HFIP TCS08: NRL P3 had Eldora (for 3D winds), GPS dropsondes, LIDAR for boundary layer winds, WC130 had dropsondes, SFMR. Learned a lot about genesis, life cycle, vertical fluxes; led to immediate improvements in models. COAMPS-TC today—first dynamical model to beat skill of statistical-dynamical forecast tools. Gives the ability to model the dynamically-driven storm changes rather than just average behavior. BUT: While it nails some storms, it busts badly on others; analysis of the BUSTS appear to be associated with processes affecting the outflow region C-130 dropsondes and in-situ measurements (ITOP), bouys and drifters: improved coupling to the ocean past experience has shown that our Methodology is robust--understand processes (theory and field expts), represent new understanding in models (State-of-the-art data assimilation and exploitation of quantitative remote sensing), integrate into prediction systems WC-130J (dropsondes/AXBTs/SFMR) Late 80’s-early 90’s: Tropical Cyclone Motion (TCM) Initiative Early 2000’s: CBLAST (surface processes) 2008: Tropical Cyclone Structure Experiment (TCS-08) (storm-scale processes) 2010: ONR Impact of Typhoons on the Ocean in the Pacific (ITOP)

Motivation These are average position and intensity errors from the National Hurricane Center from 1990-present. Track errors at 72-, 48- and 24-hours have decreased by 63% over the past 21 years, at a steadily rate of 3% per year. 72-hour forecast errors are now approximately equal to 24 –hour forecast errors 21 years ago. Since ~1993, global NWP models have improved in their ability to predict the large-scale environment through better resolution, physics, and data assimilation. Track errors have steadily declined in the past twenty years as global NWP models have improved in their ability to predict the large-scale environment (better resolution, physics, and data assimilation). We postulate that intensity is dependent both on outflow structure and evolution as well as other environmental aspects and inner-core processes which remain largely unknown. Since 2003 in WPAC, there has been little improvement in TC track prediction Since 1990 in WPAC, there has been very little improvement in TC intensity prediction 2009 New USPACOM “Goals for Tropical Cyclone Forecasting” - 50% reduction in errors 2012 ONR held a TC science workshop to identify next set of basic research questions - Problem: Upper-level outflow associated with tropical cyclones may be the key link between the environment and TC intensification - processes remain largely unexplored

Hurricane Structure and Recent Research Programs Unexplored: Outflow structure, intensity, and variability, and relationships with hurricane intensity and structure ONR TCS-08 NSF TPARC NOAA IFEX NASA GRIP NSF PREDICT Secondary [in-up-out] Circulation Primary [swirling] Wind ONR CBLAST ONR TCS-08 ONR ITOP Definition of primary and secondary circulations, inflow and outflow regions and focus of previous observational experiments on TC boundary layer. Previous process studies limited by the available observing platforms. Outflow region remains unexplored. Beyond the reach of our observing systems. There is theory, but we can’t verify or test it against obs. The new capability of observing the outflow vertical structure with dropsondes released from the NASA Global Hawk and WB-57 raise the possibility of parameterizing outflow vertical structure and associated mixing processes in much the same way as prior experiments have done in the TC boundary layer, resulting in fundamentally new knowledge and significant impact on TC model parameterizations. NSF RAINEX Image courtesy of NASA

Outflow & Intensification Pre-Rapid Intensification Upper-Level Jet Outflow & Intensification Typhoon Roke Pre-Rapid Intensification 00 UTC 19 Sep 2011 Intensity = 65 kt 150-300 mb Divergence Roke Outflow Winds: 100-250 mb, 251-350 mb, 351-500 mb Roke Outflow directed equatorward No interaction between outflow and approaching upper-level jet Weak upper-level divergence Weak typhoon

Outflow & Intensification Rapid Intensification Upper-Level Jet Outflow & Intensification Typhoon Roke Rapid Intensification 00 UTC 20 Sep 2011 (+24h) Intensity = 115 kt Outflow 150-300 mb Divergence Roke Roke Winds: 100-250 mb, 251-350 mb, 351-500 mb This is an extreme but not uncommon case And it’s hard to imagine that some relatively subtle change in surface processes (drag-wave spectra; warm eddy/increase heat and evaporation, etc) could drive such an energetic increase Not surprisingly, Models did not handle this RI well TC-Jet interaction can trigger weather impacts far downstream. Outflow shifts poleward Outflow couples with midlatitude jet Upper-level divergence triples Roke underwent Rapid Intensification, increased intensity by 50 kts in 24 hours

Model Performance during Typhoon Roke (18W) Difficulty Forecasting Rapid Intensification (From J. Doyle, NRL) Tokyo After TY Roke remained weak (and moved slowly) for days, it underwent rapid intensification (RI). Models (including COAMPS-TC) failed to capture this RI.

Hurricane Outflow Theory Secondary Circulation 15 10 5 50 100+ Radius (nm) Height (km) Inflow IN Eyewall Ascent UP Outflow OUT Eye Descent Fundamentals: Outflow as part of the secondary circulation – the “out” of the “in-up-out” (MANY UNKNOWNS AND THE LEAST UNDERSTOOD part of the cycle) Impact of the secondary circulation on the primary circulation. Important to understanding changes to the secondary circulation --changes to one branch will affect the other branches, and will subsequently affect TC intensity Secondary circulation impacts the primary circulation (and therefore TC intensity) through variations in gradient wind balance. Key factors that determine radial inflow: Surface and boundary layer characteristics Conservation of angular momentum, mass and kinetic energy Changes in one branch of the circulation will impact the other branches through continuity (must balance)

Key Science Issues and Approach Understand the coupling between all the branches of the secondary circulation (and the relationship of this coupling to intensity changes) Upper-level outflow changes lead to increased convection and intensification (Active Outflow) Upper-level outflow changes result from increased convection (Passive Outflow) What are the relative roles of the TC vortex and the environment? Observe the Evolution of outflow in relation to the environment Interaction between the outflow and the upper level environment (morphology, vertical structure, phasing, depth and strength of the outflow) Evolution of outflow channels and associated rapid intensification or weakening Are turbulent-scale mixing and shear instability processes important? Diagnose the dynamical balance within the outflow jet Employ state-of-the-art Numerical Modeling Quantify the impact of observations on TC intensity, outflow etc. Diagnose the initial state sensitivity using adjoints and ensembles. Quantify the predictability of outflow jets and TC intensity change.

Hurricane Outflow Modeling Challenges 1) Vertical Resolution There is large sensitivity of the intensity and structure to vertical resolution, particularly in the outflow layer (Zhang and Wang 2003) Typical vertical layer distribution will poorly resolve outflow layer typical level distribution 2) Model Spinup Two consistent modeling challenges related to the outflow: 1. Vertical resolution: a. the typical resolution has few layers in the outflow region and as a result the outflow is poorly resolved. Don’t know thickness or the magnitude of the flows. Satt winds give ambiguous (rel. to altitude) horizontal winds. b. the TC intensity and structure are both very sensitive to the vertical resolution. 2. Model spinup: Difficulty with accurately handling intensities early in the model runs (spin up or spin down). Initial bogus doesn’t even contain the outflow (don’t know what reasonable quantities are) “storm is poorly connected to the larger environment” … or the resolution) to accurately represent the “exhaust flow” May be other subtle things we don’t understand; interplay between processes. Nearly all dynamical models show a spin-up or spin-down of the initial vortex. The initial imbalance is likely due to a poor specification of the secondary circulation, particularly the outflow. The outflow is notoriously difficult for models to represent in the initial conditions (and hence never bogused)

NRL TC Data Assimilation and Modeling Tools (see J. Doyle, J. Moskitas and S. Chen talks in Session 5b this afternoon) COAMPS-TC COAMPS-TC Nested Adjoint COAMPS-TC EnKF COAMPS-TC Observation Impact

HS3 Observations of Leslie’s Outflow 150 mb Leslie Center X This slide shows an example of the capability of the Global Hawk (GH) to observe the upper level outflow jet structure and evolution, as exhibited by these wind observations (depicted as wind barbs where a triangle is 50 kt, a single full line 10 kt and a half line 5 kt) at 150-mb for Leslie. The ellipse shows the sondes used in the following analyses. The inset shows a visible satellite cloud image at the time of the wind observations (1013 – 1111 GMT). The thick white arrow indicates the wind direction in the outflow jet and the thin blue arrow the direction of travel of the GH. Cross Section 6 sondes

HS3 Observations of Leslie’s Outflow 7 Sep 2012 1041-1111Z Black, Red, Blue and Pink lines: Global Hawk observed wind speed and temperature profiles along jet maximum from dropsondes Green line: COAMPS-TC model wind speed profile Red line: Satellite wind speed vertical average Solid black: Tropopause Dashed: Cirrus top / jet max Dotted: Cirrus cloud base Yellow shading: Cloud Physics Lidar (CPL) domain Four dropsondes along the axis of Leslie’s outflow jet showing similar narrow wind maxima just below the tropopause height, along with layers of nearly constant wind with height just above the tropopause, below the wind max and below the cirrus cloud deck suggesting shear-induced mixing above the wind max and evaporation-driven mixing below the cirrus cloud deck. Change in temperature lapse rate from adiabatic slope to constant temperature vs height indicates tropopause transition to strong stability. The vertical red line indicates the mean value of the CIMSS 100- 250 mb winds along this segment of the outflow jet. The yellow shading indicates the region of the CPL profiles, showing details of the cirrus cloud structure in next slide. HS3 dropsondes reveal unprecedented detail in depiction of outflow jet Sharp shear zone just above the sloping tropopause (~14 km) and below outflow jet Top of outflow jet coincident with top of cirrus deck from CPL Detailed cirrus fine structure suggestive of multiple turbulent mixing mechanisms Vertical section based on 5 dropsondes from Leslie. VERY USEFUL FOR VALIDATING MODEL-DERIVED FIELDS. The cross section highlights a sharp shear zone just above the sloping tropopause and below the jet max with the top of the outflow jet coincident with the top of the cirrus deck as diagnosed from the Cloud Physics Lidar (lower panel) onboard the global hawk (AV-6). The detailed cirrus fine struCPLlent mixing mechanisms (Courtesy of Pete Black)

Impact of HS3 Dropsondes for Nadine Track Error (nm) Intensity: Max. Wind Error (kts) No drops No drops HS3 drops HS3 drops Bias (dash) Intensity: Min. SLP Error (hPa) No drops Dropsonde impact experiments performed for 19-28 Sep. (3 flights) Blue, with HS3 drops Red, No drops with synthetics COAMPS-TC Intensity and Track skill are improved greatly through assimilation of HS3 Drops HS3 drops Bias (dash) (Doyle et al. 2013)

TCI-14 Pilot Project: Hurricane Gonzalo; HDSS, HIRAD, HIWRAP HIRAD image courtesy Dan Cecil (Dan and Pete Black spoke earlier about HIRAD and HDSS)

Observational Strategy (collab. with SHOUT and HRD) 15 10 5 radius (nm) Height (km) Outflow 300 600 100 Air Force WC-130J or NOAA P-3: SFMR: Surface winds / intensity Radar: Precipitation structure GPS Dropsondes: Vertical Structure- wind, temperature, humidity SFMR Radar High Alt. A/C WB-57 HIRAD Y.E.S. Dropsondes AV-6 NCAR Dropsondes HAMSR HIWRAP Two A/C can investigate all branches of the secondary circulation, including the outflow for the first time Can really map the outflow channels in 3D detail Really covering all branches of the secondary circulation while the C130 and P-3 get the primary circulation and intensity This is the most complete sensor pkg ever assembled. Strategy: WC-130J or P-3 to monitor the TC intensity and structure Global Hawk or WB-57 to observe the outflow and environment

Summary We hypothesize that hurricane outflow is the key to unraveling the complex nature of hurricane intensity and structure Upper-level outflow is the only TC component that has not been systematically observed or studied We now have the observing capability to: investigate all branches of the secondary circulation at the same time evaluate model deficiencies improve understanding of the dynamic processes and represent them in models verify and validate model improvements Leverage the opportunity to deploy the NASA Global Hawk and WB-57 and the Hurricane Hunter A/C to observe hurricane intensity, structure and outflow interaction Employ new HDSS (high definition dropsonde system) to obtain detailed observations Expected Payoff: This effort will have two primary benefits: 1. First ever comprehensive observations of the upper-level outflow structure will provide a unique dataset that will be critical to improving our scientific understanding of the outflow evolution as well as the associated interactions with TC intensification and structure change. 2. Improved understanding of the relationship between upper-level outflow and the TC vortex, along with new techniques to assimilate field campaign observations using emerging tropical cyclone modeling systems will lead to much improved TC intensity and structure predictions (particularly for rapid intensification and rapid weakening) One of the last remINING UNEXPLORED ASPECTS OF tc’S

Science Team & Proposed Topics Kerry Emanuel, MIT Pat Harr, NPS Russ Elsberry, UCCS Sharan Majumdar, UMiami John Molinari, SUNYA Michael Bell, UH Jim Doyle, NRL Pete Black, NRL Chris Velden, UWisc Da-Lin Zhang, UMd Xuguang Wang, UOk Zhaoxia Pu, UUtah Greg Tripoli, UWisc Jason Dunnion, UMiami Beth Sanabia, USNA Collaborative efforts: Alex Reinecke, NRL Lee Harrison, SUNYA Mark Beaubien, Y.E.S. Dan Cecil, NASA Marshall Hurricane Outflow Criticality TC Intensity and Structure Changes due to Upper-Level Outflow and Environmental Interactions Environmental sensitivity of tropical cyclone outflow Outflow layer dynamics and thermodynamics Thermodynamic Constraints on TC Intensity and Structure Impact of Outflow on TC Intensification and Structure: An Observational and Numerical Modeling Perspective Coupling of TC Outflow Vents with the Environment Impact of Upper-Level Processes on TC Intensity and Structure Ensemble-based DA and multi-model ensemble simulation Impacts of Outflow on TC Formation, RI and Structure Changes An observational and numerical investigation of energy exchange between a TC and its environment at the outflow level From Ocean to Outflow: Understanding TC Circulations and Intensification Improved TC Prediction Through Unmanned Obs. Systems High Definition Dropsonde System HIRAD