Convective-scale Downdrafts in the Principal Rainband of Hurricane Katrina (2005) Anthony C. Didlake, Jr. COGS Seminar UW, Dept. Atmos Sci., Seattle, November 6, 2008

Idealized structure of a tropical cyclone Inner and Outer eyewalls Inner and Outer eyewalls Stationary Band Complex (SBC) Stationary Band Complex (SBC) principal bandprincipal band secondary bandssecondary bands Willoughby 1988 upwind downwind

Overview Background and Motivation Background and Motivation Description of RAINEX and dataset Description of RAINEX and dataset Methodology: Convective separation and cross sections Methodology: Convective separation and cross sections Characteristics of downdrafts within principal rainband Characteristics of downdrafts within principal rainband Forcing mechanisms and immediate effects of downdrafts Forcing mechanisms and immediate effects of downdrafts Possible impacts on larger tropical cyclone Possible impacts on larger tropical cyclone Summary and conclusions Summary and conclusions

Background and Motivation Dynamic role of principal rainband in the larger storm remains uncertain Dynamic role of principal rainband in the larger storm remains uncertain Several modeling studies suggest the principal rainband impacts the storm intensity Several modeling studies suggest the principal rainband impacts the storm intensity PV generation and inward advectionPV generation and inward advection Inhibiting inflow of warm, moist airInhibiting inflow of warm, moist air Formation of secondary eyewall via vortex- Rossby wave dynamicsFormation of secondary eyewall via vortex- Rossby wave dynamics Important to understand structure and dynamics of principal rainband, so that we may better address the difficulties in forecasting tropical cyclone intensity Important to understand structure and dynamics of principal rainband, so that we may better address the difficulties in forecasting tropical cyclone intensity

Houze et al. 2006, 2007

Model of Principal Rainband Convective cells embedded in stratiform rain Convective cells embedded in stratiform rain Overturning updraft, two downdrafts Overturning updraft, two downdrafts Hence and Houze 2008

Downdrafts in the Principal Rainband Low-level downdraft (LLD) Low-level downdraft (LLD) Inner-edge downdraft (IED) Inner-edge downdraft (IED) Forcing mechanisms, immediate effects, possible impacts on larger storm? Forcing mechanisms, immediate effects, possible impacts on larger storm? IED LLD

Downdrafts in ordinary convection Convective-scale saturated downdraft forced by precipitation drag Convective-scale saturated downdraft forced by precipitation drag Mesoscale downdraft due to evaporative cooling Mesoscale downdraft due to evaporative cooling Convective-scale downdraft forced by buoyancy pressure gradient force (BPGF) field Convective-scale downdraft forced by buoyancy pressure gradient force (BPGF) field Zipser 1977 Palmén and Newton 1969, Biggerstaff and Houze 1993, Yuter and Houze 1995

Hurricane Katrina (2005)

ELDORA data Reflectivity at 2 km

Convective/stratiform separation Based on local gradients in reflectivity Based on local gradients in reflectivity Similar to Steiner et al. 1995, TRMM satellite data classification Similar to Steiner et al. 1995, TRMM satellite data classification Convective Stratiform Weak echo No echo

2D frequency distributions Reflectivity data in % of height total Convective pixels Stratiform pixels Convective pixels Stratiform pixels

2D frequency distributions Vertical velocity data in % of height total Convective pixels Stratiform pixels Convective pixels Stratiform pixels

Rainband cross sections Radial cross sections at regular angular intervals Radial cross sections at regular angular intervals 0.375°  109 cross sections0.375°  109 cross sections Cross section coordinates based on classification Cross section coordinates based on classification

Average vertical velocity Updrafts (m s -1 ) Downdrafts (m s -1 ) IED LLDReflectivity (dBZ) as black contours

LLD analysis Vertical velocity (m s -1 ) at 42.2° Reflectivity (dBZ) as black contours Located in lower levels Located in lower levels Embedded in heavy precipitation Embedded in heavy precipitation

LLD forcing mechanism Average downdrafts (m s -1 ) Reflectivity (dBZ) as black contours Located in lower levels Located in lower levels Embedded in heavy precipitation Embedded in heavy precipitation Zipser’s “Convective-scale saturated downdraft” Zipser’s “Convective-scale saturated downdraft” Forced down by precipitation drag Forced down by precipitation drag Attains negative buoyancy from continuous evaporative cooling Attains negative buoyancy from continuous evaporative cooling

IED investigation area: 8.5 km  12.5 km IED investigation area: 8.5 km  12.5 km IED analysis Average downdrafts (m s -1 )

IED analysis: in kg s -1 Downward vertical mass flux 2D distribution

Condition: 4.5 km-IED ≤ 3 m s -1 or > 3 m s -1 Condition: 4.5 km-IED ≤ 3 m s -1 or > 3 m s -1 Weak mid-level IED comes with weaker low-level IED, while strong mid-level IED comes with stronger low-level IED Weak mid-level IED comes with weaker low-level IED, while strong mid-level IED comes with stronger low-level IED IED analysis: Conditional probability distribution of IED speeds at 1 km

IED analysis: Vertical velocity at 4 km Intermittent pattern of convective-scale updraft and downdraft cores Intermittent pattern of convective-scale updraft and downdraft cores

Physical relationship between IEDs and updrafts Physical relationship between IEDs and updrafts IED analysis: Autocorrelation of vertical velocity, Lag = 4 (≈ 4.5 km)

IED forcing mechanism Reflectivity (dBZ) at 35.8° Overlaid by in-plane wind vectors Originates above the melting level, outside of heavy precipitation Originates above the melting level, outside of heavy precipitation Occurs on the convective scale, rather than mesoscale Occurs on the convective scale, rather than mesoscale

IED forcing mechanism Vertical velocity (m s -1 ) at 35.8° Reflectivity (dBZ) as black contours Originates above the melting level, outside of heavy precipitation Originates above the melting level, outside of heavy precipitation Occurs on the convective scale, rather than mesoscale Occurs on the convective scale, rather than mesoscale Initially forced by the BPGF created by the updraft Initially forced by the BPGF created by the updraft

IED forcing mechanisms Reflectivity (dBZ) at 35.8° Initially forced by the BPGF created by the updraft Initially forced by the BPGF created by the updraft 2-step process!

IED forcing mechanisms Reflectivity (dBZ) at 35.8° 2-step process! Initially forced by the BPGF created by the updraft Initially forced by the BPGF created by the updraft Attains negative buoyancy by evaporating heavy precipitation of rainband Attains negative buoyancy by evaporating heavy precipitation of rainband

IED effects: Sharp inner-edge reflectivity gradient

IED effects: Sharp inner-edge reflectivity gradient Reflectivity (dBZ) at 35.8°

IED effects: Low-level wind maximum (LLWM) Tangential wind speed (m s -1 ) at 35.8°

Tangential wind speed Vertical velocity Vertical vorticity Divergence

Composite tangential wind speed from Hurricane Floyd (1981) Increased inward flux of angular momentum LLWM lies in radial inflow LLWM lies in radial inflow Increased angular momentum results in stronger vortex Increased angular momentum results in stronger vortex Possible impacts: Tangential wind speed (m s -1 ) at 35.8° Barnes et al. 1983

Conceptual model of rainband cross section

Commonly observed features of principal rainband Upwind end consists of newer, robust convective cells Upwind end consists of newer, robust convective cells Downwind end consists of older cells collapsing into stratiform precipitation Downwind end consists of older cells collapsing into stratiform precipitation Principal rainband is often stationary relative to the storm center Principal rainband is often stationary relative to the storm center Barnes et al. 1983Hence and Houze 2008

Growth and sustenance of principal rainband Possible impacts: Area of divergence near surface under LLD Area of divergence near surface under LLD Preferred region of convergence on upwind side of LLD core Preferred region of convergence on upwind side of LLD core Growth of updraft on upwind end of rainband Growth of updraft on upwind end of rainband Divergence Vertical velocity at 42.2° Plan view LLD Background flow Convergence

Growth and sustenance of principal rainband Possible impacts: Tropical storm Ophelia (2005) Tropical storm Ophelia (2005) Operational radar from Melbourne, FL Operational radar from Melbourne, FL Discrete propagation of vertical velocity cores, rainband cells Discrete propagation of vertical velocity cores, rainband cells Stationary rainband relative to storm center Stationary rainband relative to storm center Radar loop

Conceptual model of rainband at 2 km

Conclusions Principal rainband contains two repeatable convective-scale downdrafts Principal rainband contains two repeatable convective-scale downdrafts Low-level downdraft is forced by precipitation drag beneath heavy precipitation Low-level downdraft is forced by precipitation drag beneath heavy precipitation Inner-edge downdraft is initially forced by pressure perturbations created by nearby buoyant updrafts, then evaporative cooling Inner-edge downdraft is initially forced by pressure perturbations created by nearby buoyant updrafts, then evaporative cooling Vorticity dynamics of updraft and IED create a low-level wind maximum that leads to increased angular momentum of storm Vorticity dynamics of updraft and IED create a low-level wind maximum that leads to increased angular momentum of storm Interaction between updraft and two downdrafts leads to growing and sustaining convection of principal rainband Interaction between updraft and two downdrafts leads to growing and sustaining convection of principal rainband Convective-scale features allow principal rainband to continue its impact on the overall storm Convective-scale features allow principal rainband to continue its impact on the overall storm

Future Work Analyze convective-scale structures in high-resolution model output from RAINEX Analyze convective-scale structures in high-resolution model output from RAINEX Investigate outer rainbands and compare to inner core of storm Investigate outer rainbands and compare to inner core of storm

Acknowledgments Bob Houze Bob Houze Deanna Hence, Stacy Brodzik Deanna Hence, Stacy Brodzik Brad Smull, Tomislav Maric, Jian Yuan, Mesoscale Group Brad Smull, Tomislav Maric, Jian Yuan, Mesoscale Group Michael Bell, Sandra Yuter Michael Bell, Sandra Yuter Beth Tully Beth Tully Atmos Grad 2006 Atmos Grad 2006

Extra Slides

Convective/stratiform classification Technique used in Steiner et al. 1995, Yuter and Houze 1997, Yuter et al Technique used in Steiner et al. 1995, Yuter and Houze 1997, Yuter et al Algorithm separates convective regions from stratiform regions by comparing local reflectivity to background reflectivity Algorithm separates convective regions from stratiform regions by comparing local reflectivity to background reflectivity Tuning of algorithm required to recognize convective regions; the rest is designated as stratiform Tuning of algorithm required to recognize convective regions; the rest is designated as stratiform

Convective center if: Convective center if: Z  Z tiZ  Z ti Z-Z bg  Z cc (Z bg )Z-Z bg  Z cc (Z bg ) Classified convective within R(Z bg ) from convective center, remaining is classified stratiform (unless Z < Z we ) Classified convective within R(Z bg ) from convective center, remaining is classified stratiform (unless Z < Z we ) Z ti = 45 dbZ; Z we = 20 dbZ; R = (Z bg -20); R bg = 11 km; a=9; b=45 Convective/stratiform classification

4 km reflectivity

2 km reflectivity

2 km vertical velocity

6 km vertical velocity

Average downdrafts for upwind half

Statistical significance testing Two-sided Student’s t statistic Two-sided Student’s t statistic Significance level of 95% Significance level of 95% Null hypothesis that true autocorrelation is zero Null hypothesis that true autocorrelation is zero Number of independent samples determined by formula of Bretherton et al. (1999) Number of independent samples determined by formula of Bretherton et al. (1999)

Figure 16

Frequency of “strong” vertical velocities Updrafts ≥ 3.0 m s -1 Downdrafts ≥ 1.5 m s -1 IED LLDReflectivity (dBZ) as colored contours