Transcript muri_23_wimmers - Cooperative Institute for Meteorological

GEO Turbulence Detection: Tropopause Folds
and Clear Air Turbulence
Tony Wimmers
Cooperative Institute for Meteorological Satellite Studies (CIMSS), UW-Madison
MURI Workshop
Wednesday, June 8, 2005
CAT and tropopause folds
Upper-air front
stratosphere
12
200
300
14
subtropical
air mass
10
tropopause
8
400
6
500
600
700
polar air mass
Height (km)
Pressure (hPa)
150
4
(~100 km)
From Shapiro, M. A. (1980): Turbulent mixing within
tropopause folds as a mechanism for the exchange of
chemical constituents between the stratosphere and the
troposphere, J. Atmos. Sci., 37, 994-1004.
troposphere
stratosphere
Total column ozone
AWV-derived moisture
high ozone
low
ozone
tropopause
break
very low
water vapor
water
vapor
Total column ozone
and WV indicate
the tropopause
“break”
(region of
GOES WV
response)
GOESWV
specific humidity AWV
product
(WV channel)
(specific humidity product)
Building a statistical model
latitude
decreasing specific humidity
Operation of the Model
longitude
• Cloud-masked
Building a statistical model
latitude
decreasing specific humidity
Operation of the Model
longitude
• Smoothed ( = 0.30)
Building a statistical model
latitude
Operation of the Model
longitude
• Gradient magnitude
Building a statistical model
latitude
Operation of the Model
longitude
• Laplacian zero-crossing
Building a statistical model
latitude
decreasing specific humidity
Operation of the Model
longitude
• Extend out 234 km toward the warm air mass
Building a statistical model
Vertical component of the fold
stratosphere
subtropical
air mass
 surface
latitude
tropopause
polar air mass
+15K
-5K
Upper-air front
longitude
Web product: Real-time pirep validation
 Pirep data is provided
courtesy of NCAR
Aviation Digital Data
Service (ADDS)
Web product: Real-time TAMDAR validation
 TAMDAR
(Tropospheric Airborne
Meteorological Data
Report) is part of the
Great Lakes Field
Experiment
 Unfortunately, it is
mostly lower and
midtroposphere
Validation: Pilot report (pirep)
observations of turbulence
 Manual
 Subjective
 A bit of politics involved
 In-situ
 Well-developed standards of reporting among
pilots
 Record goes back for years
Validation: Details
 April 8-30, 2005 1500-2300 UTC (peak time)
 Eastern U.S. (away from mountain wave turbulence)
 Above 15,000 feet (mid- and upper troposphere)
 Areas of strong convection are filtered out (no C.A.T.)
 If the pirep is in a modeled fold and reports turbulence,
then this is a correctly classified “Yes” report. If the pirep
is outside a modeled fold and reports no turbulence, this
is a correctly classified “No” report.
 2,293 pirep observations, 62% of ALL observations are
turbulent.
Validation: Method
 Find the model’s “Probability of Detection” for
turbulence
 Next, search for any further constraints on the model
that improve the Probability of Detection
120°
… accuracy is not clearly a function of direction
… slight increase in accuracy with proximity to tropopause break
Eliminate
gradients < 5
… lower accuracy for the weaker image gradients
Eliminate folds shorter than 0.2 gcd (22 km) “long”
…Large tropopause folds are very robust
100% accuracy at 0.2 gcd (22 km) from the image gradient
Statistics for tropopause fold turbulence prediction
(N=2293, “background” rate of success=0.64)
Number of Yes
reports
Proportion of Yes reports
correctly classified
Proportion of
No reports
misclassified*
1. Initial model
296
0.77
0.63
2. Revised model:
Longer folds
240
0.78
0.63
3. Revised model (#2):
Longer folds, higher
gradients
138
0.82
0.63
* Does not purport to classify all negative reports
Preliminary conclusions: Trop folding + CAT
 The tropopause folding model shows significant skill at
predicting upper-tropospheric turbulence
 The model increases in accuracy significantly as it is made
more selective (Prob of Detection = 82%)
 The most productive area of prediction (near the tropopause
break) is the area that would benefit the most from hyperspectral
sounding
stratosphere
subtropical
air mass
Upper-air front
polar air mass
Web pages:
http://cimss.ssec.wisc.edu/asap/exper/tfoldsVer2/pirepSep.html
http://cimss.ssec.wisc.edu/asap/exper/tfoldsVer2/tamdarDisplay.html
Outline
 Method of synoptic-scale clear-air
turbulence (CAT) prediction with the
GOES water vapor channel
 Implementation of a predictive
model to a real-time web product
 Validation with pilot reports
Comparison to NCAR Turbulence prediction model
Building a statistical model
AWV gradient magnitude
above the threshold (K)
Estimating dimensions of a fold
6
(mean)
5
4
3
2
1
0
0
100
200
300
400
500
600
Length of corresponding tropopause fold (km)
• Tropopause fold size and water vapor gradient are uncorrelated
Building a statistical model
Hypothesis:
• Is flux and size of a TF proportional to the AWV gradient
magnitude above a threshold?
Introduction
GOES imagery – AWV product
(surface)
(upper
troposphere
~8 km high)
Introduction
Elements of Strat-Trop Exchange
(STE)
(upper-tropospheric air mass boundary)
150
Cut-off Low
stratosphere
12
200
300
14
subtropical
air mass
10
tropopause
8
400
6
500
600
700
polar air mass
(~100 km)
4
Height (km)
Pressure (hPa)
streamers