Access to this full-text is provided by Wiley.
Content available from Geophysical Research Letters
This content is subject to copyright. Terms and conditions apply.
Quantifying the Role of Oceanic Feedbacks
on ENSO Asymmetry
Cong Guan
1,2,3,4,5
, Michael J. McPhaden
6
, Fan Wang
1,2,4,5
, and Shijian Hu
1,2,4,5
1
Key Laboratory of Ocean Circulation and Waves, Institute of Oceanology, Chinese Academy of Sciences, Qingdao, China,
2
College of Marine Science, University of Chinese Academy of Sciences, Qingdao, China,
3
Institute of Oceanographic
Instrumentation, Shandong Academy of Sciences, Qingdao, China,
4
Center for Ocean Mega‐Science, Chinese Academy of
Sciences, Qingdao, China,
5
Qingdao National Laboratory for Marine Science and Technology, Qingdao, China,
6
NOAA/
Pacific Marine Environmental Laboratory, Seattle, WA, USA
Abstract The role of oceanic feedbacks in determining the asymmetry of El Niño–Southern Oscillation
(ENSO) magnitude, spatial structure, and duration is quantified on the basis of a novel temperature variance
budget. Results confirm previous studies that in the eastern Pacific, El Niño warm temperature anomalies
are larger in magnitude than La Niña cold temperature anomalies mainly due to stronger positive oceanic
feedbacks for El Niño. We find that La Niña cold anomalies are typically stronger than El Niño warm
anomalies in the central Pacific with a faster growth rate for cold anomalies, due to a stronger positive
thermocline feedback and weaker nonlinear damping. The thermocline feedback related to recharge
oscillator dynamics plays a dominate role and leads to asymmetry in the duration of El Niño and La Niña
events. In particular, the thermocline feedback becomes significantly negative during the late decaying
phase of El Niño and speeds up its demise.
Plain Language Summary The El Niño–Southern Oscillation (ENSO) is well known to have
profound impacts on global climate. Many asymmetric features exist between its warm phase of ENSO
(El Niño) and cold phase (La Niña), but their causes are still not fully understood. Our study examines
three aspects of ENSO asymmetry: (1) the amplitude of anomalous temperature during the mature El
Niño events is larger than La Niña in the equatorial eastern Pacific; (2) in the equatorial central Pacific,
the amplitude in mature phase of La Niña is larger than El Niño; and 3) La Niña typically lasts longer
than El Niño events. We find that the larger amplitude of El Niño than La Niña is due to stronger positive
feedback for El Niño in the eastern Pacific. In the central Pacific, La Niña has a faster growth rate
than El Niño, which may be induced by stronger positive thermocline feedback in the developing phase
and less nonlinear damping effect. The asymmetry of ENSO duration is because the decay rate for
La Niña events is slower than for El Niño, as a result of positive thermocline feedback that change sign to
negative for El Niño but not for La Niña.
1. Introduction
The El Niño–Southern Oscillation (ENSO) is the most prominent interannual phenomenon on the planet,
impacting climate patterns and their variability worldwide (McPhaden et al., 2006). Its warm phase, known
as El Niño, occurs with warm sea surface temperature (SST) anomalies in the equatorial Pacific accompa-
nied by weakened trade winds along the equator, while the converse holds true for its cold phase, La
Niña. A delicate balance of positive and negative feedbacks controls El Niño and La Niña events by either
amplifying or damping ENSO anomalies. Key positive feedbacks that determine the growth rate of ENSO
events include the thermocline feedback (TCF) due to effects of wind‐driven equatorial thermocline depth
variations on SST, the Ekman feedback (EKF) that results from anomalous vertical advection of temperature
driven by local winds, and the zonal advective feedback (ZAF) associated with anomalous thermal advection
by zonal currents (e.g., Jin et al., 2006). Net sea surface heat flux typically acts as the largest negative feed-
back that damps SST anomalies (e.g., Zhang & McPhaden, 2010).
El Niño and La Niña are like two sides of one coin, but they are not exactly symmetrical and exhibit signifi-
cant complexity (Timmermann et al., 2018). It is well known that the magnitude of El Niño SST anomalies is
usually greater than for La Niña, as evident by stronger warm SST anomalies than negative anomalies in the
©2019. The Authors.
This is an open access article under the
terms of the Creative Commons
Attribution‐NonCommercial‐NoDerivs
License, which permits use and distri-
bution in any medium, provided the
original work is properly cited, the use
is non‐commercial and no modifica-
tions or adaptations are made.
RESEARCH LETTER
10.1029/2018GL081332
Key Points:
•El Niño is stronger than La Niña in
the eastern Pacific due to stronger
positive feedbacks during El Niño
events
•La Niña is stronger than El Niño in
the central Pacific because the
thermocline feedback is stronger
and reinforced by nonlinearity
•Asymmetry in El Niño‐La Niña
duration results from asymmetric
thermocline feedback related to the
recharge oscillator dynamics
Supporting Information:
•Supporting Information S1
Correspondence to:
C. Guan,
congguan@qdio.ac.cn
Citation:
Guan, C., McPhaden, M. J., Wang, F., &
Hu, S. (2019). Quantifying the role of
oceanic feedbacks on ENSO
asymmetry. Geophysical Research
Letters,46, 2140–2148. https://doi.org/
10.1029/2018GL081332
Received 26 NOV 2018
Accepted 9 JAN 2019
Accepted article online 15 JAN 2019
Published online 18 FEB 2019
GUAN ET AL. 2140
eastern Pacific (EP; Deser & Wallace, 1987). The spatial distribution of El Niño SST anomalies is also clearly
asymmetrical with La Niña anomalous SST pattern shifted to the west compared to El Niño (Dommenget
et al., 2013; Hoerling et al., 1980). Significant differences also exist in the duration and phase transitions
of El Niño and La Niña (Choi et al., 2013; DiNezio & Deser, 2014; Dommenget et al., 2013; Kug & Ham,
2011; Larkin & Harrison, 2002; McGregor et al., 2013; Okumura & Deser, 2010). For example, El Niño events
generally decay by the summer following their mature phase, but La Niña events often persist through a sec-
ond year (Santoso et al., 2017). The frequent occurrence of a second‐year La Niña is a notable departure from
a purely linear ENSO cycle (Kessler, 2002). These asymmetric features between El Niño and La Niña can
project onto decadal or even longer time scale variability in the tropical Pacific Ocean (An, 2004;
Monahan & Dai, 2004; Rodgers et al., 2004).
Several hypotheses have been proposed to understand ENSO asymmetry. Some studies have shown that a
nonlinear relationship between SST and zonal wind stress can contribute to the El Niño‐La Niña asymmetry
(e.g., Frauen & Dommenget, 2010). Jin et al. (2003) pointed out that the vertical nonlinear thermal advection
plays a role in affecting amplitude asymmetry between El Niño and La Niña. Wang and McPhaden (2001)
found that anomalous vertical mixing at the base of ocean mixed layer enhances warm El Niño SSTs while
the tropical instability waves in the EP weaken cold La Niña SSTs. Su et al. (2010) also argued that nonlinear
horizontal thermal advection contributes to the asymmetry between El Niño and La Niña. For the same
magnitude of warm water volume anomaly in the equatorial Pacific, Meinen and McPhaden (2000) found
that positive El Niño SST anomalies were much greater than negative La Niña SST anomalies though they
could only speculate as to the reasons. Im et al. (2015) used Bjerknes stability index and pointed out that
stronger positive feedbacks contribute to a larger growth rate in the El Niño phase. However, the Bjerknes
index is by definition limited to describing impacts on EP SSTs and might overestimate TCF relative to
ZAF (Graham et al., 2014). Recent studies also emphasized the importance of ZAF on amplitude asymmetry
of ENSO (Kim et al., 2015; Santoso et al., 2017). It has been suggested that equatorial zonal transport (Chen
et al., 2016) and the meridional gradient of sea surface height anomalies (Hu et al., 2016) affect the asymme-
try of El Niño and La Niña decay phases, but precisely, how much these oceanic processes contribute to El
Niño‐La Niña asymmetry is still unanswered.
In this study, we apply a novel temperature variance budget approach as used in Guan and McPhaden (2016,
hereafter GM2016) to quantitatively estimate the role of various oceanic feedbacks in affecting the asymme-
try of ENSO with respect of amplitude, spatial structure, and event duration. Compared to a traditional tem-
perature budget analysis, the temperature variance budget can be used to estimate how different processes
contribute to the temperature variance growth or decay and to explicitly identify which terms are positive
feedbacks (creating variance) and which are negative feedback (damping variance). The temperature var-
iance budget and data we use are described in section 2. Main results are presented in section 3 followed in
section 4 by a summary and discussion of outstanding issues.
2. Temperature Variance Budget and Data
To examine individual oceanic feedbacks in a unified framework, we make use of a temperature variance
budget developed by Santoso et al. (2010) and elaborated on in GM2016. The governing equation is con-
structed by multiplying the temperature equation by temperature anomalies:
MLTvt¼TCF þEKF þZAF þMAF þMHD þTD þNL þR;(1)
where MLT
vt
is the time‐dependent temperature variance tendency and on right‐hand side are shown as the
feedback terms TCF, EKF, ZAF, meridional advective feedback, mean horizontal dynamical heating term
(MHD), thermal damping by net sea surface heat flux (TD), and nonlinear advection, respectively. The resi-
dual R represents the difference between MLT
vt
and the sum of the seven feedbacks. This residual contains
computational errors, the effects of unresolved physical processes such as vertical heat diffusion, tropical
instability waves not represented explicitly the monthly products used, and penetrative shortwave radiation,
which cannot be computed from net surface heat flux provided by the reanalysis products. In addition, since
we use reanalysis products to evaluate (1), the residual will also contain assimilation increments from the
various products (as discussed in GM2016). Detailed formulas and specificdefinitions for the temperature
variance budget are provided in the appendix and also described in GM2016.
10.1029/2018GL081332
Geophysical Research Letters
GUAN ET AL. 2141
We use three ocean reanalysis products and one numerical ocean general
circulation model simulation for budget calculations. These include the
German Estimating the Circulation and Climate of the Ocean Version 2
(GECCO2; Köhl & Stammer, 2008) from Hamburg University, the ocean
analysis/reanalysis system version 3 (ORAS3; Balmaseda et al., 2008)
and version 4 (ORAS4; Balmaseda et al., 2013) provided by the
European Centre for Medium‐Range Weather Forecasts and also the
ocean general circulation model of the Earth Simulator (Masumoto
et al., 2004) based on the MOM3. Detailed description of these four model
products can be found from the Asia‐Pacific Data‐Research Center at the
University of Hawaii (http://apdrc.soest.hawaii.edu/data/data.php).
Budget calculations are based on the monthly fields of ocean current velo-
cities, ocean temperature, and net sea surface heat flux (q
net
) from these
four products. Vertical velocity and surface heat flux are not available
for ORAS4, so for this product we calculated vertical velocities based on
mass continuity and used surface heat fluxes from the ERA‐interim reana-
lysis. Periods cover from January 1980 to December 2010. To obtain the
interannual signals of each feedback, we used Fourier low‐pass filter with
a cutoff period of 15 months (Walters & Heston, 1982). Budget calcula-
tions are carried out using the method described in Lee et al. (2004) within
a50‐m mixed layer of both the Niño3 region (5°S to 5°N, 150–90°W) and
Niño4 region (5°S to 5°N, 160°E to 150°W) for each product then averaged
to provide an ensemble mean perspective. All these four products present
reasonable good closure of heat balance (refer to Figure 3 in GM2016).
Niño3 and Niño4 indices are used to represent ENSO SST anomalies in the EP and central Pacific (CP),
respectively. These two index regions are helpful in characterizing CP/EP El Niño events (e.g., Kug et al.,
2009) and ENSO diversity (e.g., Santoso et al., 2017). Here we defined the indices based on interannual
anomalies of mixed layer temperature (MLT) in the corresponding Niño regions and then defined an event
when the index is over 0.5 °C for El Niño or below −0.5 °C for La Niña for 6 months or more. As a result, nine
El Niño and six La Niña events are identified in the Niño3 region, while nine El Niño and four La Niña
events are identified in the Niño4 region (see details in Table S1 in the supporting information). Then gen-
eral composite analysis is deployed separately for El Niño and La Niña in each Niño region, with events cen-
tered at their peaks determined as the month of maximum ENSO MLT anomaly. We tried different ways to
characterize ENSO events such as selecting events with the criterion of 0.5 °C thresholds for both regions
simultaneously or using a threshold of 1 °C. Either of these options reduces the number of events in our
ensemble and therefore increases uncertainties. Also, 0.5 °C is a widely accepted criterion for identifying
ENSO events, so we chose to use this criterion to select ENSO events separately in each region.
To estimate uncertainty, we treated each El Niño and La Niña event of each reanalysis product as separate
realizations. Thus, we have 36 realizations of El Niño in the Niño3 and Niño4 regions, 24 realizations of La
Niña in the Niño3 region, and 16 realizations in the Niño4 region. We then computed standard error and
95% confidence limits considering each event to be independent in its respective index region. These 95%
confidence limits are shown as error bars in Figures 1–3.
3. Results
The asymmetry in ENSO evolution is characterized in the composite ENSO MLT anomalies (Figure 1). It is
evident that El Niño has a larger amplitude than La Niña in the Niño3 region (Figure 1a), but the situation is
reversed in the Niño4 region (Figure 1b). The difference of ENSO amplitude between the Niño3 and Niño4
regions also indicates clear asymmetry in spatial structure. La Niña also tends to last longer than El Niño,
especially during respective decaying phases in the Niño3 region (Figures 1a and S1). On average, an El
Niño event quickly declines after its peak and returns to a neutral state within 6 months. Conversely, La
Niña decays much more slowly with negative MLT anomalies still exceeding 0.8 °C 6 months after its peak,
sometimes even followed by a weak second‐year cold event.
Figure 1. Ensemble mean mixed layer temperature (MLT) anomalies for
composite El Niño–Southern Oscillation (ENSO) events in (a) Niño3 and
(b) Niño4 regions. El Niño events are shown in red, and La Niña in blue. The
xaxis represents time in months, with negative values being months before
the peak of ENSO events, and positive values being months after the peak.
Error bars are 95% confidence levels from all the corresponding ENSO
events across the four model products.
10.1029/2018GL081332
Geophysical Research Letters
GUAN ET AL. 2142
To evaluate roles of various oceanic feedbacks in controlling these ENSO MLT asymmetries, we ana-
lyzed the temperature variance budget for El Niño and La Niña events separately in the Niño3 and
Niño4 regions. Then averages of each budget term are made over the 6 months before the ENSO peak
(i.e., ENSO developing phase) and 6 months after (i.e., ENSO decaying phase). In general, TCF and ZAF
are the largest two positive feedbacks causing El Niño and La Niña temperature anomalies to grow in
both Niño3 and Niño4 regions. During the decay phase in both regions, TCF is reduced but still on
average positive while ZAF is nearly to 0 on average (Figure 2). EKF and MHD are typically weaker
positive feedbacks particularly during the development phase. Regarding negative feedbacks, TD is
clearly the largest negative term in both developing and decaying phases in the both regions. These
results are consistent with previous studies (e.g., GM2016; Kug et al., 2009; Ren & Jin, 2013; Zhang
& McPhaden, 2010) though there are some differences as well. We find that the nonlinear feedback,
for example, is generally negative according to our formalism, while it is positive for warm ENSO events
according to Jin et al. (2003). This discrepancy may result from the particular regions chosen for analy-
sis and also the particular choice of product (GM2016). The residual R manifests itself as a
negative feedback.
Figure 2. Individual terms from the temperature variance budget averaged among the developing phase (a, c) and decay-
ing phase (b, d) in the Niño3 and Niño4 regions, respectively. Sum of the five positive feedbacks, for example, TCF,
EKF, ZAF, MAF, and MHD, is shown as Sum+, and sum of the three negative feedbacks, for example, TD, NL, and the
residual R, is shown as Sum−. Error bars are 95% confidence levels from all the corresponding El Niño–Southern
Oscillation events across the four model products. MLT = mixed layer temperature; TCF = thermocline feedback;
EKF = Ekman feedback; ZAF = zonal advective feedback; MAF = meridional advective feedback; MHD = mean hori-
zontal dynamical heating term; NL = nonlinear; TD = thermal damping; R = residual.
10.1029/2018GL081332
Geophysical Research Letters
GUAN ET AL. 2143
We next compare each of the oceanic feedbacks between El Niño and La Niña to better understand their
roles in generating ENSO asymmetries. In the Niño3 region, El Niño has a greater growth rate than La
Niña, which is attributed to stronger positive feedbacks during El Niño (Figure 2a). To quantify the relative
contribution of each positive feedback, the percentage ratio is calculated as the El Niño‐La Niña difference in
each positive feedback divided by the difference in total positive feedbacks (shown as Sum+ in Figure 2)
between El Niño and La Niña. Among the positive feedbacks, the biggest El Niño‐La Niña difference in posi-
tive feedbacks is for ZAF, EKF, and TCF, which contribute 36%, 33%, and 25% to the difference in total posi-
tive feedbacks, respectively. Together, they can explain 94% of the total difference in positive feedbacks,
among which the difference in ZAF is the largest. The sum of the negative feedbacks also shows significant
El Niño‐La Niña difference. This difference comes mainly from the difference in TD, which damps tempera-
ture anomalies once developed. Nonlinear feedback, as a damping term, is also stronger in El Niño than La
Niña. Therefore, stronger positive oceanic feedbacks during El Niño, especially the stronger ZAF, EKF and
TCF, account for larger El Niño than La Niña amplitudes in the EP. Im et al. (2015) also attributed ENSO
amplitude asymmetry to these three feedbacks but arrived at a different relative ranking of
ZAF > TCF > EKF based on a Bjerknes index analysis. We note that the percentages for these three positive
feedbacks in our analysis are not significantly different from one another in a statistical sense and that rela-
tive rankings may be sensitive to different products or methods used (Graham et al., 2014).
In the Niño4 region, the growth rate of the La Niña is stronger than that of El Niño, due to stronger positive
feedbacks for La Niña events (Figure 2c). Among the positive oceanic feedbacks, the TCF shows a prominent
difference and contributes about 46% of the difference in total positive feedbacks. The MHD contributes the
secondarily (36%), followed by EKF (16%). It is interesting that though ZAF is suggested to be the dominant
positive feedback controlling the CP ENSO events (Guan et al., 2013), there is no significant difference in
ZAF between El Niño and La Niña events in this region. Regarding the negative feedbacks, TD shows a sig-
nificant difference between El Niño and La Niña, but with a larger uncertainty for La Niña from the four
model products. We note that nonlinearity acts as a negative feedback in El Niño but is a positive feedback
in La Niña. The residual R also exhibits a notable El Niño‐La Niña difference but also shows relatively high
uncertainty. Thus, we suggest the larger growth rate for La Niña than El Niño in Niño4 is due to stronger
positive TCF reinforced (rather than opposed) by nonlinear effect.
Figure 3. TCF components superimposed on MLT variations (a, b) and thermocline depth averaged across the Pacific
basin (c, d) during the composite El Niño (a, c) and La Niña (b, d). Thick black dashed lines in (c) and (d) present the
peak months of ENSO event development. Error bars in (a) and (b) are 95% confidence levels from all the corresponding
ENSO events across the four model products. ENSO = El Niño–Southern Oscillation; MLT = mixed layer temperature;
TCF = thermocline feedback; ZG = zonal gradient; ZM = zonal mean.
10.1029/2018GL081332
Geophysical Research Letters
GUAN ET AL. 2144
We note that the composite growth rates for El Niño and La Niña in the Niño3 region are not as asymmetric
as in the Niño4 region statistically. Santoso et al. (2017) noted that some strong La Niñas, such as in
1998–1999, were preconditioned by very warm El Niño SSTs, which required very strong positive feedbacks
to flip warm condition to cold. Thus, from our composite analysis, the El Niño‐La Niña asymmetry tends to
be weak in the Niño3 region. In contrast, the Niño4 region exhibits much stronger asymmetry, perhaps
because we identified events separately for each region.
The asymmetry in El Niño‐La Niña duration is mainly determined by differences in the decaying phase,
which is significant in the Niño3 region (Figure 1a), where La Niña has an obviously slower decay rate than
El Niño (Figure 2b). This asymmetry is caused by the sum of positive feedbacks in the decaying phase of La
Niña being much larger than for El Niño. Among these feedbacks, the positive TCF shows the biggest differ-
ence, being twice as large for La Niña than El Niño. Difference can be also found in the ZAF, which though
small is also positive for La Niña but essentially 0 for El Niño. In fact, the TCF and ZAF feedbacks actually
become negative for period during the late decaying phase of El Niños (Figure S1), especially during strong
EP events (GM2016). In contrast, this tendency for TCF and ZAF to become negative during the decay phase
of La Niña events is not evident (Figure S1b). Thus on average, these feedbacks, especially TCF, are still
actively opposing the damping processes that would otherwise terminate La Niña events more quickly. In
the Niño4 region, La Niña also tends to decay more slowly than El Niño (Figures 1b and S2) but with a larger
decay rate than El Niño (Figure 2d). Unlike in Niño3, however, the differences in most feedbacks between El
Niño and La Niña are not significant, except for larger damping effects of negative TD and R in La Niña
events. This likely results from the initial condition (Santoso et al., 2017) that the negative SST anomalies
in the Niño4 region are much stronger than positive anomalies at their peaks, so that these negative feed-
backs are not sufficient to terminate cold anomalies in this region.
We find that the TCF is a major process affecting the growth and decay of ENSO events and their asymme-
tries. Based on model simulations of a nonlinear delayed oscillator, DiNezio and Deser (2014) also found that
delayed TCF contributed to the multiyear persistence of La Niña events vis a vis El Niño events. Thus, we
will next examine in detail how the TCF related to recharge oscillator dynamics affects the asymmetric decay
of ENSO events in the Niño3 region.
The TCF affects both the growth rates and phase transition of ENSO events according to Jin's (1997) recharge
oscillator theory, through the anomalous zonal gradient and zonal mean of the thermocline depth across the
Pacific basin, respectively. To further investigate the role of TCF in the asymmetric duration of El Niño and
La Niña, we first separate the anomalies in thermocline depth (defined as 20 °C isotherm depth, i.e., h20) in
the Niño3 region into two parts following GM2016 as h20
nino3
=h20
ZG
+h20
ZM
, where the subscripted ZG
and ZM represent for zonal gradient and zonal mean of the variable, respectively. Since the vertical tempera-
ture difference at the bottom of the mixed layer, which determines the TCF, is highly correlated and in phase
with the h20
Niño3
, we thus further separate the TCF into two parts, that is, TCF = TCF
ZG
+ TCF
ZM
(details
can be found in GM2016). In Figures 3a and 3b, the TCF induced by zonal gradient of thermocline variations
(TCF
ZG
) always acts as a positive feedback throughout both warm and cold ENSO events, while TCF caused
by zonal mean thermocline variations (TCF
ZM
) acts as positive feedback during the developing phase and
negative feedback during the decaying phase. To compare the differences between El Niño and La Niña,
the positive TCF
ZG
in the decaying phase is relatively small for La Niña. However, TCF
ZM
is 3 times larger
as a negative feedback (−0.3 °C/month) during the decaying phase of El Niño than that of La Niña (−0.1 °C/
month). Thus, the relatively strong negative TCF
ZM
during the decaying phase of El Niño helps to accelerate
the demise of warm events but not so much cold events. From Figures 3c and 3d it is evident that the zonal
mean thermocline quickly shoals after El Niño peaks, whereas during La Niña the corresponding deepening
is very gradual. Therefore, we conclude that the El Niño‐La Niña difference in TCF during their respective
decay phases results from asymmetric changes in zonal mean thermocline depth across the equatorial
Pacific, indicating an asymmetry in underlying recharge oscillator dynamics.
4. Summary and Discussion
In this paper, the El Niño‐La Niña SST asymmetries in respect of magnitude, spatial distribution, and
duration are interpreted from a perspective of oceanic feedbacks, based on a temperature variance analy-
sis in the Niño3 and Niño4 regions. We use a composite analysis to look for common characteristics
10.1029/2018GL081332
Geophysical Research Letters
GUAN ET AL. 2145
across all warm and cold events and find that composite El Niño events are stronger than La Niña events
in the Niño3 region due primarily to stronger positive feedbacks and in particular stronger ZAF, EKF, and
TCF. In the Niño4 region, however, La Niña has a faster growth rate and thus is stronger than El Niño.
The difference in growth rates is mainly due to the differences in positive feedbacks, among which TCF is
most prominent followed by MHD and EKF. Nonlinear processes reinforce these positive feedbacks dur-
ing La Niña but oppose them during El Niño, which also contributes to the faster development of La
Niña. Therefore, our results are consistent with previous studies (e.g., Choi et al., 2013; Im et al., 2015)
that positive oceanic feedbacks lead to asymmetries in El Niño‐La Niña amplitude in the Niño3 region
but also determined with more precise localization that TCF and nonlinear damping result in asymmetri-
cal growth rate in the Niño4 region.
Asymmetry in El Niño‐La Niña duration comes mainly from their decaying phases, with a larger decay
rate found for El Niño than La Niña. The TCF shows the largest difference among the positive feedbacks
and thus contributes to this difference in decay rates. Guided by recharge oscillator theory, we separate
TCF affects into two parts: TCF
ZG
caused by zonal gradient thermocline variations and TCF
ZM
caused
by zonal mean thermocline variations. During the decay phase of El Niño in the equatorial EP, TCF
ZM
acts as a prominent negative feedback as large as −0.3 °C/month, which is at least three times stronger
than during the decaying phase of La Niña (only −0.1 °C/month at most). This difference is induced
by zonal mean thermocline changes, which quickly shoals across the equator after El Niño peaks,
whereas a very slow deepening occurs during the decaying phase of La Niña. Therefore, ENSO asymme-
try in duration results from an underlying asymmetry in recharge oscillator dynamics. This result is con-
sistent with Kessler (2002), who stated that the recharge process is weak after the peak of La Niña events
before the onset of another El Niño.
What causes the asymmetry of recharge processes during the ENSO decaying phases? A recent study by
Neske and McGregor (2018) attributes this asymmetry to an asymmetry in oceanic processes, with the dis-
charged phase of the cycle dominated by the adjusted wave dynamics and the recharged phase dominated
by the instantaneous wind‐induced Ekman mass transport. Hu et al. (2016), on the other hand, specu-
lated the convergence of anomalous meridional currents from the off‐equatorial oceans onto the equator
may help to slow down the recharge process during the decaying phase of La Niña. From the anomalous
wind stress pattern averaged during the decaying phase of a composite La Niña event (Figure S3b), anom-
alous eastward wind stress and negative wind curl are found in the EP, which may drive the convergence
of anomalous meridional currents suggested by Hu et al. (2016). Also we note that the asymmetry of wind
stress pattern (Figure S3c) looks very similar to the asymmetric wind stress from McGregor et al. (2013),
implying the asymmetry in recharge oscillator dynamics is also related to asymmetries in the
surface winds.
In the present study we have quantitatively examined various oceanic processes affecting the El Niño‐La
Niña asymmetry. Our results confirm previous studies about the role of TCFs related to recharge oscillator
dynamics in determining the asymmetry in El Niño‐La Niña duration and provide robust evidence from the
perspective of a detailed temperature variance budget analysis for the first time. However, recent studies
show that the recharge oscillator dynamics is a less effective determinant of ENSO variability after the year
2000 (e.g., GM2016; McPhaden, 2012; Neske & McGregor; 2018), related to the more frequent occurrence of
CP El Niño events (Lee & McPhaden, 2010; Wen et al., 2014; Xiang et al., 2013). In addition, accounting for
the slowdown in global warming between about 2000 and 2013, the Pacific trade winds significantly
strengthened and an increase in subsurface ocean heat uptake was found in the equatorial thermocline
(e.g., England et al., 2014). How asymmetries in the ENSO cycle are manifest for different flavors of El
Niño and La Niña and under changing background conditions are the subject of a future study.
Appendix A.
Temperature variance budget (equation (1)) written in terms of individual feedback terms:
MLTtv¼TCF þEKF þZAF þMAF þMHD þTD þNL þR
where
10.1029/2018GL081332
Geophysical Research Letters
GUAN ET AL. 2146
MLTtv¼1
2∂Tave′ðÞ
2=∂t¼1
2MLTv
ðÞ
t
TCF ¼1
B∬wBδTB′⋅Tave0dxdy
EKF ¼1
B∬wB′δTB⋅Tave0dxdy
ZAF ¼1
B∬uW′δTW−uE′δTE
⋅Tave0dydz
MAF ¼1
B∬vS′δTS−vN′δTN
⋅Tave0dxdz
MHD ¼1
B∬uWδTW′−uEδTE′ðÞ⋅Tave0dydzþ∬vSδTS′−vNδTN′ðÞ⋅Tave0dxdz
hi
NL ¼1
B
∬uW′δTW′−uE′δTE′ðÞ⋅Tave0dydzþ∬vS′δTS
′
−vN′δTN′
⋅Tave0dxdz
þ∬wB′δTB′⋅Tave0dxdy
2
43
5
TD ¼1
ρcpB∬qnet′⋅Tave 0dxdy
References
An, S.‐I. (2004). Interdecadal changes in the El Niño–La Niña asymmetry. Geophysical Research Letters,31, L23210. https://doi.org/
10.1029/2004GL021699
Balmaseda, M. A., Mogensen, K., & Weaver, A. T. (2013). Evaluation of the ECMWF ocean reanalysis system ORAS4. Quarterly Journal of
the Royal Meteorological Society,139(674), 1132–1161. https://doi.org/10.1002/qj.2063
Balmaseda, M. A., Vidard, A., & Anderson, D. L. T. (2008). The ECMWF ocean analysis system: ORA‐S3. Monthly Weather Review,136(8),
3018–3034. https://doi.org/10.1175/2008MWR2433.1
Chen, H. C., Hu, Z. Z., Huang, B., & Sui, C. H. (2016). The role of reversed equatorial zonal transport in terminating an ENSO event. Journal
of Climate,29(16), 5859–5877. https://doi.org/10.1175/JCLI‐D‐16‐0047.1
Choi, K. Y., Vecchi, G. A., & Wittenberg, A. T. (2013). ENSO transition, duration, and amplitude asymmetries: Role of the nonlinear wind
stress coupling in a conceptual model. Journal of Climate,26(23), 9462–9476. https://doi.org/10.1175/JCLI‐D‐13‐00045.1
Deser, C., & Wallace, J. M. (1987). El Niño events and their relation to the southern oscillation: 1925–1986[J]. Journal of Geophysical
Research,92(C13), 14,189–14,196. https://doi.org/10.1029/JC092iC13p14189
DiNezio, P. N., & Deser, C. (2014). Nonlinear controls on the persistence of La Niña. Journal of Clima te,27(19), 7335–7355. https://doi.org/
10.1175/JCLI‐D‐14‐00033.1
Dommenget, D., Bayr, T., & Frauen, C. (2013). Analysis of the non‐linearity in the pattern and time evolution of El Niño southern oscil-
lation. Climate Dynamics,40(11–12), 2825–2847. https://doi.org/10.1007/s00382‐012‐1475‐0
England, M. H., McGregor, S., Spence, P., Meehl, G. A., Timmermann, A., Cai, W., et al. (2014). Recent intensification of wind‐driven
circulation in the Pacific and the ongoing warming hiatus. Nature Climate Change,4(3), 222–227. https://doi.org/10.1038/
nclimate2106
Frauen, C., & Dommenget, D. (2010). El Niño and La Niña amplitude asymmetry caused by atmospheric feedbacks. Geophysical Research
Letters,37, L18801. https://doi.org/10.1029/2010GL044444
Graham, F. S., Brown, J. N., Langlais, C., Marsland, S. J., Wittenberg, A. T., & Holbrook, N. J. (2014). Effectiveness of the Bjerknes stability
index in representing ocean dynamics. Climate Dynamics,43(9‐10), 2399–2414. https://doi.org/10.1007/s00382‐014‐2062‐3
Guan, C., Chen, Y.‐L., & Wang, F. (2013). Seasonal variability of zonal heat advection in the mixed layer of the tropical Pacific. Chinese
Journal of Oceanology and Limnology,31(6), 1356–1367. https://doi.org/10.1007/s00343‐014‐3019‐4
Guan, C., & McPhaden, M. J. (2016). Ocean processes affecting the twenty‐first‐century shift in ENSO SST variability. Journal of Climate,
29(19), 6861–6879. https://doi.org/10.1175/JCLI‐D‐15‐0870.1
Hoerling, M. P., Kumar, A., & Zhong, M. (1980). El Niño, La Niña , and the nonlinearity of their teleconnections. Journal of Climate,10(10),
1769–1786.
Hu, Z. Z., Kumar, A., Huang, B., Zhu, J., Zhang, R.‐H., & Jin, F.‐F. (2016). Asymmetric evolution of El Niño and La Niña: The
recharge/discharge processes and role of the off‐equatorial sea surface height anomaly. Climate Dynamics,49(7–8), 1–12.
Im, S., An, S.‐I., Kim, S. T., & Jin, F.‐F. (2015). Feedback processes responsible for El Niño‐La Niña amplitude asymmetry. Geophysical
Research Letters,42,5556–5563. https://doi.org/10.1002/2015GL064853
Jin, F. F. (1997). An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model.[J]. Journal of the Atmospheric Sciences,54(7),
811–829. https://doi.org/10.1175/1520‐0469(1997)054<0811:AEORPF>2.0.CO;2
Jin, F. ‐F., An, S. I., Timmermann, A., & Zhao, J. (2003). Strong El Niño events and nonlinear dynamical heating. Geophysical Research
Letters,30(3), 1120. https://doi.org/10.1029/2002GL016356
Jin, F.‐F., Kim, S. T., & Bejarano, L. (2006). A coupled‐stability index for ENSO. Geophysical Research Letters,33, L23708. https://doi.org/
10.1029/2006GL027221
Kessler, W. S. (2002). Is ENSO a cycle or a series of events? Geophysical Research Letters,29(23), 2125. https://doi.org/10.1029/
2002GL015924
Kim, W., Cai, W., & Kug, J.‐S. (2015). Migration of atmospheric convection coupled with ocean currents pushes El Niño to extremes.
Geophysical Research Letters,42, 3583–3590. https://doi.org/10.1002/2015GL063886
Köhl, A., & Stammer, D. (2008). Variability of the meridional overturning in the North Atlantic from the 50‐yr GECCO state estimation.
Journal of Physical Oceanography,38(9), 1913–1930. https://doi.org/10.1175/2008JPO3775.1
Kug, J., & Ham, Y. (2011). Are there two types of La Niña? Geophysical Research Letters,38, L16704. https://doi.org/10.1029/2011GL04 8237
10.1029/2018GL081332
Geophysical Research Letters
GUAN ET AL. 2147
Acknowledgments
We thank three anonymous reviewers
for their constructive comments on the
original version of this manuscript. We
acknowledge Hamburg University,
JAMSTEC APL, and ECMWF for their
valuable reanalysis and model
products. This study was supported by
the National Natural Science
Foundation of China (Grants 41806016
and 41730534), the China Postdoctoral
Science Foundation (2017M622289),
Qingdao postdoctoral application
research project, the National Natural
Science Foundation of China (Grant
41776018), the National Program on
Global Change and Air‐Sea Interaction
(GASI‐IPOVAI‐01‐01), and the Key
Research Program of Frontier Sciences,
CAS (QYZDB‐SSW‐SYS023). PMEL
contribution number is 4814. The four
model products data are available from
the Asia‐Pacific Data‐Research Center
at the University of Hawaii (http://
apdrc.soest.hawaii.edu/data/data.php).
Kug, J. S., Jin, F. ‐F., & An, S. ‐I. (2009). Two types of El Niño events: Cold tongue El Niño and warm pool El Niño. Journal of Climate,22(6),
1499–1515. https://doi.org/10.1175/2008JCLI2624.1
Larkin, N. K., & Harrison, D. (2002). ENSO warm (El Niño) and cold (La Niña) event life cycles: Ocean surface anomaly patterns, their
symmetries, asymmetries, and implications. Journal of Climate,15(10), 1118–1140. https://doi.org/10.1175/1520‐0442(2002)01 5<1118:
EWENOA>2.0.CO;2
Lee, T., Fukumori, I., & Tang, B. (2004). Temperature advection: Internal versus external processes. Journal of Physical Oceanography,
34(8), 1936–1944. https://doi.org/10.1175/1520‐0485(2004)034<1936:TAIVEP>2.0.CO;2
Lee, T., & McPhaden, M. J. (2010). Increasing intensity of El Niño in the central‐equatorial Pacific. Geophysical Research Letters,37, L14603.
https://doi.org/10.1029/2010GL044007
Masumoto, Y., Sasaki, H., Kagimoto, T., Komori, N., Ishida, A., Sasai, Y., et al. (2004). A fifty‐year eddy‐resolving simulation of the world
ocean—Preliminary outcomes of OFES (OGCM for the Earth simulator). Journal of the Earth Simulator,1,35–56.
McGregor, S., Ramesh, N., Spence, P., England, M. H., McPhaden, M. J., & Santoso, A. (2013). Meridional movement of wind anomalies
during ENSO events and their role in event termination. Geophysical Research Letters,40, 749–754. https://doi.org/10.1002/grl.50136
McPhaden, M. J. (2012). A 21st century shift in the relationship between ENSO SST and warm water volume anomalies. Geophysical
Research Letters,39, L09706. https://doi.org/10.1029/2012GL051826
McPhaden, M. J., Zebiak, S. E., & Glantz, M. H. (2006). ENSO as an integrating concept in Earth science. Science,314(5806), 1740–1745.
https://doi.org/10.1126/science.1132588
Meinen, C. S., & McPhaden, M. J. (2000). Observations of warm water volume cha nges in the equatorial Pacific and their relationship to El
Niño and La Niña [J]. Journal of Climate,15(20), 3551–3559.
Monahan, A. H., & Dai, A. (2004). The spatial and temporal structure of ENSO nonlinearity. Journal of Climate,17(15), 3026–3036. https://
doi.org/10.1175/1520‐0442(2004)017<3026:TSATSO>2.0.CO;2
Neske, S., & McGregor, S. (2018). Understanding the warm water volume precursor of ENSO events and its interdecadal variation.
Geophysical Research Letters,45. https://doi.org/ 10.1002/2017GL076439
Okumura, Y. M., & Deser, C. (2010). Asymmetry in the durati on of El Niño and La Niña. Journal of Climate,23(21), 5826–5843. https://doi.
org/10.1175/2010JCLI3592.1
Ren, H., & Jin, F. (2013). Recharge oscillator mechanisms in two types of ENSO. Journal of Climate,26(17), 6506–6523. https://doi.org/
10.1175/JCLI‐D‐12‐00601.1
Rodgers, K. B., Friederichs, P., & Latif, M. (2004). Tropical Pacific decadal variability and its relation to decadal modulations of ENSO.
Journal of Climate,17(19), 3761–3774. https://doi.org/10.1175/1520‐0442(2004)017<3761:TPDVAI>2.0.CO;2
Santoso, A., McPhaden, M. J., & Cai, W. (2017). The defining characteristics of ENSO extremes and the strong 2015/2016 El Niño. Reviews of
Geophysics,55,1079–1129. https://doi.org/10.1002/2017RG000560
Santoso, A., Sen Gupta, A., & England, M. H. (2010). Genesis of Indian Ocean mixed layer temperature anomalies: A heat budget analysis.
Journal of Climate,23(20), 5375–5403. https://doi.org/10.1175/2010JCLI3072.1
Su, J., Zhang, R., Li, T., Rong, X., Kug, J.‐S., & Hong, C.‐C. (2010). Causes of the El Niño and La Niña amplitude asymmetry in the equa-
torial eastern Pacific. Journal of Climate,23(3), 605–617. https://doi.org/10.1175/2009JCLI2894.1
Timmermann, A., An, S. I., Kug, J. S., Jin, F. F., Cai, W., Capotondi, A., et al. (2018). El Niño‐Southern Oscillation complexity. Nature,
559(7715), 535–545. https://doi.org/10.1038/s41586‐018‐0252‐6
Walters, R. A., & Heston, C. (1982). Removing tidal‐period variations from time‐series data usi ng low‐pass digital filters. Journal of Physical
Oceanography,12(1), 112–112, 115. https://doi.org/10.1175/1520‐0485(1982)012<0112:RTPVFT>2.0.CO;2
Wang, W., & McPhaden, M. J. (2001). Surface layer temperature balance in the equatorial Pacific during the 1997–98 El Niño and 1998‐99
La Niña. Journal of Climate,14(16), 3393–3407. https://doi.org/10.1175/1520‐0442(2001)014<3393:SLTBIT>2.0.CO;2
Wen, C., Kumar, A., Xue, Y., & McPhaden, M. J. (2014). Changes in tropical Pacific thermocline depth and their relationship to ENSO after
1999. Journal of Climate,27(19), 7230–7249. https://doi.org/10.1175/JCLI‐D‐13‐00518.1
Xiang, B., Wang, B., & Li, T. (2013). A new paradigm for the predominance of standing central Pacific warming after the late 1990s. Climate
Dynamics,41(2), 327–340. https://doi.org/10.1007/s00382‐012‐1427‐8
Zhang, X. B., & McPhaden, M. J. (2010). Surface layer heat balance in the eastern equatorial Pacific Ocean on interannual time scales:
Influence of local versus remote wind forcing. Journal of Climate,23(16), 4375–4394. https://doi.org/10.1175/2010JCLI3469.1
10.1029/2018GL081332
Geophysical Research Letters
GUAN ET AL. 2148
