Decadal Sea-Level Variability in the South Pacific

Yoshi N. Sasaki1, Shoshiro Minobe1, Takashi Kagimoto2, Masami Nonaka2 and Hideharu Sasaki3

(Graduate School of Science, Hokkaido University1, Frontier Research Center for Global Change/JAMSTEC2

and Earth Simulator Center/JAMSTEC3)

Outline•Background and Purpose•OFES vs Observations comparison from 1993 to 2003•Sea-level variability in OFES from 1970 to 2003•Relation with velocity fields and atmospheric fluctuations•Summary

Background

• The advent of satellite altimeter makes it possible to obtain a global distributed dataset of Sea Level Anomalies (SLAs). The data provide important information about the upper ocean gyre circulation (e.g. Häkkinen and Rhines, 2004) and western boundary current(e.g. Qiu, 2003) variations.

Decadal sea-level variability in the South Pacific

• Recently, interesting decadal timescale sea-level variability in the South Pacific is reported by Roemmich et al. (2005, submitted) and Qiu and Chen (2006).

• Roemmich et al. showed that this variability accompanies a spin-up of the subtropical gyre.

• However, the studies of large-scale sea-level variability are limited the last decade (after the satellite era).

Qiu and Chen [2006]

Purposes• Purposes of this study are

– to identify the dominant sea level variability in the South Pacific using OFES hindcast outputs with emphasis on the period before the satellite era.

– to know its relation with the western boundary currents and atmospheric circulation.

– to get some hint for the mechanism of its sea level variability through comparison with that of linear model.

Method• We employ the first Empirical Orthogonal Function mode (EOF1) .

• We consider the reliability of the forcing data from the NCEP/NCAR reanalysis over the South Pacific, we use OFES outputs from 1970 to 2003 (Marshall and Harangozo, 2000; Bromwich and Fogt, 2004).

Data• Satellite altimetry data

– The data are provided by AVISO from 1993 to 2003– Both SLA datasets of the satellite and OFES are interpolated

to a horizontal resolution of 0.5º and averaged monthly intervals for consistency. In addition, we remove monthly climatology to exclude seasonal variations.

• Linear model– We adopted a long-Rossby wave model with a reduced

gravity. The model is almost the same as that of Qiu and Chen (2006), except for the value of the reduced gravity. The Tasman Sea, ignored by Qiu and Chen (2006), is taken into account according to Liu et al. (1999), who showed that the sea-level in the west of the island can be calculated using Kelvin wave response to incoming Rossby waves to the island.

OFES vs observations comparison

• The spatial patterns and time series are very similar.

• OFES can simulate well the observed sea level variability.

The first EOF mode of OFES

cm

The first EOF mode of Satellite

cm

OFES，Satellite

The first EOF mode from 1970 to 2003

• This spatial pattern resembles that of the first EOF mode from 1993 to 2003.

• Positive sea level anomalies are located in the central South Pacific and the Tasman Sea. In particular, the amplitudes are large along the Tasman Front and the northeastern coast of the New Zealand.

The first EOF mode of Satellite

cm

cm

EOF1 of OFES from 1970 to 2003

The temporal structure of EOF1

• The temporal structure also resembles that of the first EOF mode from 1993 to 2003.

• The phase during the periods 1970–76 and 1999–2003 is positive. The period during 1977–96 is almost negative phase.

• The timescale of this variability is decadal.• The dominant sea level variability from 1970 to 2003 is similar to that

from 1993 to 2003.

OFES，Satellite

cm

Regression of velocity fields from 1993 to 2003

• The negative zonal velocity anomalies along the Tasman Front are common of the two datasets.

OFES (200m velocity)Satellite (surface velocity)

cm s-1

Regressions of zonal velocity (shading) + climatological velocity (vector)

Velocity change across the Tasman Front

• The northern part of the zonal velocity along the Tasman Front is weakened.

• The anomalous velocity of OFES is located northward compared to that of observations.

cm s-1 cm s-1 OFES (200m velocity)Satellite (surface velocity)Zonal velocity averaged over 150º–175ºE

1992–1996，1999–2003

1993–1996，1999–2003

Vertical structure of zonal velocity of OFES across the Tasman Front

• The zonal velocity change in the northern part of the Tasman Front has baroclinic vertical structure.

cm s-1

Dep

th (m

)

1992–1996 1999–2003Zonal velocity averaged over 150º–175ºE

Velocity change across the Tasman Front

• This relation between EOF1 and the zonal velocity change along the Tasman Front holds well from 1970 to 2003.

→The sea level variability of EOF1 are accompanies zonal velocity changes along the Tasman Front.

1971–1975，1982–1986，1993–1996，1999–2003

cm s-1 OFES (200m velocity)Zonal velocity averaged over 150º–175ºE

+ – – +

Correlation of SST in EOF1

• OFES also can reproduce realistic SST variability.

• The time series of EOF1 is well correlated the SST variation averaged in the Tasman Sea.

OFES　2.5m temperature

MOHSST

℃

EOF1, MOHSST

MOHSST anomaly averaged in the Tasman Sea

OFES vs linear model comparison

• The overall spatial structure of the linear model shows good agreement with that of OFES, consistent with Qiu and Chen (2006).

• However, the detailed structure, especially in the Tasman Sea and the northeastern coast of the New Zealand, is different. This result suggests that other mechanisms such as a topographic effect are important in these region. →Future study

Regressions of linear model Regressions of OFES

cmcm

Regressions of Z500 in EOF1

• The 500-hPa geopotential height pattern is like the Pacific South American (PSA) pattern, which is the major atmospheric response to the El Niño-Southern Oscillation (ENSO) (Karoly, 1989).

(m)Mo (2000)

Pacific South American pattern

Relation with ENSO

• The time series of EOF1 is positively and significantly correlated with the Southern Oscillation Index (SOI).

• Although Roemmich et al. (2005, submitted) suggests the Southern Hemisphere annular mode affected the sea level trend between 1993 and 2004 using satellite observations, our results suggest the PSA pattern also affected the dominant sea level variability in the South Pacific.

OFES，SOI

Summary• The dominant sea level variability of OFES from 1970

to 2003 is identified by EOF analysis.• The first EOF mode has large sea level changes in the

central and western South Pacific and accompanies zonal velocity changes along the Tasman Front.

• This sea level variability is strongly affected by the ENSO-related atmospheric fluctuation on the decadal timescale.

• The overall spatial pattern of the sea level variability is explained by the linear model, but the detailed structure in the Tasman Sea and the northeastern cost of the New Zealand is not explained.