Evaluation of sea surface salinity
observed by Aquarius
Hiroto Abe, Naoto Ebuchi
(Hokkaido University, Japan)
1
8th Aquarius/SAC-D
Science Team Meeting
(November 13th, 2013)
Objective of this study(1/1)
Validate SSS observed by Aquarius using
various salinity data including Argo observation.
2
Data (1/2) - Level 2 -
Satellite salinity
Aquarius SSS (beam1, 2, 3)
1) V2.3.1 : NASA/JPL PO.DAAC
2) CAP V2.3.1 : NASA/JPL, Dr. Simon Yueh
3) RSS V3 : Remote Sensing Systems
In situ salinity
1) Argo salinity : Global Data Assembly Center,
real-time mode data
2) TAO/TRITON, PIRATA, RAMA buoys
3
Data (2/2) - Level 3 -
Satellite salinity
Aquarius Level 3 SSS (V1.3, V2.2, V2.3.1, CAP V2.3.1)
NASA/JPL PO.DAAC
Salinity data
1) JAMSTEC Argo optimal interpolation system : MOAA/GPV
(Argo salinities are interpolated based on World Ocean Atlas 2001)
2) JMA data assimilation system : MOVE/MRI.COM
(In-situ and satellite altimeter data are assimilated into it)
4
Comparison with in situ data (1/3)
5
Aquarius (Level2) v.s. Argo float
rms difference =
0.44 psu
6
Comparison with in situ data (1/3)
1) wind speed < 15m/s
2) Argo temperature > 5C)
3) Argo depth < 12.5 dbar
4) rad_land_frac < 0.0005
5) rad_ice_frac < 0.0005
Period : 25 Aug 2011 – 31 Oct 2013
Matchup condition : 200 km, 12 h
Argo salinity (psu)
Aquarius S
SS
(psu)
V2.3.1, beam1
7
Comparison with in situ data (2/3)
Rms difference =
0.44 0.55 0.44
Aquarius S
SS
(psu)
Argo salinity (psu)
V2.3.1 CAP V2.3.1 RSS V3
beam1
Comparison with in situ data (3/3)
8
Aquarius (Level2) v.s. moored buoys
9
Comparison with in situ data (3/3)
V2.3.1 CAP V2.3.1 RSS V3
beam1
Aquarius S
SS
(psu)
buoy salinity (psu)
Rms difference =
0.43 0.49 0.47
10
V2.3.1 CAP V2.3.1 RSS V3
-0.8 +0.8 0.0
Sep. 2011
Mar. 2012
psu
Ascending minus descending seasonality (1/2)
11
CAP V2.3.1
RSS V3
60N
60S
0
The bias is small
for the CAP product.
Ascending minus descending seasonality (2/2)
J M M J S N N S
2011 2012
+0.4
-0.4
0.0
(psu)
V2.3.1 la
titu
de
J M
2013
Residual analysis (1/6) -SST-
12
Argo temperature(degC)
resid
ual (p
su)
2
-2 0
30 20 10 0
V2.3.1
RSS V3
CAP V2.3.1
residual = AQ - Argo
Residual analysis (2/6) -SST-
13
bias stddev 1.0
-1.0
0.0
0
1.0
10 20 30 Argo temperature(degC)
The stddev is large under low SST condition.
2.0
0.0
Argo temperature(degC) 0 10 20 30
V2.3.1 CAP RSS V2.3.1 CAP RSS
Residual analysis (3/6) -wind speed-
14
scatt. wind speed (m/s)
2
-2 0
20 10 0
V2.3.1
RSS V3
CAP V2.3.1
residual = AQ - Argo
resid
ual (p
su)
Residual analysis (3/6) -wind speed-
15
bias 1.0
-1.0
0.0
0 10 20
V2.3.1 CAP
stddev
1.0
2.0
0.0 0 10 20
V2.3.1 CAP RSS RSS
scatt. wind speed (m/s) scatt. wind speed (m/s)
The stddev is large under strong wind condition.
16
Rms difference is large under low SST and strong wind conditions.
The CAP product is sensitive to SST and wind speed.
V2.3.1 CAP V2.3.1 RSS V3
Argo temperature(degC)
Win
d s
pee
d (
m/s
)
rms difference
0.0
0.4
0.8
1.2
beam1
Residual analysis (5/6) -SST and wind speed-
rms d
iffe
ren
ce
(p
su
)
Residual analysis (6/6) -SST and wind speed-
17
V2.3.1 CAP RSS
Argo temperature(degC)
Win
d s
pe
ed
(m
/s)
beam1
beam2
beam3
Rms difference depends on beam for the CAP product.
Comparison with JAMSTEC Argo OI (1/3)
18
V2.3.1
JAMSTEC Argo OI (10m)
V2.3.1 – JAMSTEC Argo OI
Residual SSS is negative in
low and mid latitudes
rms is 0.25 psu (>0.2 psu)
Jan. 2012
19
CAP V2.3.1
JAMSTEC Argo OI (10m)
CAP – JAMSTEC Argo OI
Comparison with JAMSTEC Argo OI (2/3)
Negative bias is small
for the CAP product.
Jan. 2012
rms is 0.18 psu (<0.2 psu)
20
2011
Comparison with JAMSTEC Argo OI (3/3)
2012 2013
Jan. 2012
Rms difference is the smallest for the CAP product.
Rms difference (psu) (40S-40N)
Comparison with JMA assimilation system (1/3)
21
V2.3.1
MRI.COM (1m)
V2.3.1 – MRI.COM
Negative bias
rms is 0.23 psu (>0.2 psu)
Jan. 2012
22
CAP V2.3.1
CAP – MRI.COM
MRI.COM (1m)
Less negative bias
Comparison with JMA assimilation system (2/3)
rms is 0.18 psu (<0.2 psu)
Jan. 2012
23
2011 2012 2013
Rms difference is the smallest for the CAP product.
Comparison with JMA assimilation system (3/3)
Jan. 2012 Rms difference (psu) (40S-40N)
Summary (1/1)
Aquarius SSSs, retrieved by recent algorisms, were validated
using various salinity data (V2.3.1, CAP V2.3.1, RSS V3).
24
Rms difference was 0.43 – 0.55 psu.
The ascending – descending bias was small for the CAP product.
Rms difference depended on beam for the CAP product.
The negative biases found in the low/mid latitudes were small for
the CAP product.
Rms difference was 0.21 psu (CAP) and 0.23 psu (the others).
Level-2 SSS v.s. Argo, moored buoy
Level-3 SSS v.s. JAMSTEC OI, JMA/MRI.COM
25
Backup Slide
26
Comparison with in situ data (1/4)
Argo salinity (psu)
V2.3.1 CAP V2.3.1 RSS V3
beam1
Rms difference
0.44 0.55 0.44
Aquarius S
SS
(psu)
27
Comparison with in situ data (2/4)
Argo salinity (psu)
V2.3.1 CAP V2.3.1 RSS V3
Rms difference
0.46 0.52 0.47
beam2
Aquarius S
SS
(psu)
28
Comparison with in situ data (3/4)
V2.3.1 CAP V2.3.1 RSS V3
Aquarius S
SS
(psu)
Argo salinity (psu)
beam3
Rms difference
0.48 0.50 0.51
29
Comparison with in situ data (4/4)
Rms difference is large for the CAP product.
But those are mostly the same in beam3.
CAP
V2.3.1
RSS
Statistical summary
xy-plot of the collocation points (1/1)
30