Does mesoscale instability control sting jet variability?

advertisement
Does mesoscale instability control sting jet
variability?
Neil Hart, Suzanne Gray and Peter Clark
Martínez-Alvarado et al 2014, MWR
Instability and Predictability
Convective
Instability
~1km
~30mins
Symmetric
Instability
~100km
~6hours
Baroclinic
Instability
~1000km
~day
St Jude Forecast:Global Ensemble
Courtesy: ECMWF
St Jude Forecast: MetOffice 4km
Courtesy: MetOffice
St Jude Forecast: MetOffice UKV
IR satellite image at 0600 UTC
Courtesy: MetOffice & EUMETSAT
Why mesoscale instability?
• Browning 2004 hypothesized
that Conditional Symmetric
Instability (CSI) in the cloud head
cloud is an explanation for the
fingering seen in satellite
imagery at the tip of some cloud
heads
• The resulting slantwise
circulation would see ascent into
the cloud head with descent near
the cloud head tip
• This hypothesizes a
mechanism for sting jet
descents, as seen in
1987 Great Storm
Fig. 14 Browning 2004, QJRMS
Why mesoscale instability?
T -10hrs
Shading is number of
pressure levels
between 800hPa and
600hPa, that have CSI
(MPVS*<0)
Blue circle indicates
position of air parcels
manually identified as
part of the sting jet
descent
Moist Baroclinic LC1 experiment
Fig. 7 Baker et al 2013, QJRMS
Why mesoscale instability?
T -6hrs
Shading is number of
pressure levels
between 800hPa and
600hPa, that have CSI
(MPVS*<0)
Blue circle indicates
position of air parcels
manually identified as
part of the sting jet
descent
Moist Baroclinic LC1 experiment
Fig. 7 Baker et al 2013, QJRMS
Why mesoscale instability?
T -2hrs
Shading is number of
pressure levels
between 800hPa and
600hPa, that have CSI
(MPVS*<0)
Blue circle indicates
position of air parcels
manually identified as
part of the sting jet
descent
Moist Baroclinic LC1 experiment
Fig. 7 Baker et al 2013, QJRMS
Friedhelm, Robert and Ulli
Friedhelm 8 Dec ‘11
Robert 27 Dec ’11
Martínez-Alvarado et al
2014, MWR
Identified with DSCAPE
diagnostic applied to
ERA-Interim
(after Martínez-Alvarado
2012)
Ulli 3 Jan ‘12
Smart & Browning 2013
Cyclone Robert
Courtesy: EUMETSAT, Sat24.com
Methodology
Friedhelm 8 Dec ‘11
Robert 27 Dec ’11
Ulli 3 Jan ‘12
1.
Produce 24 member ensemble simulations of each storm
2.
Compute back trajectories from low-level jet region of each member
3.
Cluster analysis to classify trajs. to identify descending airstreams
4.
Explore link between these descents and CSI across ensemble
Model Setup
•
MetOffice Unified Model vn8.2
•
MOGREPS-Global ETKF
24 Init. Pert. Members
(Bowler et al, 2008)
•
MOGREPS-Regional
•
N. Atl. & Europe Domain
•
•
12km, 70 Levels
All storms initialised at 18 UTC
the day before maximum intensity
•
Results analysed further are T+10 to T+24 forecasts
Synoptic Overview
Synoptic Overview
Small spread in synoptic scale evolution between ensemble members:
Good, since can now focus on mesoscale differences
Compute Back Trajectories
Control Run from Cyclone Robert ensemble
Compute Back Trajectories
Cloud Top Temperature
850hPa 45m/s Isotach
Control Run from Cyclone Robert ensemble
Compute Back Trajectories
Trajectories Computed with Lagranto (Wernli & Davies, 1997)
Control Run from Cyclone Robert ensemble
Classification of Airstreams
Cluster Class Mean
Trajectories: Each
trajectory described by
x,y, P, θw for 5 hours
preceding arrival in
low-level jet
Identify class means that
descend
Use Relative Humidity to
remove descents that
started outside cloud
head
Ward’s Hierarchical
Clustering Algorthim
Classification of Airstreams
Identify class means that
descend
Use Relative Humidity to
remove descents that
started outside cloud
head
Ward’s Hierarchical
Clustering Algorthim
Classification of Airstreams
Identify class means that
descend
Use Relative Humidity to
remove descents that
started outside cloud
head
Classification of Airstreams
Use Relative Humidity to
remove descents that
started outside cloud
head
Classification of Airstreams
Classification of Airstreams
Each Class contains a
population of individual
trajectories that arrive at
given time.
Next slide
Size of these populations
are gathered for all
descent classes at all
times for each ensemble
member
1600
# of Traj. Arriving in LLJ
1600
# of Traj. Arriving in LLJ
Majority of of ensemble members
have peak in # trajs at 12UTC
Ensemble Sensitivity
Control run 281K θw 850hpa
Control run cloud head
X
Interpret as change in # trajs for 1 s.d change in CSI metric
Ensemble Sensitivity
X
Methodology
after Torn &
Hakim 2008
Ensemble Sensitivity
X
Methodology
after Torn &
Hakim 2008
Ensemble Sensitivity
X
Methodology
after Torn &
Hakim 2008
Ensemble Sensitivity
X
Methodology
after Torn &
Hakim 2008
Conclusions

Consistent synoptic development across ensemble

Considerable variability in mesoscale wind features

Demonstrated method to classify descending airstreams

Large variability in number of descending trajectories across
ensemble

Does mesoscale instability control sting jet variability?
Strength of sting jet descent is associated
with CSI in the cloud head
(in Robert as simulated with MetUM)
Cyclone Friedhelm
Cyclone Friedhelm Comparison to
Martinez- Alvarado et al 2014 manual
classification
# trajs
Ensemble Sensitivity
CSI across ensemble
If correlation > threshold
(0.5 used here), good!
# trajs
Ensemble Sensitivity
∆y
∆x
CSI across ensemble
Calculate Gradient
# trajs
Ensemble Sensitivity
∆y
∆x
CSI across ensemble
Ens. Sensitivity = ∆y
(∆x = 1 s.d.)
Download