_{1}

^{*}

Using satellite-based 10-m surface wind (SW), wind stress (WS) and sea surface temperature (SST) anomalies, trends and inter-annual variability during 1993 and 2008 over the Southern Ocean (SO) are addressed. The climatological mean (16 years average) indicates that negative wind stress curl diminished (enhanced) between 40°S and 60°S zonal strap region coincide with weak (strong) SW and warm (cold) SST anomaly during January (July). Annual climatology indicates that strong region of SW divides warmer waters northward with positive wind stress curl (WSCL) and colder waters southward with negative WSCL. The time series anomalies are smoothened with a 12-month running mean filter. The filtered area-mean time series anomalies of zonal and meridional component of SW and SST have linear trends of -0.0005 ± 0.0003 m/s/decade, 0.0012 ± 0.0002 m/s/decade and -0.00005°C ± 0.0001°C/decade, respectively. The SW anomalies show an increasing trend of 0.0013 ± 0.0002 m/s/decade, with the meridional (zonal) component exhibiting an increasing (decreasing) trend. The meridional component plays a critical role in heat transfer through atmospheric circulation. The WS and wind stress divergence exhibit increasing trends whereas wind stress curl shows a decreasing trend. The SST fluctuates close to zero with repeated high and low peaks at an interval of 2 - 3 years. We address the interannual variability by performing EOF analysis on SW, WS, WSCL and SST anomalies which have been passed through a 12-month running mean filter. EOF-1 spatial pattern portrays low variances south of 40°S in SW between 25°E and the Drake Passage whereas in WS it is confined to Pacific Ocean sector of the SO. EOF-2 pattern exhibits high variability along the ACC, which is pronounced in the central Indian sector of the SO in SW and Pacific Ocean sector of the SO in WS. The time coefficient (SW, WS and WSCL) of EOF-1 (EOF-2) is correlated with reversed (actual) Antarctic Oscillation (AAO) index. The EOF-1 of SST shows high variance in the Indian sector south of 55°S and in the south Pacific sector of the SO. The corresponding time coefficient function indicates an inverse relationship with AAO index. EOF-2 of SST shows dipole structure of high variance in the Pacific Ocean sector of the SO; high positive variance is also evident in Indian sector. The time coefficient of EOF1 (EOF-2) is correlated with reversed AAO (Nino 3.4) index, with the earlier (latter) leading the former by 4 (9) months. Based on the EOF analyses, it can be inferred that AAO and Nino indices play an important role in modulating SW and SST changes in the SO.

The Southern Ocean (SO) is one of the largest sectors of the world oceans. It plays a key role in the climatic system on the Earth through its complex circulation induced by interconnection of three ocean basins at high southern latitude and strong air-sea interaction on seasonal and interannual time scales. The SO encompasses the Antarctic Circumpolar Current (ACC) around Antarctic continent, and also is characterized by basins and ridges with complex bottom topography (

The West Wind Drift reveals strong seasonality according to recent investigations; the NCEP-NCAR sea level atmospheric pressure at 65˚S exhibits a climatic tendency of −0.166 ± 0.039 hPa・yr^{−1} from 1957 to 1998 and −0.177 ± 0.062 hPa・yr^{−1} from 1969 to 1998. The negative tendency weakens with time to −0.123 ± 0.221 hPa・yr^{−1} from 1979 to 1993 [^{−1} [

In recent decades the importance of climate signals in the SO has been narrated in many studies by the researchers. For instance the trend for the Southern Hemisphere is reported to be −0.3% ± 0.5% per decade for sea

ice thickness and −1.2% ± 0.6% per decade for sea ice area from 1979 to 2003 [^{−}^{2}・decade^{−1} in January at 55˚S to small values in austral winter months. There are indications that this is linked to the strengthening in the Southern Annular Mode (SAM), which demonstrates an upward trend. Later, [^{−1}・decade^{−1} (from 1987 to 2006) using special sensor microwave/imager (SSM/I) data.

The above studies reveal that the changes do occur in recent decades and the close relationship between the surface winds (SW), wind stress (WS), and sea surface temperature (SST) in the SO to the climate signals is significant to understand the variability in the study region. Furthermore, the trends of these parameters have not been reported earlier. An attempt is made in the study region to address the seasonal, trends and interannual variability in the SO which covers swath of 30˚ width from 40˚S to 70˚S globally that is from 0˚E to 360˚E, spanning 16 years (from 1993 to 2008) of SO in recent decades. For this purpose we have used the satellite observed SW, WS over the ocean and SST in last decades. In this work we focus on interannual variability of the annual mean fields; the interannual variability in seasonal field is not considered in the current study.

To understand interannual variability of the SO we need to have temporal and spatial data which is possible with the technology of satellite remote sensing. SST and Winds are directly or indirectly associated with the global climate changed. Presently long term temporal parameters are available with the CERSAT to understand climate signals in the recent decades. We used monthly mean wind speed field produced by CERSAT at the French Research Institute for exploitation of the sea (IFREMER). The product is derived from the observations of scatterometers AMI-Wind onboard the European Space Agency satellites ERS-1 (1991-1996) and ERS-2 (1996-2001), NASA scatterometers NSCAT onboard the ADEOS-1 (1996-1997), and Sea Winds onboard the Quik-SCAT (1999-2008). The CERSAT product provides wind stresses (WS) which are derived from the wind speed using the bulk formula of [

The monthly-mean SST data are derived from NCEP Optimum Interpolated (OI) SST Analyses product, which has a resolution of 1˚ × 1˚ covering the period from January 1993 to December 2008. It is merged satellite and in situ SST observations [

To explain the significance and delineate the role of temporal modes of Empirical Orthogonal Function (EOF), we use monthly mean Antarctic Oscillation (AAO) index (http://www.cpc.ncep.noaa.gov/), which is the first leading mode from the EOF analysis of monthly mean geopotential height anomalies at 700 hPa over the southern hemisphere. Monthly mean anomaly data were used to obtain the loading patterns. Also, monthly mean Nino 3.4 data which is the average SST anomaly in the region bounded from 5˚N to 5˚S, from 170˚W to 120˚W, were used. This region has large variability on El Niño time scales, and is close to the region where changes in local SST are important for shifting the large region of rainfall typically located in the far western Pacific [

To study the inter-annual variability using global time series of SW, WS and SST are averaged on 5˚ latitude by 10˚ longitude grid. The simple linear regression model is calculated from its observed values y_{i} at the time t_{i},

where the observed values of the y_{i}’s constitute the responses or values of the dependent variable, the t_{i}’s are the settings of the independent variable, b_{0} and b_{1} are the intercept and slope (trend) parameters, respectively, and the ε_{i}’s are independently distributed normal errors each with mean zero and variance s^{2} based on a method described by [

To study the interannual variability, EOF was performed on the mean values and linear trends were first removed from the time series which was smoothed values through the 12-month running mean [

To construct the M × N data matrix, D, using the M rows (locations x_{m}) and N column (time t_{n}) of the normalized data series and from this derive the symmetric covariance matrix, C, by multiplying D by its transpose D^{T}.

The data series

the i^{th} orthogonal mode at time _{i}, whose amplitude are weighted by M times depended coefficient, _{im} varies with time.

As a result, the variance in each orthogonal mode is given by

This mode is arranged in order of maximum magnitude. The total of M empirical orthogonal function corresponds to M eigenvalues. The 1^{st} mode contains the highest percentage of the total variance λ_{i} and so on. If we add up all the total variances in all the time series we get

The total variances in the M time series equation. The total variances contains in the statistical modes. The time dependent amplitude of the i^{th} statistical mode is given by

The eigen values of the two leading modes for the SW, WS and SST are listed in

In addition, climatological mean (16 year averaged) monthly fields and the annual mean field for the SW, WSCL, and SST is also calculated. As a background fields for the trends and interannual variability, the January, July, and annual climatological mean fields are explained.

The seasonal climatological monthly mean fields and annual (ensemble) mean fields of the winds and SST are described respectively as a background field for the trends and interannual variability. The January, July, and annual climatological mean fields are given in Figures 2-4 for the SW, WSCL, and SST, respectively. Complete climatological monthly mean fields for 12 months can be found from, for example, [

south in January and north in July. Above 30˚S latitude the wind patterned are more complex and build a conflict (divergence) between polar air and subtropical air known as the polar fronts, which is present between 50˚S and 60˚S zonal belt. In this region polar easterlies and westerlies together give rise to the convective current. The magnitude of the SW anomalies ranges between 0 to −3.0 m/s during summer and 0 to 3.0 m/s during winter. The major features known to exist in the SW field are the subtropical anticyclonic gyres in the South Pacific, Atlantic, and Indian Oceans, year-round strong westerly surface winds over the SO that becomes positive in July. South of 40˚S, NCEP99 wind stress magnitude is typically about 35% stronger in the Indian Ocean sector of the SO [

The climatological distributions of seasonal SST anomalies are shown in

The linear trend of a variable is calculated from its anomaly values at the times by fitting to a linear relation through least squares method. The time series anomaly mean of parameters averaged over study area is shown in

of −0.0005 ± 0.0003 m/s/decade, 0.0012 ± 0.0002 m/s/decade and −0.00005 ± 0.0001˚C/decade, respectively. The SST does not have obvious trend during 1993 and 2008 period as compared with wind parameters. Time series curve of SST shown continuous high and low peaks in 2 to 3 year interval (may be heat balanced in the SO). The SW anomalies shown increasing trend of 0.0013 ± 0.0002 m/s/decade, with the MC (ZC) exhibiting an increasing (noticeable decreasing) trend. The MC plays a critical role in heat transfer through atmospheric circulation. The WS and WSDV exhibit increasing trend whereas WSCL shown a decreasing trend could be due to decreasing trend of ZC. Correlation coefficient of ZC (

In order to understand the interannaul variability, the mean values and linear trends are first removed from the smoothed values through the 12-month running averages. The EOF analyses performed on winds parameters and SST are explained in details. The variances (eigen values) of the two leading modes for SW, WS, WSDV, WSCL and SST are listed in

Figures 6-8 show the Empirical Orthogonal Functions (EOFs) modes and the associated Time Correlation Function (TCFs) of the SW, WS and WSCL, respectively. The EOF-1 of SW and WS resembles each other and demonstrates convergence event which explains 19% and 25% of their total variances respectively. An eigen value of WS is higher than that of the SW. The convergence occurred above sea surface in sub-polar region due

P | Trends | STD ER (95%) | STD Variances | CC | I |
---|---|---|---|---|---|

decade^{−1} | |||||

SW | 0.0013 | ±0.0002 | 0.1067 33% | 0.654 | −0.178 |

ZC | −0.0005 | ±0.0003 | 0.1269 36% | −0.206 | 0.066 |

MC | 0.0012 | ±0.0002 | 0.0882 30% | 0.728 | −0.167 |

WS | 0.0001 | ±0.00001 | 0.0074 8% | 0.809 | −0.015 |

WSDV | 0.0080 | ±0.0012 | 0.5999 77% | 0.695 | −1.194 |

SST | −0.00005 | ±0.0001 | 0.0375 19% | −0.066 | 0.009 |

Parameters | EOF No. | Eigen Values | Cumult. % | % | Standard Errors |
---|---|---|---|---|---|

SW | 1 | 57.26 | 0.199 | 19 | 8.096 |

2 | 44.59 | 0.354 | 15 | 6.305 | |

WS | 1 | 72.48 | 0.252 | 25 | 10.25 |

2 | 41.59 | 0.396 | 14 | 5.882 | |

WSDV | 1 | 30.91 | 0.108 | 10 | 4.370 |

2 | 26.58 | 0.200 | 09 | 3.759 | |

WSCL | 1 | 32.99 | 0.115 | 11 | 4.665 |

2 | 30.00 | 0.219 | 10 | 4.240 | |

SST | 1 | 64.04 | 0.222 | 22 | 9.050 |

2 | 45.97 | 0.382 | 16 | 6.502 |

to the collision of westerlies and polar easterlies winds developed a low pressure system. Due to collision and chilly conditions air mass rises upward and forms a convective cell. The spatial pattern in EOF-1 of SW in

11% of their total variances with zero lag month and lower correlation coefficient of 0.4771 and it is coincides with negative WS in

In EOF-2 the spatial pattern of SW (

The spatial distribution of SST in EOF-1 (

after removing seasonal variability using 12 months running mean. [

In the current study we investigated seasonal changes, trends and interannual variability on the basis of satellite data during 1993 and 2008 by performing EOF analysis on SW, WS, WSCL and SST anomalies which had been passed through a 12-month running mean filter. January (July) climatology indicates weaker (stronger) SW which is coincident with slightly warmer (cooler) SST. The climatological mean of SW (16 years average) indicates zonal frontal region due to westerlies associated with positive WSCL and polar easterlies associated with negative WSCL. The area-mean time series anomalies of zonal and meridional component of SW and SST have linear trends of −0.0005 ± 0.0003 m/s/decade, 0.0012 ± 0.0002 m/s/decade and −0.00005 ± 0.0001˚C/decade, respectively. The SW anomalies show an increasing trend of 0.0013 ± 0.0002 m/s/decade, with the meridional (zonal) component exhibiting an increasing (decreasing) trend. The WS and WSDV exhibit an increasing trend whereas WSCL shows a decreasing trend. The SST fluctuates close to zero with repeated high and low peaks in an interval of 2 - 3 years.

EOF-1 spatial pattern of SW is characterized by low variances in the polar frontal region along 60˚S zonal section with maximum node confined to Pacific sector of SO. EOF-2 pattern exhibits shifting of polar frontal region more southward. The time coefficient of EOF-1 of SW, WS and WSCL is correlated with reversed AAO and EOF-2 with AAO. EOF-1 TCFs of WS, SST and SW explain 25%, 22% and 19% of their total variances respectively whereas EOF-2 TCFs of SST, SW and WS explain 16%, 15% and 14% of their total variances respectively. The spatial distribution EOF-1 of SST shows high variances in the Indian sector and in the Pacific sector of the SO. The corresponding time coefficient function indicates a reversed relationship with AAO index. EOF-2 of SST shows dipole structure in the Pacific Ocean sector of the SO associated ENSO event; high positive variance is also evident in the Indian sector. The time coefficient of EOF1 (EOF-2) is correlated with inversed AAO (Nino 3.4) index, with the earlier (latter) leading the former by 4 (9) months. Based on the EOF analyses, it can be inferred that AAO and Nino 3.4 indices play an important role in modulating SW and SST changes in the SO. This work signifies that the interannual variability do exist filtering seasonal changes which could be due to the AAO and the ENSO related event as revealed by the EOF analyses of SST in the SO region of the Global Ocean.

The surface winds and SST were obtained from CERSAT/IFREMER and NCEP/NOAA, respectively. We acknowledge the encouragement of the NCAOR Director and Secretary, Ministry of Earth Sciences.