1
2 Geophysical Research Letters
3 Supporting Information for
4
5
Multi-core structure of the North Pacific subtropical mode water from enhanced
Argo observations
6
7 Wendian Gao 1 , Peiliang Li 1 , Shang-Ping Xie 2,1 , Lixiao Xu 1 , Cong Liu 1
8
9
1 Physical Oceanography Laboratory, Ocean University of China, Qingdao, China, 2 Scripps Institution of
Oceanography, University of California San Diego, La Jolla, California
10
11
12
13
14
15
Discussions about the DO data processing and Figures S1 to S11.
16
17
18
19
This supporting information provides DO data processing and the figures which are helpful to understand the main article.
26
27
28
29
20
21
22
23
24
25

56
57
58
59
60
61
52
53
54
55
67
68
69
70
71
62
63
64
65
66
45
46
47
48
41
42
43
44
49
50
51
34
35
36
37
30
31
32
33
38
39
40
DO data processing:
PROVOR-DO-I float is equipped with the Oxygen Optode AANDERAA 4330 sensor for measuring dissolved oxygen. The oxygen sensors had been tested before deploying the floats according to the user manual (http://www.nke-corporate.fr/). The
DO data obtained from Argo floats which failed the range test, spike test or gradient test according to the "Argo quality control manual (version 2.9)" [ Wong et al.
, 2014] had been excluded. Data of one float, whose values are out of the mean value by minus or plus three standard deviations at each pressure during the study period, were also removed. Finally, the selected data were interpolated at 1m resolution vertically.
During the shipboard hydrographic survey, we first deployed 17 Argo floats in two days (26-27 March), and then carried out several zonal stations around the deploying area in a week. The conductivity-temperature-depth profiler (CTD) was equipped with water samplers, and we analyzed the dissolved oxygen content at depths of 100 200
300 500 700 1000m using the "colorimetric determination by iodometry" method.
Figure S1 shows the locations of CTD stations and Argo floats on the first few days.
The two nearest stations and floats (C4 station and Argo float 16 on 28 March, D4 station and Argo float 9 on 29 March) are chosen to check the DO at the beginning
(Figure S2). The two measurements match very well in the vertical.
We also compared the DO data of Argo float 2901552 with nearby CTD stations
(C01, C03, C06, and C12-17) from the KS-14-7 cruise in May 2014 on R/V
Shinsei-maru of the Japan Agency for Marine-Earth Science and Technology (Figure
S3). The cruise report can be seen at http://ocg.aori.u-tokyo.ac.jp/member/eoka/ks-14-7/report/KS-14-7report.pdf. The rate of sensor drift is estimated at the deep water layer (


=27.0 kg/m 3 ) where DO values should be stable and are rarely influenced by the upper ocean. The mean values and standard deviations of the CTD stations and Argo float 5 on the 27.0 kg/m 3 isopycnal are (82.41
± 1.54
 mole/kg) and (72.56
± 0.71
 mole/kg), respectively. The drift seems significant beyond the time and spatial variations. The drift rate is
(82.41-72.56)/53=0.1858
 mole kg -1 day -1 . We applied a linear calibration to all the
Argo observations. Figure S4 and S5 show the time series of the uncalibrated and calibrated DO on isopycnals, respectively. The calibrated DO values increase sharply especially in deep layers. For example, it changes from 86
 mole/kg to 109
 mole/kg on the 26.9-27.0 kg/m3 layer from 27 March to 10 August 2014 . Since the water in the deep ocean is isolated from the sea surface and has no oxygen source, the DO is expected to decrease gradually (as in Figure S4). This linear calibration is thus judged unsuccessful. Some meso-scale or micro-scale processes may also contribute to the differences of the CTD and Argo observations. With only DO measurements of CTD stations at the beginning and the middle of the study period, we are forced to use the
DO data uncalibrated. We note that the instrument drift only causes errors in the absolute values but not in the vertical variations, which are the focus of our study.

72
73
74
75
Figure S1 . Locations of CTD stations (red dots) and Argo floats (black triangles) on the first few days. Number 1~17 denote the Argo floats WMO ID 2901550~2901566.
76
77
78
79
80
Figure S2 . Vertical profiles of DO obtained from Argo floats (blue lines) and water samples (red stars) at CTD stations on (a) 28 March, and (b) 29 March 2014. The
WMO ID 2901566 for Argo 16 and 2901558 for Argo 9.

81
82
83
84
85
86
87
Figure S3.
(a) Locations of CTD stations (red dots) of KS-14-7 cruise on 15-23 May
2014 and the nearby Argo floats (triangles) on 10-27 May. (b) Average of DO of the chosen stations (C01, C03, C06, and C12-17) and Argo float 5 (WMO ID 2901554) are shown by red and black curves, respectively. The standard deviation is shown by shading.
88
89
90
91
92
Figure S4 . Time series of the average DO (uncalibrated) on the isopycnals. Color lines represent the 16 floats. Average DO of all the floats are shown by thick black lines, and the mean value plus or minus one standard deviation are shown by thick grey lines. The dashed red line points out the date of 10 August 2014.

93
94
95
96
Figure S5 . Same as Figure S4 but for calibrated DO.
97
98
99
Figure S6 . Same as Figure 1a in the paper, but for non-smoothed potential vorticity.

100
101
102
103
104
105
106
Figure S7. Profiles of vertical band-passed (a) PV and (b) AOU near the AOU minima from all the Argo floats (thin curve). The thick curves denote the mean value of PV
(magenta) and AOU (green). The Y-axis represents the distance (-50 ~ 50m) relative to the AOU minima (a multi-core profile is used more than once in forming the composite). The mean PV minimum is at -5m.
107
108
109
110
111
112
113
Figure S8. Depth-Time map of PV from Argo float WMO ID 2901566 (shading).
Magenta curve denote the MLD. Potential density contours are drawn in white color with an interval of 0.2 kg/m 3 .
The erosion of the seasonal thermocline [
., 2006] and the strong forcing events inferred from the MLD [
2014 ] are obvious.

114
115
116
117
118
Figure S9. Locations of (a) light, (b) medium, (c) young dense, and (d) old dense core from 27 March to 10 August 2014. The color dots denote the date and the background shading is the bathymetry.
119
120
121
122
Figure S10. (a) Average of PV maximum in the 25.25-25.35 kg/m
3
layer of all the profiles from 16 Argo floats (red line). (b) Average of
∂
∂𝑧
(
∂𝑄
) at the depth of PV
∂𝑧 maximum of all the profiles (red line). Blue lines show the mean value minus or plus

123
124
125
126
127
128
129 one standard deviation and the dashed lines are the approximating polynomials of degree 3 for red lines.
The PV value decreases by 0.5
×10 -10 m -1 s -1 in about 60 days with the second derivative of PV at -1~-0.5
×10 -11
m
-3 s
-1
. Based on the vertical diffusion equation for
PV (
):
∂𝑄
∂𝑡
=
∂
∂
(𝐾 𝑣
∂𝑄
) , the vertical eddy diffusion (
v
) is estimated to be 0.96~1.93
∂𝑧
×10 -6
m
2 s
-1
at the 25.3 kg/m
3
isopycnal.
130
131
132
Figure S11. T-S diagram of the STMW cores.
133
134 Qiu, B., P. Hacker, S. Chen, K. A. Donohue, D. R. Watts, H. Mitsudera, G. H. Nelson,
135
136 and S. R. Jayne (2006), Observations of the subtropical mode water evolution from the Kuroshio Extension System Study.
,
(3), 457-473,
137
139 doi: http://dx.doi.org/10.1175/JPO2849.1.
138 Rainville, L., S. R. Jayne, and M. F. Cronin (2014), Variations of the North Pacific subtropical mode water from direct observations.
,
(8): 2842-2860, doi:

140 10.1175/JCLI-D-13-0022 7.1.
141 Wong, A., R. Keeley, T. Carval, and Argo Data Management Team (2014), Argo
142 quality control manual, doi:10.13155/33951.