Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
1
Eddy Analysis in the Subtropical Zonal Band of 1
the North Pacific Ocean 2
Yu Liu1,2, 3
Changming Dong3, 4
Yu Ping Guan1, 5
Dake Chen4, 6
James McWilliams3 7
1Key Laboratory of Tropical Marine Environmental Dynamic (LED), South China Sea 8
Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, 510301, China 9 2Graduate University of the Chinese Academy of Sciences, Beijing, 100049, China 10
3Institute of Geophysics and Planetary Physics 11
University of California, Los Angeles, CA 90095, USA 12 4State Key Laboratory of Satellite Oceanic Environment and Dynamics, 13
SIO/SOA, Hangzhou, 310012, China 14
Submitted to JGR-Ocean 15
1 Corresponding author address: Dr. Yu Ping Guan, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou 510301, China. E-mail: [email protected]
2
ABSTRACT 16
There are two zonal bands of eminently high eddy kinetic energy (EKE) in the North 17
Pacific Ocean. The highest one is located in the Kuroshio Extension and the second one 18
is in the subtropical area. This paper is focused on the latter. The following observational 19
data are used: satellite measured sea surface height anomalies (SSHA), sea surface 20
temperature (SST), Argo data and QuikSCAT wind. An eddy detection scheme based on 21
velocity geometry is applied to the SSHA-derived geostrophic currents to identify and 22
track eddies, and generate an eddy dataset in the band, including spatial and temporal 23
information of eddy generation, evolution and termination. Through the analysis of the 24
eddy data set, a series of eddy characteristic parameters are investigated. The eddy 25
location and time information is used to track observed Argo vertical profiles falling in 26
eddy areas, which exposes how eddies impact the thermocline and halocline. The frontal 27
intensity derived from the SST data and wind curls estimated from QuikSCAT wind data 28
are used to explain the mechanism of temporal and spatial eddy variations in the zonal 29
band. 30
3
1. Introduction 31
Mesoscale eddy activity is an outstanding phenomenon in the upper ocean. However, 32
oceanic eddy generation is not spatially uniform, and concentrated in certain special 33
regions. In the Northern Pacific Ocean, particularly, there are two zonal bands of strong 34
eddy activities [Aoki and Imawaki, 1996; Wunsch and Stammer, 1998; Qiu, 1999; Qiu 35
and Chen, 2010]. The spatial distribution of eddy kinetic energy calculated from 36
altimetry-measured sea surface height anomalies (SSHA) with respect to the long-term 37
mean and after high-pass filtering (shorter than 90 days and longer than 7 days) is shown 38
in Fig. 1. These two high EKE zonal bands can be easily identified. The bands are also 39
clearly emerged in high-pass root-mean-square sea surface height anomalies, e.g., Qiu 40
and Chen [2010]. The northern band is collocated with the Kuroshio Extension, and this 41
high EKE band is closely linked to the instability of the Kuroshio after the jet leaves the 42
coastal restraint area and flows to the open ocean, see Qiu and Chen [2010] for a 43
summary. The southern band is located in the subtropical area, extending from east of 44
Luzon Strait all the way to the Hawaii Islands, and this zonal band is the focus of this 45
study. 46
In most previous studies on eddy activities the southern band is further divided into 47
two areas along about 170ºE: the western zone, west of 170°E, is the subtropical counter-48
current zone [Qiu, 1999; Hwang et al., 2004; Liu et al., 2005; Qiu and Chen, 2010; Kang 49
et al., 2010]; the eastern zone, east of 170°E, is the lee side of Hawaii Islands [e.g., Yu et 50
al., 2003; Dong et al., 2009; Yoshida et al., 2010]. There is one exception, Kobashi and 51
Kawamura [2002], studied eddy variations in the entire southern band. It can be easily 52
4
understood that such separation is based on distinctively different mechanisms of eddy 53
generation: in the former area eddies are generated through frontal instability associated 54
with the subtropical front and in the latter area they are due to wind curl in the wakes of 55
wind and oceanic current past the Hawaii islands. However, a significant number of 56
eddies generated in the eastern zone propagate westward and enter the western zone (see 57
Section 3); moreover, wind curl also works as an important driving force generating 58
eddies in western zone (Sec. 5). Therefore, in this study we examine eddy properties in 59
the entire band as a whole. 60
On the other hand, previous studies shown that eddies in the zonal band have impacts 61
on the subtropical gyre [Qiu, 1999], North Pacific Subtropical Mode Water [Uehara et al., 62
2003; Qiu et al., 2007], North Pacific intermediate water [Qiu and Chen, 2011], 63
subtropical ventilation [Endoh et al., 2006; Nishikawa et al., 2010], vertical mixing [Pan 64
and Liu, 2005], and even biological processes [Vaillancourt et al., 2003; Johnson et al., 65
2010]. Several approaches have been used to study the eddy variability in this zonal band. 66
Qiu [1999] and Qiu and Chen [2010] used SSHA anomalies to examine eddy variability 67
on the interannual and seasonal time scales. Noh et al. [2007] and Tsujino et al. [2010] 68
applied numerical models to investigate eddy variability. Kang et al. [2010] generated an 69
eddy dataset through identifying eddies from SSHA data, and provided preliminary 70
statistical results. Two global eddy data mappings and analysis by Chelton et al [2007, 71
2011] include eddy activity information in the area. However, more comprehensive 72
analysis is required for better understanding of characteristic features of eddies and their 73
variability in this band. 74
5
The present study is focused on the area of [15°N-28°N] x [112°E-140°W], see Fig. 75
1; the data used in this study consists of the following observational data: SSHA, SST, 76
Argo and QuikSCAT wind data. An eddy detection scheme based on velocity geometry is 77
applied to SSHA-derived geostrophic current anomalies to detect and track eddies, and 78
then an eddy dataset is set up. A series of statistical analysis is applied to the eddy 79
dataset to explore features of eddy dynamics. 80
In addition, all vertical profiles from Argo data in the study area from 1995-2009 are 81
collected. Information of location and time of eddies detected from the eddy dataset is 82
used as indices to identify Argo vertical profiles falling within these eddies, and these 83
profiles are used to examine the eddy effect on the thermocline and halocline. 84
The rest of the paper is organized as follows: the data and eddy detection scheme used 85
are described in Section 2; statistical analysis of the eddy dataset is presented in Section 86
3; the application of Argo profiles to study eddy effects on the thermocline and halocline 87
is discussed in Section 4; Section 5 discusses eddy generation mechanisms. Finally, the 88
summary is presented in Section 6. 89
2. Data and Eddy detection Scheme 90
91
2.1 Data 92
The following observational data are used in the present paper: satellite measured 93
sea surface height anomalies (SSHA), Argo float-measured temperature (T) and salinity 94
(S) vertical profiles, sea surface temperature (SST) and QuikSCAT wind vectors. 95
Surface geostrophic velocity anomalies are derived from SSHA data using the 96
6
following formula: and , where and are the zonal 97
and meridional components of the geostrophic velocity anomalies, and is the sea level 98
anomaly (SLA) ; is gravitational acceleration and is Coriolis parameter. The data 99
used in this study are taken from AVISO multiple satellite-merged SSHA with a spatial 100
resolution of 1/3°x1/3° and 7-day temporally sampling over the period from January 1993 101
to December 2009, downloaded from www.aviso.oceanobs.com. The derived geostrophic 102
velocities are used for eddy detection and eddy characteristic parameter estimation, 103
including eddy location, size, intensity, path and temporal evolution. 104
Argo float T/S vertical profiles are downloaded from 105
ftp://www.usgodae.org/pub/outgoing/argo/geo/pacific_ocean/. There are totally over 106
42,000 floats in the area of study. The data are sampled daily and the maximum depth 107
reachable by Argo floats is about 2000 meters. Spatial (center location and eddy radius) 108
and temporal information from detected eddies are used as indices to identify Argo 109
vertical profiles falling inside these eddies, and these profiles are used to examine eddy 110
impacts on the thermocline and halocline. 111
The Advance Very High Resolution Radiometer (AVHRR) SST data downloaded 112
from ftp://podaac.jpl.nasa.gov/GHRSST/data/L4/GLOB/NCDC/AVHRR_OI/ are used to 113
examine the subtropical front . The data have a horizontal resolution of 25 km. 114
The monthly QuikSCAT wind data are used to explore possible relationship between 115
wind curl and surface vorticity derived from the satellite measured SSHA data. The wind 116
data have a horizontal resolution of 0.5 degree. They are downloaded from 117
http://www.ifremer.fr/dodsG/CERAST/quickscat_daily. 118
7
2.2 Eddy Detection Scheme 119
Several schemes of eddy detection have been developed in literature; see Nencioli et 120
al. [2010] for a review. In this paper, an eddy detection scheme based on velocity 121
geometry [Nencioli et al., 2010] is used. For an observer moving with the mean velocity, 122
an eddy can be defined as a flow feature where the relative velocity vectors rotate around 123
a center. The velocity fields associated with mesoscale cyclones and anticyclones are 124
characterized by following common features: a velocity minimum in proximity of their 125
centers, tangential velocities increasing linearly proportional to the distance from the 126
center and deceasing after reaching a maximum value. Furthermore, because of the 127
rotational nature of the motion, the (u,v) components of velocity reverses in sign cross the 128
center of the eddy. In this scheme, four constraints were derived in conformance with the 129
general characteristics associated with the eddy’s velocity field and eddy centers at those 130
grid points where all the constraints are satisfied. In addition, an eddy tracking scheme is 131
also included. For more details about the eddy detection scheme the reader is referred to 132
Nencioli et al. [2010]. 133
3. Eddy Analysis 134
135
In the subtropical zonal band, the EKE estimated from the SSHA-derived 136
geostrophic currents varies on seasonal and interannual time scales, see Fig. 2. The EKE 137
in later spring and early summer (May and June) is much higher than that in other 138
seasons. The level of EKE in 1993-1995 and 1999-2002 is much lower than normal. 139
Anomalies of both the sea surface EKE and vorticity propagate westward with a speed of 140
8
about 10 cm/s, see Fig. 3. These results are similar to those obtained by Qiu and Chen 141
[2010], whose study is focused on a smaller domain west of 165°W. 142
A primary contribution to the EKE variability is from nonlinear eddies, which can 143
transport both energy and mass [e.g., Chelton et al., 2007]. To better understand activities 144
of these nonlinear eddies, we apply the eddy detection scheme introduced in Sect. 2.2 to 145
seventeen years (1993-2009) SSHA-derived geostrophic current anomalies. Fig. 4 shows 146
an example of detected eddies on April 28, 2008 (only eddies with radius larger than 50 147
km are displayed). With the eddy detection scheme introduced in Sect 2.2 applied to the 148
above velocity anomaly data, an eddy dataset is generated, and it consists of the 149
following characteristic parameters: eddy size (radius) and boundary, eddy center 150
location (longitude and latitude), eddy polarity, eddy vorticity (averaged over each eddy), 151
eddy kinetic energy (averaged over each eddy). 152
3.1 Eddy Number, Size and Lifetime 153
154
The total number of eddy detected is 136237, including 70493 cyclonic eddies and 155
65744 anticyclonic eddies. However, this total number of eddy includes counting 156
repetition of same eddies at different times. If we count each eddy for its whole life time 157
as a single item, the total number of individual eddy detected is 17756: including 9070 158
cyclonic eddies and 8686 anticyclonic eddies. The total number of cyclonic eddies is 159
about 3% larger than that of anticyclonic eddies. Taking in consideration of possible 160
errors associated with processing SSHA data, in the following analysis we consider 161
eddies with signals lasting longer than or equal to four weeks only. Subject to this 162
9
criterion, there are 4883 cyclonic eddies and 4542 anticyclonic eddies detected in the 163
zonal band. In terms of eddy size, eddy number distribution is a near-Gaussian 164
distribution with peak at 50km for the both cyclonic and anticyclonic eddies, and this is 165
approximately the same as the mean first baroclinic radius of deformation of the zonal 166
band of the ocean, as shown in the upper panel of Fig. 5, where a symmetric shape of the 167
histograms is clearly visible. The histogram of the lifetime of eddies, which longer than 4 168
weeks, is shown in the lower panel of Fig. 5. The numbers of both anticyclonic and 169
cyclonic eddies are approximately symmetric in terms of the lifetime distribution. It is to 170
note that some eddies can survive more than one year. 171
Fig. 6 shows the eddy size distribution, and eddy size is defined as the average for 172
all eddies whose center falls in each 1°x1° bin. Fig. 6 shows the eddy size is maximal at 173
the central latitude and decrease northward and southward. It might be explained with the 174
evolution of eddy radius in Sect. 3.4, where eddies at mature stages always have their 175
largest eddy size during their life time. 176
3.2 Eddy Vorticity 177
Relative vorticity within an eddy varies from its center (the magnitude of the relative 178
vorticity is maximum at the center theoretically) to its boundary (reaches zero). We 179
define the vorticity of an eddy as the maximum vorticity value within the eddy area 180
confined by its boundary. The spatial distribution of vorticity of detected eddies 181
normalized by the background planetary vorticity, i.e., the Coriolis coefficient, at eddy 182
centers in each 1°x1° bin is plotted in Fig. 7. A high vorticity band can be clearly seen in 183
an area centered with the largest values appearing near the Kuroshio region and the lee 184
10
side of Hawaii Islands. 185
An outstanding phenomenon can be identified from Fig. 7: the eddy normalized 186
vorticity increase westward, except at the vicinity of the wakes of the Hawaii Islands. 187
This implies two potential factors: eddies become stronger on their way westward, and 188
eddies generated in the western part of the latitudinal band are stronger than those 189
generated in the eastern part of the latitudinal band. Further to display the varying trend, 190
the meridional averaged normalized eddy vorticity as a function of the longitude is 191
plotted on the bottom panel of Fig. 7, which shows clearly two peaks near Kuroshio and 192
in the lee of Haiwaii and the westward increase in vorticity magnitude. 193
The histogram of eddy vorticity with eddy lifetimes equal to or longer than 4 weeks 194
is shown in Fig. 8; the symmetrical distribution of vorticity intensity against the eddy size 195
for both positive (cyclonic) and negative (anticyclonic) eddies vorticity can be clearly 196
seen. The peaks of normalized vorticity are near 0.2, i.e. the most popular eddies are with 197
a rather weak relative vorticity, on the order of 20% of the mean planetary vorticity. 198
Another noticeable difference between the cyclonic and anticyclonic eddies is the 199
following: the normalized vorticity for the anticyclonic eddies has a slightly higher peak 200
and slightly narrower band of distribution compared with the cyclonic eddies, i.e. the 201
anticyclonic eddies are slightly stronger or more nonlinear than the cyclonic eddies. 202
3.3 Eddy Generation and Termination 203
The information about where and when most eddies are generated and terminated 204
can help us identify mechanisms for eddy generation and termination. We define the first 205
(last) record in the time series of each eddy lifetime as the eddy generation (termination); 206
11
however, for a specified region, some eddies can move into (out of) the area, so that they 207
are not generated (terminated) locally. To exclude those eddies, we remove the first (last) 208
records are found within 0.5° strips along four boundaries from the generation 209
(termination) records. The eddy generation number variations with latitude and longitude 210
are plotted on the upper panel of Fig. 9. In general, eddy generation rate is nearly uniform 211
in this latitudinal band, with a slightly enhancement in the southern and northern 212
boundaries. There is a peak of cyclonic eddy generation near 141oE. A detailed 213
discussion of eddy generation mechanisms will be presented in Sect. 5. The distribution 214
of eddy termination is shown in the lower panel of Fig. 9. The meridional distribution is 215
nearly uniform, which is consistent with the facts that the meridional distribution of eddy 216
generation rate is nearly uniform and these eddies move mostly in the westward direction 217
and eventually die within the same latitudinal band. For eddies generated close to the 218
western boundary, they dissipate their energy and western boundary thus works as a 219
graveyard for these eddies. As a result, the number of eddy termination has a maximum 220
near the western boundary (near the Kuroshio), as shown in the lower-right panel of Fig. 221
9. 222
Fig. 10 shows seasonal variations in eddy generation and termination. There is a 223
peak in early spring (February and March) and a trough in summer (July) for eddy 224
generation. Eddy termination peaks are delayed by one or two months. There is no 225
significant difference between cyclonic and anticyclonic eddies in terms of seasonal 226
variation; however, there are slightly more cyclonic eddies generated than anticyclonic 227
eddies. There is a minimum in eddy generation around the year of 1999-2002 with 228
smaller seasonal variation than other years, and a similar pattern can be seen for the eddy 229
12
termination. 230
3.4 Eddy Evolution 231
An eddy can be characterized by its parameters such as: eddy radius, vorticity, 232
kinetic energy and deformation rate. The eddy kinetic energy is defined as the averaged 233
kinetic energy within an eddy area (defined from the eddy center to its boundary). The 234
eddy deformation rate is defined as , where and are 235
shear deformation rate and stretching deformation rate [Carton, 2001; Hwang et al., 236
2004]. These parameters evolve with time during the lifetime of each eddy. To describe 237
the mean evolution of eddies, we consider eddies longer than 20 weeks, and the total 238
number of such eddies are: 941 cyclonic eddies and 859 anticyclonic eddies. We 239
introduce an eddy age normalized by its lifespan. For each eddy, the time evolution of 240
four basic parameters discussed above can be represented by the time evolution of the 241
non-dimensionized variable based on the corresponding maximum in its lifespan of each 242
eddy. Averaging over all eddies with lifespan longer than 20 weeks, we obtained the 243
normalized temporal evolution for these four parameters, as shown in Fig. 11. It is readily 244
seen that eddy size, vorticity magnitude and kinetic energy increase in its first 1/5 of life 245
cycle (youth) and then stay stable for next 3/5 of its life cycle (adult). In the last 1/5 of the 246
mean life cycle (aged), these parameters decrease sharply. The deformation rate shows 247
the opposite trend, in its first 1/6 of life cycle (youth), the rate decreases and then stays 248
roughly constant for next 2/3 of the life cycle and finally increases sharply before eddies 249
die eventually. The life cycle of vorticity has a similar feature for both the cyclonic and 250
13
anticyclonic eddies. However, the magnitude of the mean vorticity for anticyclonic eddies 251
is larger than that of cyclonic eddies. In addition, the magnitude of the mean vorticity of 252
the anticyclonic eddies is larger than that of cyclonic eddies, and this is consistent with 253
the information presented in Fig. 8. 254
3.5 Eddy Movement 255
Fig. 12 plots trajectories of eddies with lifetime longer than 50 weeks. All these 256
eddies move westward. Some eddies generated near Hawaii islands can move all the way 257
to the area near the Kuroshio. Thus, it is more meaningful to combine the zone of 258
subtropical counter-current and the lee side of Hawaii Islands as a united area in this 259
study as we stated in Introduction. The westward velocity of eddies varies with the 260
latitude. The left panel of Fig. 13 shows that the westward velocity averaged within a 261
band of 1° decreases with the latitude, which is due to the β effect and self advection 262
[McWilliams and Flierl, 1979]. The right panel of Fig. 13 shows the northward velocity 263
averaged a band of 1° varies with latitude, it is equatorward north of 21°N and poleward 264
south of 21°N for both cyclonic and anticyclonic eddies with the speed in 1 cm/s. As the 265
eddy movement is affected by both the mean flow and the β effect, the meridional eddy 266
moving velocity could be affected by the regional mean circulation, which results in a 267
deflection in the meridional direction. The combination of the equatorward movement in 268
the northern half of the band and the poleward movement in the southern half of the band 269
may induce high concentration of eddies of mature stage when eddies reach their largest 270
sizes during their life spans (Fig. 11) in the middle latitude of this zonal band, which 271
results in the large eddy size along the middle latitude as shown in Fig.6. 272
14
273
3.6 Eddy Interaction with the Kuroshio 274
As discussed above, eddies move westward and some long-lived eddies can be 275
traced back to the lee size of Hawaii Islands. When the westward propagating eddies 276
encounter Kuroshio, what will happen? Using a numerical model and SSHA data, Zhai et 277
al. [2010] postulated that the western boundary area is a graveyard for westward 278
propagating eddies. On the other hand, Numerical solution by Sheu et al. [2010] suggests 279
that some eddies can penetrate the Kuroshio and enter the South China Sea. In the 280
following subsection, we will use the eddy dataset to explore this issue in detail. 281
Two cases (one cyclonic eddy and one anticyclonic eddy) are selected to show how 282
an eddy interacts with the Kuroshio. Figure 14 shows a cyclonic eddy moves 283
northwestward and then towards the Luzon Strait. The eddy stays southeast of Taiwan for 284
about three weeks. Finally, it merges into the Kuroshio near southeast of Taiwan. Figure 285
15 displays the movement of an anticyclonic eddy first seen southeast of Taiwan. This 286
eddy eventually moves across the Luzon Strait. 287
To have a more accurate accounting for the eddy interaction with the Kuroshio, we 288
choose a box (120°E ~126°E and 18°N ~23°N) shown in Fig. 16, and the number of 289
eddies moving into or out of the box is listed along each boundary. Note that some eddies 290
might cross boundaries for a few of times. Thus, if an eddy is born outside the box and 291
dies within the box, we record the first time when it enters the box; however, if it dies 292
outside the box, we record the last time when it leaves the box. On the other hand, if an 293
eddy is born in the box, we will only record the time when it leaves box finally. In total, 294
162 cyclonic (146 anticyclonic) eddies enter the box and 105 cyclonic (88 anticyclonic) 295
15
eddies leave the box across the four boundaries. Note that the total numbers of eddies 296
entering and leaving the box are not exactly balanced because some eddies are born/die in 297
the box. In 17 years, 251 cyclonic (265 anticyclonic) eddies are generated in the box and 298
308 cyclonic (323 anticyclonic) eddies terminate in the box. Therefore, when eddy 299
numbers are balanced, about 74% cyclonic (78% anticyclonic) eddies die in the box. The 300
rate of eddy termination in the box seems to support the idea that the region near the 301
Kuroshio is a graveyard for westward eddies as suggested by Zhai et al. [2010]. 302
However it should be noted that in the 17 years, only about 100 eddies from over nine 303
thousands of eddies can reach the region near the Kuroshio when they are generated in 304
the zonal band and propagate westward. In other words, most of them die on the way 305
towards the western boundary region. So in this sense, the western boundary area is not a 306
“graveyard” for westward-propagating eddies. Moreover, it can be seen in the next 307
paragraph that many eddies continue to move westward (cross the Kuroshio) or advect 308
downstream (along the Kuroshio). 309
Among eddies crossing four boundaries, 45 cyclonic (28 anticyclonic) eddies leave 310
the box on the northern boundary, advected by the Kuroshio, in contrast to 8 cyclonic (8 311
anticyclonic) eddies move against the Kuroshio to cross the southern boundary. 49 312
cyclonic eddies (50 anticyclonic) eddies pass through the Luzon Strait into the South 313
China Sea. 97 cyclonic (93 anticyclonic) eddies cross the eastern boundary into the box, 314
in contrast to 3 cyclonic (2 anticyclonic) eddies leave the box eastward. 23 cyclonic 315
eddies (26 anticyclonic) eddies enter the box through the southern boundary. To test the 316
sensitivity of the selection of the western line of the box, we move the western line of the 317
box backward to 121°E, 32 cyclonic (29 anticyclonic) eddies pass through the Luzon 318
16
Strait, that implies some eddies are generated within the strait. Though Li et al. [2007] 319
and He et al. [2010] argued that eddies did not pass the Luzon Strait from the Western 320
Pacific into the South China Sea, and Sheu et al. [2010] argued that under certain 321
conditions eddies can penetrate the Kuroshio into the South China sea, and the statistical 322
results seems to support the latter one. 323
Figure 17 shows the seasonal variation in the number of eddies passing the Luzon 324
strait and those advected northward by the Kuroshio. It is shown that the number of 325
cyclonic eddies moving across the Kuroshio and passing the strait is minimum in the 326
summer when the Kuroshio is the strongest which is in phase with the summer monsoon. 327
However the trend for anticyclonic eddies is quite different from that of the cyclonic 328
eddies. 329
The northward movement of eddies, which is apparently induced by the Kuroshio, 330
does not show clear seasonal variation. Sheu et al. [2010] suggested that whether an eddy 331
can cross the Kuroshio and pass through the Luzon Strait or advected northward by the 332
Kuroshio depends on both the strength and the relative horizontal potential vorticity 333
profile of the Kuroshio. It is clear that a detailed explanation of the above statistical result 334
requires more observational data. 335
4. Eddy Impact on Thermocline and Halocline 336
337 With the altimetry data, we can only see the eddy activities at the sea surface. Argo 338
T/S vertical profiles provide much needed information for the subsurface ocean. In total, 339
36382 Argo vertical profiles are found in the studying zonal band from Sep. 1995 to Dec. 340
17
2009, most of which were deployed after 2000. First, we interpolate recorded temperature 341
and salinity vertical profiles into vertical levels evenly separated from 10 meters to 1000 342
meters with an interval of 10 meters. Temporal and spatial information of the detected 343
eddies are used as indices to select the vertical profiles falling in eddy areas. Two criteria 344
are used: since the SSHA data are weekly-sampled and the Argo record is daily-recorded, 345
we select all Argo profiles whose recording time are within a period of 3 days before and 346
after the time when an eddy is presented in the SSHA data and whose locations are within 347
1.2 time the radius from the eddy center. We identify 1640 vertical profiles within 348
anticyclonic eddies and 1656 within cyclonic eddies. The number of profiles within 349
cyclonic and anticyclonic eddies are almost equal, that is amazing. 350
The mean temperature and salinity vertical profiles for cyclonic and anticyclonic 351
eddies from all Argo profiles in the study area are shown in the panels of Fig. 18. The 352
temperature decreases with depth but the salinity has subsurface maximum at 353
approximately 150 meters below the sea surface. These curves are very close to each 354
other with small differences. To demonstrate the eddy impact on the thermocline and 355
haloclines, the profiles of temperature/salinity anomalies in eddy areas with respect to the 356
mean T/S profiles are shown in the lower panels of Fig. 18. The temperature anomaly 357
profiles show that cyclonic (anticyclonic) eddies induce negative (positive) temperature 358
anomaly which reaches maximum at a depth of 150 meters; the impact of eddies can 359
reach the depth of about 1000 meters. The salinity anomalies profiles show a 360
complicated situation because the salinity maximum is located at 150 meters. Within a 361
cyclonic eddy, high salinity water is pulled upward and water becomes saltier, meanwhile 362
the fresher water below the depth of salinity maximum is also pulled upward and lowers 363
18
the salinity, which results in a thicker layer of fresh water. When an anticyclonic eddy is 364
presented, the fresh water is pushed downward and moved the salinity maximum from 365
150 meters to 200 meters. 366
Temperature (salinity) anomalies in the thermocline (haloclines) discussed above can 367
be carried by westward-propagating eddies, which could affect the heat and salt balance 368
in the ocean [Roemmich and Gilson, 2001]. 369
5. Eddy Generation Mechanisms 370
371 What mechanisms drive eddy generation? As discussed in the introduction, this band 372
can be separated into two regions regulated by different dynamics. The western region is 373
the subtropical frontal zone associated with a weak eastward counter current. Using T/S 374
vertical profile along one section of 137°E, Qiu and Chen [2010] suggested that eddies 375
generation in the western part of the zonal band is due to the baroclinic instability 376
associated with the front. The eastern region is coincident with the lee side of Hawaii 377
Islands where wind curl is strongly affected by the presence of islands. Using SSHA 378
data, the close correlation between wind curl and eddy generation in this region was 379
discussed by Yoshida et al. [2010]. 380
In order to examine eddy generation mechanisms in this zonal band, we analyze the 381
1993-2009 AVHRR SST data to estimate the correlation between SST front and eddy 382
generation variations. The monthly SST meridional gradient ∂T/∂y (averaged for the 383
zonal band) is calculated. The upper panel of Fig. 19 plots the time series of the zonal 384
averaged SST meridional gradient, which shows strong variability in seasonal and 385
19
interannual time scales. When the cross-latitude averaged, we can see clearly that the 386
seasonal variation in the meridional gradient of SST matches the seasonal variation in the 387
number of eddy generation (Fig. 10): a greater magnitude in SST gradient corresponds to 388
a larger number of eddies generated in the early spring; a smaller magnitude in SST 389
gradient corresponds to a less number of eddies generated in summer. The interannual 390
variation in the bottom panel shows that the magnitude of SST gradient was relatively 391
smaller from 1999-2001, and this is consistent with the lower rate of eddy generation 392
shown in Fig. 10. It should be noted that our analysis was carried over the whole band. 393
This further confirms the argument by Qiu and Chen [2010] that baroclinic instability is 394
responsible for the eddy generation. 395
In addition to the baroclinic instability, the vorticity distribution in lee side of 396
Hawaii Island is spatially well correlated with the distribution of wind stress curl: positive 397
vorticity of eddies and positive local wind curls in the northwest of Hawaii Islands, and 398
negative vorticity and wind curl in the southwest of Hawaii Islands (in lee side), by 399
comparing Fig. 20 with Fig. 7. The role of wind curl in the eddy generation in lee side of 400
Hawaii Islands has been extensively discussed in previous literatures, e.g., Calil et al. 401
[2008], Yoshida et al. [2010]. From the distribution of the wind curl in the whole zonal 402
band, we can see there is a persistent patch of positive wind curl on the southern part. The 403
seasonal variation in wind curl agrees very well with that of generation for both cyclonic 404
and anticyclonic eddies. Such agreement implies the wind curl plays a direct or indirect 405
role in eddy generation in the western part of the band. The interannual variation for wind 406
curls is not discussed here because QuikSCAT wind data do not cover the period prior to 407
year 2000. 408
20
6. Summary 409
Using the observational data: SSHA, SST, QuikSCAT wind, Argo T/S vertical 410
profiles, this paper analyzes cohesive eddy activities in one zonal band in the subtropical 411
North Pacific Ocean with the second largest eddy activities (the largest one is located in 412
the Kuroshio extension region). A geometry-based eddy detection scheme by Nencioli et 413
al. [2010] is applied to the SSHA-derived geostrophic currents to identify and track 414
eddies. An eddy dataset is set up, which includes spatial and temporal information of 415
eddy generation, termination, evolution, and a series of eddy characteristics parameters. 416
Eddy properties are presented through a series of statistical analysis. The eddy location 417
and time data are used to track vertical profiles of eddies from the Argo data, which 418
exposes how eddies impact the thermocline and halocline in the area. The SST gradient 419
(frontal intensity) derived the SST data is in association with the eddy generation number 420
in both seasonal and interannual scales. The wind curl variation in the area shows a good 421
relationship with eddy generation not only in the lee side of Hawaii Islands but also in 422
western part of the band, which implies wind curl might play a role in the eddy 423
generation either directly or indirectly 424
Acknowledgments: YL and YPG appreciate supports from National 425
Basic Research Program of China (2007CB411801) and the Knowledge Innovation 426
Program of the Chinese Academy of Sciences (Grant KZCX1-YW-12-4). CD appreciates 427
the support from the National Aeronautics and Space Administration (grant 428
NNX08AI84G). The work was partially done when YL visited CD at UCLA and working 429
21
with CD in 2010. YL and CD appreciate the support from the State Key Laboratory of 430
Satellite Oceanic Environment and Dynamics, Second Institute of Oceanography, SOA, 431
China. YPG thanks Joint Institute for Regional Earth System of UCLA for the host of 432
YPG’s visit at UCLA. We thank Dr. Rui Xin Huang from Woods Hole Oceanographic 433
Institute for his careful reading of and comments on the manuscript. 434
22
References 435
436
Aoki, S. , and S. Imawaki (1996), Eddy Activities of the Surface Layer in the Western 437
North Pacific Detected by Satellite Altimeter and Radiometer, J. Oceanogr., 52, 438
457- 474. 439
Calil, P. H. P., K. Richards, Y. Jia, and R. Bidigare (2008), Eddy activity in the Lee of the 440
Hawaiian Islands, Deep Sea Res., II, 55, 1179-1194, 441
doi:10.1016/j.dsr2.2008.01.008. 442
Carton, X. (2001), Hydrodynamical modeling of oceanic vortices, Surv. Geophys., 22, 443
179-263, doi: 10.1023/A:1013779219578. 444
Chelton, D. B., M. G. Schlax, R. M. Samelson, and R. A. deSzoeke (2007), Global 445
observations of large oceanic eddies, Geophys. Res. Lett., 34, L15606, 446
doi:10.1029/2007GL030812. 447
Chelton, D. B., M. G. Schlax, and R. M. Samelson (2011), Global observations of 448
nonlinear mesoscale eddies, Progr. Oceanogr., in press. 449
Dong, C., T. Mavor, F. Nencioli, S. Jiang, Y. Uchiyama, J. C. McWilliams, T. D. Dickey, 450
M. Ondrusek, H. Zhang, and D. K. Clark (2009), An oceanic cyclonic eddy on the 451
lee side of Lanai Island, Hawai'i, J. Geophys. Res., 114, C10008, 452
doi:10.1029/2009JC005346. 453
Endoh, T., Y. Jia, and K. J. Richards (2006), Sensitivity of the ventilation process in the 454
North Pacific to eddy-induced tracer transport, J. Phys. Oceanogr., 36, 1895-1911, 455
doi: 10.1175/JPO2941.1. 456
He, Y., S. Cai, and S. Wang (2010), The correlation of the surface circulation between 457
the Western Pacific and the South China Sea from satellite altimetry data, 458
23
International J. Remote Sensing, 31, 4757-4778, doi: 459
10.1080/01431161.2010.485137. 460
Hwang, C., C.-R. Wu, and R. Kao (2004), TOPEX/Poseidon observations of mesoscale 461
eddies over the Subtropical Countercurrent: Kinematic characteristics of an 462
anticyclonic eddy and a cyclonic eddy, J. Geophys. Res., 109, C08013, 463
doi:10.1029/2003JC002026. 464
Johnson, Gregory C., Kristene E. McTaggart (2010), Equatorial Pacific 13°C water 465
eddies in the eastern subtropical south Pacific Ocean, J. Phys. Oceanogr., 40, 466
226-236, doi: 10.1175/2009JPO4287.1. 467
Kang, L., F. Wang, and Y. Chen (2010), Eddy generation and evolution in the North 468
Pacific Subtropical Countercurrent (NPSC) zone, Chinese J. Ocean. Limn., 28, 469
968-973, doi: 10.1007/s00343-010-9010-9. 470
Kobashi, F., and H. Kawamura (2002), Seasonal variation and instability nature of the 471
North Pacific Subtropical Countercurrent and the Hawaiian Lee Countercurrent, J. 472
Geophys. Res., 107(C11), 3185, doi:10.1029/2001JC001225. 473
Li, L., C.S. Jing, and D.Y. Zhu (2007), Coupling and propagating of mesoscale sea level 474
variability between the western Pacific and the South China Sea, Chinese Sci. 475
Bull., 52, 1699-1707, doi: 10.1007/s11434-007-0203-3. 476
Liu, Q., D. Souza, Y. Jia, and W. Liu (2005), Eddies in the Northwest Subtropical Pacific 477
and Their Possible on the South China Sea, J. Ocean University of China, 4, 329-478
333. 479
McWilliams, J. C., and G. R. Flierl (1979), On the evolution of isolated non-linear 480
vortices. J. Phys. Oceanogr., 9, 1155-1182, doi: 10.1175/1520-‐481
24
0485(1979)009<1155:OTEOIN>2.0.CO;2. 482
Nencioli, F., C. Dong, T. Dickey, L. Washburn, and J. McWilliams (2010), A vector 483
geometry based eddy detection algorithm and its application to high-resolution 484
numerical model products and High-Frequency radar surface velocities in the 485
Southern California Bight, J. Atmos. Ocean. Technol., 27, 564-579, 486
doi:10.11.1175/2009JTECHO725.1. 487
Nishikawa, S., H. Tsujino, K. Sakamoto, and H. Nakano (2010), Effects of mesoscale 488
eddies on subduction and distribution of subtropical mode water in an eddy-489
resolving OGCM of the Western North Pacific, J. Phys. Oceanogr., 40, 1748-490
1765, doi: 10.1175/2010JPO4261.1. 491
Noh, Y., B. Y. Yim, S. H. You, J. H. Yoon, and B. Qiu (2007), Seasonal variation of 492
eddy kinetic energy of the North Pacific Subtropical Countercurrent simulated by 493
an eddy-resolving OGCM, Geophys. Res. Lett., 34, L07601, 494
doi:10.1029/2006GL029130. 495
Pan, A., and Q. Liu (2005), Mesoscale eddy effects on the wintertime vertical mixing in 496
the formation region of the North Pacific Subtropical Mode Water, Chinese Sci. 497
Bull., 50, 1949—1956. 498
Qiu, B. (1999), Seasonal eddy field modulation of the north pacific subtropical 499
countercurrent: TOPEX/Poseidon observations and theory, J. Phys. Oceanogr., 29, 500
2471-2486, doi: 10.1175/1520-0485(1999)029<2471:SEFMOT>2.0.CO;2. 501
Qiu, B., and S. Chen (2010), Interannual variability of the North Pacific Subtropical 502
Countercurrent and its associated mesoscale eddy field, J. Phys. Oceanogr., 40, 503
213-225, doi: 10.1175/2009JPO4285.1. 504
25
Qiu, B., and S. Chen (2011), Effect of decadal Kuroshio extension jet and eddy 505
variability on the modification of North Pacific intermediate water, J. Phys. 506
Oceanogr., 41, 503-515, doi: 10.1175/2010JPO4575.1. 507
Qiu, B., S. Chen, and P. Hacker (2007), Effect of Mesoscale Eddies on Subtropical Mode 508
Water Variability from the Kuroshio Extension System Study (KESS), J. Phys. 509
Oceanogr., 37, 982–1000. 510
Roemmich, D., and J. Gilson (2001), Eddy transport of heat and thermocline waters in the 511
north Pacific: A key to interannual/decadal climate variability? J. Phys. 512
Oceanogr., 31, 675-687, doi: 10.1175/1520-513
0485(2001)031<0675:ETOHAT>2.0.CO;2. 514
Sheu, W.-J., C.-R. Wu, and L.-Y. Oey (2010), Blocking and westward passage of eddies 515
in the Luzon Strait, Deep Sea Res., II, 57, 1783–1791, 516
doi:10.1016/j.dsr2.2010.04.004. 517
Tsujino, H., S. Nishikawa, K. Sakamoto, H. Nakano, and H. Ishizaki (2010), Mesoscale 518
eddy statistics and implications for parameterization refinements from a diagnosis 519
of a high resolution model of the North Pacific, Ocean Modell., 33, 205-223, 520
doi:10.1016/j.ocemod.2010.02.004. 521
Uehara, H., T. Suga, K. Hanawa, and N. Shikama (2003), A role of eddies in formation 522
and transport of North Pacific Subtropical Mode Water, Geophys. Res. Lett., 523
30(13), 1705, doi:10.1029/2003GL017542. 524
Vaillancourt, R. D., J. Marra, M. P. Seki, M. L. Parsons, and R. R. Bidigare (2003), 525
Impact of a cyclonic eddy on phytoplankton community structure and 526
photosynthetic competency in the subtropical North Pacific Ocean, Deep Sea 527
26
Res., I, 50, 829–847, doi: 10.1016/S0967-0637(03)00059-1. 528
Wunsch, C., and D. Stammer (1998), Satellite altimetry, the marine geoid and the oceanic 529
general circulation, Annu. Rev. Earth Planet. Sci., 26, 219-54, doi: 530
10.1146/annurev.earth.26.1.219. 531
Yoshida, S., B. Qiu, and P. Hacker (2010), Wind generated eddy characteristics in the lee 532
of the island of Hawaii, J. Geophys. Res., 115, doi:10.1029/2009JC005417. 533
Yu, Z., N. Maximenko, S.-P. Xie, and M. Nonaka (2003), On the termination of the 534
Hawaiian Lee Countercurrent, Geophys. Res. Lett., 30(5), 1215, 535
doi:10.1029/2002GL016710. 536
Zhai, X., H. L. Johnson, and D. P. Marshall (2010), Significant sink of ocean-eddy 537
energy near western boundaries, Nature Geosci., 3, 608-612, 538
doi:10.1038/ngeo943. 539
27
Figure captions 540
541
Figure 1. The spatial distribution of the high-pass (shorter than 90 days and longer than 7 542
days) EKE (in unit of cm2/s2) spatial distribution in the Northern Pacific Ocean, 543
calculated from altimeter SSHA and averaged over the period of 1993~2009. The AVISO 544
data with a resolution of 1/3 x 1/3 degree spatial resolution and 7 day temporal resolution 545
are used. The rectangular area marked by black lines is focus of the present study 546
(15°N~28°N, 115°E-150°W). 547
Figure 2. Upper Panel: time series of EKE in the study area. The solid line is the 7 days 548
sampled data and the dashed line is the result after 52-week smoothing, which shows the 549
interannual variability. Lower panel: seasonal variation of the EKE obtained from the 550
upper panel through monthly averaging. 551
Figure 3. Hovmoeller plots for EKE (left panel) and normalized vorticity (right panel) in 552
a band of 21°N~23°N. Vorticity is normalized by the background planetary vorticity 553
averaged over the band of 21°N~23°N. 554
Figure 4. A snapshot of eddy distribution on April 30, 2008 for eddy sizes larger 50km. 555
Red and blue dots are denoted to centers of anticyclonic and cyclonic eddies, 556
respectively. The flow field is velocity anomalies derived from SSHA data. 557
Figure 5. Upper panel: histogram of eddy number (for each 10-km bin) against eddy size, 558
where the positive (negative) eddy sizes denote cyclonic (anticyclonic) eddies, 559
respectively. Lower panel: the histogram of eddy number against eddy lifetime. 560
Figure 6. Eddy size distribution: cyclonic (upper panel) and anticyclonic (lower panel). 561
The eddy sizes averaged over 1° x1° bins are displayed in the figure. Unit: km. 562
Figure 7 Top panel: the same as Fig. 6, except for the normalized cyclonic eddy vorticity; 563
28
middle panel: the same as upper panel but for anticycloninc eddies; bottom panel: the 564
meridional mean normalized eddy vorticity as a function of the longitude. 565
Figure 8. Histogram of eddy normalized vorticity with a bin width of 0.02 (only eddies 566
with lifetime equal to or longer than 4 weeks are selected). 567
Figure 9. Left panels: number of eddy generation/termination for each 0.5-degree latitude 568
bin (zonally averaged). Right panels: number of eddy generation/termination for each 2-569
degree longitude bin (meridionally averaged). 570
Figure 10. Upper two panels: seasonal variation of the number of eddy 571
generation/termination. Lower two panels: interannual variation of the number of eddy 572
generation/termination. 573
Figure 11. The time evolution of mean eddy characteristic parameters: radius (upper-574
left), vorticity (upper-right), kinetic energy (lower-left) and deformation (lower-right). 575
Each eddy’s age is normalized by its life span. Each parameter of each eddy is 576
normalized its maximum magnitude of the parameter, and the mean eddy parameters 577
obtained by averaging over eddies with lifespan longer than 20 weeks are plotted in the 578
figure. Dashed and solid lines denote cyclonic and anticyclonic eddies, respectively. 579
Figure 12. The eddy trajectories (for eddies with lifetime ≥ 50 weeks); red lines depict 580
trajectories of cyclonic eddies and blue lines for trajectories of anticyclonic eddies; the 581
solid points are the starting positions, and the star points for the ending positions. 582
Figure 13. Left panel: westward speed (cm/s) of eddies; right panel: northward speed 583
(cm/s) of eddies; the solid line for cyclonic eddies and the dashed line for anticyclonic 584
eddies. These curves represent the mean for eddies with lifetime ≥ 4 weeks. 585
Figure 14. The time evolution of a westward cyclonic eddy, which is eventually blocked 586
29
by the Kuroshio. 587
Figure 15. The time evolution of an eddy, which eventually passes through the Luzon 588
Strait. 589
Figure 16. Eddy number budget for a square area enclosed the Luzon Strait: left panel for 590
the cyclonic eddies and right panel for anticyclonic eddies. Two western boundaries are 591
selected to test the sensitivity of the calculation: the solid line is located at 120°E and the 592
dashed line at 121°E. 593
Figure 17. Seasonal variation in number of eddies crossing the western boundary 594
(120°E) (left panel) and the northern boundary (23 °N) (right panel) leaving the box 595
depicted in Fig. 16. 596
Figure 18. MeanT/S profiles from Argo data within the study area. Upper-left panel: 597
mean temperature; upper-right: mean salinity; lower-left: mean temperature anomaly 598
(deviation from the mean temperature profile); lower-right: mean salinity anomaly. 599
Figure 19. Upper panel: Monthly averaged SST meridional gradient (in unit of 600
ºC/110km, average the band of 115°E~150°W, 15°N~28°N) from AVHRR 1993 to 601
2009 with 25 km in resolution. Bottom panel: seasonal variation in the meridional SST 602
gradient. 603
Figure 20. Upper panel: mean wind curl (calculated from QuikSCAT wind data) 604
distribution in the zonal band. Lower panel: monthly mean wind curl and number of eddy 605
generated. 606
30
Figure 1. The spatial distribution of the high-pass (shorter than 90 days and longer than 7
days) EKE (in unit of cm2/s2) spatial distribution in the Northern Pacific Ocean,
calculated from altimeter SSHA and averaged over the period of 1993~2009. The AVISO
data with a resolution of 1/3 x 1/3 degree spatial resolution and 7 day temporal resolution
are used. The rectangular area marked by black lines is focus of the present study
(15°N~28°N, 115°E-150°W).
31
Figure 2. Upper Panel: time series of EKE in the study area. The solid line is the 7 days
sampled data and the dashed line is the result after 52-week smoothing, which shows the
interannual variability. Lower panel: seasonal variation of the EKE obtained from the
upper panel through monthly averaging.
32
Figure 3. Hovmoeller plots for EKE (left panel) and normalized vorticity (right panel) in
a band of 21°N~23°N. Vorticity is normalized by the background planetary vorticity
averaged over the band of 21°N~23°N.
33
Figure 4. A snapshot of eddy distribution on April 30, 2008 for eddy sizes larger 50km.
Red and blue dots are denoted to centers of anticyclonic and cyclonic eddies,
respectively. The flow field is velocity anomalies derived from SSHA data.
34
Figure5. Upper panel: histogram of eddy number (for each 10-km bin) against eddy size,
where the positive (negative) eddy sizes denote cyclonic (anticyclonic) eddies,
respectively. Lower panel: the histogram of eddy number against eddy lifetime.
35
Figure 6. Eddy size distribution: cyclonic (upper panel) and anticyclonic (lower panel).
The eddy sizes averaged over 1° x1° bins are displayed in the figure. Unit: km.
36
Figure 7. Top panel: the same as Fig. 6, except for the normalized cyclonic eddy vorticity;
middle panel: the same as upper panel but for anticycloninc eddies; bottom panel: the
meridional mean normalized eddy vorticity as a function of the longitude.
37
Figure 8. Histogram of eddy normalized vorticity with a bin width of 0.02 (only eddies
with lifetime equal to or longer than 4 weeks are selected).
38
Figure 9. Left panels: number of eddy generation/termination for each 0.5-degree
latitude bin (zonally averaged). Right panels: number of eddy generation/termination for
each 2-degree longitude bin (meridionally averaged).
39
Figure 10. Upper two panels: seasonal variation of the number of eddy
generation/termination. Lower two panels: interannual variation of the number of eddy
generation/termination.
40
Figure 11 The time evolution of mean eddy characteristic parameters: radius (upper-left),
vorticity (upper-right), kinetic energy (lower-left) and deformation (lower-right). Each
eddy’s age is normalized by its life span. Each parameter of each eddy is normalized its
maximum magnitude of the parameter, and the mean eddy parameters obtained by
averaging over eddies with lifespan longer than 20 weeks are plotted in the figure.
Dashed and solid lines denote cyclonic and anticyclonic eddies, respectively.
41
Figure 12. The eddy trajectories (for eddies with lifetime ≥ 50 weeks); red lines depict
trajectories of cyclonic eddies and blue lines for trajectories of anticyclonic eddies; the
solid points are the starting positions, and the star points for the ending positions.
42
Figure 13. Left panel: westward speed (cm/s) of eddies; right panel: northward speed
(cm/s) of eddies; the solid line for cyclonic eddies and the dashed line for anticyclonic
eddies. These curves represent the mean for eddies with lifetime ≥ 4 weeks.
43
Figure 14. The time evolution of a westward cyclonic eddy, which is eventually blocked
by the Kuroshio.
44
Figure 15. The time evolution of an eddy, which eventually passes through the Luzon
Strait.
45
Figure 16. Eddy number budget for a square area enclosed the Luzon Strait: left panel for
the cyclonic eddies and right panel for anticyclonic eddies. Two western boundaries are
selected to test the sensitivity of the calculation: the solid line is located at 120°E and the
dashed line at 121°E.
46
Figure 17. Seasonal variation in number of eddies crossing the western boundary (120°E)
(left panel) and the northern boundary (23 °N) (right panel) leaving the box depicted in
Fig. 16.
47
Figure 18. MeanT/S profiles from Argo data within the study area. Upper-left panel:
mean temperature; upper-right: mean salinity; lower-left: mean temperature anomaly
(deviation from the mean temperature profile); lower-right: mean salinity anomaly.
48
Figure 19. Upper panel: Monthly averaged SST meridional gradient (in unit of
ºC/110km), average the band of 115°E~150°W, 15°N~28°N) from AVHRR 1993
to 2009 with 25 km in resolution. Bottom panel: seasonal variation in the meridional
SST gradient.
49
Figure 20. Upper panel: mean wind curl (calculated from QuikSCAT wind data)
distribution in the zonal band. Lower panel: monthly mean wind curl and number of
eddy generated.