1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Intraseasonal latent heat flux based on satellite observations
Semyon A. Grodsky1, Abderrahim Bentamy2, James A. Carton1, and Rachel T. Pinker1
Submitted to Journal of Climate
October 22, 2008
1
Department of Atmospheric and Oceanic Science University of Maryland, College Park,
MD 20742
Institut Francais pour la Recherche et l’Exploitation de la Mer (IFREMER), Plouzane,
France
2
Corresponding author:
senya@atmos.umd.edu
41
Abstract
42
Weekly average satellite based estimates of latent heat flux (LHTFL, positive if the ocean
43
losses heat) are used to characterize spatial patterns and temporal variability in the
44
intraseasonal band (periods shorter than 3 months). The strongest zonally averaged
45
intraseasonal variability of LHTFL in excess of 30 Wm-2 is observed at midlatitudes
46
between 400 S to 100 S and 100 N to 450 N. Intrasesonal variability of LHTFL is locally
47
stronger in the regions of major SST fronts (like the Gulf Stream, Agulhas) where the
48
standard deviation of intraseasonal LHTFL is up to 50 Wm-2. The amplitude of the
49
intraseasonal LHTFL decreases at high latitudes and in the regions of equatorial
50
upwelling, reflecting the effect of decreased SST. In midlatitudes the intraseasonal
51
variability of LHTFL is forced by passing storms and is locally amplified by unstable air
52
stratification over warm SSTs. Although weaker in amplitude, but still significant,
53
intraseasonal variability is observed in the tropical Indian and Pacific Oceans due to the
54
eastward propagation of Madden-Julian Oscillations. In this tropical region the
55
intraseasonal LHTFL and incoming solar radiation are out-of-phase, namely, evaporation
56
increases just below the convective clusters. Over much of the global Ocean anomalous
57
LHTFL provides a negative feedback on the underlying intraseasonal SST anomaly,
58
although there are considerable geographical variations. The feedback exceeds 20 Wm-
59
2 o
60
tropical Pacific and Atlantic where the time average LHTFL is weak.
/ C in regions around 20oS and 20oN, but decreases at high latitudes and in the eastern
61
1
62
1. Introduction
63
Latent heat flux (LHTFL) links the air-sea heat exchange with the hydrological cycle.
64
This evaporative heat loss makes up a significant portion of the ocean net surface flux
65
and compensates in part for the ocean heat gain by solar radiation (da Silva et al., 1994).
66
It has been demonstrated that satellite sensors can measure sea surface temperature
67
(SST), near-surface winds, and humidity, and thus provide data for estimating sea surface
68
evaporation. Currently, several satellite-based global ocean latent heat flux products are
69
available (e.g. Chou et al., 2003 and references therein).
70
71
Most observational examinations of LHTFL focus on its behavior on monthly and longer
72
timescales (e.g., da Silva et al., 1994; Yu et al., 2006). Recent studies of midlatitudes
73
(Qiu et al. 2004) and tropics (Zhang and McPhaden, 2000) have shown that intraseasonal
74
variations of LHTFL associated with synoptic meteorological disturbances can alter SST
75
by up to 1oC. Modeling studies (Maloney and Sobel, 2004) suggest that these SST
76
variations may in turn organize intraseasonal atmospheric convection and thus provide an
77
air-sea interaction mechanism for phenomena such as the 30-60 day Madden-Julian
78
Oscillations and may contribute to the year-to-year variability. Since LHTFL is also
79
proportional to evaporation its intraseasonal variations contribute to variations of surface
80
salinity, thus increasing LHTFL impact on surface density. In this study we exploit the
81
availability of a global 16-year (1992 – 2007) record of weekly turbulent fluxes of
82
Bentamy et al. (2008) to examine the observed geographic distribution of intraseasonal
83
LHTFL and its role in air-sea interactions.
84
2
85
The tropical atmosphere is subject to a variety of synoptic-scale disturbances including
86
30-60 day fluctuations known as the Madden-Julian Oscillations (MJO) (Madden and
87
Julian, 1994) that are driven in part by the evaporation-wind feedback that involves
88
impacts of the zonal wind perturbations on evaporation (Neelin et al., 1987). MJO is a
89
feature of all tropical sectors, although it is most pronounced over the eastern Indian
90
Ocean and western Pacific Ocean in boreal winter and is strongly modulated by ENSO.
91
MJO is characterized by strong fluctuations of surface winds of 2-4m/s and precipitation
92
(Araligidad and Maloney, 2008). As a result it produces correlated fluctuations of both
93
LHTFL and shortwave radiation (SWR) with amplitudes of 30-50 Wm-2 and is observed
94
to result in 0.5oC fluctuations of SST (Krishnamurti et al. 1988; Shinoda and Hendon,
95
1998; Zhang and McPhaden, 2000). Wind-induced surface heat exchange, in which
96
increases in wind-induced LHTFL cause reductions in SST, which further increase heat
97
loss, has been implicated in the maintenance of the MJO (Maloney and Sobel, 2004; Han
98
et al., 2007). Moreover, recent research suggests that intraseasonal fluctuations may
99
actively interact with lower frequency climate variations, just as in the Pacific, where the
100
westerly wind bursts trigger the evolution of the El Niño/Southern Oscillation (ENSO)
101
cycle (McPhaden, 2004). Foltz and McPhaden (2004) examine the intraseasonal (30-70
102
day) oscillations in the tropical and subtropical Atlantic and similarly find a close
103
relationship (correlation of 0.75) between SST change and LHTFL variability in this
104
basin.
105
106
The subtropics and midlatitudes are subject to additional synoptic meteorological forcing
107
originating in the mid-latitude storm systems. This additional variability has a strong
3
108
seasonal component and varies from year-to-year. In the Kuroshio extension region
109
Bond and Cronin (2008) find that in late fall through early spring cold air outbreaks
110
associated with synoptic events lead to intense episodes of LHTFL and sensible heat loss.
111
In summer and fall cloud shading effects accompanying synoptic disturbances become
112
important sources of intraseasonal flux variations. Based on experiments with a mixed
113
layer model Qiu et al. (2004) suggest that these summertime intraseasonal flux variations
114
can induce SST variations with climatologically significant ±1oC amplitudes. This and
115
other observational evidence suggest significant contributions by LHTFL variability in
116
the intraseasonal band to the state of the climate system. In this study we focus on
117
geographical patterns of LHTFL, consistency with SST, interplay with incoming solar
118
radiation, as well as modulation by longer period processes.
119
120
This study is made possible due to several improvements to the climate observing system.
121
Beginning in the early 1990s a succession of three satellite scatterometers provides high
122
resolution surface winds. Brightness temperature estimates from the Special Sensor
123
Microwave Imager provide an estimate of relative humidity. When combined with
124
estimates of surface temperature it is possible to estimate LHTFL at weekly resolution
125
(Bentamy et al., 2008). Clouds and aerosols, the main factors affecting SWR, are
126
available from a variety of sensors flying in both geostationary and polar orbits (Rossow
127
and Schiffer, 1999; Pinker and Laszlo, 1992). Finally, an array of more than 90 moorings
128
distributed across all three tropical oceans (McPhaden et al., 1998) provides ground truth
129
at high temporal resolution which can be used to explore the accuracy of the remotely
130
sensed estimates.
4
131
132
2. Data and method
133
This research is based on the recent update of weekly satellite-based turbulent fluxes of
134
Bentamy et al. (2003, 2008). The three turbulent fluxes, wind stress ( τ ), LHTFL ( Q E ),
135
and sensible heat flux ( QH ) are estimated using the following bulk aerodynamic
136
parameterizations (Liu et al., 1979):
137
138
τ
 C D u a  u s (u a  u s )

139
QE
 CE u a  u s ( qa  qs )
L
140
QH
 C H u a  u s (Ta  Ts ) ,
C p
(1)
141
142
where  is the air density, L =2.45*106 J/kg is the latent heat of evaporation, C p =1005
143
J/kg is the specific heat of air at constant pressure. The turbulent fluxes in (1) are
144
parameterized using wind speed ( ua  u s ) relative to the ocean surface current, the
145
difference of specific air humidity and specific humidity at the air-sea interface ( qa  qs ),
146
and the difference of air temperature and SST ( Ta  Ts ). The lower indices (a) and (s)
147
indicate atmosphere at the reference level (normally 10m) and at the sea surface,
148
respectively. The bulk transfer coefficients for wind stress ( CD , drag coefficient), latent
149
heat flux ( C E , Dalton number), and sensible heat flux ( CH , Staton number) are estimated
150
from wind speed, air temperature, and SST using the Fairall et al. (2003) algorithm
5
151
(COARE3 version). LHTFL is positive if the ocean loses heat, while QH is positive if the
152
ocean gains heat.
153
154
The variables needed for the evaluation of (1) are obtained from satellite measurements.
155
Wind speed relative to the ocean surface current ua  u s is measured by scatterometers
156
onboard the European Research Satellites ERS-1 (1992-1996), ERS-2 (1996-2001), and
157
QuikSCAT (1999-2007) (e.g. Liu, 2002). The humidity ( q a ) is derived from the Special
158
Sensor Microwave Imager multi channel brightness temperatures using the Bentamy et al.
159
(2003) method, while the specific surface humidity ( qs ) is estimated from daily averaged
160
SST. This version of LHTFL uses the new Reynolds et al. (2007) daily SST while the
161
previous version of LHTFL (Bentamy et al., 2003) is based on the Reynolds and Smith
162
OIv2 weekly SST.
163
164
The air temperature is determined from remotely sensed data based on the Bowen ratio
165
( B  QH / QE ). This method has been suggested by Konda et al. (1996) and validated
166
by comparisons to buoy data in the tropics and midlatitudes. For the gradient-based K-
167
parameterization of fluxes and equal eddy diffusivity for heat and water vapor, the
168
Bowen ratio reads
169
170
B
C p Ta
C p C H (Ta  Ts )

,
L qa
L C E ( qa  q s )
(2)
171
6
172
where the right hand side term is the Bowen ration from (1). Using the Clausius-
173
Clapeyron law to relate the saturated humidity ( qsat ) and air temperature and neglecting
174
dependence
175
 ln( qsat ) / T
of
T Ta
the
relative
>>  ln( r ) / T
T Ta
humidity
(r )
on
air
temperature,
, (2) takes the form:
176
qa qsat
qsat (Ta ) T
177

T Ta
CE ( qa  qs )
CH (Ta  Ts )
(3)
178
179
and is solved for Ta .
180
181
The turbulent fluxes are calculated using the COARE3.0 algorithm from daily averaged
182
values binned onto a 1° global grid over satellite swaths. Due to differences in sampling
183
by different satellite radars and radiometers, the final flux estimate is further averaged
184
weekly and spatially interpolated on a regular 1° grid between 80° S and 80° N using the
185
kriging method described in Bentamy et al. (2003). The accuracy of the resulting weekly
186
fluxes is assessed by comparisons with in-situ measurements from moored buoys in the
187
tropical Atlantic and Pacific (PIRATA and TAO/TRITON), the northeastern Atlantic and
188
northwestern Mediterranean (UK Met Office and Météo-France), and the National Data
189
Buoy Center (NDBC) network off the U.S. coast in the Atlantic and Pacific Oceans1.
190
Quite high correlations (ranging from 0.8 to 0.92) are found between satellite and in-situ
191
LHTFL, while biases and standard deviations are generally low. Standard deviations of
192
satellite and in-situ LHTFL vary from 18 Wm-2 and 25 Wm-2. The highest bias is found
See ‘New Release of Satellite Turbulent Fluxes 1992 – 2007’ at
ftp://ftp.ifremer.fr/ifremer/cersat/products/gridded/flux-merged/documentation/flux.pdf
1
7
193
in comparisons with the NDBC buoys in the Gulf Stream region where the time mean
194
satellite LHTFL is 10 Wm-2 below in-situ values (or 7% of the NDBC regional LHTFL
195
mean). In the tropics satellite LHTFL overestimates in-situ LHTFL by 8 Wm-2. These
196
comparisons indicate significant improvements of the new LHTFL product over the
197
previous release described in Bentamy et al. (2003).
198
199
Intraseasonal signal is evaluated in few steps. First, the annual cycle is calculated from
200
the weekly data as a sum of the first three harmonics (Mestas-Nuñez et al., 2006). Next,
201
the anomaly is calculated by subtracting the annual cycle from the original signal.
202
Finally, the intraseasonal signal is calculated as the difference between the anomaly and
203
its 13 week running mean. This procedure retains periods shorter than 3 months that are
204
referred to as intraseasonal in this study. The variability of intraseasonal fluxes is
205
characterized by the running standard deviation that mimics the upper envelope of the
206
intraseasonal signal. Running standard deviation of intraseasonal signal is calculated
207
using the same 13 week running window. Comparisons of the satellite intraseasonal
208
LHTFL with in-situ data from the TAO/TRITON moorings in the tropical Pacific, the
209
PIRATA moorings in the tropical Atlantic, and the RAMA moorings in the tropical
210
Indian Ocean are presented in the Appendix.
211
212
The LHTFL from this study is compared with LHTFL provided by the National Center
213
for Climate Prediction/ National Center for Atmospheric Research (NCEP/NCAR)
214
reanalysis (Kalnay et al., 1996), the Woods Hole Oceanographic Institution objectively
215
analyzed air-sea fluxes of Yu et al. (2004), and with ship borne estimates collected by the
8
216
International Comprehensive Ocean-Atmosphere Data Set (ICOADS) of Worley et al.
217
(2005). Mean sea level pressure for this study is provided by the NCEP/NCAR
218
reanalysis. In-situ measurements from the TAO/TRITON moorings in the tropical Pacific
219
Ocean (McPhaden et al., 1998), the PIRATA moorings in the tropical Atlantic (Bourles
220
et al., 2008), and the RAMA moorings in the tropical Indian Ocean (McPhaden et al.,
221
2008) are also used for comparisons.
222
223
For several years now, uniform, long-term data from observations made from numerous
224
satellites relevant for inferring surface shortwave radiation (SWR) have been prepared
225
into homogeneous time series. The satellites that are being used for SWR retrieval
226
usually have between two to five channels in spectral intervals that are relevant both for
227
inferring SWR (visible) and for detecting clouds. Cloud data are provided by the
228
International Satellite Cloud Climatology Project (version D1) at a nominal resolution of
229
2.5◦ at 3hr time intervals (Rossow and Schiffer, 1999). The original version of the SWR
230
retrieval scheme is described in Pinker and Laszlo (1992) and has been used at
231
NASA/Langley for generating the GEWEX/SRB product2. Since, several modifications
232
have been introduced as related to aerosols (e.g. Liu and Pinker, 2008), data merging
233
(Zhang et al., 2007), and elevation correction (Ma and Pinker, 2008).
234
235
3. Results
236
Mean LHTFL and seasonal variations
237
First, presented are global patterns of the LHTFL and its annual and semiannual
238
harmonics. These components form the annual cycle that is used as a reference for
2
http://gewex-srb.larc.nasa.gov
9
239
evaluating anomalies and intraseasonal signal. Spatial patterns of magnitude and phase of
240
these harmonics are similar to Mestas-Nuñez et al. (2006) analysis that is based on the
241
three year long record (1996-1998) from the previous release of the LHTFL archive of
242
Bentamy et al. (2003). Comparison of the time mean LHTFL from this study with the
243
time mean LHTFL provided by alternative analyses (NCEP/NCAR Reanalysis, WHOI
244
air-sea fluxes, and ICOADS) indicates reasonable correspondence of spatial patterns
245
(Figs. 1 a-d). The time mean latent heat flux is dominated by evaporation in the trade
246
wind regions and resembles the time average wind speed in the 30o S to 30o N belt (Fig.
247
2a). SST impacts are evident across the subtropical fronts where temperature decreases
248
with latitude. Poleward decrease in SST is accompanied by decrease in Ta . Hence, the
249
humidity contrast, q  qs  qa , also decreases sharply poleward of 30o S and 30o N (Fig.
250
2b). These meridional changes of  q explain weak LHTFL in the extratropical oceans in
251
spite of rather strong winds in the northern and especially southern hemisphere storm
252
track corridors. SST impacts are also noticeable in the equatorial eastern Pacific and
253
Atlantic where the mean LHTFL is weak due to the presence of cold tongues of SST
254
maintained by the equatorial upwelling. Local minimum of evaporation over the cold
255
tongue regions is explained by direct impact of cool SST on the air humidity as well as
256
by indirect impact of SST on the near surface atmospheric boundary layer that tends to
257
decelerate over cold water and vice-versa, accelerate over warm water (Wallace et al.,
258
1989; Beal et al., 1997).
259
260
Regardless of the good correspondence of the geographical distribution of the time mean
261
LHTFL, the four analyses are somewhat different in magnitude. In current analysis
10
262
LHTFL (Fig. 1a) has higher values in the trade wind regions in comparison to the other
263
three analyses. This analysis is closer to in-situ ship observations from the ICOADS (Fig.
264
1d) and NCEP/NCAR reanalysis (Fig. 1b), but exceeds the WHOI estimates by 20 to 40
265
Wm-2 (Fig. 1c). Quantification of the LHTFL discrepancy may be achieved only through
266
a poleward extension of the tropical ocean buoy network and better sampling of
267
evaporation in the trade wind regions.
268
269
Strong time mean latent heat loss (exceeding 80 Wm-2) is drawn from the warm Gulf
270
Stream waters off the east coast of the US. Similarly strong time mean LHTFL is
271
observed near Japan over the warm Kuroshio (Fig. 1 a-d). In both these regions the
272
LHTFL experiences the strongest annual variation peaking during the winter, when cold
273
dry continental air off-shore of North America and Japan crosses the Gulf Stream north
274
wall in the Atlantic or the Kuroshio SST front in the Pacific, respectively (Fig. 1e, 1f).
275
Semiannual LHTFL variations are prominent in the Arabian Sea and Bay of Bengal due
276
to annual reversal of winds forced by South Asian Monsoon (Figs. 1g, 1h). The monsoon
277
flow in the Arabian Sea low level westerly jet intensifies in boreal summer while
278
northeasterly winds spread over the region in boreal winter when the monsoon ceases.
279
Weaker semiannual variability is observed in the Caribbean low level jet where the
280
easterly winds also intensify twice a year in February and again in July (Munoz et al.,
281
2008).
282
283
Magnitude of intraseasonal variation
11
284
To characterize the intraseasonal variations of LHTFL we consider the intraseasonal
285
variation of qw (Fig.2e). This variable closely corresponds to the LHTFL as the Dalton
286
number, C E , has weak dependence on wind speed (for winds ranging from 4 ms-1 to 14
287
ms-1, Large and Pond, 1982) though it depends on the stratification of the near surface
288
atmospheric layer. The spatial patterns of the intraseasonal variability of qw bear only
289
partial correspondence to intraseasonal winds (Fig. 2c). In particular, the decrease in
290
variance of intraseasonal qw towards the equator reflects relatively weak variability of
291
intraseasonal wind at low latitudes. In contrast to low latitudes, the intraseasonal
292
variability of LHTFL decreases at high latitudes despite stronger wind variability there.
293
This behavior is explained by low humidity over cold SSTs (and cold Ta ) poleward of
294
40o - 50o in each hemisphere (Fig. 2b). As it is expected from the bulk formulation (1),
295
the regions of strong intraseasonal variability of LHTFL are spatially collocated with the
296
regions of strong intraseasonal variability of winds (Fig. 2c) and of air humidity (Fig. 2d).
297
Although linear decomposition of the intraseasonal variance of qw suggests that wind
298
component ( qw' , Fig. 2f) accounts for a major portion of variability of the intraseasonal
299
LHTFL, neither components dominates globally. In particular, the air humidity
300
variability term ( wqa ' , Fig. 2g) peaks along the major SST fronts and reflects an impact
301
of moisture transport across the ocean SST fronts by synoptic weather systems. SST itself
302
( wqs ' , Fig. 2h) also impacts the intraseasonal LHTFL along the major western boundary
303
current fronts and in the Agulhas current area. Both, the mean LHTFL (Fig.1a) and its
304
variability (Fig.2e) weaken over cold SSTs where mean values of LHTFL are also low.
12
305
This is particularly evident in the cold sector of the Gulf Stream, in the Brazil-Malvina
306
confluence region, in the subpolar north Pacific, and in the Southern Ocean.
307
308
Standard deviation of LHTFL in the intraseasonal band (  1 , Fig. 3a) exceeds the
309
standard deviation in the low frequency band (  2 , Fig. 3b) over much of the global ocean
310
except in the equatorial Pacific, where the low frequency variability (due to the
311
interannual ENSO) slightly exceeds the intraseasonal variability. Although intraseasonal
312
variation of LTHFL is stronger than low frequency variation of LHTFL, the relative
313
impact of intraseasonal LHTFL on the mixed layer temperature is mitigated by the
314
difference in characteristic periods. Anomalous LHTFL defines the rate of change of
315
anomalous SST, H(SST ) / t ~  (LHTFL ) , where H is the mixed layer depth. Hence,
316
the ratio of SST responses to intraseasonal and low frequency variations of LHTFL is
317
defined by the standard deviation of LHTFL in these two bands and is scaled by the
318
ration of characteristic periods that are  1 =1 month and  2 =1 year, respectively. Relative
319
impact of intraseasonal and low frequency variations of LTHFL on SST is roughly
320
evaluated in Fig. 3c as ( 1 1 ) /( 2 2 ) . SST response to intraseasonal LHTFL variation
321
does not exceed ~30% of the SST response to low frequency LHTFL variation because of
322
the difference in the characteristic periods. This ratio has local minimum of ~ 10% in the
323
equatorial regions and increases poleward of 20o S - 20o N band reflecting stronger
324
transient (synoptic) variability at mid-latitudes.
325
326
Intraseasonal LHTFL in midlatitudes
13
327
The strongest variability of the intraseasonal LHTFL occurs in midlatitudes where the
328
regional maxima are linked to areas of major SST fronts (Fig. 3a). In particular, in the
329
Atlantic sector the highest intraseasonal variance is observed along the Gulf Stream front.
330
Similarly high intraseasonal variability is observed in the Agulhas Current and in the
331
Brazil-Malvina confluence region. This suggests important roles the stratified
332
atmospheric boundary layer plays in amplifying intraseasonal air-sea interactions. The
333
intraseasonal LHTFL variance changes seasonally and peaks in winter (not shown)
334
suggesting an association with midlatitude storms which also intensify in cold season.
335
We next identify weather systems that are responsible for strong intraseasonal variability
336
of LHTFL in these regional maxima areas by projecting the intraseasonal LHTFL time
337
series spatially averaged over particular index area on atmospheric parameters elsewhere
338
(Fig. 4a). This regression analysis reveals correspondence between strengthening of
339
intraseasonal LHTFL in the Gulf Stream region and midlatitude storm systems. Increase
340
in LHTFL drawn from the region is associated with the area of mean sea level pressure
341
low and corresponding cyclonic anomalous winds centered east of the region. The air
342
pressure pattern is similar to that deducted by Foltz and McPhaden (2004) in their
343
analysis of the intraseasonal variability of the Atlantic trade winds. In fact, the anomalous
344
wind in Fig. 4a decelerates the northern flank of the northeasterly trades (where
345
anomalous LHTFL is somewhat weaker) and significantly accelerates off-shore winds
346
over the Gulf Stream (Fig. 4b). Maximum increase of wind speed is observed over warm
347
sector of the Gulf Stream where winds further accelerate due to the atmospheric
348
boundary layer adjustment. In addition to intensification of mean winds, the anomalous
349
northwesterly wind outbreaks cold and dry continental air toward the sea. Spreading of
14
350
dry continental air lowers air humidity thus increasing the air-sea humidity contrast (Fig.
351
4c). This, in turn, compliments the LHTFL increase due to stronger winds. The ocean
352
responds to continental air outbreak by cooling SST north of the Gulf Stream northern
353
wall that is seen in decreasing values of qs (Fig. 4c). Intraseasonal winds have a weak
354
impact on SST south of the Gulf Stream temperature front where the ocean mixed layer is
355
deep and its thermal inertia is relatively strong.
356
357
Similar response is seen in the Agulhas current south of the Cape of Good Hope (Fig. 5).
358
Like in the Gulf Stream area, increase of LHTFL over the warm Agulhas Current is
359
linked to a passing storm. When the storm center locates to the east of the index area the
360
anomalous southerly winds bring cold and dry sub-Antarctic air northward. This
361
amplifies the latent heat loss due to increasing wind speed and increasing air-sea surface
362
humidity contrast. Although storm systems are similarly strong as they propagate around
363
the globe in the South Atlantic and the Southern Oceans, the intraseasonal LHTFL is
364
stronger in the Agulhas region and in the Brazil-Malvina confluence (Fig. 3a) in
365
comparison to values observed at similar latitude in the ocean interior. Both these areas
366
host sharp SST fronts that promotes higher  q and stronger LHTFL. It is interesting to
367
note that the regression analysis in Fig. 5 reveals a sequence of propagating storms over
368
open spaces of the South Atlantic and South Oceans. The mean sea level pressure troughs
369
in the regression pattern are separated by approximately 90o in longitude suggesting the
370
zonal wavenumber of 4.
371
372
Intraseasonal surface fluxes and SST
15
373
Variability of LHTFL and SST are related. LHTFL affects SST by affecting the net ocean
374
surface heat balance. But, SST also affects LHTFL directly through qs and indirectly by
375
affecting near-surface winds that accelerate over warmer SSTs. We next characterize
376
interplay between these intraseasonal variations. As expected, the LHTFL feedback to
377
underlying anomalous SST is generally negative (Fig. 6a), i.e. LHTFL increases in
378
response to increased SST. This suggests a damping of the underlying SST anomalies,
379
although there are considerable geographical variations (Park et al., 2005). The feedback
380
exceeds 20Wm-2 in the regions around 20oS and 20oN, but decreases at high latitudes and
381
in the eastern tropical Pacific and Atlantic where the time average LHTFL is also weak.
382
In contrast to the LHTFL response to underlying SST that is positive, the SST response to
383
intraseasonal variation of LHTFL is negative over much of the ocean (Fig. 6b). But, in
384
several regions SST warms up in response to LHTFL increase. In particular, this behavior
385
occurs in the cold tongue regions of the eastern tropical Pacific and Atlantic Oceans. The
386
relationship between intraseasonal LHTFL and SST depends on the relative role the
387
LHTFL plays in the mixed layer heat balance. If this balance is local and governed by the
388
surface flux, the SST cools down in response to increasing latent heat loss (negative
389
correlation when LHTFL leads). This negative relationship dominates away from the cold
390
tongue regions and strong currents. In contrast, in the cold tongue regions the mixed layer
391
temperature balance is governed primarily by the vertical (upwelling) or horizontal
392
(Tropical Instability Waves, e.g. Grodsky et al., 2005) heat transports. Here the positive
393
correlation between LHTFL and SST is explained by the stratified atmospheric boundary
394
layer adjustment and associated wind acceleration over warm SSTs. Therefore, in the
16
395
cold tongue regions the LHTFL increases in response to increasing wind and SST rather
396
than SST responses to change in LHTFL.
397
398
Over the regions where the surface heat flux dominates the mixed layer heat budget, the
399
variations of LHTFL force variations of the mixed layer temperature and, thus, should be
400
correlated with the SST rate of change, T / t , as seen in Fig. 7a. The time correlation of
401
intraseasonal LHTFL and T / t is statistically significant over much of the ocean3. It
402
decreases at high latitudes where the upper ocean stratification is weak, the mixed layer is
403
deep, and SST response is weak. The time correlation is also weak in the tropical Pacific
404
and Atlantic Oceans in the regions where vertical and horizontal heat transports dominate
405
the mixed layer heat budget. Similar but weaker correlation is found for the short wave
406
radiation (Fig. 7b). If the two components (LHTFL and SWR) of the surface flux are
407
combined, the correlation increases (Fig. 7c) suggesting reasonable correspondence of
408
intraseasonal flux variations with intraseasonal SST.
409
410
Intraseasonal LHTFL and SWR
411
Both, the intraseasonal LHTFL and SWR agree reasonably with independent
412
measurements of the rate of change of intraseasonal SST. We next explore the global
413
correspondence between intraseasonal variations of the two surface flux components.
414
They are weakly correlated over much of the global ocean with an exception of the
415
tropical Indian Ocean and the western tropical Pacific where the intraseasonal LHFL and
416
SWR are negatively correlated (Fig. 8). Lagged correlation indicates that LHTFL
417
increases in phase with decrease in SWR, suggesting stronger latent heat loss just below
3
The 99% confidence level of zero correlation is 0.1.
17
418
convective systems. This phase relationship is similar for satellite flux data and in-situ
419
TAO/TRITON mooring data (Fig. 8, inlay). It is consistent with Zhang and McPhaden
420
(2000) analysis of the TAO/TRITON surface fluxes who also have found near in-phase
421
relationship among maxima in latent heat flux and minima in solar radiation during
422
passage of the MJO events. From one side, the out-of-phase variations of intraseasonal
423
LHTFL and SWR support a hypothesis that the evaporation affects the humidity and
424
therefore the cloudiness and thus solar radiation at the sea surface. But theoretical
425
considerations (see Zhang and McPhaden, 2000 for a summary of existing approaches)
426
suggest a lagged relationship between intraseasonal LHTFL and SWR forced by MJO. In
427
particular, in the Neelin et al. (1987) model the maximum LHTFL is shifted to the east of
428
the convective center, if mean wind is easterly. Explanation of the phase relationship
429
between LHTFL and SWR variations on the intraseasonal timescales is not clear,
430
although a qualitative description based on the observed variations of air humidity is
431
briefly discussed below.
432
433
Coherent variations of the intraseasonal LHTFL and SWR are apparent in the time-
434
longitude diagrams in Figs. 9e, 9f. These accorded intraseasonal variations propagate
435
eastward between 600 E and the dateline at an average speed of 7.5 ms-1 typical of the
436
MJO. East of the dateline the correlation between LHTFL and SWR is weak (Fig. 8).
437
This zonal change of correlation is explained by the lack of cloudiness east of the dateline
438
that is the only major source of SWR variability. Intraseasonal LHTFL variations are
439
mostly driven by the intraseasonal variations of zonal wind. In fact, this is illustrated in
440
Fig. 9a that shows substantial correlation of LHTFL and zonal wind along the whole
18
441
equatorial belt. The sign of correlation switches depending on the zonal wind direction.
442
In the equatorial Indian Ocean and western equatorial Pacific the time mean zonal wind is
443
westerly. In this region a superposition of the mean eastward wind with an eastward
444
anomalous wind enhances the wind speed and LHTFL resulting in a positive correlation
445
of the zonal wind velocity, U , and LHTFL as seen in Fig. 9a. In contrast, a negative zero
446
lag correlation is observed east of the dateline where the mean zonal wind is easterly.
447
448
In distinction from the correlation of U and LHTFL that is observed along the whole
449
equatorial belt, coherent variations of intraseasonal zonal wind and SWR occur only west
450
of the dateline (Fig. 9b) with a gap in correlation over the maritime subcontinent. In the
451
Indian Ocean where the time mean westerly wind is stronger the anomalous eastward
452
wind lags by ~1 week the anomalously low SWR (see negative correlation at positive
453
lags in Fig. 9b). In the western Pacific where the mean zonal wind is weak the anomalous
454
zonal wind is almost out-of-phase with anomalous insolation. In spite of zonally varying
455
phase relationship of intraseasonal U and SWR, the intraseasonal SWR and LHTFL are
456
firmly out-of-phase west of the dateline (Fig. 9c). Possible explanation for this out-of-
457
phase behavior may include an impact of air humidity that varies in phase with
458
intraseasonal SWR (Fig. 9d). Specific air humidity decreases below convective systems
459
in response to cooling of the near-surface atmosphere while the sea surface saturated
460
humidity doesn’t change much because of the thermal inertia of the ocean mixed layer.
461
Difference in responses of q a and qs leads to an increase in the vertical gradient of
462
specific air humidity below convective cloud clusters that, in turn, enhances evaporation
463
and LHTFL.
19
464
465
Intraseasonal and longer period variability of LHTFL
466
The interannual evolution of the ocean surface fluxes has been extensively studied. But, it
467
appears that amplitude of intraseasonal fluxes is not stationary and experiences
468
significant modulation by longer period variability. Noting that our dataset is only 16
469
years long, the consideration is limited to the tropical Pacific Ocean that hosts the ENSO
470
and, thus, displays significant interannual variability that is resolved by relatively short
471
records. Interannual SWR anomaly is modulated by ENSO through zonal displacements
472
of convection. These interannual displacements of convection between the western
473
tropical Pacific and the central tropical Pacific produce SWR anomalies that are well
474
detected by satellite techniques (Rodriguez-Puebla et al., 2008). Because clouds are the
475
only physical mechanism driving the intraseasonal SWR, the amplitude of intraseasonal
476
SWR also shifts zonally following anomalously low SWR. In the central equatorial
477
Pacific the magnitude of intraseasonal SWR increases in-phase with warming of the
478
Nino3 SST (Fig. 10a, inlay). Here, the standard deviation of intraseasonal SWR increases
479
by up to 5 Wm-2 in response to a 1 oC rise of SST in the Nino3 region (Fig. 10a). As such,
480
interannual variation of the amplitude of intraseasonal SWR reaches 15 Wm-2 during a
481
mature phase of El Niño when anomalous Nino3 SST warms up by 3 oC. This interannual
482
modulation of amplitude of the intraseasonal SWR is comparable to the characteristic
483
amplitude of SWR variation by MJO (Shinoda et al., 1998).
484
485
In distinction to the amplitude of intraseasonal SWR that varies in-phase with El Niño,
486
the magnitude of intraseasonal LHTFL doesn’t have similarly significant in-phase
20
487
variation. Impact of El Niño on the intraseasonal LHTFL differs from the impact on the
488
total anomalous LHTFL that enhances in the eastern tropical Pacific, around the
489
Maritime Continent, and the equatorial Indian Ocean (Mestas-Nunez et al., 2006). In
490
contrast, the magnitude of intraseasonal LHTFL amplifies over the western tropical
491
Pacific approximately 8 months in advance of the mature phase of El Niño (Fig. 10b and
492
inlay). This amplification reflects impacts of the westerly wind bursts that precede the
493
onset of El-Nino, which were evident in advance of the 2002/03 El Niño and particularly
494
noticeable in advance of the 1997-98 event (McPhaden, 2004).
495
496
4. Conclusions
497
Although the major portion of the intraseasonal variability of LHTFL is accounted for by
498
winds, neither components (wind, air humidity, or sea surface humidity) dominates the
499
variability globally. In particular, contributions of q a and qs are significant along major
500
SST fronts due to moisture transport across the ocean SST fronts by synoptic weather
501
systems. Both the mean LHTFL and its intraseasonal variability weaken over cold SSTs
502
due to low air-sea humidity contrast. In contrast, the strongest intraseasonal LHTFL is
503
observed over the warm sectors of SST fronts.
504
505
The strongest variability of the intraseasonal LHTFL (in excess of 50 Wm-2) occurs at
506
middle latitudes where the regional maxima are linked to areas of major SST fronts. In
507
particular, in the Atlantic sector the highest intraseasonal variance is observed along the
508
Gulf Stream. Similarly high variability is observed in the Agulhas Current and in the
509
Brazil-Malvina confluence. Coincidence of the regional maxima of intraseasonal LHTFL
21
510
with SST fronts suggests important roles the stratified atmospheric boundary layer plays
511
in amplifying intraseasonal air-sea interactions. Temporal variation of the intraseasonal
512
LHTFL in these regional maxima is linked to passing midlatitude storms. The
513
intraseasonal variability of LHTFL forced by these passing storms is locally amplified by
514
unstable atmospheric stratification over warm SSTs.
515
516
Although weaker in amplitude but still significant intraseasonal variability of LHTFL
517
(standard deviation of 20 to 30 Wm-2) is observed in the tropical Indian and Pacific
518
Oceans. This variability is linked to the eastward propagating Madden-Julian
519
Oscillations. In this tropical region the intraseasonal LHTFL and incoming solar radiation
520
vary out-of-phase, i.e. evaporation enhances just below the convective clusters while the
521
SWR low leads by approximately a week the eastward wind anomaly. The out-of-phase
522
relationship is observed west of the dateline, while east of the dateline both, intraseasonal
523
LHTFL and SWR, are weak and their relationship is not significant. Intraseasonal
524
variation of q a that varies in phase with intraseasonal SWR contributes to this out-of-
525
phase relationship. Specific air humidity decreases below convective systems following
526
cooling of the near-surface atmosphere while qs doesn’t change much because of the
527
ocean thermal inertia. Difference in responses of q a and qs increases the vertical
528
gradient of air humidity below convective cloud clusters and thus enhances evaporation
529
and LHTFL.
530
531
Amplitude of intraseasonal LHTFL and SWR displays significant interannual variations
532
in the tropical Pacific Ocean. Amplitude of intraseasonal SWR increases in the central
22
533
equatorial Pacific by 15 Wm-2 during mature phase of El Niño following the eastward
534
shift of convection. In distinction to the amplitude of intraseasonal SWR that varies in-
535
phase with El Niño, the amplitude of intraseasonal LHTFL doesn’t have similarly
536
significant in-phase variation. In contrast, the intraseasonal LHTFL amplifies over the
537
western tropical Pacific approximately 8 months in advance of the mature phase of El
538
Niño. This amplification reflects impacts of the westerly wind bursts that normally
539
precede the onset of El-Nino.
540
541
Over much of the global ocean anomalous LHTFL provides negative feedback on the
542
underlying intraseasonal SST anomaly, although there are considerable geographical
543
variations. The feedback exceeds 20 Wm-2/oC in the regions around 20o S and 20o N, but
544
decreases at high latitudes and in the eastern tropical Pacific and Atlantic where the time
545
average LHTFL is weak.
546
547
Appendix
548
Comparisons of in-situ LHTFL with satellite-derived LHTFL in the intraseasonal band is
549
shown in Fig. 11. This comparison is based on in-situ buoy measurements in the tropics
550
including 68 TAO/TRITON buoys in the Pacific, 21 PIRATA buoys in the Atlantic, and
551
10 RAMA buoys in the Indian Ocean. During the 1992-2007 the data set has 30592
552
concurrent buoy-satellite weekly measurements in the Pacific, 3044 weeks of data in the
553
Atlantic, and 318 weeks of concurrent buoy and satellite data in the Indian Ocean. The
554
aggregate time series of buoy and satellite LHTFL have statistically significant
555
correlation around 0.6. The 99% confidence level of zero correlation is corr99% <0.1 for
23
556
the Pacific and Atlantic while it is slightly higher corr99% =0.14 for the Indian Ocean due
557
to shorter time series. Time series of intraseasonal LHTFL at each buoy location also
558
indicate significant correlation (Figs. 11a, 11c). Time correlation (TCORR) exceeds 0.6
559
over much of the tropical Pacific where average length of the LHTFL time series at
560
particular buoy is around 450 weeks ( corr99% =0.12). TCORR increases towards the west
561
following the westward increase of the mean LHTFL in the tropical Pacific (Fig. 1a). In
562
contrast, somewhat weaker TCORR is observed along 50 N where LHTFL is weaker due
563
to weaker winds and higher specific humidity in the ITCZ. The impact of the ITCZ is
564
better seen in the Atlantic where TCORR decreases below 0.5 in the ITCZ area (Fig.
565
11c). These comparisons suggest that the LHTFL retrieval should be rectified in the
566
ITCZ area. The air relative humidity has regional maximum in the ITCZ area. Therefore,
567
the meridional displacement of the ITCZ could produce variations of the relative
568
humidity strong enough that need to be accounted for in the Kondo et al. (1996) Bowen
569
ration approach.
570
571
Satellite intraseasonal LHTFL compares well with in-situ LHTFL to within the scatter of
572
the data (Figs. 11b, 11d, 11f). However, the magnitude of satellite intraseasonal LHTFL
573
is weaker than in-situ data. This bias is more evident in the Pacific where the
574
intraseasonal variations of satellite LHTFL are 15% to 20% weaker than those from
575
buoys, while this bias is less than 10% in the Atlantic and is not evident in the Indian
576
Ocean. We attribute this bias to the spatial and temporal smoothing of satellite data that
577
inevitably results in loosing of a portion of variance observed at fixed location and high
578
temporal resolution.
24
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
Acknowledgements. SAG and JAC acknowledge support from the NASA Ocean Wind
Vector Science Team (OWVST). AB is grateful for the technical and financial supports
provided by CERSAT/IFREMER and MERCATOR. RTP gratefully acknowledges
support by NASA grant NNG04GD65G from the Radiation Sciences Program and NSF
grant ATM0631685 to the University of Maryland. The ISCCP data were obtained from
the NASA Langley Research Center EOSDIS Distributed Active Archive Center.
References:
Araligidad, N. M., and E. D. Maloney, 2008: Wind-driven latent heat flux and the
intraseasonal oscillation. Geophys. Res. Lett., 35, L04815,
doi:10.1029/2007GL032746.
Beal, R. C., V. N. Kudryavtsev, D. R. Thompson, S. A. Grodsky, D. G. Tilley, V.A.
Dulov, and H. C. Graber, 1997: The influence of the marine atmospheric boundary
layer on ERS 1 synthetic aperture radar imagery of the Gulf Stream. J. Geophys.
Res., 102, 5799-5814.
Bentamy, A., K.B. Katsaros, A.M. Mestas-Nuñez, W.M. Drennan, E.B. Forde, and H.
Roquet, 2003: Satellite Estimates of Wind Speed and Latent Heat Flux over the
Global Oceans. J. Climate, 16, 637–656.
Bentamy, A., L-H. Ayina, W. Drennan, K. Katsaros, A. M. Mestas-Nuñez, R. T. Pinker,
2008: 15 Years of Ocean Surface Momentum and heat Fluxes from Remotely
Sensed Observations, FLUXNEWS, 5, 14-16. Available online at
http://sail.msk.ru/newsletter/fluxnews_5_final.pdf
Bond, N.A., and M.F. Cronin, 2008: Regional Weather Patterns during Anomalous Air–
Sea Fluxes at the Kuroshio Extension Observatory (KEO). J. Climate, 21, 1680–
1697.
Bourlès, B., R. Lumpkin, M.J. McPhaden, F. Hernandez, P. Nobre, E. Campos, L. Yu, S.
Planton, A. Busalacchi, A.D. Moura, J. Servain, and J. Trotte, 2008: The Pirata
Program: History, Accomplishments, and Future Directions. Bull. Amer. Meteor.
Soc., 89, 1111–1125.
Chou, S.H., E. Nelkin, J. Ardizzone, R.M. Atlas, and C.L. Shie, 2003: Surface Turbulent
Heat and Momentum Fluxes over Global Oceans Based on the Goddard Satellite
Retrievals, Version 2 (GSSTF2). J. Climate, 16, 3256–3273.
da Silva, A., C. C. Young, and S. Levitus, 1994: Algorithms and Procedures. Vol. 1,
Atlas of Surface Marine Data 1994, NOAA Atlas NESDIS 6, 83 pp.
25
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
Fairall, C.W., E.F. Bradley, J.E. Hare, A.A. Grachev, and J.B. Edson, 2003: Bulk
Parameterization of Air–Sea Fluxes: Updates and Verification for the COARE
Algorithm. J. Climate, 16, 571–591.
Foltz, G. R., and M. J. McPhaden, 2004: The 30–70 day oscillations in the tropical
Atlantic. Geophys. Res. Lett., 31, L15205, doi:10.1029/ 2004GL020023.
Grodsky, S. A., J. A. Carton, C. Provost, J. Servain, J. A. Lorenzzetti, and M. J.
McPhaden, 2005: Tropical instability waves at 0N, 23W in the Atlantic: A case
study using Pilot Research Moored Array in the Tropical Atlantic (PIRATA)
mooring data. J. Geophys. Res., 110, C08010, doi:10.1029/2005JC002941.
Han W., D. Yuan, W. T. Liu, D. J. Halkides, 2007: Intraseasonal variability of Indian
Ocean sea surface temperature during boreal winter: Madden-Julian Oscillation
versus submonthly forcing and processes. J. Geophys. Res., 112, C04001,
doi:10.1029/2006JC003791.
Jones, C., and B. C. Weare, 1996: The role of low-level moisture convergence and ocean
latent heat fluxes in the Madden-Julian oscillation: An observational analysis using
ISCCP data and ECMWF analyses. J. Climate, 11, 1057– 1072.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-year reanalysis project, Bull.
Amer. Meteorol. Soc., 77, 437-471.
Konda, M., N. Imasato, and A. Shibata, 1996: A new method to determine near-sea
surface air temperature by using satellite data, J. Geophys. Res., 101(C6), 1434914360.
Krishnamurti, T., D. Oosterhof, and A. Mehta, 1988: Air–Sea Interaction on the Time
Scale of 30 to 50 Days. J. Atmos. Sci., 45, 1304–1322.
Large, W., and S. Pond, 1982: Sensible and Latent Heat Flux Measurements over the
Ocean. J. Phys. Oceanogr., 12, 464–482.
Liu, H. and R. T. Pinker, 2008: Radiative fluxes from satellites: Focus on aerosols. J.
Geophys. Res., 113, D08208, doi:10.1029/2007JD008736.
Liu, W. T., K. B. Katsaros, and J. A. Businger, 1979: Bulk parameterization of air-sea
exchanges of heat and water vapor including the molecular constraints at the sea
surface. J. Atmos. Sci., 36, 1722-1735.
Liu, W.T., 2002: Progress in scatterometer application, J. Oceanogr., 58(1), 121-136.
Madden, R.A., and P.R. Julian, 1972: Description of Global-Scale Circulation Cells in
the Tropics with a 40–50 Day Period. J. Atmos. Sci., 29, 1109–1123.
26
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
Maloney, E. D., and A. H. Sobel, 2004: Surface fluxes and ocean coupling in the tropical
intraseasonal oscillation, J. Climate., 17, 4368–4386.
McPhaden, M.J. , and Coauthors, 1998: The Tropical Ocean-Global Atmosphere observing
system: A decade of progress, J. Geophys. Res., 103(C7), 14,169-14,240.
McPhaden, M. J., G. Meyers, K. Ando,Y. Masumoto, V. S. N. Murty, M. Ravichandran,
F. Syamsudin, J. Vialard, L. Yu, W. Yu, 2008: RAMA: The Research Moored
Array for African-Asian-Australian Monsoon Analysis and Prediction, subm. Bul.
Amer. Meteor. Soc., http://www.pmel.noaa.gov/tao/oceansites/RAMA_BAMS.pdf
McPhaden, M.J., 2004: Evolution of the 2002/03 El Niño. Bull. Amer. Meteor. Soc., 85,
677–695.
Mestas-Nuñez, A.M., A. Bentamy, and K.B. Katsaros, 2006: Seasonal and El Niño
Variability in Weekly Satellite Evaporation over the Global Ocean during 1996–98.
J. Climate, 19, 2025–2035.
Muñoz, E., A.J. Busalacchi, S. Nigam, and A. Ruiz-Barradas, 2008: Winter and Summer
Structure of the Caribbean Low-Level Jet. J. Climate, 21, 1260–1276.
Neelin, J.D., I.M. Held, and K.H. Cook, 1987: Evaporation-Wind Feedback and LowFrequency Variability in the Tropical Atmosphere. J. Atmos. Sci., 44, 2341–2348.
Park, S., C. Deser, and M.A. Alexander, 2005: Estimation of the Surface Heat Flux
Response to Sea Surface Temperature Anomalies over the Global Oceans. J.
Climate, 18, 4582–4599.
Pinker, R. T., and I. Laszlo, 1992. Modeling surface solar irradiance for satellite
applications on a global scale. J. Appl. Meteor. Clim., 31 (2), 194-211.
Qiu, B., S. Chen, and P. Hacker, 2004: Synoptic-Scale Air–Sea Flux Forcing in the
Western North Pacific: Observations and Their Impact on SST and the Mixed
Layer. J. Phys. Oceanogr., 34, 2148–2159.
Reynolds, R. W., T. M. Smith, C. Liu, D. B. Chelton, K. S. Casey and M. G. Schlax,
2007: Daily High-resolution Blended Analyses for sea surface temperature. J.
Climate, 20, 5473-5496.
Rodriguez-Puebla, C., R. T. Pinker, and S. Nigam, 2008: Relationship between
downwelling surface shortwave radiative fluxes and sea surface temperature over
the tropical Pacific: AMIP II models versus satellite estimates, Ann. Geophys., 26,
785–794.
Rossow, W. B., and R. A. Schiffer, 1999: Advances in understanding clouds from
ISCCP. Bull. Amer. Meteor. Soc., 80, 2261-2287.
27
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
Shinoda, T., and H.H. Hendon, 1998: Mixed Layer Modeling of Intraseasonal Variability
in the Tropical Western Pacific and Indian Oceans. J. Climate, 11, 2668–2685.
Wallace, J.M., T.P. Mitchell, and C. Deser, 1989: The influence of sea surface
temperature on surface wind in the eastern equatorial Pacific: seasonal and
interannual variability. J. Climate, 2, 1492-1499.
Worley, S.J. S. D. Woodruff, R. W. Reynolds, S. J. Lubker, and N. Lott, 2005: ICOADS
release 2.1 data and products, Int. J. Climate, 25, 823-842, doi:10.1002/joc.1166
Yu, L., R.A. Weller, and B. Sun, 2004: Mean and Variability of the WHOI Daily Latent
and Sensible Heat Fluxes at In Situ Flux Measurement Sites in the Atlantic Ocean.
J. Climate, 17, 2096–2118.
Yu, L., X. Jin, and R.A. Weller, 2006: Role of Net Surface Heat Flux in Seasonal
Variations of Sea Surface Temperature in the Tropical Atlantic Ocean. J. Climate,
19, 6153–6169.
Zhang, B. L., R. T. Pinker, and P. W. Stackhouse, 2007: An empirical orthogonal
function iteration approach for obtaining homogeneous radiative fluxes from
satellite observations. J. Appl. Meteor. Clim., 46, 435-444.
Zhang, C., and M.J. McPhaden, 2000: Intraseasonal Surface Cooling in the Equatorial
Western Pacific. J. Climate, 13, 2261–2276.
28
740
741
742
743
744
745
746
Figure 1. 1992-2007 mean LHTFL (Wm-2) from (a) this study, (b) NCEP/NCAR
reanalysis, (c) WHOI objectively analyzed air-sea fluxes, and (d) ICOADS. The means
are based on whatever part of this time interval is available. Annual harmonics (e)
magnitude and (f) phase. Semiannual harmonics (g) magnitude and (h) phase. Zero phase
corresponds to January.
29
747
748
749
750
751
752
Figure 2. Time mean (a) wind speed, w , (b) sea-air specific humidity difference,
q  qs  qa . Standard deviation of intraseasonal (c) wind speed, w' , (d) humidity
difference, q ' . (e) Standard deviation of intraseasonal ( qw)' and contribution to it
from intraseasonal variation of (f) wind speed, (e) specific humidity, and (h) saturated
near surface humidity.
30
753
754
755
756
757
758
759
Figure 3. Standard deviation of latent heat flux STD(LHTFL) in (a) intraseasonal band
(periods shorter than 3 months, HF) and (b) low frequency band (periods longer than 3
months, LF). (c) Relative impact of HF and LF variations of LHTFL on SST. Only areas
with at least 5 years of data are shown.
31
760
761
762
763
764
765
766
767
768
769
Figure 4. Time regression of intraseasonal latent heat flux index averaged over the Gulf
Stream area on (a) mean sea level pressure and winds, (b) latent heat flux elsewhere and
wind speed, (c) saturated near surface humidity and humidity. The index area is defined
as the area where STD of intraseasonal LHTFL exceeds 40 Wm-2 (see Fig. 4a) and is
dotted in panel (a). The index is defined as the index area average intraseasonal LHTFL
normalized by its standard deviation (25 Wm-2). Only wind arrows exceeding 0.4 ms-1 are
shown.
32
770
771
772
773
774
775
776
777
Figure 5. Time regression of intraseasonal latent heat flux index averaged over the
Agulhas Current area on intraseasonal mean sea level pressure (contours) and winds
(arrows). The index is defined as an area average intraseasonal LHTFL normalized by its
standard deviation (37 Wm-2). The index area is shown by the shaded rectangle. Only
wind arrows exceeding 0.4 ms-1 are shown.
33
778
779
780
781
782
783
784
785
Figure 6. Lagged regression of intraseasonal latent heat flux (LHTFL) and SST. (a) SST
leads LHTFL by 1 week, (b) LHTFL leads SST by 1 week. Areas where time correlation
exceeds the 99% confidence level of zero correlation are dotted. Inlay in panel (b) shows
lagged correlation spatially averaged over the equatorial east Pacific (black) and the
midlatitude Pacific (red) areas shown in panel (b). Negative lags are LHTFL lead in
weeks.
34
786
787
788
789
790
791
Figure 7. Time correlation of the rate of change of intraseasonal SST, T / t , with (a)
intraseasonal LHTFL, (b) intraseasonal short wave radiation, SWR, (taken with the minus
sign), and (c) sum of the two, LHTFL-SWR. Correlation exceeding 0.1 is significant at
the 99% confidence level.
35
792
793
794
795
796
797
Figure 8. Time correlation of intraseasonal LHTFL and SWR. Inlay shows lagged
correlation of LHTFL and SWR averaged over the rectangle area (solid with symbols)
and from the TAO/TRITON mooring at 0N, 165E (dashed). The mooring location is
shown by the cross. Lags are in weeks. Positive lags imply that SWR leads LHTFL.
36
798
799
800
801
802
803
804
805
806
Figure 9. Lagged correlation of intraseasonal (a) zonal wind and LHTFL, (b) zonal wind
and SWR, (c) SWR and LHTFL, and (c) specific air humidity and SWR.
Positive/negative correlations are shown in solid/dashed. Correlations below 0.2 in
magnitude are not shown. Positive lags imply that the second variable leads the first.
Time-longitude diagrams of intraseasonal (e) LHTFL and (f) SWR averaged 5S to 5N in
Wm-2. Shading in panel (a) is the time mean zonal wind (ms-1).
37
807
808
809
810
811
812
813
814
815
816
Figure 10. Time regression of anomalous Nino3 SST (210E-270E, 5S-5N) with running
standard deviation,  , of intraseasonal (a) SWR and (b) LHTFL. Correlation for SWR is
instantaneous, while LHTFL is correlated with Nino3 SST that lags it by 35 weeks.
Inlays show lagged correlation of Nino3 SST with the intraseasonal variance of flux
spatially averaged over the rectangle shown in each panel. Positive lags imply flux
leading Nino3 SST. Areas where the time regression is significant at the 99% level are
cross-hatched.
38
817
818
819
820
821
822
823
824
825
Figure 11. Comparison of intraseasonal buoy and satellite-derived latent heat flux. Spatial
maps of time correlation (left) and scatter diagrams (right) for (a,b) tropical Pacific, (c,d)
tropical Atlantic, and (e,f) tropical Indian Ocean. Gray shading in (b,d) shows the
standard deviation of satellite intraseasonal LHTFL in 2 Wm-2 intervals of in-situ data.
TCORR is the time correlation evaluated from the aggregate satellite/buoy comparisons
for each basin, NUM is the length of the aggregate record (in weeks). Buoy locations are
shown by closed circles in (a,c,e).
39