1
Combining Ocean Velocity Observations and
Altimeter Data for OGCM Verification

• Peter Niiler
• Scripps Institution of Oceanography
• with original material from
• N. Maximenko, M.-H.Rio, L. Centurioni, C.
  Ohlmann, B. Cornuelle, V. Zlotnicki,, D.-K. Lee

2
Method of Calculating Ocean Surface
Circulation Combines Drifter and Satellite
Observations

• Between 1/1/88 and 12/1/06 1988 10,561 drifters
  drogued to 15m depth were released in the global
  ocean, with array of 1250 since 9/18/05

3
Satellite Observationssea level altimeter
height - geoid height

• Altimeters GEOS, T/P, JASIN, ERS III
• Data from 1992 - Present (rms noise /- 4cm
  relative to geoid)
• GRACE04 Estimated accuracy of geoid /-3 cm at
  400 km horizontal scale
• Sea level gradient, or geostrophic velocity,
  depends upon method and scale of averaging, or
  mapping, of sea level data

4
Drifter velocity observations are accurate (/-
0.015 m/sec daily averages), but spatial
distribution of data can result in biased
averages in space and time
5
Altimeter data is used to calculate geostrophic
velocity with smoothing scales (and amplitude
correction) consistent with drifter data e.g.
AVISO










N/S E/W AVISO Correlation Scales

B. Cornuelle
6
Drifter observed rms velocity variance
ltu2gtltv2gt1/2
N.Maximenko
7
Log10 (Eddy Energy/ Mean Energy)1/2
N. Maximenko
8
The simple method of obtaining a velocity map
9
Vector Correlation between drifter and altimeter
derived AVIO geostrophic velocity anomalies
N.Maximenko
10
East Sea 3 day average velocity from simple
method vs drifter obs. 8/01-11/03
D.-K. Lee
11
Comparison drifter and ECCO 15m zonal velocity
components in tropical Pacific
B. Cornuelle
12
Vector correlation and scatter plots of
geostrophic velocity residuals from drifters
and AVISO in California Current
L.Centurioni
13
HYCOM NLOM POP ROMS
spatial domain global global global 1000 x 2000 km (USWC)
vertical coordinates hybrid layers levels sigma (ETOPO5)
horizontal resolution 1/12 (7 km) 1/32 (3.5 km) 1/10 (10 km) 5 km
vertical layers/levels 26 6 ML 40 20
time step 6 hour 6 hour 6 hour 15 minute
mixed layer KPP Kraus-Turner KPP KPP
wind forcing ECMWF NOGAPS/HR NOGAPS COADS (seasonal)
heat forcing ECMWF NOGAPS ECMWF COADS (seasonal)
buoyancy forcing COADS (restored to Levitus) Levitus (restoring) Levitus (restoring) COADS (seasonal) parameterization for Columbia River outflow
integration time 1990-2001 1991-2000 1990-2000 9 years
assimilation none SST, SSH none none
other Low computational cost open boundaries
C. Ohlmann
14
Unbiased drifter and satellite derived
geostrophic 15m velocity (on left) and ROMS 5km
resolution sea level (right)
L.Centurioni and C. Ohlmann
15
Geostrophic zonal velocity from drifter and
altimeter data
L. Centurioni
16
Decadal MEAN SEA LEVEL (cm) in models of the
California Current
C. Ohlmann
17
OGCM (Eddy Energy)1/2 California Current
ROMS
18
Geostrophic EKE0.5 ROMS (left) corrected AVISO
(right) (0-20 cm s-1)
L. Centurioni and C. Ohlmann
19
Ageostrohic 15m velocity and MSL in 5km
resolution ROMS of California Current
C. Ohlmann
20
THE GLOBAL SOLUTIONS1. Time mean surface
momentum balance for surface sea level gradient

• Observed drifter D
• Computed Ekman E

21
2. Compute sea level that minimized the global
cost function in least square

• The solution is also minimized relative to
  parameters of Ekman force and GRACE altimeter
  referenced sea level, Go, is averaged on 1000km
  scales.
• Maximenko-Niiler

22
3. Perform an objective mapping of sea level,
with mesoscale based, geostrophic, correlation
functions, as a linear combination of
Levitus 1500m relative steric level, GRACE
referenced altimeter derived sea level Drifter
geostrophic velocity.RIO(05), Knudsen-Andersen
23
1992-2002 Mean Sea Level Maximenko (05)
24
Zonal, unbiased geostrophic velocity (-10,10
cm/sec)
25
1993-1999 Mean Sea Level RIO (05)
M.-H. Rio
26
Difference between Maximenko(05)-Rio(05) MSL
with both data adjusted to 1993-1999 period
M.-H. Rio RMS difference of 5cm
27
Comparison of 15m velocity from SURCOLF and
MERCATOR near real time maps of Gulf Stream
region with drifter data
M.-H. Rio
28
Mean Sea Level Knudsen-Anderson
V. Zlotnicki
29
ECCO-2 CUBE49 (18km horizontal, global
assimilation with flux and diff.par.optim.)
V. Zlotnicki
30
ECCO-2 cube 37 and 49 east velocity difference
from Maximenko (05) and Knudsen-Anderson
V. Zlotnicki
31
CONCLUSIONS

• Combined drifter and altimeter derived velocity
  anomalies can be used to make regional,
  realistic, near real time maps of 15m ocean
  circulation.
• Global, absolute sea level on 50km scale from
  combined data displays new circulation features.
• OGCM solutions are most stringently tested with
  velocity fields derived from combined drifter and
  altimeter observations.

32
Why does our view of ocean circulation always
have such a dreamlike quality...Henry Stommel

• THE DREAM HAS COME TRUE
• we are observing the circulation
• peter niiler

33
East-west average vorticity balance at 15m depth
(black line is from Ekmans 1906 model, shaded is
drifter data 100 km coastal and western boundary
currents excluded)
34
The eddy transport of vorticity

• The eddy transport vector of vorticity is
  computed around the Gulf Stream eddy energy
  maximum.

35

• North Atlantic
• 0.25º resolution sea level (upper) and
  simple geostrophic velocity (lower).

36
The 1992-2000 time average quasi-geostrophic eddy
vorticity flux vector in the Gulf Stream region.
37
The mean kinetic energy at 15m depth from
drifters. This quantity graphed is (ltvvgt/2g)
and represents the sea level change caused by
Bernoulli effect of ocean time variable eddies.
38
SST convergence (x10-7Cºsec-1) at 15m depth
39
1978-2003 Average drifter velocity with QSCT/NCEP
blended wind-stress divergence
40
Conservation of vorticity in the Agulhas
Extension Current
41
Drifter geostrophic velocity compared with ocean
circulation model sea-level in California Current
in POP (left) and UCLA/ROMS (right)
42
Global streamlines of 1992-2002 average 15m depth
velocity
43
Zonal, unbiased geostrophic velocity
(-40,40cm/sec)