1 operational use of ocean surface vector winds at tpc/nhc noaa operational satellite svw...
TRANSCRIPT
1
Operational Use of Ocean Surface Operational Use of Ocean Surface Vector Winds at TPC/NHCVector Winds at TPC/NHC
NOAA Operational Satellite SVW Requirements Workshop5 June 2006
Michael J. BrennanMichael J. BrennanUCAR Visiting Scientist
Tropical Prediction Center/National Hurricane Center
2
OutlineOutline
• Importance of satellite derived SVW at TPC
• Current use of SVW in TPC operations (QuikSCAT)– Tropical cyclones– Marine forecasting and analysis
• Advances made possible by QuikSCAT
• Shortcomings of QuikSCAT
• Current and future needs
3
Area of ResponsibilityArea of Responsibility
• Tropical cyclone forecasts in Atlantic and Eastern Pacific basins• Marine forecast, warning, and analysis over 12 million nm2 area
4
Typical 12Z Surface ChartTypical 12Z Surface Chart
• Compare observation density in Eastern Pacific to Gulf and Caribbean
• Only 2 ship/buoy wind observations in southeast Pacific area
• Need exists for better spatial coverage of surface wind data
5
QuikSCAT CharacteristicsQuikSCAT Characteristics
• 1800-km wide swaths of data
• One ascending and descending pass per day in sun synchronous orbit
• ~500 km gaps equatorward of 20° and ~1000 km near Equator
6
Use of QuikSCAT Winds at TPCUse of QuikSCAT Winds at TPC• Winds from NASA QuikSCAT
available at TPC in near-real time since 2000
• 25-km and 12.5-km winds available
• QuikSCAT integrated into TPC operations
• Display on NMAP workstations – Used for analysis and product
generation– Overlay with other data (obs,
satellite, etc.)
• Grids of 1/3°, 1/4°, and 1/8° QSCAT winds created for improved visualization and computation of derived quantities
Streamlines and vorticity
MGDR-Lite winds
7
Tropical CyclonesTropical CyclonesCenter FixingCenter Fixing
• Ambiguity removal errors makes automated QuikSCAT center fixes unreliable
• Small errors in center location potentially detrimental to numerical track guidance
• Tropical Storm Lisa 21:15 UTC 28 Sep 2004
Automated solution depicts closed circulation centered near 25.3°N 46.3°W
8
Tropical CyclonesTropical CyclonesCenter FixingCenter Fixing
• Manual ambiguity analysis located center farther north
• This type of analysis useful in real time and post-storm analysis
• Centers from the automated QuikSCAT solutions often biased to the SW
• Manual analysis of ambiguities by forecasters used to help locate true center (e.g., Edson 2004)
Manual analysis supports center location further north near 25.9°N, 46.2°W.
Westerly ambiguities north of QSCAT solution center
9
Tropical CyclonesTropical CyclonesClosing off a systemClosing off a system
• 0937 UTC 3 August 2004
• Automated solution indicates open wave
No closed circulation evident in automated solution
10
Tropical CyclonesTropical CyclonesInitiation of AdvisoriesInitiation of Advisories
• Manual ambiguity analysis indicates closed circulation
• Advisories initiated on TD#2 (later Bonnie) partly based on this analysis:
Tropical Cyclone Discussion:
“IT IS DIFFICULT TO ASCERTAIN IF THE SYSTEM...AN ESPECIALLY FAST MOVING ONE...HAS A CLOSED CIRCULATION WITHOUT DATA FROM A RECONNAISSANCE PLANE. YOU COULD MAKE THE CASE THAT A SMALL CIRCULATION EXISTS USING QUICKSAT AMBIGUITY ANALYSIS.”
Several northerly and northwesterly ambiguities to W and SW of apparent center
Manual analysis supports surface circulation centered near 12.9°N, 52.4°W
11
Tropical CyclonesTropical CyclonesMaximum WindsMaximum Winds
• Estimation of maximum winds in TCs using QuikSCAT complicated by
– Rain contamination • Over- or under-estimation
depends on rain rate and wind speed
• Lack of co-located rain rate makes interpretation difficult
– Saturation of instrument at high wind speeds (> 100 kt)
– Spatial resolution • 25-km and 12.5-km winds unable
to resolve inner core of hurricane
• QSCAT winds sometimes used in determining intensity in tropical storms & marginal hurricanes
– Real time and best track analysis
12
Tropical CyclonesTropical CyclonesMaximum WindsMaximum Winds
• All available 25-km and 12.5-km QuikSCAT passes over 2005 Atlantic TCs examined
– 171 25-km passes– 139 12.5-km passes
• Maximum QSCAT wind from w/in 200 nm of center compared to
– Best track intensity – TAFB Dvorak intensity
• Bias and mean absolute error (MAE) computed
– Averaged by best track and Dvorak intensity bins
13
Average QS Bias Average QS Bias Binned by Best Track IntensityBinned by Best Track Intensity
QS Bias vs. BT Intensity
-80.0
-70.0
-60.0
-50.0
-40.0
-30.0
-20.0
-10.0
0.0
10.0
20.0
30.0
40.0
25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140 145
BT intensity (kt)
Bia
s (k
t)
12.5km QS Bias
25km QS Bias
QS maximum wind has high bias at TD/weak TS stage
12.5 km bias ~10 kt higher than 25 km
Bias reverses sign around 50 kt (25 km)
70 kt for 12.5 km
Bias increases rapidly > 70 kt
12.5-km bias generally smaller at high winds
Relatively few passes over very intense hurricanes
14
Average MAEAverage MAEBinned by Best Track IntensityBinned by Best Track Intensity
QS max wind has high bias at TD/weak TS stage
Bias of 12.5 km significantly larger than 25 km
Bias reverses sign around 50 kt for 25 km data
70 kt for 12.5 km
Bias increases rapidly > 70 kt
12.5-km bias generally less at high winds
QS MAE vs. BT Intensity
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140 145
BT intensity (kt)
MA
E (
kt)
12.5 km MAE
25 km MAE
Same overall trend seen as in bias
MAE smallest in 40–60 kt range
25 km MAE generally smaller than 12.5 km
MAE values larger than bias in tropical storm stage cancellation due to errors of opposite sign
15
Average QS Bias Average QS Bias Binned by TAFB Dvorak CIBinned by TAFB Dvorak CI
QS max wind has high bias at TD/weak TS stage
Bias of 12.5 km significantly larger than 25 km
Bias reverses sign around 50 kt for 25 km data
70 kt for 12.5 km
Bias increases rapidly > 70 kt
12.5-km bias generally less at high winds
QS Bias vs. TAFB Dvorak CI
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
20
30
25 30 35 45 55 65 77 90 102 115 127 140 155
Dvorak CI (kt)
12.5
km
QS
Bia
s (k
t)
12.5km QS Bias
25km QS Bias
Similar trends in QSCAT bias when compared to Dvorak intensity estimates
Positive bias for weak systems
Negative bias for hurricanes
16
QuikSCAT and TC Maximum WindsQuikSCAT and TC Maximum Winds
• Comparison of QuikSCAT maximum wind to best track and Dvorak intensity estimates shows limited skill– Smallest bias/MAE in tropical storms and weak hurricanes– Some low bias values due to cancellation errors of opposite sign
• Large positive bias in TD stage likely due to rain enhanced backscatter– Weak winds at surface overwhelmed by effects of rain
• Negative bias at high winds due to attenuation from rain, saturation of signal, and limited resolution
• QuikSCAT winds must be used with care for intensity analysis – Interpretation ambiguous at best– Lack of co-located rain rate information
• Evaluation of QuikSCAT winds compared to aircraft recon & SFMR, and with TRMM and 88-D radar underway
17
Tropical CyclonesTropical CyclonesWind RadiiWind Radii
• QuikSCAT can provide valuable information on the radius of 34 and 50 kt winds in tropical cyclones outside range of reconnaissance aircraft
• 34-kt radii often outside the area of rain in major hurricanes, limiting rain contamination
• Wind radii analyses and forecasts critical to timing of watch/warning issuance and for marine interests
• Gridded QuikSCAT makes radii analysis easier
– Development of automated routine underway
18
Exercises! - WilmaExercises! - Wilma
• Estimate 34-kt wind radii in NE quadrant from QuikSCAT pass over Wilma – 2332 UTC 18 October 2005
Estimated center from best track
Which (if any) of these areas of 34+ kt vectors do you believe?
19
Compare with Recon, ObsCompare with Recon, Obs
34 kt reduced flight level wind
Surface/ship obs suggest winds < 34 kt here
20
Enhanced Satellite ImageryEnhanced Satellite ImageryHigher winds here
associated with patch of colder cloud tops
higher rain rates
21
WilmaWilma
• 34+ kt QuikSCAT winds well north and east of center can be discounted due to rain contamination
• Operational 34-kt wind radius in 03Z advisory in NE quadrant were 135 nm– Taking QuikSCAT as truth would have resulted in
radii > 250 nm
22
Exercises! - ZetaExercises! - Zeta
• Estimate maximum wind and 34-kt wind radii from QuikSCAT pass over Zeta – 752 UTC 30 December 2005
– No aircraft reconnaissance available
Estimated center from interpolated
best track
Which (if any) of these areas of 34+ kt vectors do you believe?
Your answer drastically impacts 34-kt radii in NE and SE
quadrants
23
Exercises! - ZetaExercises! - ZetaNo ship/buoy observations
nearby
All of these 34 kt + areas in areas of enhanced convection
Which, if any to trust?
24
ZetaZeta
• Particularly difficult estimation of current intensity and 34-kt radii– Higher winds would be expected in areas of
convection – No ship/buoy obs to serve as “ground truth”
• Zeta not classified operationally at this time– Best track intensity set at 40 kt at 06Z– 34-kt radii set to 125 nm in NE and SE quadrants
discounting 34 kt wind areas farthest from center in each quadrant
25
TAFB QuikSCAT ApplicationsTAFB QuikSCAT Applications• Provides valuable wind
information for analyses, forecasts, and warnings
• Useful in placement of fronts, high/low centers, trough/ridge axes, tropical waves, and ITCZ
• Verification for forecasts, warnings, and NWP model output– Differentiation between gale
and storm force events– Utility limited due to rain
contamination effects
H
26
ITCZ AnalysisITCZ Analysis
• QuikSCAT allows for analysis of ITCZ even in absence of convection
• Increased confidence and continuity enabled placement of ITCZ on TPC and Unified surface analysis
27
Surface Analysis ApplicationsSurface Analysis Applications• QuikSCAT provides
surface winds in regions typically void of dense observations
• Improves analysis of features w/o distinct cloud signatures
– Weak fronts (especially in Tropics)
– Ridge axes and surface high centers
– Tropical waves
• Improve isobar packing in areas of enhanced winds
Coverage of 12Z ship/buoy wind observations
(TPC analysis area south of pink line)
Coverage of 12Z ship/buoy wind observations w/ QSCAT
(TPC analysis area south of pink line)
28
Tehuantepec ClimatologyTehuantepec Climatology• QuikSCAT has allowed
development of climatology of high-wind events in Eastern Pacific associated with gap flow (e.g., Cobb et al. 2003)
– Only non-TC storm force winds in Eastern Pacific forecast area
• From 1999-present– 128 gale force events
– 29 storm force events
• Average of 17 (4) gale (storm) force winds each cold season
29
Tehuantepec EventsTehuantepec Events• Ship observations
sparse and many events would go unverified w/o QuikSCAT
• “Ground truth” of QuikSCAT winds allows forecasters to adjust model guidance
– Improve forecast of onset and intensity of events
– GFS model often too slow and too weak with onset of high winds at 10-m
• Rain contamination non-factor
GFS model 10-m wind forecast for Tehuantepec events in Feb. 2006
41 kt45 kt
43 kt 58 kt
72 h 48 h
24 h QSCAT
30
Advances at TPC Made Possible by Advances at TPC Made Possible by QuikSCATQuikSCAT
• Tropical Cyclones– TC center location (and identification) possible with manual ambiguity
analysis– Limited skill in intensity analysis at tropical storm stage– Outer wind radii in absence of reconnaissance aircraft
• Improved surface analysis– Feature placement, especially features lacking significant cloud
signatures– Inclusion of ITCZ on surface map– Improved pressure analysis
• Improved forecast and warnings for gale/storm force winds– Verification of forecasts/warnings/models– Differentiate between gale and storm force winds in rain-free areas – Climatology of East Pacific gap wind events
• Model biases recognized
– Recognition of areas where model under-forecasts winds (gaps, passes, Columbian coast, etc.)
31
Limitations of QuikSCATLimitations of QuikSCAT• Rain contamination greatly complicates interpretation of QuikSCAT
– All Tropical Cyclone applications
– Other features accompanied by precipitation
• Lack of co-located rain rate estimate empirical rain flag– Determination of rain bias due requires a priori knowledge of rain rate
and wind speed
• Unable to resolve inner core of TC• Saturation of signal at high wind speeds • Maximum of two passes daily & gaps between swaths
– TCs and other features can be missed by several consecutive passes
• Delay in receipt of real time data– 1.5–3 hr for 25 km
– 45–60 min longer for 12.5 km
32
QuikSCAT and BeyondQuikSCAT and Beyond
• New real time QuikSCAT retrieval algorithm being evaluated and tested
• Initial WindSat evaluation underway – Data available in near real time– Comparisons underway to QuikSCAT and ship/buoy
observations– Forecaster training in development
33
AcknowledgementsAcknowledgements
• James Franklin and Rick Knabb (TPC) provided ambiguity analysis examples
• Greg McFadden (SAIC/OPC) provided code for gridding QuikSCAT data
• Hugh Cobb (TPC) provided Tehuantepec climatology data and figures
34
Thank YouThank You