Satellite Applications in Tropical Weather Forecasting

Mark DeMaria

Regional and Mesoscale Meteorology Team

NESDIS/CIRA

Colorado State University, Ft. Collins CO

Lenny 11/17/99

Acknowledgments• RAMMT

– Roger Phillips, Ray Zehr, Jack Dostalek, John Knaff, Bernadette Connell, Stan Kidder

• TPC– Jiann-Gwo Jiing (SOO), Richard Pasch,

Michelle Huber, Bill Frederick

• CIMSS– Chris Velden, Gary Wade

• NESDIS ORA– Roger Weldon

Outline

• General Circulation in the Tropics– ITCZ– Subtropical Ridge– Tropical Upper Tropospheric Trough

• Synoptic-Scale Weather Systems– Upper-level lows– Tropical Waves

• Tropical Cyclones– Dvorak method– Environmental interactions– Measurements from Polar orbiting satellites

Terminology

• Channel 1 - Visible - .6 m ( .52- .72)

• Channel 2 - Shortwave IR - 3.9 m (3.78- 4.03)

• Channel 3 - Water Vapor - 6.7 m (6.47-7.02)

• Channel 4 - Longwave IR - 10.7 m (10.2-11.2)

• Channel 5 - Split Window - 12.0 m (11.5-12.5)

• Microwave frequencies 20-90 Ghz (1.5-0.3 cm)

1 2 3 4 5 1 2 3 4 5

Hadley Cell

Walker Cells

P (mb)

-90 -60 -30 0 30 60 90-90 -60 -30 0 30 60 90

DJF JJA

1979-1996 January Average 200 mb Streamlines

1979-1996 July Average 200 mb Streamlines

1979-1996 January Average 850 mb Streamlines

1979-1996 July Average 850 mb Streamlines

GMS, GOES, METEOSAT IR Imagery Composite

GMS, GOES, METEOSAT IR Imagery Composite

Global WV Imagery Composite 22 August 1998

0000 UTC

2100 UTC

(Real-time data available from CIMSS web site)

Tropical Upper-Level Lows

• Typically form within TUTT

• Cold-core systems

• Shallower circulation than mid-latitude lows (little circulation below 500 mb)

• Often produce precipitation

• Can influence intensity and track of tropical cyclones

• Can be tracked using water vapor imagery

300 mb 700 mb

Cold-low in NCEP Analysis 8/19/96 0000 UTC

Perturbation Temperature Cross-Section from NCEP Analysis 8/26/96 0000 UTC

Schematic Representation of Cold Low Structure

(from Whitfield and Lyons, WF, 1992)

Radiation Analysis of WV Imager Channel for Idealized Sounding

16-Frame Water Vapor Imagery Loop7/26/98 02:45 to 7/29/98 20:45

Formation of a Cold-Low

Presentation

TPC Tropical Analysis and Forecast BranchCold-Low Study

• Period of Study: Aug. 12-Oct. 1, 1996

• Domain: 0-35 N, 110-20 N

• Cold-low centers tracked using GOES WV imagery

• 47 lows during 50 days

• Average Duration of 3 days, Max of 12 days

• Up to 8 lows in domain at once (average of 3)

• Vorticity center present in NCEP 200 mb analyses for nearly all cold lows identified by satellite observations– Average NECP analysis location error of 100 nm

NCEP Aviation Model Cold-Low Average Track Error

Extrapolation

Persistence

Cold Lows

1996 Tropical Cyclones

200

400

600

Tropical Waves

• Formation near western Africa

– Barotropic/baroclinic instability, PV gradient changes sign

– Secondary instability region in Western Caribbean

• Maximum amplitude near 700 mb

• Period of 2-5 days, wavelength 2-3000 km, May-Dec

• Precipitation associated with waves

• About 2/3 of Atlantic TC genesis associated with waves

• Role in east Pacific tropical cyclogensis

Seasonal Mean PV at 750 mb (Molinari et al, 1997)

Mean meridional wind fromGATE wave composite(Reed et al, 1977)

TAFB Methods for Tracking Waves

• TAFB tropical surface analysis generated 4 times per day, includes surface wave positions

• Rawindsone time series• Surface data when available • Satellite analysis over west Africa

– Animation of channels 1, 2 and 4 indicates rotation

• Hovmoller satellite diagrams for continuity across tropical Atlantic

• NCEP aviation model analyses and forecasts

Dakar Sounding (15 N, 18 W) Time Series August 1995-97

95

96

97

20 W 40 E20 W40 E

08/01/96

08/15/96

08/16/96

08/31/96

Tropical Strip Time Series40 E-40 W 0 N-15 N

Tropical Cyclogenesis

• Eastern Pacific – Interaction of tropical waves/monsoon trough– Includes region of maximum global storm

formation

• Atlantic– 2/3 of storms from tropical waves– Some baroclinic, subtropical and monsoon trough

developments– Very “peaked” season– Experimental genesis parameter under development

Climtological Tropical Cyclone Frequency(From WMO Global Guide To TC Forecasting)

Formations per 5 days (67-98)

0

5

10

15

20

25

30

M J J A S O N D

No.

Sto

rms

Atl.BasinTrop.Atl.

(Tropical Atlantic: 0-20 N, 0-60 W)

Tropical Atlantic Genesis Parameter

• Developed from cases since 1991• Genesis variables

– zonal shear (NCEP analyses)

– vertical stability (NCEP analyses)

– Mid-level moisture (Cloud-cleared GOES-8 water vapor brightness temperature, 1996-99)

– 5-day running means 8-18 N, 30-50 W

• Variables scaled from -1 to 1• Genesis parameter = shear x stability x moisture

Tropical Atlantic Genesis Locations 1991-99

45 Total Cases: 7 Un-named TDs 13 TS 9 Non-major Hurricanes 16 Major Hurricanes

Climatological Shear, Stability, WVBT (Scaled)

Marginal

Unfavorable

Favorable

Genesis Parameter in 1999

Shear Stability

WVBT GP

(www.cira.colostate.edu/ramm/gparm/genesis.asp)

5-Day Running Mean Cloud Cleared Water Vapor Imagery for August 1999

LOOP

Tropical Cyclone Classification

• NHC has responsibility for Atlantic and east Pacific basins– Atlantic: 10 storms, 6 hurricanes

– East Pacific: 16 storms, 10 hurricanes

• Aircraft recon available only for Atlantic west of 55 W

• Majority of center and intensity estimates from GOES satellite data

• TAFB, SAB, AFWA provide classifications

Overview of the Dvorak Technique

• Visible and Infrared Technique• Simplified Visible Technique given here (See

Technical Report for full details)• Uses patterns and measurements as seen on

satellite imagery to assign a number (T number) representative of the cyclone’s strength.

• The T number scale runs from 0 to 8 in increments of 0.5.

Overview of the Dvorak Technique Cont’d

• In the following examples, only the Data T Number (DT) will be calculated, the final (official) T number assigned to a tropical cyclone includes further considerations.

• DT computations familiarize one to various tropical cyclone patterns.

Four Basic Patterns

• Curved Band Pattern

• Shear Pattern

• Central Dense Overcast (CDO) Pattern

• Eye Pattern– Pattern is not always obvious– Pattern typically varies with time

Patterns and Associated T Numbers

Empirical relationship between T number and wind speed

Finding the Cloud System Center (CSC)

• First step in the Dvorak technique• From Dvorak (1985):

“The cloud system center is defined as the focal point of all the curved lines or bands of the cloud system. It can also be thought of as the point toward which the curved lines merge or spiral.”

• Center not always obvious, especially at night• TPC technique combines channel 2 and 4

T.S. Lisa Channel 4

T.S. Lisa

Curved Band PatternTS Ivan 9/23/98 11:15 UTC

Curved Band Pattern Cont’d

1.0 2.0 2.5 3.0 3.5 4.0 4.5

DT Number

Shear PatternHurricane Bertha 7/11/96 2015 UTC

Shear Pattern DT Numbers

1° latitude = 60 nautical miles (nmi) = 111 km

Central Dense Overcast (CDO)Hurricane Georges 9/21/98 1545 UTC

CDO

• No eye

• DT number determined by CF+BF=DT– CF=CENTRAL FEATURE– BF=BANDING FEATURE– DT=DATA T NUMBER

CDO Central Feature (CF)• Measure Diameter of CDO in degrees latitude• For a well defined CDO

– 3/4 ° CF=2– 1 1/4 ° CF=3– 1 3/4 ° CF=4– >2 1/4 ° CF=5

• For an irregular CDO– 1° to 1 1/2 ° CF=2– >1 1/2 ° CF=3

CDO - Banding Feature (BF)

Eye PatternHurricane Georges 9/19/98 1345 UTC

Eye Pattern

• DT number determined by CF+BF=DT– CF=CENTRAL FEATURE– BF=BANDING FEATURE– DT=DATA T NUMBER

Banding Eye Hurricane Bonnie 8/25/98 2131 UTC

Infrared (IR) Technique

• Can be used during night as well as during day

• At times, more objective than visible technique

• Fully objective version developed at CIRA

• Updated objective technique from CIMSS

Example Digital IR: Hurricane Erika 1515 UTC 8 September 1997

• Warmest eye pixel 16 °C

• Warmest pixel 30 nmi (55

km) from center -71 °C• Nomogram gives Eye no.

=7

Operational Dvorak Technique Verificationfor 1997-98 Atlantic Seasons

Position (nm)

x-bias (nm)

y-bias (nm)

Max wind (kt)

Wind bias (kt)

TAFB 97 18.8 -3.8 -0.4 4.6 -1.2

SAB 97 19.7 -3.9 -3.0 5.9 -2.2

TAFB 98 21.1 -1.6 0.3 6.5 0.7

SAB 98 21.3 0.0 -2.1 6.5 -1.7

NHC Official Forecast Skill (1997-99)

-40

-30

-20

-10

0

12 24 36 48 60 72

Forecast Interval (hr)

No

rmal

ized

Err

or

(%)

Track

Intensity

Input for NHC Track and Intensity Forecasts

• Track– 70 % Numerical model guidance– 15 % Synoptic reasoning– 15 % Recent trends

• Intensity– 50 % Recent trends– 40 % Synoptic reasoning– 10 % Numerical model guidance

Applications of Satellite Data to TC Forecasting

• Track– Inclusion of remotely sensed data in NWP models– Diagnosis of model initial state– Evaluation of synoptic situation

• Intensity– Evaluation of factors affecting intensity

• SST changes

• vertical shear (especially cloud track winds)

• trough interaction

– Inclusion in NWP models

300 mb AVN Winds and WV Image 19 Sept 1998

NEW GOES WINDS PRODUCTS ARE BEING PRODUCED BY CIMSS/NESDIS:

- HIGH - DENSITY WINDS DERIVED FROM IR AND HIGH-RESOLUTION VISIBLE CLOUD MOTIONS AS WELL AS WATER VAPOR MOTIONS, USING AUTOMATED ALGORITHMS

- DISPLAYS OF THESE WINDS FOR UPPER- & LOWER-LEVEL LAYERS OVER THE TROPICS ARE ROUTINELY AVAILABLE VIA THE INTERNET

-UWISC/CIMSS TC WEB SITE: http://cimss.ssec.wisc.edu/tropic/tropic.html

NEW GOES WINDS PRODUCTS ARE BEING PRODUCED BY CIMSS/NESDIS:

- HIGH - DENSITY WINDS DERIVED FROM IR AND HIGH-RESOLUTION VISIBLE CLOUD MOTIONS AS WELL AS WATER VAPOR MOTIONS, USING AUTOMATED ALGORITHMS

- DISPLAYS OF THESE WINDS FOR UPPER- & LOWER-LEVEL LAYERS OVER THE TROPICS ARE ROUTINELY AVAILABLE VIA THE INTERNET

-UWISC/CIMSS TC WEB SITE: http://cimss.ssec.wisc.edu/tropic/tropic.html

Vertical Wind Shear Analysis from GOES High Density Winds

Hurricane-Trough PV Interaction During Hurricane Elena 1985 (Molinari et al 1995)

8-Frame Water Vapor Imagery Loop 9/20/98 23:45 to 9/21/98 23:45

Hurricane/Trough Interaction

Presentation

Storm-Scale Structure

• Convective transients, asymmetries– Velden and Olander (1998) technique

• IR BT usually warmer than WV

• Deep convection transport into stratosphere

• WV BT warmer than IR

• Concentric eye walls, eye wall cycles

• Mesovortices within the eye– High spatial and time resolution imagery (RSO, SRSO)– Extra-high density satellite winds

Channel 3, 4 Convective Parameter (CP) for Hurricane Opal 1995 (From Bosart, et al 1999)

Min P P (mb)CP

GOES-East Scanning Strategies

• Routine Scanning– Conus hr+01,31– Extended NH hr+15,45– (2 or 4 per hr)

• Rapid Scan– Conus, NH 5-10 min– (8 per hr)

• Super Rapid Scan– Selected Sector 1-5 min– (22 per hr)

10-frame Rapid-Scan Visible Imagery Loop

9/21/98 19:02 to 20:10Hurricane Georges Approaching Puerto Rico

Presentation

Comparison of Operational and SRSOWinds for Hurricane Floyd

Hurricane Luis 9/6/95 1345 UTC

TC Measurements from Microwave Frequencies

• Special Sensor Microwave Imager (SSM/I)– DMSP polar orbiting satellites

– 19, 22, 37, 85 GHz, 25 km resolution

• Advanced Microwave Sounding Unit (AMSU)– NOAA polar orbiting satellites (NOAA 15 +)

– 15 channels 23-89 GHz, 50 km resolution

• Can see “through” clouds• Depicts rainband, eye structure• Rainfall and surface wind algorithms• AMSU temperature retrievals

Hurricane Jeanne

9/23/981021 UTC

From NRL web site

SSM/I Imagery for Hurricane Floyd13-14 September 1999

IR ImageryMarch 1, 1999

AMSU TemperatureRetrieval (570 mb)

Hurricane Floyd Analysis 14 Sept 1999 12 UTC

AMSU Swath IR Image Liquid Water

(www.cira.colostate.edu/ramm/tropic/amsustrm.asp)

WithoutCorrection

WithCorrection

Ta

V

Ta

V

Applications of Balanced Winds:Asymmetric Vortex Structure

850 mb Isotachs (m/s) from Floyd 14 Sept 99 12 UTC

-0-10-20-30-40

Gale Force (34 kt) Wind Radii Prediction

5 predictors selected, AE=18 nm, r2=83%

AMSU Variable Normalized Coefficient0-100 km Liquid water 29.1Max Temp anomaly 26.5 Storm latitude 25.3600-0 km pressure drop -17.2Radius of 15 kt wind at 3 km 8.1

R2 = 0.8331

0

50

100

150

200

250

300

0 50 100 150 200 250 300

Predicted Gale Radius (nm)

Ob

se

rve

d G

ale

Ra

diu

s (

nm

)

AMSU-A Rainfall Rate for Hurricane Georges (.01 inches/hr)

TRaP for Key West = 6.7 inches

GOES IR and SSM/I 85 GHzHuricane Floyd 10 UTC 11 Sep 1999

(from NRL web site)IR 85 GHz

Auto-Estimator>

<Multi-Spectral

Rainfall EstimationFrom GOES

Summary

• Multispectral GOES imagery provides synoptic overview– ITCZ, Subtropical Ridges, TUTT and cold lows, waves, TCs

• WV Imagery is especially useful for tracking cold lows

• Continuity in imagery is primary tool for tropical waves

• Dvorak method provides quantitative intensity estimates– Also provides framework for operational forecasting

– Shortwave IR aids center location, especially at night

• Sat. winds: environmental interactions, model intialization

• WV- IR difference isolates tropical deep convection

• Rapid-scan imagery: storm-scale fluctuations

• Microwave imagery is useful for low-level storm structure

• New AMSU data shows promise for hurricane analysis