Hydrometeorological Prediction Center Tropical Cyclone Responsibilities and Tropical Cyclone QPF

1
Hydrometeorological Prediction Center Tropical
Cyclone Responsibilities and Tropical Cyclone QPF

• Michael J. BrennanScience and Operations Officer
• Hydrometeorological Prediction Center
• NOAA in the Carolinas Hurricane Conference 6
  May 2008

2
Outline

• History of HPCs Tropical Cyclone (TC)
  Responsibilities
• Synoptic Reasoning and Track Forecasting
• Tropical Cyclone QPF
• Inland Advisories
• NHC COOP

3
History of HPCs Tropical Cyclone Program
Responsibility

• HPC (and its predecessor, NMC) have provided
  track forecast guidance to NHC for at least the
  last 30 years
• Initially done because NMC had access to much
  more global model guidance than NHC
• NHC COOP program
• HPC/NMC has been the official NHC backup since at
  least 1978
• Overhauled and modernized in 2002-2003
• HPC began providing the rainfall statement for
  NHCs public advisories in 2005

4
Synoptic Reasoning and Track Forecasting

• HPC provides track forecast guidance (points) to
  NHC for TCs west of 60W
• HPCs synoptic reasoning provided on Hurricane
  Hotline conference call
• Day 6 7 TC positions for HPC medium range
  charts coordinated with NHC in daily Noon
  conference call

Day 6 Forecast of Ernesto
HPC Medium Range Prog Valid 12Z 31 August 2006
5
Synoptic Reasoning and Track Forecasting

• HPC takes advantage of expertise in large-scale
  pattern diagnosis to compliment NHC track
  forecast
• HPC track forecast not constrained by continuity
  and can react more to run-to-run changes in model
  guidance
• Ensures HPC has working forecast in place for all
  TCs west of 60W in case backup for NHC needed

6
Tropical Cyclone Rainfall
7
U.S. Weather Fatalities
8
TC Rainfall Major Contributor to Annual
Precipitation in Southeastern U.S.

• 2005 Annual Rainfall (mm)

2005 TC Rainfall (mm)
10-20 of annual rainfall from TCs
Images Courtesy Frank Marks (AOML/HRD)
9
Factors Impacting Rainfall Distribution in
Landfalling TCs

• Storm size (positive) the larger the TC, the
  more it rains at any given spot
• Storm track (location)
• Diurnal cycle core rainfall overnight, outer
  band rainfall during the day
• Topography enhances rainfall in upslope areas,
  but decreases past the spine of the mountains
• Wind shear (negative) leads to a quicker
  drop-off in rainfall for inland TCs
• Interaction with nearby features (troughs,
  fronts/jets), extratropical transition can
  greatly modify rainfall distribution

10
Depth of Upper Trough Causing Recurvature Key

• Storms which drop most of their rain left of
  track recurve due to significant upper troughs in
  the westerlies
• Rainfall streaks out well north of the system due
  to jet streaks moving around the upper trough and
  frontogenesis at the troughs leading edge
• Trough transitions from positive to negative tilt
  during interaction with TC
• Storms where most rain occurs to the right of
  track are steered predominantly by shear lines or
  through a break in the subtropical ridge
• Rainfall tends to be concentrated near and right
  of track

Atallah et al. (2007) Mon. Wea. Rev.
11
Bertha (1996)
250-mb height and wind
Bertha moving around break in subtropical
ridge Upper trough and jet well northwest of
Berthas track
00Z 13 July
15Z 13 July
Norman W. Wes Junker
12
Floyd (1999)
250-mb height and wind
Floyd interacts with positively-tilted upper
trough more directly Precipitation forms in
pronounced jet entrance region
00Z 16 Sep
12Z 16 Sep
Norman W. Wes Junker
13
Bertha (1996) vs. Floyd (1999)
14
Predecessor Rainfall Events

• Coherent area of rain displaced poleward of TC
• Maximum rainfall rates exceeded 100 mm ( 4 in.)
  in 24 h
• Moisture transport from TC toward PRE

47 PREs associated with 21 TCs were identified
during the period from 1998-2006 (2 PREs per
TC) 1/3 of all U.S. landfalling TCs produced at
least one PRE Five cases where TC did not make
U.S. landfall
PRE
Slides courtesy Matt Cote Lance Bosart SUNY
Albany
Gaston
1800 UTC 040830 WSI NOWRAD Radar Mosaic
15
Southeast Recurvatures

• 7/11 (64) produced at least one PRE
• 16 PREs from 7 TCs
• Influential geographical features
• - Gulf of Mexico - Atlantic Ocean
  - Appalachians

• Approximate point of PRE formation

Slides courtesy Matt Cote Lance Bosart SUNY
Albany
16
PRE Statistics
Separation Distance 1086 482 km
Median 935 km Event Duration 14 7 h
Median 12 h Time
Lag 45 29 h Median 36
h
Bosart and Carr (1978) conceptual model of
antecedent rainfall from Agnes (1972)
Slides courtesy Matt Cote Lance Bosart SUNY
Albany
17
Track-Relative Location of PREs
PRE Locations Relative to TC Track 19982006
Potential for flooding in areas not directly
impacted by TC rainfall
Potential for excessive flooding beginning before
arrival of TC rainfall
26
12
9
Slides courtesy Matt Cote Lance Bosart SUNY
Albany
18
Southeast Recurvatures CompositeTime of PRE
InitiationMatt Cote Lance Bosart SUNY Albany
700 hPa Ht (dam) and UVM (µb s-1)
925 hPa Ht (dam), ?e (K), and 200 hPa wind speeds
(m s-1)

• Significant midlevel trough with weak UVM well
  poleward of TC
• Deep meridional flow transports tropical
  moisture up East Coast

• PRE forms
  - in right-entrance region of
  intensifying upper-level jet
• - on western edge of ?e ridge

19
Tropical Cyclone QPF
20
Tropical Cyclone QPF

• HPC provides rainfall guidance for all TCs with
  land impacts
• Rainfall statements included in NHCs public
  advisories and HPCs advisories on inland systems
• Detailed QPF guidance to WFOs and RFCs for river
  and flood forecasting

TROPICAL STORM ERNESTO ADVISORY NUMBER 30 NWS
TPC/NATIONAL HURRICANE CENTER MIAMI FL 1100 PM
EDT THU AUG 31 2006 RAINFALL TOTALS OF 5 TO 10
INCHES ARE POSSIBLE FROM NORTHEASTERN SOUTH
CAROLINA INTO THE MID-ATLANTIC STATES...AND THE
SOUTHERN AND CENTRAL APPALACHIANS THROUGH
SATURDAY...WITH ISOLATED MAXIMUM AMOUNTS OF 15
INCHES.
21
How Much QPF?

• Depends on multiple factors
• Availability of moisture
• Moisture at T0
• Moisture flux more important than the initial
  moisture!
• Intensity of Precipitation
• Speed and direction of movement
• Training cells
• Structured banding
• Enhancement by boundaries and/or topography
• Precipitation efficiency
• No proven Quantitative method for QPF Yet!

22
HPC TC QPF

• HPC TC QPF process incorporates
• Global models
• GFS, ECMWF
• Mesoscale models
• NAM
• Ensembles
• SREF
• Hurricane models
• GFDL, HWRF
• Climatological models
• R-CLIPER
• Satellite techniques
• TRaP
• Extensive in-house TC rainfall climatology and
  research
• Official NHC track forecast

23
HPC TC QPF Methodology

• Forecasts made in 6-h increments from 0-84 h and
  in one 48-h period for days 4-5 by 3 forecast
  desks
• Day 1 QPF
• Day 2/3 QPF
• Medium Range (Days 4-5)
• Starting point is model closest to NHC track
  forecast
• Usually GFS
• Locate relevant synoptic scale and mesoscale
  features (troughs, fronts, etc.)
• Use conceptual models, current structure, and
  pattern recognition to modify/shift QPF
• Use TRaP and recent satellite and radar imagery
  to analyze current structure and rainfall rates

24
HPC TC QPF Methodology

• Look at forecast storm-relative shear and 200-mb
  winds to further shift/limit QPF
• Use climatology (PRISM, R-CLIPER, TC Rainfall
  Climatology) to
• Adjust amounts and numerical guidance biases
• Depict areas of terrain that could be
  significantly impacted
• Provide a reality check
• Create TC rainfall statements for the Public
  Advisories
• Day 1-3 forecasts issued by 06z/10z/18z/22z
• Day 4-5 forecasts and 5-day total graphic issued
  at 00z/12z

Day 1-3 QPF
Day 1-5 QPF
25
TC Rainfall ClimatologyUnited States TC Rain
Maxima By State
26
Ensembles

• Can help quantify degree of confidence in a
  forecast and identifying range of possible
  solutions
• Examine spaghetti plot of precipitation isohyets
  from ensemble members to see areas of overlap and
  spread
• Dont use ensemble mean QPF dampens extreme
  events

1 isohyets from SREF members
27
HPC QPF Verification

• 24-hour Day 1 QPFs have been verified since 1961
• Verified on common 32-km grid based on manual
  observational precipitation analysis performed at
  HPC
• TC QPF verification has been performed on a
  regional scale for the past 3 seasons
• Precipitation from a TC may straddle more than
  one region
• Other systems may influence precipitation in a
  region
• New for 2007 ECMWF QPF verification

HPC QPF Verification Regions
28
Verification Metrics
Threat Score Correct
. (Forecast Observed -
Correct) Measures fraction of observed and/or
forecast events that were correctly predicted
Sensitive to hits, penalizes both misses and
false alarms For a perfect forecast Correct
Forecast Observed to yield a TS of
1 The worst possible forecast Correct 0,
yields a TS of zero
29
Verification Metrics
Bias Forecast _
Observed Gauges accuracy of areal/station
coverage of specified threshold amount,
regardless of accuracy in location Ideal
forecast Forecast Observed ? Bias
1
30
Seasonal Verification

• Verification of all Day 1 QPF for TCs and TC
  remnants with precipitation impacts in the CONUS
• 2005 Arlene, Cindy, Dennis, Katrina, Ophelia,
  Rita, Tammy, Wilma (8)
• 2006 Alberto, Ernesto (2)
• 2007 Andrea, Barry, Erin, Gabrielle, Humberto,
  Henriette (East Pacific), TD 10, Noel (8)
• Seasonal threat score and bias from HPC, GFS,
  NAM, and ECMWF computed
• Major changes occurred to all three models during
  this period
• Difficult to identify consistent biases from
  season to season

31
2005 Seasonal Verification
Threshold (.in)

• NAM tends to under-forecast TC precipitation,
  especially at higher thresholds
• GFS has best model bias from 1-6 in. NAM best at
  0.5 in.
• ECMWF TS competitive with GFS across all
  thresholds ECMWF bias slightly lower than GFS
• HPC Bias generally too high, especially at higher
  thresholds
• HPC TS better than the NAM and GFS at all
  thresholds for tropical systems in 2005

32
2005 Seasonal VerificationHPC D1 TS Improvement
over NAM/GFS
1313.79
3393.15
NAM
GFS
33
2006 Seasonal Verification
Threshold (.in)

• ECMWF had highest model threat score across all
  thresholds through 4 in.
• GFS and ECMWF had biases near 1 through 3 in.
  threshold
• GFS had pronounced high bias at 4 in. and above
• HPC showed most improvement in TS at higher
  amounts at the expense of high bias
• NAM showed lowest TS and worst bias for amounts
  of 3 in. or greater

34
2007 Seasonal Verification
Threshold (.in)

• GFS provided best overall model QPF guidance
• HPC showed low bias at highest thresholds,
  reversing trend of 2005 and 2006
• ECMWF performance worse than previous years,
  especially above 2 in. threshold
• NAM continued to have lowest TS and severe low
  bias at high amounts
• HPC had a better or the same TS as the best model
  for all 2007 TC total stats Days 1-3 (except
  amounts gt 3 on Day 2)

35
Gabrielle

• Isolated area of heavy rainfall along central NC
  coast
• Not well-predicted by models which struggled with
  track and structure of the system
• HPC showed significant improvement on heavier
  amounts, particularly on day 2

3
7
5
1
Gabrielle Rainfall Analysis by David Roth - HPC
36
GabrielleDay 3 QPF valid for 24-h ending 12Z 10
September 2007
HPC
GFS
NAM
ECMWF
Threshold (.in)

• HPC showed some skill at 1 threshold
• All models had zero bias for amounts greater than
  0.5

37
GabrielleDay 2 QPF valid for 24-h ending 12Z 10
September 2007
HPC
GFS
NAM
ECMWF
Threshold (.in)

• HPC successfully forecasted location of maximum
  rainfall and showed major improvement over the
  models for amounts gt 0.50
• Higher HPC TS at 3 came at expense of high bias
• NAM had best model forecast only model to
  depict any amounts gt 1

38
GabrielleDay 1 QPF valid for 24-h ending 12Z 10
September 2007
HPC
GFS
NAM
ECMWF
Threshold (.in)

• HPC and model QPF displaced too far northeast
• None of the models hit the amounts gt 1
• The NAM had the best TS for 0.50 1
• All of the models had a high bias through 2
  except ECMWF

39
HPC Inland TC Advisories
40
HPC Inland Advisories

• HPC issues advisories on TCs that have moved
  inland and weakened below tropical storm strength
  but still pose a threat of heavy rains and flash
  floods in the CONUS or adjacent areas within
  Mexico
• Advisories continue even if the low-level TC
  remnant center dissipates as long as flood threat
  remains
• HPC advisories cease when threat of flash
  flooding has ended
• Advisories issued at 03, 09, 15, and 21 UTC under
  the following headers
• Atlantic TCPAT(1-5) WTNT3(1-5) KWNH
• East Pacific TCPEP(1-5) WTPZ3(1-5) KWNH
• Links to HPC advisories can be found on HPC
  webpage, NHC webpage and graphical TWO

41
HPC Inland Advisories

• Primarily focus on potential for heavy rain and
  flash flooding
• Contents of HPC advisories include
• Headline describing the current event
• General summary of current watches, warnings, and
  advisories
• Description of the current location, maximum
  sustained winds, movement, and minimum central
  pressure of the system
• Table listing precipitation induced by the system
  
• Description of the evolution and forecast for the
  system
• Statement including time of next HPC advisory or,
  if the final HPC advisory, where to find
  subsequent information
• Forecast for the position of the surface center
  (if it still exists) as long as the system is a
  flash flooding threat for the CONUS

42
HPC Inland Advisories Issued Annually 2002-2007
43
NHC Continuity of Operations Plan
44
NHC COOP

• HPC is the designated backup for NHC (since at
  least late 1970s)
• HPC will assume forecast responsibility for
  Atlantic and East Pacific TCs in the event of
  communications outage or other problems at NHC
• Approximately 20 HPC forecasters trained in
• TC analysis and forecasting
• Use of ATCF software
• Hurricane Program Policy

45
NHC COOP

• Major technical effort to maintain software,
  scripts, code and to update with annual changes
  to products, graphics, etc.
• Annual in-house exercise, planned and unannounced
  backup drills performed to maintain proficiency
• Implementation of COOP designed to be seamless
  and transparent to users
• Operational NHC products reproduced
• NHC website maintained

46
NHC COOP

• COOP program overhaul began in 2002-2003
• 2002
• Simple communications tests to ensure NHC
  products could be transmitted from HPC
• 2003
• All GS-14 and many GS-13 forecasters trained for
  NHC backup
• Began back-up drills on active TCs (issuing full
  forecast package)
• 2005
• Began issuing East Pacific TWO on the 1st Tuesday
  of every month to test communications and
  proficiency

47
NHC COOP

• HPC has never officially taken over for NHC, but
  there have been at least two close calls
• 2003 or 2004 NHC requested backup due to
  communications problems, but recovered before HPC
  had to issue products
• Hurricane Andrew (1992) NHC sent forecasters to
  NMC for backup prior to the storm hitting Miami,
  but NHC remained up during and after the storm

48
Acknowledgements

• HPC Staff, especially
• Mark Klein
• Jessica Clark
• David Roth
• Wes Junker
• Chris Lauer of TPC

49
Questions??
NESDIS
Media Room
Tropical/Winter Wx
QPF
Basic Wx
Surface Analysis
Model Diag.
Medium Range
http//www.hpc.ncep.noaa.gov/
Michael.J.Brennan_at_noaa.gov