Diapositive 1

1
Regional atmosphere/ocean simulations in Central
Chile during October 2000 Impact of mesoscale
wind variability on upwelling Lionel Renault
(1), Mark Falvey(2), Boris Dewitte(1, 3),
Vincent Echevin(4), Jose Rutllant(2), Rene
Garreaud(2) (1)LEGOS, Toulouse, France (2)
DGF, Santiago, Chile (3) IMARPE, Callao, PERU ,
(4)LOCEAN, Paris, France ()lionel.renault_at_legos.
cnes.fr
Main results with WRF
Introduction The study of regional
oceanographic processes has undergone
considerable development in recent years due to
the expansion of in-situ observation networks,
the increased availability of satellite data, and
the development of high resolution numerical
models. Coastal zones are of special interest in
regional applications as they are often the place
of intense ocean mesoscale circulations that play
a key role in coastal and deep-ocean exchange.
Thus, an accurate representation of the coastal
atmospheric forcing is likely to be important in
studies seeking to evaluate coastal to deep-ocean
transfer and to clarify the mechanisms associated
with coastal wind variability. In this study we
use the WRF (Weather Research and Forecasting)
regional atmospheric model to simulate the near
surface atmospheric circulation along the South
American (SA) coast between 15S - 40S using a
multiply nested domain with grid spacing as low
as 6 km. The simulations were performed for a
sustained coastal jet event in October 2000
during which there was significant
atmosphere-ocean interaction. We also present the
results of preliminary experiments in which
atmospheric fields produced by WRF at various
resolutions were used to force the ROMS (Regional
Oceanic Modeling System) ocean model simulations
of coastal upwelling processes.
COMPARISON WITH QUIKSCAT The figure 3 (right)
presents the temporal correlations between the
QuikSCAT product (0.5) and the WRF1. The WRF
model is capable of reproducing the observed wind
variability during the study period. The mean
temporal correlation for each of the three
simulations is about 70. The figure 2 (below)
shows an example of a point in the coastal jet
core location (72.5W 30.1S). The Coastal Jet
event is well reproduced by WRF.
Fig. 3 The temporal correlations between WRF and
QuikSCAT for the October 2000 month.
The spatial resolution has an impact on the
simulated fields The figure 4 (below left)
shows the wind variability for the different
domains we observe that the variability
increases with higher spatial resolution. Near
the coast, both the wind curl and the mean wind
are different on each domain (figure 5, below
right) and we can therefore expect differences in
the Ekman transport with a corresponding impact
on upwelling and offshore transport.
Fig 1 The mean wind speed during October 2000
for the three WRFs domains

• Models and methodology

ROMS model The ROMS (Regional Ocean Modeling
System) is a free surface, hydrostatic model
which resolves the primitive equations using
stretched, terrain-following coordinates in the
vertical and orthogonal curvilinear coordinates
in the horizontal. We use 32 sigma levels on two
ROMS domains linked by a a two way nesting
scheme. The respective spatial resolutions are
1/6 (18km) and 1/18 (6km). We performed a 8
month spin up using a WRF simulation (54km) and
horizontal boundary conditions from the ORCA
model. To force the model, the following WRF
outputs are used Wind speed, Air density, Heat
Flux, Sea level air temperature, Relative
humidity, Precipitation rate.
WRF model The WRF (Weather Research and
Forecasting) Model is a next-generation mesoscale
numerical weather prediction system designed to
serve both operational forecasting and
atmospheric research needs. It features multiple
dynamical cores, a 3-dimensional variational
(3DVAR) data assimilation system, and a parallel
software architecture. WRF is suitable for a
broad spectrum of applications across scales
ranging from meters to thousands of
kilometers. The WRF model configuration
consisted of three nested grids each with 57
vertical sigma levels. The domains have
resolutions of 54km., 18km and 6km, respectively,
with a Lagrangian map projection and a two way
interaction between nests. Boundary conditions
are from NCEP 4x reanalysis and SODA for the SST.

WRF1 WRF2 WRF3
Fig. 4 The contours represent the average wind
speed during October 2000. The colors represent
the RMS of the wind speed.
Fig 5 Wind speed (Top) and Ekman Pumping
(bottom) along two zonal sections at 28.5S
(left) and 30.1S (right).
Preliminary results with ROMS Impact of
atmospheric model resolution
The study takes place during the October month
2000. We present three runs
RUN 1
We use a spin up of 8 months using the first run
of WRF, but we still believe the solutions
presented here are degenerated due to the fact
our spin up is not long enough to be free from
spin-up problem, we do not present here any
verification of data against observations.
Tab.1 horizontal resolution of ROMS and WRF for
the different runs
Impact on the Sea Surface Temperature and surface
currents RUN 2 is characterized by more eddy
activity than RUN1 (Figure 6, right). The SST
pattern near the coast is also different, the
WRF2 forcing resulting in cooler temperatures
than the WRF1 forcing through the enhancement of
Ekman pumping. Results with the WRF3 (not shown)
were very similar to those of WRF2.
Impact on the Sea Surface Temperature and on the
suface currents The figure 6 presents for the
mean Sea Surface Temperature (SST) and the mean
currents during the month of October for two
different forcing WRF1 (54km) and WRF2 (18km) in
the same grid (1/18). The high spatial
resolution of the winds seams to generate more
eddies than the low resolution. Moreover, we
dont have the same SST pattern near the coast.
The WRF2 forcing is more cooler than the WRF1
forcing, maybe with an enhancing of the Ekman
transport or Ekman pumping. The forcing with the
high spatial resolution of 6km. is not shown
here, but we found similarly results than the
WRF2 forcing.
Temperature (C)
Impact on the Mixed Layer Depth (MLD) Figure 8
highlights the impact of the high resolution
local forcing on the ocean MLD. The MLD is the
most impacted with much larger variability for
RUN2 and RUN3. This effect could have some impact
on the local biology. The changes in MLD are most
likely due to the differences in the wind stress
curl between model runs (see fig 5). The SST
pattern is also shown and again, differences are
a observed in the three runs RUN3 is generally
cooler than SST RUN1 because of stronger Ekman
pumping.
Impact on the Mixed Layer Depth (MLD) The figure
7 highlights the impact of the local forcing high
resolution on the ocean depth. Near the coast,
where we had significantly Ekman pumping
difference with the wind resolution, we have a
different mixing and then a deepening of the MLD
which depends of the local forcing resolution.
This effect could have some impact on the local
biology.
Fig 6 Mean SST surface currents during October.
Top panel WRF1 forcing, bottom panel WRF2 forcing.
Fig 6 The average SST for the October month and
the mean sea surface currents. Top pannelWRF1
forcing, bottom pannel WRF2 forcing.
The pattern of the SST are not the same, the
ROMS3 is more cold than ROMS1, we can suppose
that the increase of the Ekman pumping and of the
Wind Speed caused an upwelling intensification
and so a cooling of the ocean.
RUN 3
RUN 2
RUN 1
Impact on the Peru/Chile Undercurrent The
Peru/Chile undercurrent is confined near the
depth of 200m. and between the Rossby rayon and
the coast (30km). The figure 8 shows a section a
30.1S, near Coquimbo of the currents for ROM1,
ROMS2 and ROMS3. ???
RUN 2
RUN 1
RUN 2
RUN 3
Fig. 7 The meridional current at 30S. We can see
a current intensification in RUN2 and RUN3.
Temperature (C)
Depth (m)
Depth (m)
Depth (m)
Impact on the Peru/Chile Undercurrent Profiles
of the meridional currents at 30.1S show a clear
impact of model resolution, including
intensification of both the southward Peru/Chile
undercurrent (blue region along coast between
30-200m) and a near-surface northward current
some 70 km offshore.
Bibliography http//www.wrf-model.org/ and
http//www.brest.ird.fr/ressources/roms.htm Capet,
X., P. Marchesiello, and J. McWilliams
Upwelling response to coastal wind profiles.
Geophys. Res. Lett, 31, 13 L13309
10.1029/2004GL020303 , 2004. Garreaud, R. and R.
Munoz, 2005 The low-level jet off the
subtropical west coast of South America
Structure and variability. Mon. Wea. Rev., 133,
2246-2261. Munoz. R. and R. Garreaud, 2005.
Dynamics of the low-level jet off the subtropical
west coast of South America. Mon. Wea. Rev., 133,
3661-3677 Acknowledgements The QuikSCAT data
were obtained from CERSAT, at IFREMER, Plouzan
(France). We would like to thank ECOS Sud for
their financial contribution via project C0998766
as well as the ATUPS project of the UPS of
Toulouse which provided financial support and the
Centre National dEtude Spatiale (CNES) and the
Institut Français de Recherche pour
lExploitation de la Mer (IFREMER) for doctoral
grant support.
Conclusion We have performed three simulations
of the atmosphere and ocean near Coquimbo, Chile
(30S) during October 2000. The regional
atmospheric model WRF was used to simulate the
surface winds and heat fluxes, which were then
used to force a regional oceanic model (ROMS). We
use three different spatial resolutions of the
WRF model to study the impact of a better
resolution of atmospheric mesoscale structures on
the oceanic model. Our results suggest the
existence of important mesoscale atmosphere-ocean
interactions. The mixed layer depth of the ocean
and the Chile-Peru sub-current show significant
differences depending on the resolution of the
wind forcing. In particular, the high resolution
forcing generates more eddies and induces
stronger Ekman pumping near the coast resulting
in changes in the thermodynamical budgets of the
mixed layer. Future wok will concentrate on the
analysis of the different advection and
entrainment terms in the mixed layer in order to
document the developing and decaying stages of a
coastal Jet event off Central-Chile.