POSSIBLE INFLUENCE OF FEBRUARY-APRIL ARCTIC OSCILLATION ON THE
ITCZ ACTIVITY OF WESTERN-CENTRAL PACIFIC
HU Miao1, 2, GONG Dao-yi1, MAO Rui1
(1. State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University, Beijing 100875, China;
2. Academy of Disaster Reduction and Emergency Management, Beijing Normal University, Beijing 100875, China)
Abstract: The daily patterns and activity of Intertropical Convergence Zone(ITCZ) in the Western-Central
Pacific Ocean are analyzed using NOAA interpolated Outgoing Longwave Radiation dataset during the period
from 1979 to 2008, and the relationships between ITCZ patterns and Arctic Oscillation(AO) is investigated in
this paper. In accordance with the central activity region the daily ITCZ can be divided into six
patterns—north, south, equator, double, full and weak pattern, respectively. The statistic result shows that the
north (accounts for 30.98% of the total observations), south (31.11%) and weak (24.05%) ITCZ patterns are
the most active daily patterns within a 30-year period, while the other three ITCZ patterns occur infrequently.
Results show that the Feb.—Apr. AO index has a significant positive (negative) correlation with the
frequency of the north (weak) ITCZ pattern from Mar.—May to Aug.—Oct., with the strongest correlation in
Apr.—Jun. (Mar.—May). At the same time, the lower tropospheric atmosphere circulation (850hPa wind field)
and SST anomalies corresponding to the AO change significantly in the tropical Pacific Ocean. When AO is
in the positive phase, there is an anomalous westerly from Equator to 15°N and warmer SST in the critical
north ITCZ active region, while there is an anomalous easterly and insignificant change of SST from Equator
to 15°S. The wind and SST anomalies share the same characteristics of the equatorial asymmetry and thus
enlarge the gradient between the south and north of equator, which would help reinforce convection in north
of equator and imply more frequent north type ITCZ.
Key words: Climatology; statistical characterization; Intertropical Convergence Zone; Arctic Oscillation;
Pacific Ocean
1. Introduction
As one of the essential components of atmospheric circulation in the tropics, Intertropical Convergence
Zone (ITCZ) is a narrow east-west-oriented zonal band and mainly featured by enhanced cumulonimbus
convection and strong precipitation, which is formed by the northeast and southeast trade winds converging
along the equatorward boundary regimes[1]. ITCZ is geographically a global belt and its distribution is not
zonally uniform due to the land-sea thermal contrast. Over oceanic areas (Pacific, Atlantic and Indian Ocean),
the preferred latitude and seasonal shift in the latitudinal position of ITCZ are also different [1-2]. Some
studies analyzed the ITCZ with focus on its seasonal migration and anomalous north-south shift, and
investigated the corresponding reasons employing both observation analysis and numerical simulation [3-6].
On the monthly, seasonal and annual time scales, the ITCZ is a nearly steady zonal band circulating around
the tropics; in contrast, on the synoptic scale, there exists eastward and westward propagating disturbances in
the ITCZ, suggesting that ITCZ is also a multi-feature system on different time scales[4, 7-9].
Previous studies showed the change of ITCZ and or tropical circulation have been modulated by some
factors, including the annual variation of solar radiation, solar elevation angle, sea-land orographic diffidence
and sea surface temperature (SST) meridional distribution. There is an issue that meteorologists have paid
increasing attention on: whether the extratropical climate changes can have some influence on ITCZ variation?
Barreiro and Philander[10] pointed out that a decrease in cloud cover over high latitudes, which means a
decrease in extratropical albedo, can cause an extratropical and tropical warming with the less intensity of the
equatorial cold tongues in the Pacific and Atlantic Oceans by using a coupled ocean-atmosphere model. They
found that the tropical adjustment might be made through the constraint of a balanced heat budget for the
ocean. Over high latitude region the albedo decrease causes the ocean to lose less heat, thus the ocean ought to
obtain less heat in the tropics, with the intense of cold tongues reducing as a consequence, especially over
eastern Pacific. According to the study of Limpasuvan and Hartmann [11], a strong Hadley circulation is
accompanied with positive AO and the anomalous meridional circulations associated with AO are evident.
Miller et al. [12] have studied the relationship between AO and global outgoing longwave radiation(OLR),
and found that in the positive AO years, the OLR are relatively lower in the tropical Pacific and Atlantic
Ocean; however, they did not address the possible mechanisms. Zhou and Miller[13] also indicated that the
eastward-propagating of the convectively active phase of the Madden–Julian oscillation (MJO) appears to
have a relationship to the AO variability. Suo et al.[14] displayed that the strength of winter AO can influence
the frequencies of extreme temperature in China. In the Northern Hemisphere the AO is the predominant
circulation system in the mid-high latitude, while the ITCZ is the largest circulation in the tropics. It is unclear
whether AO has some connection with the ITCZ activity.
In this study, we investigate the possible AO-ITCZ relationship and the ITCZ activity and pattern are
defined based on the daily OLR. This paper is organized as follows: section 2 describes the data and the
methodology for categorizing daily ITCZ patterns. Section 3 presents distribution characteristics of ITCZ
patterns and the statistic relationship between AO and ITCZ patterns. Section4 provides the association of
different ITCZ patterns with SSTs and winds. A discussion and conclusion are given in sections 5 and 6.
2. Data and methodology
2.1 Data
Monthly AO indices are derived from the Climate Prediction Center, and AO is defined as the leading
mode of Empirical Orthogonal Function analysis of monthly mean geopotential height anomalies poleward of
20° latitude for the Northern Hemisphere at 1000 hPa. In this study we use boreal winter mean AO index (the
mean value of January-February-March), and the index is normalized by the standard deviation with the base
period of 1971-2000. The monthly horizontal and vertical winds are obtained from National Centers for
Environment Prediction-National Central for Atmospheric Research reanalysis I with the resolution of 2.5x2.5
degree. The monthly SST data is derived from the Hadley Center Sea Ice and Sea Surface Temperature data
set (HadISST) with the resolution of 1x1 degree. Monthly anomalies of wind and SST are computed related to
its monthly mean of the 1971-2000 base periods. The OLR dataset is derived from National Center for
Atmospheric Research with sophisticated spatial and temporal interpolation, which extends from January
1979 to December 2008 with resolution of 2.5x2.5 degree.
Since ENSO exerts significant influence on tropical and mid-high latitude atmospheric circulation, this
may bring a misleading AO-ITCZ relation. To address this problem, we remove the ENSO signals in the
climate fields using a linear regress method. The ENSO-induced climate anomalies are determined by simple
linear regressing the time series of variables at each grid against the Nino3.4 SST time series, and then
subtract the estimated ENSO-related component parts from original variables. The regression residuals could
2
be ‘ENSO-free’ parts and all data in our study are handled in this way.
2.2 Method
To identify the spatial pattern and position of ITCZ this paper has taken the method of Chen et al.[7]. In
Chen et al.[7] work, they first defined six distinct ITCZ spatial patterns, and then identified daily ITCZ pattern
based on GPCP (Global Precipitation Climatology Project) daily rainfall dataset according to the location of
the convective precipitation center. Due to the short time span of GPCP (about 10 years), it is difficult for
analyzing the variability of ITCZ on long-time scale. Here we take OLR data as a substitute for GPCP data.
With respect to the definition of ITCZ activity, we first compared the patterns derived from the GPCP
and OLR data. Taking their common time period (from October 1996 to August 2006) and the similar
approach used in Chen et al.[7], we tested different threshold of OLR values to match the features of patterns
and frequencies as close as possible to GPCP-derived features.. As the threshold of OLR determined ,they are
employed to defined the daily ITCZ patterns for the last 30 years.
To identify the spatial pattern of the ITCZ in western-central Pacific, we confine the study region in
130°E-150°W and divide the region into three sub-regions—the northern region (5-12.5°N), the equatorial
region (2.5°N-2.5°S) and the southern region (5-12.5°S), and compute the coverage percentage of the strong
convection (low OLR value) area. A binary “convection intensity index” L is defined for each region for each
day as follows: at a given day in a particular region, when grid points with OLR≤205 W·m-2 (the threshold of
strong convection, which is approximately equivalent to 10mm day-1 precipitation rate in Chen et al. [7]) are
more than 25% of total points, we define Lx = 1, otherwise Lx = 0 (where x = n, s or e, representing the
northern, southern, and equatorial regions, respectively). The spatial patterns of daily ITCZ can be identified
with different combinations of Lx as in Table 1. For example, the daily ITCZ type is set to North ITCZ (NI for
short) when Ls=0 and Ln=1 at the same time,; the daily ITCZ type is set to double ITCZ(DI) only when Le=0
and Ln=Ls=1. Therefore consistent with the definition of different ITCZ patterns based on daily precipitation,
the main OLR minimum can exhibit a distinctive strong convection distribution for each ITCZ type
correspondingly.
Table 1: Definitions of spatial patterns of the ITCZ. L(n, e, s) index: at a given day in a particular region
(north, equator or south), when grid points with OLR(convection) <= 205 W·m-2
North
South
Equator
Double
Full
Weak
Ln
1
0
0
1
1
0
Le
0 or 1
0 or 1
1
0
1
0
Ls
0
1
0
1
1
0
Table 2: Percentage of six ITCZ patterns for the period of 1979-2008
ITCZ Patttern
NI
SI
EI
DI
FI
WI
Number
3395
3409
559
315
646
2634
of
3
days
percentage（%） 30.98
31.11
5.10
2.87
5.9
24.05
Based on the definition of each ITCZ daily patterns, a particular daily ITCZ pattern can be identified
from January 1979 to December 2008. Thus, composite spatial patterns are obtained by averaging OLR value
on the identified ITCZ type, and their frequencies are calculated as a percentage of the number of a certain
ITCZ days over the total 30 years.
3. ITCZ spatial patterns and the relationship with AO
3.1 Six ITCZs’ spatial patterns and frequencies
The composite daily OLR distributions for the north, south, equator, double, full and weak ITCZs can be
obtained by categorizing ITCZ daily patterns over western-central Pacific, as shown in Fig.1. The OLR
minimum presents a distinctive convection center and distribution characteristic for each ITCZ type, which is
consistent with the definition of ITCZs (Table 1). Table 2 shows the long-term mean frequency of six ITCZ
types in the 30 year period.
NI (Fig. 1a) is characterized by a single strong convection band north of the equator around 7.5°N, which
expands from the western Pacific to the eastern-central Pacific, with one local maxima located at the western
Pacific with the low value of about 190 W·m-2. The convection band in the Southern Hemisphere is rather
weak with the average OLR ranging from 240 to 250 W·m-2. The zonal mean plot (omitted) also shows a
single OLR peak (210 W·m-2) at 7.5°N. The NI pattern occurs frequently (3395 days), occupying about
30.98% of the total number of days over 30 years.
For the south ITCZ (SI for short, Fig 2a), the OLR values in the Northern Hemisphere are larger,
indicating the weakened convection; however there is a well-defined strong convection belt south of the
equator, with two centers at 7.5°S in western-central Pacific one located at New Guinea Island and the another
located in oceanic area between 160°-170°E. The SI pattern occurs also frequently, and the frequency of the
occurrence is about 31.11%. The mean position and distribution feature of the NI and SI are similar with
previous studies based on monthly and seasonal time scale.
The equator ITCZ (EI for short, Fig 1c) accounts for only 5.1% occurrence in the total 30 years. It
features a local strong convection center (OLR less than 195 W·m-2) over the equatorial western Pacific
between 130° and 160°E. The convection weakens and slightly north-dipping as it extends eastward into the
eastern Pacific. The peak of its zonal mean is located around 2.5°N with a value of 225W·m-2. Figure 1d
shows a distinct DI structure with convection center on each side of the equator and weakened convection at
the equator. In north branch, the local convection center can be found between 130° and 150°E and in south
branch is about 140°E. From the zonal mean distribution (omitted), in the north of equator the low values can
be found clearly at about 7.5°N with the minimum about 212 W·m-2 and in the south at 7.5°S with the
4
minimum about 222 W·m-2.This DI pattern only occurs about 2.8% of the total days (Table 2). For the full
ITCZ (FI for short, Fig 1e) strong convection covers a broad equatorial band between 7.5°N and 7.5°S across
the whole western Pacific and the convection weakens and is divided into two branches, with a northern
branch extending to 160°W and a southern branch to 170°W. Its zonal mean illustrates an overall strong and
wide convection, with two local valleys of about 208 W·m-2at 5°N and 217 W·m-2 between 5° and 7.5°S. The
frequency of occurrence for FI is about 5.9% (Table 2).
The weak ITCZ (WI for short, Fig 1f) is a kind of ITCZ of no distinctive strong convection center on and
off the equator over the western-central Pacific. The distribution pattern is similar to long-term mean OLR,
although the intensity is weak
Its zonal mean (omitted) has two small valleys around 7.5°N and 7.5°S, with
the maximum larger than 235 W·m-2. The WI occurs frequently and the frequency of occurrence is about
24%.
Overall, the NI, SI and WI patterns represent the three major ITCZ daily patterns over western-central
Pacific, and they account for almost 85% of the total number, indicating they are the active ITCZ patterns.
The EI, DI and FI patterns are the three inactive ITCZ daily patterns.
Figure1 The composite daily OLR for a)north, b) south, c) equator, d) double, e) full, f) weak ITCZs for the period of 1979-2008.
Units: W·m-2
Figure 2 Time series of normalized AO index (Feb-Apr) and North/Weak ITCZs frequency (Apr-Jun). Grey bar: AO index;
5
Solid lines: North ITCZs; Dashed lines: Weak ITCZs.
Table3: The correlation coefficients of Feb-Apr AO index and North/Weak ITCZs for the period of 1979-2008
Month
North
North(ENSO
Weak
removed)
2-4
0.29
b
0.26
0.41
b
Weak(ENSO
removed)
-0.13
-0.42 b
3-5
0.41
4-6
0.49 c
0.49 c
-0.32 a
-0.34 a
5-7
0.47 c
0.41 b
-0.36 b
-0.30 a
6-8
0.43 b
0.34 a
-0.35 a
-0.27
7-9
0.42
b
a
b
-0.28
8-10
0.33 a
0.29
-0.35 c
-0.23
9-11
-0.04
0.04
-0.17
-0.02
0.33
-0.38
-0.18
b
-0.37
a：90%, b：95%, c：99% significance level
3.2 Correlations between AO and ITCZ
In this section we investigate the possible linkage between AO and ITCZ by simply computing AO
correlations with the frequencies of daily ITCZ type. We considered only the active types, namely the NI, SI
and WI types.
Because AO is strongest in January-March (JFM), we use the time-lag correlation analyses to investigate
the relationship between JFM AO and six ITCZ patterns. The simultaneous correlations of ITCZ and Jan-Mar
AO are weak, with correlation coefficients definitely insignificant (between 0.1 and -0.1). AO is also active in
spring season; therefore we analyzed the statistic relationships between ITCZs frequencies and AO index of
February-April (FMA) and March-May (MAM), respectively. In contrast, FMA and MAM AO index have
very significant relationship with simultaneous ITCZs. Interestingly, the most significant relationship is found
when ITCZ lags AO by 1-2 months. Since the correlations for FMA AO are stronger than MAM, in this paper
we focus on the FMA AO index to explore the connections between AO and ITCZs activity. Accordingly to
the correlation analysis, major results obtained as follow:1)among six ITCZ types, only NI and WI
frequencies have evident correlations with AO; 2)the frequency of SI has insignificant negative correlation
with AO, and AO-SI correlation disappears from May-July; 3)the three inactive ITCZs have relatively weaker
correlations with AO.
Table 3 shows the correlations of FMA AO index with contemporary and lag NI and WI frequencies,
including results from original data and ENSO-removed data. FMA AO has a stable positive correlation with
simultaneous (FMA) and lag (from MAM to August-October) NI frequency, indicating that when FMA AO is
in the positive (negative) phase, the probability that ITCZ center locating in the north of equator tends to be
6
higher(lower) in February to October. By comparing the original and ENSO-removed data, we found that two
groups are consistent with each other in term of magnitude and value. Thus, we can consider the NI frequency
is significantly associated with AO variations, being independent of ENSO. Moreover, this positive
correlation gradually increases from FMA and to the peak in April-June (AMJ) with a maximum of 0.49
(significant at the 99% level), and then decreases afterwards. In July-September the AO-NI correlation is 0.33
(significant at the 90% level), in ASO the positive correlation decreases remarkably and disappears in
September-November. It is also found that AO has a stable negative correlation with WI and the correlation
becomes stronger after removing ENSO signals. The highest correlation occurs in MAM with a correlation
coefficient of -0.41, significant at the 95% level. Overall, when FMA AO is strong, the NI occurs more
frequently from February to October and WI occurs less. The lag correlations are more significant than
contemporary correlation, especially for the AO-NI and AO-WI correlations.
We further investigated the interannual to decadal variations of AO, NI and WI. Figure 2 illustrates the
standardized time series of FMA AO indices, AMJ NI and WI frequencies during the time period of 1979 to
2008. Note that ENSO signals have been linearly removed from AO and ITCZ timeseries. Here we considered
the peak phases of ENSO and its possible time lag influence and used the mean Nino 3.4 SST from previous
December to current months as the predictor when regressing ENSO signals with the AO and ITCZ . The
most remarkable decadal change of AO happens around 1989. From 1979 to 1988, the mean value is -0.55
and there are 9 negative phase years. In contrast, from 1989 to 2008 the mean value increases to 0.29 and
there are 12 positive phases years, occupying 60% of the total 20 years. Accompanying the AO decadal
change, the AMJ NI index has changed similarly. Before 1989 the negative NI anomalies accounts for 60% of
the observation, and the mean value is -0.15; meanwhile after 1989 the positive values occupies 55% and the
mean value increases to 0.07. At the same time, the WI index has the opposite decadal change. In around 1989,
it changes from positive phase (positive anomalies accounts for 70%) to negative phase (negative anomalies
accounts for 60%), and mean values decrease from 0.48 to -0.24 correspondingly. Of the year-to-year
variations, we also note that during positive (negative) AO years, the NI types tend to occur more (less) and
WI types tend to occur less (more). The in-phase AO-NI relationship appears in 24 years, and the out-of-phase
AO-WI relationship appears in 15 years.
Furthermore, the co-variations of AO, NI and WI are also found in extreme years. .For example, in the
year of 1984 the AO index is extremely low, at the same time the NI frequency and WI frequency are
extremely low and high, correspondingly. However, for some years they cannot correspond with each other
well, this may be due to the fact that AO is not the only factor influencing ITCZ activity. Anyway, above
analysis show that FMA AO has significant relationship with Pacific ITCZ variations. Their correlations can
be supported by the associated atmospheric circulation changes as analyzed in the next section.
4. Atmospheric circulation changes
Since FMA AO index has a stable correlation with NI/WI frequency, this kind of stable relationship
should be reflected in the related large-scale circulation or SST field. Thus, to investigate the associated
7
features in atmospheric circulation, the mean AMJ winds at 850 hPa are regressed against FMA AO index and
AMJ NI frequency, respectively. Figure 3 shows the distribution of the winds anomalies, and the composite of
winds of high AO and low AO.
4.1 Low-level winds changes
As shown in Figure 3a in association with positive phase AO there is an obviously anomalous cyclonic
wind in north of the 65°N, and an anti-cyclonic wind located over mid-high latitude of North Pacific.
Importantly, in the lower latitude of North Pacific, there is a huge anomalous cyclonic circulation. In the west
flank of the cyclonic circulation there are anomalous northeastern flows extending southwards along about
140°E, turning eastward at 15°N, and then becoming westerly between 15°N and equator. South of the
equator there is an anomalous easterly winds between the equator and 15°S, which converge with the
anomalous westerly after crossing the equator. In the ITCZ composites, the OLR spatial feature for NI types
(Fig 1a) show a strong convection belt north of equator, covering the region 0°-15°N, 120°E-150°W with a
center located at about 7.5°N. Comparing the regression map of wind and the composite of NI, it is found that
the region of anomalous westerly is coincident with NI precipitation center.
Figure 3 (a) Regression coefficients of 850hPa wind upon Feb-Apr AO (ENSO removed); (b) Composites of 850hPa wind
anomalies in Apr-Jun. Years of composites: 1982, 1990, 1997 and 2002 for high AO; 1979, 1980 and 1984 for low AO. Units
are ms-1. Shading: 90% level.
8
Figure 4 (a) Regression coefficients of 850hPa wind upon Apr-Jun NI (ENSO removed); (b) Composites of 850hPa wind
anomalies in Apr-Jun. Years of composites: 1980, 1982, and1990 for high NI; 1984 and 1993 for low NI. Units are ms-1.
Shading: 90% level.
We should pointed out that regression analysis only can represent the linear relationship and is likely to
underestimate the nonlinear relationship. Thus here we additionally employed composite analysis to examine
the AO-NI connection. First step is to pick out composite years from original time series of AO based on a
threshold of +/-1.0 standard deviation. By this standard 5 high AO years (1982, 1989, 1990, 1997 and 2002)
and 7 low AO years (1979, 1980, 1983, 1987, 1996 and 2005) are identified. The next step is to exclude
ENSO years. If an ENSO event occurs during the period from December in previous year to present June,
then this year is excluded. Here the ENSO events are defined after Climate Prediction Center, i.e., when 3
month running mean of Nino 3.4 SST anomalies above +0.5C(or below -0.5C) during 5 consecutive months it
is an ENSO event (http://www.cpc.noaa.gov). After ENSO years excluded, we finally have 4 high AO years
(1982, 1990, 1997 and 2002) and 3 low AO years (1979, 1980 and 1984) for analysis.
The difference in winds at 850 hPa between high and low AO years is shown in Fig. 3b. It can be found
that the composite features over the North Pacific are consistent with those of Fig. 3a, including an anomalous
anti-cyclonic winds north of the 45°N, the cyclonic winds in the middle latitude and the significant anomalous
westerly between the equator and 15°N. Joseph and Sijikumar[17] studied the variability of 850hPa winds in
association with strong convection and found when the low OLR axis between the equator and 10°N has
moved northward to the latitude between 10°N and 20°N, the maximum wind and convection are located in
10°-20°N correspondingly. This is consistence with our analysis for positive AO cases (Fig 3b), i.e., the
strongest anomalous winds and maximum convection are co-located between 10°N and 20°N over
northwestern Pacific. This suggests a more-than-normal occurrence of NI ITCZ type.
In reverse, we investigated the lower tropospheric wind anomalies in association with NI activities. Fig 4
shows the regression map of winds against NI, the wind composites of high minus low NI years are also
shown together for comparison. The simultaneous regression patterns of the wind anomalies with NI
frequency index in late spring (AMJ, Figure 4a) are highly similar to those with AO index in early spring
(FMA, Figure 3a). They share the feature on the meridional structures, an anomalous anti-cyclonic circulation
over mid-high latitude and cyclonic winds over low latitudes. But we note there are difference in the
significant regions.. The most significant changes in AO-related winds is cyclonic anomalies and
anti-cyclonic anomalies north of the 45°N, and the anomalous westerly in the equator region is less significant;
however, the most significant changes in NI-related winds is the anomalous western flows in north of the
equator. These differences might results from the wind difference directly related to them. The meridional
dipole pattern over mid-high latitude is the essential feature of AO. And in the term of NI, the cyclonic
circulation located in equatorial western Pacific and the anomalous westerly north of the equator might be
more important. Figure 4b shows the difference in winds at 850 hPa between high and low NI years. The high
NI years include 1980, 1982 and 1990, and low years are 1984 and 1993. The anomalous cyclonic winds and
9
tropical westerly in low latitude could be also found in the NI composites. These features are consistent with
the regression coefficients (Fig 4a). All these results indicate that when FMA AO is in positive phase, an
anomalous westerly appears between equator and 15N over western-central Pacific in April-June; the western
flows might enhance convection in north of equator, resulting in more NI type ITCZ occurrence.
Limpasuvan and Hartmann (2000) have shown the meridional circulations associated with the AO, and
found that the high AO tends to enhance the zonal mean raising motion in the tropics and subsidence in the
mid-latitudes. It turns out that the high AO is associated with a stronger Hadley circulation As the rising
branch of Hadley circulation, the ITCZ will consequently strengthen over north of the equator. In conclusion,
when AO is in positive phase, convection could be reinforced in the north of the equator, as well as a more
frequent NI ITCZ pattern.
Figure 5 (a) Regression coefficients of SST upon Feb-Apr AO (ENSO removed); (b) Composites of SST anomalies in Apr-Jun.
Solid lines: positive correlations/anomalies; Dashed lines: negative correlations/anomalies; the contour interval is 0.2 for (a) and
0.4 for (b). Unit is degree. Shading: 90% level.
4.2 SST changes
SST is one important factor that can have impact on ITCZ, and normally ITCZ is preferred to occur at
the latitude of warmer SST[3,19]. Pike[3] investigated the establishment and migration of ITCZ and its
relationship to the profile of SST based on a couple two-layer ocean model and primitive equation
atmospheric model, and found oceanic ITCZ is encouraged to occur near the maximum in surface temperature.
Alapati and Raman[20] constructed a zonally averaged global energy balance model to study the seasonal
migration of ITCZ and they found it depends on the meridional distribution of SST.
Figure 5a shows the regression distribution of SST over North Pacific with the AO index, and Figure 5b
displays the composite differences of SST between positive and negative phase AO years. The AO-related
SSTs (Figure 5a) exhibit a zonal “positive-negative-positive” pattern from mid-high latitudes to low latitude
in Northern Hemisphere. The significant correlation regions locate around the latitude of 45°N and 15°N. In
the positive phase AO year, the tropical SSTs increase over large areas from 120°E to 90°W and from equator
to 15°N. The whole central-eastern Pacific SST changes significantly, with the maximum above 0.3°C at
around 180°E in central Pacific. There is very little change of SST in south of equator. The changes of SST
are not significant between equator and 15°S in the composite differences. Compared AO-related SSTs and
AO-related winds, we find the distributions are very similar and the significant changes are located in the
10
north of equator. That could enlarge the gradient between the south and north of equator and is beneficial to
convergence. In the regression distribution of SST on NI index (figure is omitted), the equatorial asymmetry
of SST anomalies can be seen more clearly. SST anomalies are significant positive from equator to 15°N, but
weak (or negative) from equator to 15°S. Such asymmetry feature is more significant and consistent with
regressed SST against AO, which suggests NI has a close relation with SST anomalies.
Our above analysis indicates that in a positive phase AO year, SSTs are warmer in the north of the
equator and changes little in the south. Such AO-related SST pattern enlarges meridional gradient of SST
between the south and north of equator which would help reinforce convection in the north of the equator, as
well as the more frequencies of north type ITCZ. The situation is reversed in a negative phase AO year. There
are mainly two factors involving in this equatorial asymmetry SST pattern. One is the changes of atmospheric
circulation which directly leads to SST anomalies; another is air-sea interaction. How does AO-related
asymmetry SST pattern occur, however, need further numerical experiments for clarification.
5. Discussion
Xie and Philander[6] studied why ITCZ prefers to stay in the Northern Hemisphere and they proposed
that wind-evaporation-SST (WES) feedback mechanism can contribute to this asymmetry about the equator
over the eastern Pacific. Lin et al.[21] examined the strength of WES feedback over the tropical Pacific Ocean
using multiple long-term observation datasets, and considered the WES feedback has some influence over the
region west of 120°W. According to the WES feedback mechanism, first suppose the SST is slightly warmer
in the north of the equator than in the south, and then this north-south SST gradient will cause north-south sea
level pressure gradient, which will drive southerly winds across the equator. These southerlies turn to west in
the south of equator because of the Coriolis force and eastward north of the equator. Therefore, the anomalous
westerly winds will occur in the north of the equator and easterly winds in the south. Superimposed on the
background easterly winds, in the north of the equator the easterly winds decrease, evaporation decreases,
latent heat flux increases and SSTs become warmer; otherwise, in the south of the equator, the easterly winds
increase, evaporation increases, latent heat flux decreases and SSTs become cooler. We examined the mean
winds at 850 hPa in April-June. In section 4 we show that in the positive phase AO years between equator and
15°N there are anomalous western winds and warmer SST, and anomalous eastern winds exists in south of
equator where SST changes are not significant. All the changes of SST and winds are consistent with WES
feedback processes, implying that the increased NI occurrences may be related to the mechanism of the
positive feedback of WES.
The SST pattern over tropical central-western Pacific, as shown in Figure 5, has some similarity with
ENSO pattern. Thus an arisen question is whether the AO has some influence on the following ENSO event.
Nakamura et al.[22,23] examined the possible influence of the AO in spring on ENSO events in the following
winter. And found an anomalous low tropospheric westerly winds burst over the western tropical Pacific in
the positive AO phase, which may help trigger an ENSO.
We also notice that the correlation between Feb-Apr AO and simultaneous NI/WI frequencies is
11
insignificant, but FMA AO has evident correlations with Mar-May NI/WI frequencies, which implies one to
three months’ lag correlations is more important in AO-ITCZ relation. There seems two reasons for such
phenomenon. One reason is about the seasonal migration of ITCZ itself. Three active ITCZ patterns (NI, SI
and WI) have obvious seasonal dependence. In Feb-Apr the mean ITCZ maintains in Southern Hemisphere,
with SI occupying 60% of whole ITCZ and NI only 10%. Such low occurrence of NI may be one reason for
the insignificant AO-NI relationship. When the mean ITCZ moves northward in MAM, AO-NI relations
become notable. Another reason may be related to the response of NI/WI to AO. If NI/WI responds to AO
changes through the air-sea interactions, there would be time lags.
As the weaker convection pattern over the whole region, WI has negative with AO. We suppose the
reason is because NI and WI are main patterns from April to October, these two occupy 80% of the total
number. This implies that the changes of NI frequency can influence the WI notably given the confined
observation days, .i.e., when one type occurs more frequently the other should be less . Meanwhile, AO is not
the only factor impacting ITCZ. We examined the simultaneous correlation of AO-SI in Jan-Mar and found
the correlation is still weak when JFM AO is the most active among all months and SI frequency is at its peak
(occupying 70% of total), implying that AO’s influence on Southern Hemisphere may be weak.
6. Summary
Based on our observation data analysis, we summary our results as follows. (1) The three major active
ITCZ patterns (NI, SI and WI) have seasonal migration. NI and WI have similar migration features, with peak
in the boreal summer and valley in the boreal winter. The SI migration is opposite, with peak in boreal winter
and valley in boreal summer. Among three minor inactive types, only EI has seasonal migration. (2) The AO
correlations with the frequencies of six type ITCZs are analyzed, and found that in association with positive
AO the NI frequency increases and WI deceases in Apr-Jun, and in negative AO years the situation reverses.
There significant correlation is independent of ENSO. The relations between AO and other four type ITCZ are
not evident. (3) The NI frequencies variations have closely linked to AO-related tropospheric circulation and
SST anomalies. In positive AO years, an anomalous cyclonic wind occurs in north of equator and anomalous
western winds appear between the equator and 15°N and eastern winds in south of the equator. Meanwhile,
SSTs are warmer in north of equator but change little in south of equator. The wind and SST anomalies may
play roles in enhance the convective in north of equator by forming the equatorial SST asymmetry. But the
details need further study.
Acknowledgements. This study was supported by the 973 project (grant no. 2012CB955401) and NSFC (grant
no. 41375071).
Reference
[1] HOLTON J R, WALLACE J M, YOUNG J A. On Boundary layer dynamics and the ITCZ[J]. Journal of
the Atmospheric Science, 1971, 28: 275-280.
12
[2] JIANG S C. The climatic characteristics of the ITCZ all over the global[J]. Acta Meteorologica Sinica,
1988, 46(2): 241-245.
[3] PIKE A C. Intertropical convergence zone studied with an interacting atmosphere and ocean model[J].
Monthly Weather Review, 1971, 99: 469-477.
[4] WALISER D E, SOMERVILIE R C J. Preferred Latitude of the Intertropical Convergence Zone[J].
Journal of the Atmospheric Sciences, 1993, 51(12): 1 619-1 639.
[5] XIE S P. On the genesis of the equatorial annual cycle[J]. Journal of Climate, 1994, 7(12): 2 008–2 013.
[6] XIE S P, PHILANDER S G H. A coupled ocean-atmosphere model of relevance to the ITCZ in the
eastern Pacific[J]. Tellus A, 1994, 46(4), 340-350.
[7] CHEN B D, LIN X, BACMEISTER J T. Frequency Distribution of Daily ITCZ Patterns over the
Western–Central Pacific[J]. Journal of Climate, 2008, 21(17): 4 207-4 222.
[8] GU G, ZHANG C. A spectrum analysis of synoptic-scale disturbances in the ITCZ[J]. Journal of Climate,
2001, 14: 2 725–2 739.
[9] WALISER D E, GAUTIER C. A satellite-derived climatology of the ITCZ[J]. Journal of Climate, 1993,
6: 2 162–2 174.
[10] BARREIRO M, PHILANDER S G H. Response of the tropical Pacific to changes inextratropical
clouds[J]. Climate Dynamics, 2008, 31: 713-729.
[11] LIMPASUVAN V, HARTMANN D L. Wave-maintained annular modes of climate variability[J].
Journal of Climate, 2000, 13(24): 4 414-4 429.
[12] MILLER A J, ZHOU S, YANG S K. Relationship of the Arctic and Antarctic Oscillations to the
outgoing longwave radiation[J]. Journal of Climate, 2003, 16(10): 1 583-1 592.
[13] ZHOU S T, MILLER A J. The Interaction of the Madden-Julian Oscillation and the Arctic
Oscillation[J]. Journal of Climate, 2005, 18(1): 143-159.
[14] SUO L L, HUANG Y J，TAN B K. The influence of winter Arctic Oscillation on maximum and
minimum air temperature over China in winter[J]. Journal of Tropical Meteorology, 2008，24(2)：163-168.
[15] KALNAY E, KANAMITSU M, KISTLER R, et al. The NCEP/NCAR 40-year reanalysis project[J].
Bulletin of the American Meteorological Society, 1996, 77: 437-470.
[16] RAYNER N A, PARKER D E, HORTON E B, et al. Global analyses of SST, sea ice and night marine
air temperature since the late nineteenth century[J]. Journal of Geophysical Research, 2003, 108:
4407-4435, doi:10.1029/2002JD002670.
[17] LIEBMANN B, SMITH C A. Description of a complete (interpolated) outgoing longwave radiation
dataset[J]. Bulletin of the American Meteorological Society, 1996, 77: 1 275-1 277.
[18] JOSEPH P V, SIJIKUMAR S. Intraseasonal Variability of the Low-Level Jet Stream of the Asian
Summer Monsoon[J]. Journal of Climate, 2004, 17(7): 1 449-1 458.
[19] GOSWAMI B N, SHUKLA J, SCHNEIDER E K, et al. Study of the dynamics of the Intertropical
Convergence Zone with a symmetric version of the GLAS climate model[J]. Journal of the Atmospheric
Sciences, 1984, 41(1): 5-19.
[20] ALAPATI K, RAMAN S. A Study of the Seasonal Migration of ITCZ and the Quasi-Periodic
Oscillation in a Simple Monsoon System Using an Energy Balance Model[J]. Meteorology and
Atmospheric Physics, 1989, 41(4): 191-211.
[21] LIN J L, HAN W Q, LIN X. Observational analysis of the wind-evaporation-SST feedback over the
tropical Pacific Ocean[J]. Atmospheric Science Letters, 2008, 9(4): 231-236.
[22] NAKAMURA T, TACHIBANA Y, SHIMODA H. Importance of cold and dry surges in substantiating
the NAM and ENSO relationship[J]. Geophysical Research Letters, 2007, 34, L22703,
doi:10.1029/2007GL031220.
[23] NAKAMURA T, TACHIBANA Y, HONDA M, et al. Influence of the Northern Hemisphere annular
mode on ENSO by modulating westerly wind bursts[J]. Geophysical Research Letters, 2006, 33, L07709,
doi:10.1029/2005GL025432.
13
14