A Novel Hurricane OVW Retrieval Technique for QuikSCAT W. Linwood - - PowerPoint PPT Presentation

A Novel Hurricane OVW Retrieval Technique for QuikSCAT

• W. Linwood Jones1, Peth Laupattarakasem1, Suleiman Alsweiss1,

Christopher C. Hennon2, and Svetla Hristova-Veleva3

1Central Florida Remote Sensing Laboratory

University of Central Florida Orlando, Florida 32816

2University of North Carolina Asheville

Asheville, NC 28804

3Jet Propulsion Laboratory

Pasadena, CA

Ocean Vector Winds Science Team Meeting Annapolis, MD May 9-11, 2011

Outline

• SeaWinds OVW measurements in hurricanes
• Radar backscatter geophysical model function
• X-Winds OVW retrieval algorithm

– wind direction retrieval – wind direction retrieval – wind speed retrieval

• X-Wind Comparisons with H*Wind

SeaWinds Hurricane OVW Measurements

• Historically Ku-band scatterometers have

consistently under estimated hurricane wind speeds

• Issues

– Inadequate spatial resolution – Inadequate spatial resolution – Rain contamination – Geophysical OVW algorithms

• GMF – relationship between radar backscatter

and surface wind speed

• Rain correction

Extreme Winds Sigma-0 Geophys Model Function (XW-GMF)

• Special GMF developed for hurricanes

– Training dataset of 35 QScat hurricane overpasses – 3-D GMF: σo = f(ws, relative wdir, atmos transmis)

• WS: one-minute sustained 10m wind speeds from NOAA
• WS: one-minute sustained 10m wind speeds from NOAA

HRD H*Wind surface wind analysis

• Relative wind direction χ: from multi-radar az looks and

from H*Wind analysis

• Atmos transmissivity: inferred from simultaneous
• bservations of QRad H-pol brightness temperatures

Hurricane Geophysical Model Function (GMF) QuikSCAT H-pol, 30 m/s

• 10
• 8
• 6

)

σ

• = C0(ws)+C1(ws)∗cos χ +C2(ws)∗cos2χ

50 100 150 200 250 300 350

• 20
• 18
• 16
• 14
• 12

χ χ χ χ (° ° ° °) σ σ σ σ

( d B )

σo , dB

GMF C0 Coeff Dependence on Wind Speed

• 12
• 10
• 8
• 6
• l

( d B ) Extreme Winds GMF C0 H-pol coefficient

, dB

10

1

• 22
• 20
• 18
• 16
• 14

Wind Speed (dB) C H

• p
• C0 , dB

10 20 40 60 Wind Speed, m/s

GMF Anisotropy with Wind Speed

0.7 0.8 0.9 1

• l

Extreme Winds GMF C1 H-pol coefficient 1.4 1.6 1.8 2

• l

Extreme Winds GMF C2 H-pol coefficient

r units r units

∆σ o = C1(ws)∗cos χ +C2(ws)∗cos2χ

10

1

0.1 0.2 0.3 0.4 0.5 0.6 Wind Speed (dB) C

1

H

• p
• 10

1

0.2 0.4 0.6 0.8 1 1.2 Wind Speed (dB) C

2

H

• p
• C1 , linear uni

C2 , linear uni

10 20 40 60 Wind Speed, m/s 10 20 40 60 Wind Speed, m/s

XW-GMF Atmospheric Transmissivity

• 11
• 10.5

170

• Rain attenuation is corrected implicitly through use of the

QRad H-pol brightness temperature Tbh – GMF = f(ws,rel_wdir, Tbh)

Tbh

50 100 150 200 250 300 350

• 13.5
• 13
• 12.5
• 12
• 11.5

χ χ χ χ(° ° ° °) σ σ σ σ0 H-pol (dB)

140 K 155 K

bh

X-Winds OVW retrieval Algorithm

• Attributes

– Assumes cyclonic wind direction rotation about TC center – Assumes rain effects are primarily absorptive

• Rain volume backscatter is neglected
• Rain volume backscatter is neglected

– Empirical 3-D XW-GMF accounts for backscatter saturation with wind speed & rain absorption – Uses scalar wind direction and wind speed estimation

• Not traditional maximum likelihood estimation

Scalar Wind Direction Estimation

• Relies on anisotropy of measured difference between

forward and aft looking backscatter measurements ∆σo

meas = (σo fore – σo aft)

@ top-of-the-atmos

its Latitude Index Longitude Index ∆σo - linear units

Wind Direction Modeling of ∆σo

Sigma-0 anisotropy model for single radar azimuth look

∆σ mod

• = cosχ fore −cos χaft

( )+C2 cos2χ fore −cos2χaft ( )

∆σ o = C1(ws)∗cos χ +C2(ws)∗cos2χ

Taking the difference between fore & aft radar looks yields

∆σ mod = cosχ fore −cos χaft

( )+C2 cos2χ fore −cos2χaft ( )

Wind direction solutions (aliases) are roots of

∆σ meas

• − ∆σ mod
• (

) = 0

Yields ambiguous wind direction solutions typically ~ 6 – 8

Wind Direction Alias Removal

• Keep ambiguities within window ±30° from CCW spiral
• Multi-pass median filter and populate missing pixels

through interpolation

• Use smoothed TC wind field for wind speed estimation

Wind Directions Ambiguities Initial Wind Direction Field

X-Winds Scalar Wind Speed Estimation

• Use smooth TC wind field as “estimated true” wind

direction and calculate relative wind direction χ = (radar az – “est-wind dir”) Find 1-D wind speed solution for each σo flavor (pol & direction) that satisfies this relationship direction) that satisfies this relationship σ meas

• −(XW −GMF)

( ) = 0

X-Winds Wind Speeds Retrievals (4- σo flavors)

HF HA VF VA Wind Speed (m/s)

X-Winds Comparisons

• L2B-12.5km OVW product
• H*Wind surface wind analysis

Hurricane Fabian Rev#21898

X-Winds Previous Q-Winds X-Winds Previous Q-Winds H*Wind L2B-WS 12.5km

Hurricane Ivan Rev#27217

X-Winds Previous Q-Winds X-Winds Previous Q-Winds H*Wind L2B-WS 12.5km

40 50 m 220 240 260

Wind Speeds Comparison (10 revs)

X-Winds QuikSCAT L2B-12.5km

40 50 220 240 260

TbH-pol TbH-pol

10 20 30 40 50 10 20 30 H*Wind L2B-12.5km 100 120 140 160 180 200 10 20 30 40 50 10 20 30 H*Wind X-Winds 100 120 140 160 180 200

Wind Direction Comparison (10 revs)

300 350 n ( °

° ° °)

240 260 300 350 ion ( °

° ° °)

240 260

TbH-pol

X-Winds QuikSCAT L2B-12.5km

TbH-pol

100 200 300 50 100 150 200 250 H*Wind Wind Direction (°

° ° °)

X-Winds Wind Direction 100 120 140 160 180 200 220 100 200 300 50 100 150 200 250 H*Wind Wind Direction (°

° ° °)

L2B-12.5km Wind Directio 100 120 140 160 180 200 220

Conclusion

• A new OVW retrieval algorithm has been

developed for SeaWinds

– Specifically tailored to to tropical and extra- tropical cyclones – Retrieves wind speeds that are approx 10 m/s higher than the standard L2B-12.5km OVW product

• Performs better in comparison with H*Wind

surface wind analyses

• A new L-3 SeaWinds TC OVW data set will be

produced starting this summer 2011