6.7
THE SEA AND LAND BREEZE SIMULATION WITH DATA ASSIMILATION
IN URBAN AREA BY THE MM5 MODEL
S. Berkovic and Y. Feliks
Department of Mathematics, Israel Institute for Biological Research
P.O.B 19, Ness-Ziona, Israel
1. INTRODUCTION
resolution of 6 hours. The model was initialized at
This work simulates the circulation in the Tel-Aviv
0Z and run for 36 hours every day during July and
area during July and August 1998 by the MM5
August 1998. The physical parameterizations used
model. High resolution of 0.67 km of the finest
were: Grell Cumulus scheme, MRF boundary layer
domain and observational nudging from 3 stations
scheme,
were employed. The results were verified against
Dudhia simple ice microphysical scheme and
measurements from 15 stations 6 m above the
Cloud-radiation scheme. No objective analysis or
roof height, 0.5-1 km apart. The monthly average
data
errors (RMS) per each station for the wind speed
Newtonian relaxation (nudging) of the wind from 3
and direction were calculated. The improvement of
stations in the Tel-Aviv area was performed every
the data assimilation results was verified against
10 minutes, every day, from 0Z. Optimization for
simulations without data assimilation.
the nudging coefficients and nudging radius was
Five-layer
assimilation
performed.
-3
The
soil
was
temperature
done.
optimized
model,
Observational
parameters
were:
-1
2. MODEL SETUP
2*10 sec for the relaxation factor and 30 km for
The standard PSU/NCAR MM5 model version 3.6
the relaxation radius.
(Dudhia 2000) was run for 4 one-way nested
domains starting at a grid resolution of 18 km over
3. RESULTS
the eastern coast of the Mediterranean Sea down
The most prominent flow during the summer
to a grid resolution of 0.67 km for the Tel-Aviv area
(June-September)
in the finest domain (Fig 1). 26 vertical levels were
circulation. At the beginning of the night the land
applied, 11 levels under 2 km above the ground.
breeze develops along the coast due to sea-land
The model initial and boundary conditions were
differential cooling. During this period the wind is
taken from the NCEP reanalysis data provided by
weak ( ≤ 1 m/s) and has no strongly defined
the NOAA-CIRES Climate Diagnostics Center,
direction. Later, as the katabatic wind arrives from
Boulder, Colorado, USA
the mountains (30 km inland) the wind intensity
(http://www.cdc.noaa.gov), having a time
increases and becomes southeast. The model
is
the
sea-land
breeze
*Corresponding author address: Sigalit Berkovic,
results were compared with measurements every
IIBR, Dept. of Mathematics, Ness-Ziona,
10 minutes at 15 stations 1-2 km apart in the Tel-
Israel.
Aviv area. Study of the wind field showed better
Email: [email protected]
H
TA
45 km
R
Urban
50cm
dry crop 15cm
irr crop
15cm
mix dry irr crop
crop/grass
crop/wood
Grassland 12cm
shrubland
shrub/grass 11cm
savana
broadleaf
needleaf
ever-green
ever-green
mixed forest
WATER
wetland
Fig 1:Above: 18, 6, 2 and 0.67 km MM5 modeling domains.
Below, the land-use of the studied area and locations of important points.
TA == Tel-Aviv, H == Herzeliya and R == Rishon-LeZiyyon
have
east
components
(E-and-W).
Better
prediction after the katabatic wind had joined the
prediction of wind direction was obtained during
land-breeze flow and crossed the city. For
the E-or-W period. The RMSE of the wind
example, figures 2 and 3 present the measured
direction during the E-or-W period for weak winds
and predicted wind field at 25.7.98 22:00 LT
(1m/s < speed< 2 m/s) is 22-43 degrees and for
before the arrival of the katabatic flow and 4 hours
strong winds (speed> 2m/s) is: 10-27 degrees.
afterwards at 26.7.98 2:00LT.
The RMSE of the wind direction during the E-and-
The average RMS error of the wind speed and
W period for weak winds is 27-64 degrees and for
direction was calculated for July and August 1998
strong winds is: 16-38 degrees. During the E-or-W
at each station for the following categories:
period the wind direction prediction for weak winds
1. Events when all the stations have east or west
is somewhat better during the day time (06:00-
wind component (E-or-W) 2. Events when some
18:00) (11-40 degrees) than during the night time
stations have west wind component and others
(18:10-05:50) (20-50 degrees). For strong winds
25.7.98 22:00 LT
measurements
45
ramt
35
bel
betz
30
dol
2 m/s
25
25
30
35
40
45
coast line
FDDA
grid points 2/3 km apart
Fig 2: Calculated and measured surface wind on 25.7.98 22:00 LT.
The blue line indicates the coastline. The x,y axes indicate
the model grid points 2/3 km apart.
FDDA
26.7.98 2:00LT
45
ramt
40
grid points 2/3 km apart
grid points 2/3 km apart
40
35
bel
30
betz
dol
2 m/s
25
25
30
35
grid points 2/3 km apart
Fig 3: As in Fig 2 on 26.7.98 2:00LT
40
45
coast line
mesurments
the prediction for the day and night times is the
assimilation shows improvement relative to no
same. The RMSE of the wind speed is the same
assimilation simulations particularly during the
(difference of ~ 0.3 m/s between the stations) for
night period. The improvements of the wind
the E-and-W, E-or-W, day and night periods. It is
direction predictions are between 6-10 degrees for
0.5-1.0 m/s for weak winds and 0.4-1.4 m/s for
the day and 15-35 degrees for the night, for the
strong winds.
speed the improvements were between 0.1-0.3
In order to evaluate the contribution of the data
m/s for the day and 0.2-0.4 m/s for the night.
assimilation
the
errors
without
and
with
assimilation were calculated at each station during
5. REFERENCES
July and August 1998 for each day, during the day
Dudhia J.,D. Gill, Y. Guo and K. Manning 2000:
and the night periods separately. The data
PSU/NCAR Mesoscale modeling system tutorial
assimilation improves the prediction of the wind
class notes and User’s guide.
direction especially during the night. A maximum
improvement of 60 degrees was obtained at all the
stations during the night. During the day period the
maximum improvement was 10 –20 degrees. The
average improvement at all the stations is 6-35
degrees for the night period and 6-10 degrees for
the day period. The wind speed prediction is
slightly improved as a result of the assimilation
(average improvement of 0.0-0.4 m/s for the day
and 0.1-0.6 m/s for the night).
4. SUMMARY
Surface wind data assimilation from 3 stations in
the Tel-Aviv area was performed for July and
August 1998. The surface wind predictions were
compared with measured surface winds at 15
stations. It was shown that the model follows the
sea-land breeze circulation and better predicts the
surface winds during the E-or-W periods.
The monthly average error (RMSE) per each
station for the wind direction is between 14-36
degrees, and for the speed is between 0.5-1.1
m/s. The error for the day period is 13-34 degrees
and 16-42 degrees for the night period, the speed
error is about the same (~ 1 m/s). The data