Tropical (Cyclone) Applications of Satellite Data Andrea Schumacher Cooperative Institute for Research in the Atmosphere (CIRA) Fort Collins, Colorado Mark DeMaria NESDIS Center for Satellite Applications and Research (StAR) Regional and Mesoscale Meteorology Branch Fort Collins, Colorado COMET Faculty Course August 10, 2010
57
Embed
Tropical (Cyclone) Applications of Satellite Data Andrea Schumacher Cooperative Institute for Research in the Atmosphere (CIRA) Fort Collins, Colorado.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Tropical (Cyclone) Applications of Satellite Data
Andrea SchumacherCooperative Institute for Research in the Atmosphere (CIRA)
Fort Collins, Colorado
Mark DeMariaNESDIS Center for Satellite Applications and Research (StAR)
Regional and Mesoscale Meteorology BranchFort Collins, Colorado
COMET Faculty CourseAugust 10, 2010
Track Forecasting Applications
• Initial position/storm structure analysis• Improvement of numerical models
– Assimilation of satellite data• “Evaluation” of numerical models
– Synoptic feature identification for qualitative prediction
– Model analysis/satellite loop overlays provide assessment of t=0 hr accuracy
Center Fixing• Accurate positions necessary for estimation
of storm motion• Animation of imagery
– Visible/IR during day– GOES IR window and shortwave IR at night
• Microwave imagery (SSM/I and AMSU-B)– Formal center fixes began in 2003
Storms with eyes are easiest…
Storms with eyes are easiest…
but still can have errors due to parallax
10.7 m
Tropical Storm Lee 2005
Storms without an eye more difficult
10.7 m 3.9 m
Tropical Storm Lee 2005 Example of the Utility of 3.9 m (GOES Channel 2) data
Storms without an eye more difficult
Helps to use combination of satellite data types
Multi-Spectral Views of Hurricane Katrina(from www.nrlmry.navy.mil/tc_pages/tc_home.html)
VisibleGOES
IR (10.7 m)GOES
Microwave:DMSP SSMI
85 Ghz/H
QuikSCATOcean Surface
Winds
Satellite Data Assimilation
• Satellite Radiances (IR and microwave sounders)– T , water vapor and trace gas (e.g. ozone) profiles– Indirect impact on wind through assimilation
AMSU-A Temperature/Gradient Wind Retrievals(Demuth et al 2006, JAM)
Uncorrecte
d
Correcte
d
T(r,z) Ps(x,y) V(r,z)
R2 = 0.72
0
20
40
60
80
100
120
140
0 20 40 60 80 100 120 140
Predicted Max Wind (kt)
Ob
se
rve
d M
ax
Win
d (
kt)
AMSU Predicted vs. Observed Maximum Winds(Statistical relationships between AMSU retrievals and Intensity)
Single or multiple channel methods also developed by Brueske et al 2003And Spencer and Braswell 2001
Presentation
Visible Imagery Loop 9/21/98 19:02 to 20:10Hurricane Georges During Rapid Intensification
Intensity Trends
Animation of 6-hr Motion- Relative IR Average Images
(Averaging helps separate short/long term intensity changes)
• Loop 1 – just prior to onset of Mitch’s rapid intensification
• Loop 2 – during Mitch’s rapid intensification
Eyewall Cycle of Hurricane Floyd Seen in SSM/I Data
Environmental Interactions
• Vertical Shear of Horizontal Wind– Limits intensification– Prevents establishment of vertically aligned circulation– Increases ventilation of eyewall circulation
• Trough Interaction– Sometimes leads to intensification– Positive momentum flux convergence in upper levels– Increases vertical depth of cyclonic flow– Possible trigger of eye-wall cycle
• SST has strong influence on intensity Change
• Geo and Polar data used in SST products
• Multi-sensor approach to correct for aerosol effects
Improved Sea Surface Temperature
Ocean Heat Content Retrievals from Satellite Altimetry
Statistical Intensity Forecast Improvements Using Satellite Data
(RAMM Branch Joint Hurricane Testbed Project)
• Goal: To determine if satellite data (GOES and satellite altimetry) can improve the intensity forecasts from the statistical-dynamical SHIPS model
• Method: Parallel version of SHIPS with satellite input was run in real-time for 2002-03– Satellite SHIPS made operational in 2004
• Evaluation: Compare operational and parallel SHIPS forecasts for Atlantic and east Pacific
Input from GOES Imagery and OHC Analysis
Hurricane Floyd 14 Sept 1999 OHC 26 Sept 2002
SHIPS Model Improvements with Satellite Input(2002-2003 Experimental Forecasts)
-8
-4
0
4
8
12
16
12 24 36 48 60 72 84 96 108 120
Forecast Interval (hr)
Fo
reca
st Im
pro
vem
ent
(%) Atlantic W of 50 W
East Pacific
Impact on SHIPS Forecasts for Category 5 Storms since OHC was added
• Isabel (03), Ivan (04), Emily, Katrina, Rita, Wilma (05)• Verify only over-water part of forecast
-10.0
-5.0
0.0
5.0
10.0
15.0
20.0
25.0
12 24 36 48 60 72 84 96 108 120
Forecast Interval (hr)
Pe
rce
nt
Imp
rov
em
en
t
All Cases
Isabel 2003
Ivan 2004
Emily 2005
Katrina 2005
Rita 2005
Wilma 2005
Ocean Heat Content for Hurricane Ivan
Convective Response of Ivan to OHC
Wind Structure Applications
• Operational requirement for radii of 34, 50 and 64 kt surface winds
• Satellite techniques– Feature tracked winds for outer circulation– SSM/I surface wind speeds– Scatterometer observations (ASCAT, former QuikSCAT– AMSU-A wind nonlinear balance retrievals– Empirical relationships with IR data
• Experimental, K. Mueller MS Thesis
Cloud drift and water vapor windsHurricane Ivan 12 Sept 2004From CIMSS
x
QSCAT
( Plot provided by Remote Sensing Systems available at www.remss.com )
QuikSCAT and AMSU Nonlinear Balance Winds for Hurricane Ivan
(Plot provided byRemote Sensing Systems
available at www.remss.com )
SSM/I Wind Speeds for Hurricane Ivan
Properties of Satellite Winds for TC Analysis
• IR/WV/Vis cloud drift winds• High quality, but far from the center• Spatial coverage often limited• Height assignment errors
• ASCAT (active)– surface wind vectors• Speeds good to ~50 kts• Rainfall effects winds • Directions sometimes unreliable
• Windsat (passive) - surface wind vectors• Properties similar to ASCAT
• SSM/I surface wind speeds• Rain free areas• Speed only
• AMSU nonlinear balance winds• Temporal coverage limited• Not at the surface• Can not resolve inner core due to 50 km resolution
Properties of Satellite Winds for TC Analysis
• IR/WV/Vis cloud drift winds• High quality, but far from the center• Spatial coverage often limited• Height assignment errors
• ASCAT (active)– surface wind vectors• Speeds good to ~50 kts• Rainfall effects winds • Directions sometimes unreliable
• Windsat (passive) - surface wind vectors• Properties similar to ASCAT
• SSM/I surface wind speeds• Rain free areas• Speed only
• AMSU nonlinear balance winds• Temporal coverage limited• Not at the surface• Can not resolve inner core due to 50 km resolution
What’s Missing?
Properties of Satellite Winds for TC Analysis
• IR/WV/Vis cloud drift winds• High quality, but far from the center• Spatial coverage often limited• Height assignment errors
• ASCAT (active)– surface wind vectors• Speeds good to ~50 kts• Rainfall effects winds • Directions sometimes unreliable
• Windsat (passive) - surface wind vectors• Properties similar to ASCAT
• SSM/I surface wind speeds• Rain free areas• Speed only
• AMSU nonlinear balance winds• Temporal coverage limited• Not at the surface• Can not resolve inner core due to 50 km resolution
What’s Missing?
TC CoreWinds
(especially for smaller TCs)
Hurricane FLOYD – 1515 UTC 14 Sep 99
Hurricane IRIS – 0015 UTC 9 Oct 01
MSLP 932mb
MAX Sustained Winds 125 kt
NE SE SW NW
64 kt 110 75 60 90
50 kt 180 140 105 150
34 kt 250 190 150 190
MSLP 954 mb
MAX Sustained Winds 120 kt
NE SE SW NW
64 kt 15 15 10 15
50 kt 25 25 15 25
34 kt 125 50 40 60
Inner Core TC Winds from IR Imagery(Mueller et al 2007, WF)
• Model wind field by sum of storm motion and symmetric
• Assume Vmis know from Dvorak or other methods
• Estimate x and Rm from IR imagery, Vm and latitude
mxm
m
mm
m
Rrr
RVrV
RrR
rVrV
)()(
)()(
Vm=100 kts, Rm=55 km, x=0.5
Putting Satellite Structure Data Together
Experimental RAMMB Product Satellite-Only Wind analysis• Combine all available satellite inputs in variation analysis• Find Uij Vij to minimize cost function C: C = wk[(uk-Uk)2 + (vk-Vk)2] + wm(sm-Sm)2
+ [(rUij2 +rVij
2) + (Uij2 + Vij
2]
• Uij Vij are gridded radial and tangential wind• u k,vk = obs, Uk Vk= model counterpart of ukvk• sm,Sm are observed wind speeds and model counterpart• Wk and Wm are data weights , terms are smoothness constraints• For wind analysis, “model” is gridded function interpolated to observation point• azimuthal smoothing >> radial smoothing• Based on Thacker and Long (1990)• Could also add other constraints if necessary
• Large and small scales fundamental– More difficult forecast problem
• Satellite radiances/winds improve model analysis• Satellite data improve statistical intensity models• Dvorak used world-wide to estimate storm intensity• SST, altimetry data used for ocean heat content• Satellite data helps identify large-scale shear, and storm response to shear• WV Imagery helpful for identification of trough interaction
– Wind structure • Multi-platform analysis needed
– QuikSCAT, ASCAT, AMSU, SSM/I and IR– Rainfall
• GFDL model has some rainfall forecast skill • IR and microwave data for QPE• Extrapolation (TRaP) and rainfall CLIPER for QPF
Tropical Satellite Data Resources• Tropical RAMSDIS
– TC-centered satellite imagery and derived products– Global TCs, archived online through 2006– http://rammb.cira.colostate.edu/products/tc_realtime/
• CIMMS TC Webpage– http://tropic.ssec.wisc.edu
• Tropical Cyclone Formation Probability Product– Current and climatological TC formation probabilities and input parameters– NESDIS Operational Product (N. Atlantic, NE Pacific and NW Pacific):