r

1
Determining the underlying structures in modelled
orographic flow
R. R. Burton1, S. B. Vosper2 and S. D. Mobbs1 1
Institute for Atmospheric Science, School of
Earth and Environment, University of Leeds,
Leeds, UK 2 Met Office, Exeter, UK
r
Numerical Modelling
Motivation severe turbulence in the Falklands

• The model used to simulate the flow over the
  Falklands was 3DVOM 2, 3
• Linear model for turbulent flow over hills
• Terrain-following coordinates
• Incorporates boundary-layer model
• Mixing length scheme for turbulence
• Semi-implicit finite difference scheme
• Real orography dataset

A series of 725 model runs has been completed,
initialised with daily radiosonde data from MPA.
Of these runs, cases where the measured average
wind direction in the lowest 1km of the
atmosphere was from North-Westerly to
North-Easterly were selected, leading to 94 model
runs. These are then subjected to a principal
components analysis.
Mount Pleasant Airport (MPA), Falkland Islands,
suffers from bouts of extreme turbulence 1 and
any insight into predicting severe weather events
would be highly desirable. Identifying the
surface pressure patterns associated with severe
turbulence and windstorms could be a useful aid
in this regard.
dirn
speed
MPA
5/11/00
Location of the Falkland Islands and MPA
Turbulence and rotor cloud at MPA
PCA of model runs
Principal Component Analysis (PCA) is an
objective method for determining underlying
patterns in data. The significant structures are
known as empirical orthogonal functions, or EOFs.
Shown below are the significant structures
present in the modelled pressure and 400m
vertical velocity fields.

• The first EOF should account for as much of the
  variability as possible the second EOF should
  account for as much of the remaining variability
  as possible and so on.

EOF 2 (P_SURF)
EOF 1 (P_SURF)
EOF 4 (P_SURF)
EOF 3 (P_SURF)
P_SURF
Event B 8/9/01
Event A 5/11/00
SURFACE PRESSURE
PC score
Run
P_SURF
W_400m
normalised eigenvalue
EOF 4 (W_400m)
EOF 3 (W_400m)
W_400m
EOF 1 (W_400m)
EOF 2 (W_400m)
EOF
A
B
VERTICAL WINDSPEED AT 400M
PC score
Run
These EOFs show distinct gravity wave patterns,
as shown in both the pressure and vertical
velocity structures. EOF1(P) and EOF2(P) show
high drag situations, with slightly differing
orientations when either the PC scores of EOF1
and EOF 2, or both, are large and positive, we
would expect a very large pressure gradient to
exist around the area of MPA.
This is indeed the case (see the actual model
outputs corresponding to events A and B). Output
corresponding to other peaks in the time series
(not shown) also display significant effects at
the surface. Note that severe turbulence was
observed at MPA during event A (see the photo and
anemograph trace above.)
Event A positive EOF1, negative EOF 2 gives a
N-S wave pattern
Event B positive EOF1, positive EOF 2 gives a
NE-SW wave pattern
Actual model output surface pressure and wind
vectors
Correlation with radiosonde profiles
In order to determine a relationship (if any)
between the input profile used to drive the model
and the output from the model, a similar set of
PCA analyses was performed on the radiosonde
profiles, using the lowest 2km of each profile.
The actual profiles used are derived from the
radiosonde data and are produced by the
boundary-layer model.
EOF v(d?/dz) P_SURF
1 47 47
2 24 17
3 10 13
4 6 8
26/02/01
05/11/00
height (m)
It was found that the vertical profile of
v(d?/dz) corresponds very well to the model
output, in the sense that the first PC scores for
the two cases show a marked degree of correlation
(r 0.72) and the first EOFs explain the same
proportion of variance.
First PC score for P_SURF
Best fit r0.72
EOF value (Ks-1)
20/08/01
First EOF for d?/dz
Proportion of variance explained by the EOFs for
v(d?/dz) and surface pressure
The presence of inversion has already been
connected with turbulence at MPA 1.
A strong inversion and large negative (northerly)
meridional winds will lead to large v(d?/dz)
This correlation suggests that an inversion,
together with with a strong meridional wind at
the inversion level, is linked to turbulent
effects at the ground the stronger the v(d?/dz)
signal, the stronger the response.
First PC score for v(d?/dz)
A marked correlation between the first EOFs for
v(d?/dz) and surface pressure
High values of the first PC score for v(d?/dz)
corresponded to actual severe weather at MPA (see
anemograph traces to the right).
Conclusions
Acknowledgments
Ralph Burton was funded by a NERC grant.

• Dominant structures have been found the first
  three EOFs account for 70 of the variance in the
  data for both surface pressure and vertical wind
  speed
• Approximately the same amount of variance is
  contained in the first three EOFs for the
  vertical profile of v(d?/dz) derived from the
  input profile

• A marked degree of correlation is found between
  the first EOFs for surface pressure and for
  v(d?/dz)
• This suggests that the vertical profile of
  v(d?/dz) may be useful as a diagnostic in the
  prediction of severe weather events.

References
1 Mobbs, S. D., Vosper, S. B., Sheridan, P. F.,
Cardoso, R., Burton, R. R., Arnold, S. J., Hill,
M. K., Horlacher, V. and Gadian, A. M. (2005)
Observations of downslope winds and rotors in
the Falkland Islands, Q. J. Roy. Meteor. Soc.,
131, 329-351 2 Vosper, S. (2003) Development
and testing of a high resolution mountain-wave
forecasting system, Meteorol. Appl. 10,
75-86 3 King, J. C., Anderson, P. S., Vaughan,
D. G., Mann, G. W., Mobbs, S. D. and Vosper, S.
B. (2004) Wind-borne redistribution of snow
across an Antarctic ice rise, J. Geophys. Res.,
109, D11104
For more information about this poster please
contact Dr Ralph Burton, School of Earth and
Environment, Environment, University of Leeds,
Leeds, LS2 9JT Email ralph_at_env.leeds.ac.uk