RESEARCH ARTICLE
10.1029/2019JD030595
Key Points:
• Mesoscale oceanic eddies in the
Kuroshio region support stronger
storm growth rate and intensified
cyclones with shorter life cycle
• Presence of mesoscale oceanic
eddies increases moisture supply
significantly and is crucial for
cyclogenesis
Supporting Information:
• Supporting Information S1
Correspondence to:
X. Ma,
maxiaohui@ouc.edu.cn
Citation:
Zhang, X., Ma, X., & Wu, L. (2019).
Effect of mesoscale oceanic eddies on
extratropical cyclogenesis: A tracking
approach. Journal of Geophysical
Research: Atmospheres, 124, 6411–6422.
https://doi.org/10.1029/2019JD030595
Received 6 MAR 2019
Accepted 20 MAY 2019
Published online 25 JUN 2019
©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.
ZHANG ET AL.
Effect of Mesoscale Oceanic Eddies on Extratropical
Cyclogenesis: A Tracking Approach
Xingzhi Zhang1 , Xiaohui Ma1 , and Lixin Wu1
1
Key Laboratory of Physical Oceanography/Institute for Advanced Ocean Studies, Ocean University of China and
Qingdao National Laboratory for Marine Science and Technology, Qingdao, China
Abstract Oceanic eddies populated in the western boundary current regions in the midlatitude have
been found to exert significant influence on atmospheric boundary layer, storm tracks, and large‐scale
atmospheric circulation. However, mechanisms governing how mesoscale sea surface temperature (SST)
anomalies associated with oceanic eddies affect extratropical cyclogenesis remains unclear. Here, we
investigate the influence of Kuroshio oceanic eddies on cyclogenesis in the North Pacific in high resolution
climate model simulations using a cyclone tracking approach. Based on cyclone tracking and composite
analyses, we find that presence of mesoscale SST anomalies almost doubles water vapor supply, leading
to significant increase of diabatic heating release and eddy potential energy to eddy kinetic energy
conversion and thus supporting stronger storm growth rate and intensified cyclones. This finding implies
that moisture process is the key linking mesoscale oceanic eddies in the western boundary current regions
with storm tracks in the midlatitude.
Plain Language Summary Extratropical cyclones transport tremendous amount of heat and
moisture poleward in the midlatitude and are important for climate variability. Mesoscale oceanic eddies,
circulations with horizontal scales of hundreds kilometers, rich in the frontal regions are found to
potentially influence atmospheric storm track and extratropical cyclones. However, the dynamics behind
remains unclear. By identifying and tracking cyclones in high‐resolution model simulations, we
demonstrate that the presence of mesoscale oceanic eddies in the Kuroshio region can support stronger
storm growth rate and intensified cyclones with shorter life cycle. We also provide direct evidence for the
modification effect of mesoscale oceanic eddies on water vapor, heat release, and energy conversion during
cyclone development. The findings have important implications for improving extratropical cyclone
forecasts and climate prediction in the midlatitude.
1. Introduction
It has been recognized that active air‐sea coupling in the midlatitude occurs at frontal scale and mesoscale
(Chelton & Xie, 2010; Small et al., 2008). Recent high‐resolution satellite observations and numerical models
have confirmed the forcing effect of mesoscale SST anomalies associated with oceanic eddies on atmospheric
boundary layer and deep troposphere variables including surface wind, turbulent heat fluxes, boundary
layer height, cloud, and rainfall (Bryan et al., 2010; Chelton et al., 2004; Frenger et al., 2013; Minobe
et al., 2008; Putrasahan et al., 2013). Emerging research also began to notice the storm track and large‐scale
atmospheric circulation response to oceanic fronts and mesoscale oceanic eddies in the western boundary
current regions (Kuwano‐Yoshida & Minobe, 2017; Piazza et al., 2016; Révelard et al., 2016; Sheldon
et al., 2017; Smirnov et al., 2015). However, mechanisms connecting mesoscale oceanic eddies and extratro-
pical storm track dynamics is still not fully understood.
The relationship between sharp oceanic fronts along western boundary currents and extratropical storm
tracks has been widely discussed. Earlier studies stressed the anchoring role of oceanic fronts on storm tracks
through low‐level baroclinicity, emphasizing the importance of SST gradient on cyclogenesis. (Nakamura &
Shimpo, 2004). The following studies further confirmed that variability of atmospheric storm track is closely
related to underlying oceanic front change like shift of Kuroshio and Gulf Stream extensions (Brayshaw
et al., 2008; Joyce et al., 2009; O'Reilly & Czaja, 2015). Until very recently, a new mechanism is proposed,
indicating that mesoscale oceanic eddies can modulate storm track genesis and development via moist bar-
oclinic instability (Ma et al., 2015, 2017), drawing our attention to moist dibabatic process in affecting
6411
 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

cyclogenesis (Willison et al., 2013). The sensitivity of diabatic heating associated with synoptic storms to
mesoscale SST anomalies is also verified in the idealized atmospheric simulations (Foussard et al., 2019).
All the previous work is carried out under Eulerian framework, which obscures the impacts of oceanic
eddies on cyclogenesis at different developing stages. In this study, using a tracking approach to detect
and follow cyclones simulated in a regional atmospheric model in the North Pacific, we attempt to
gain further understanding of different storm track response during evolution of extratropical cyclones
and factors contributing to cyclogenesis in the presence of mesoscale oceanic eddies in the Kuroshio
extension region.
The paper is organized as follows. Section 2 describes model configuration and experiment design, cyclone
detection and tracking algorithms, composite analyses, and significance test applied in this study. Section 3
presents the main findings including influence of mesoscale oceanic eddies on statistical characteristic of
cyclones, cyclogenesis response to mesoscale oceanic eddies, and dynamics behind. Discussion and conclu-
sions are given in Section 4.

2. Data and Methods

2.1. Model Configuration and Experiment Design
Weather Research and Forecasting (WRF) model with 27‐km horizontal resolution is used in this study
(Skamarock et al., 2008). Though the focus of the study is in the Kuroshio extension region, the computa-
tional domain covers the entire North Pacific ([3.6°N–66°N], [99°E–270°E]) to avoid boundary effect. A twin
set of ensemble experiments with and without mesoscale SST forcing are conducted in boreal winter season
from 1 October 2007 to 31 March 2008. The 2007/2008 winter is chosen because the Kuroshio was in an
unstable state and eddy activity was strong in this year. At the same time, it was a relatively neutral El
Niño–Southern Oscillation and Pacific decadal oscillation year that helps to minimize the influence from
the tropics and large‐scale climate mode of variability. (Ma et al., 2017). Each of the twin ensembles contains
10 members with different initial conditions. The initial conditions are derived from NCEP2 reanalysis with
the same day in 10 different years, that is, 1 October in 2002, 2003, … , 2011 (Ma et al., 2015). All settings of
the twin ensembles are identical except for the SST forcing. One ensemble is forced with daily high‐
resolution (0.09°) Microwave InfraRed Optimal Interpolated satellite SST (Gentemann et al., 2004, 2010),
which resolves mesoscale oceanic eddies in the Kuroshio extension and is referred to as control (CTRL)
simulation. The other ensemble is forced with low‐pass spatial filtered SST, which removes mesoscale SST
anomalies and is referred to as filter (FILT) simulation. Figures 1a–1c show averaged SST field in
2007/2010 in CTRL and FILT and the mesoscale SST anomalies removed. We note that the largest mesocale
SST anomalies removed are confined in the Kuroshio extension region, while the large‐scale SST gradient
remains similar in CTRL and FILT. Detailed model configuration and parameterization schemes applied
in the experiments can be found in Ma et al. (2015). Considering the model spin‐up time, only the last five
months (November–March, NDJFM) of WRF simulation is used in the following analyses. In addition, 20
years (1979–1999) NCEP‐CFSR (the National Centers for Environmental Prediction‐Climate Forecast
System Reanalysis; Saha et al., 2010) data set is used for comparison of cyclone characteristics with
model simulations.

2.2. Cyclone Detection and Tracking

Considering different key features of extratropical cyclones, various detection and tracking algorithms have
been developed for trajectory analysis. A comprehensive comparison of commonly used tracking algorithms
of extratropical cyclones is given by Neu et al. (2013). Single or multiple atmospheric variables such as sea
level pressure (SLP), geopotential height, and vorticity can be used as metrics. In this study, 6‐hourly SLP
anomalies (SLPa) derived by removing climatology winter season (NDJFM) mean of 10 ensemble members
are used to identify cyclones in WRF simulations. The center of a cyclone is defined as the minimum of the
innermost closed SLPa contour. The edge of a cyclone is defined as the outmost closed SLPa contour, which
encloses only one cyclone center (Wernli & Schwierz, 2006). For cyclone tracking, the position of cyclone at
time step n + 1 is determined by searching the nearest SLPa minimum to previous cyclone center (at time
step n) within certain distance (800 km/6 hr) along the direction of background wind. The direction of
background wind is estimated by averaged total wind over the area of cyclone at previous time step n,
adjusted by propagation path of cyclone (Hoskins & Pedder, 1980; Wernli & Schwierz, 2006). Tracking

ZHANG ET AL. 6412

 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

Figure 1. Sea surface temperature (°C) averaged in 2007/2010 in CTRL (a), FILT (b), and difference between CTRL and FILT (c). (d) Examples of two cyclone tra-
jectories detected in CTRL. The two cyclones are generated in the Kuroshio extension region. Red dots indicate centers of cyclone at different time steps (n = 1, 2, 3,
… , corresponding to t = 0, 6, 12 hr, …). For each cyclone detected, black box outlines the area (24° × 24°) to conduct composite analysis.

stops if cyclone at time step n + 1 cannot be found. The first point detected along a cyclone track is marked as
the genesis site and defined as Time 0. Only cyclones living longer than 1 day are included in our analyses.
For twin‐core cyclones, we choose the minimum SLPa of the two cores as the center of detected cyclones. If
two cyclones merged into one, then the merged cyclone will be considered as the continuation of the
stronger cyclone while the weaker cyclone is discarded. If one cyclone separates into two, the stronger
one of the separated two cyclones will be considered as the continuation of the previous cyclone. To
minimize the uncertainties from tracking method, we visually checked each of the cyclone tracks to make
sure the tracks identified are as accurate as possible. Examples of two cyclone trajectories identified in
CTRL simulation are shown in Figure 1d. Alternative tracking method using geopotential height field is
tested and gives similar results.

2.3. Composite Analyses and Significance Test

In order to investigate the impact of mesoscale SST anomalies on cyclogenesis, we focus our study on
cyclones generated in Kuroshio extension region where mesoscale eddies are most active and SST anomalies
are strongest ([30°N–45°N], [135°E–175°E]). As shown in Figure 1c, there are also mesoscale SST differences
in marginal seas along the east coast of Asia, which may exert influence on atmospheric cyclones (Kuwano‐
Yoshida & Minobe, 2017). To minimize the influence of mesoscale SST forcing in these regions, the compo-
site analyses are conducted in the following way. Though cyclones over the whole computational domain are
detected and tracked, only cyclones generated in the Kuroshio extension region are selected for composite
analyses, that is, centers of cyclones at first time step locate in [30°N–45°N, 135°E–175°E]. Cyclones gener-
ated in the marginal seas and other regions are excluded. Futhermore, for each cyclone detected in CTRL, a
paired cyclone in FILT is searched and determined if its genesis time/location is close to the cyclone center
detected in CTRL, that is, genesis time/location difference between CTRL and FILT is less than 18 hr/300 km.
Cyclones that are already notably different between CTRL and FILT before they enter the Kuroshio
extension region are excluded. This is to minimize the initial difference between cyclones, assuring that
the different cyclone response between CTRL and FILT primarily comes from mesoscale SST forcing in
the Kuroshio extension region. A total of 151 paired cyclones are identified and tracked in CTRL and
FILT, respectively. In the following of this study, all composite analyses are performed based on these
paired cyclones.
Composites of specific variables are computed in cyclone coordinate with zero indicating the center of
cyclone (see Figure 3a for an example). For each cyclone identified at certain time step, the corresponding
atmospheric field within a 24° × 24° box is aligned relative to the cyclone center and composite is then made
over all cyclones detected at this time step. The anomalous fields for composite analyses are also derived by

ZHANG ET AL. 6413

 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

Figure 2. Winter season (November–March) averaged cyclone track density derived from National Centers for Environmental Prediction‐Climate Forecast System
Reanalysis (NCEP‐CFSR; shading) and CTRL (contours) (a). Difference of winter season averaged cyclone track density between CTRL and FILT (b). Differences
significant at the 90% confidence level based on T test are shaded by black dots. Intensity of cyclones as a function of time derived from NCEP‐CFSR (black),
CTRL (blue), and FILT (red) (c). The intensity is computed by averaging minimum sea level pressure anomalies of all cyclones detected at time T (T = 0, 6, 12,
14 hr, ...). Histogram of number of cyclones pre winter season as a function of cyclone lifetime derived from NCEP‐CFSR (black), CTRL (blue), and FILT (red) (d).

removing climatology winter season mean. A student's t test is applied for given variables when comparing
the difference between CTRL and FILT, assuming each of the cyclone identified is an independent sampling.

3. Results
3.1. Statistical Characteristics of Cyclones
To assess the performance of WRF model in reproducing extratropical cyclones, we compare the statistical
characteristics of cyclones identified and tracked in CTRL simulation against that in NCEP‐CFSR. Similar
detection and tracking algorithm described in section 2.2 is applied to 6‐hourly NCEP‐CFSR SLPa in 20
winter seasons (NDJFM). Figure 2a illustrates the seasonal averaged track density incorporating full life
cycles of all cyclones detected in CTRL and NCEP‐CFSR in the North Pacific. Here, track density of cyclones
is computed by counting the number of cyclones detected in per 5° × 5° unit area and then applying a spatial
smoothing (2° × 2° boxcar filter) for better visualization. Extratropical cyclones mostly occur in the Kuroshio
extension region with an average maximum of 12 cyclones per winter and have a southwest‐northeast
orientation in both CTRL and NCEP‐CFSR. The spatial distribution is consistent with the near‐surface storm
track pattern computed based on band‐pass filtered variables in Eulerian view (Chang et al., 2002), validat-
ing the accuracy of detection and tracking algorithm applied. Figure 2c shows the time evolution of cyclone
intensity measured by the minimum SLPa at cyclone center. On average, cyclones experience relatively
rapid growth during the first 3 days, followed by a slow decay thereafter. Compared with NCEP‐CFSR,
the model correctly reproduces the development of cyclones at each stage with the maximum strength
occurring on day 3, except that the amplitude is slightly underestimated. Probability density functions of
cyclone lifetime also show similar distribution between CTRL and NCEP‐CFSR (Figure 2d). Lifetime of
detected cyclones varies from 1 to 9 days with an averaged value of 3.7 (3.9) days for CTRL (NCEP‐CFSR).
Overall, WRF simulated cyclone track, intensity, evolution, and lifetime agree well with reanalysis, support-
ing the fidelity of model results.
Difference of track density between CTRL and FILT displays that the occurrence of cyclones in the Kuroshio
extension region remains almost unchanged, while a significant change arises downstream in the eastern
Pacific (Figure 2b). The downstream track density change is dominated by increased occurrence in the north
and decreased occurrence in the southeast. This is consistent with storm track and mean flow shift in this
area revealed by previous studies (Ma et al., 2015, 2017). Furthermore, the eastward extension of cyclone

ZHANG ET AL. 6414

 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

Figure 3. Composite of sea level pressure anomalies in CTRL (a) and difference of that between CTRL and FILT (b) at T = 0, 12, 24, 36 hr. Composite of
vertical section of geopotential height anomalies in CTRL (c) and difference of that between CTRL and FILT (d) at T = 0, 12, 24, 36 hr. The vertical profile is
constructed in the following way: A zonal average (4°) across the cyclone center (minimum geopotential height anomalies) at each isobaric layer is first calculated
and then the averaged lines are aligned with increasing height. The x axis of the vertical profile is oriented from south to north. Differences significant at 90%
confidence level based on T test are shaded by black dots.

ZHANG ET AL. 6415

 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

trajectory seems to suggest longer lifetimes of cyclones in FILT. In fact, this is substantiated by probability
density functions of cyclone lifetime (Figure 2d), which clearly exhibit increased number of cyclones toward
longer lifetime (>4 days) in FILT. These results indicate that mesoscale oceanic eddies in the Kuroshio
extension region exert a potential impact on the evolution and lifetime of cyclones, which can in turn modify
the downstream cyclone trajectory.

3.2. Cyclogenesis Response to Oceanic Eddies

The explosive growth of cyclones takes place when cyclones pass over the Kuroshio extension region
(Kuwano‐Yoshida & Minobe, 2017). To evaluate the direct impact of mesoscale SST anomalies on local
cyclogenesis, we compare the composite of the 151 paired cyclones picked in CTRL and FILT during their
genesis period, that is, from time = 0 to time = 42 hr with cyclone deepening rate less than 0. The cyclone
deepening rate is computed as the difference of minimum SLPa at cyclone centers between two time steps
based on SLPa composite. As mentioned in section 2.3, the initial growth of picked cyclones primarily occurs
in the Kuroshio extension region. For all cyclones selected for composite analyses, there is no systematically
position shift between CTRL and FILT in the Kuroshio extension region as can be inferred by cyclone track
density difference west of 175°E in Figure 2b. Figure 3a shows the development of cyclones at different stages
at surface. Cyclones gradually grow with deepening SLPa and increasing size. The composite difference
between two simulations reveals intensified deepening of cyclones in CTRL than in FILT, moreover, the
difference increases with time (Figure 3b), suggesting stronger growth rate and cyclone intensity in CTRL.
On average, the cyclone intensity in CTRL is ~20% stronger than that in FILT.
The stronger cyclone growth is further collaborated by the composite of vertical profile of geopotential
height anomalies. To capture the cyclone center (minimum geopotential height anomalies) at different
levels, a 4° zonal average across the cyclone center at each isobaric layer is first calculated and then the aver-
aged lines are aligned with increasing height (Figure S1 in the supporting information). The titled geopoten-
tial height anomalies from surface to upper level is supportive of baroclinicinic cyclone growth (Hoskins
et al., 2009). Consistent with SLPa, cyclones in the lower troposphere develop with weak amplitude at the
initial stage and gradually evolves into deep structure with increased size and amplitude. Correspondingly,
the difference of geopotential height originates from surface and extends to upper troposphere (Figure 3d).
We note that the significant change of cyclone intensity is confined below 500 hPa, while cyclone intensity
change at upper level is weaker and insignificant, confirming the underlying SST forcing effect
on cyclogenesis.
A further analysis of the cyclone deepening rate (Yoshiike & Kawamura, 2009) clearly shows the different
evolution of storms in CTRL and FILT (Figure 4). Among all 151 paired cyclones selected, about 15% of them
undergo explosive growth, with SLP drop greater than 24 hPa/24 hr. On average, there is about 8.4 (6.2) hPa
SLP drop from 0 to 24 hr in CTRL (FILT) and 6.9 (7.6) hPa drop from 24 to 48 hr. The results reveal that while
the abrupt growth of cyclones in CTRL occurs in relative shorter time and leads to quick saturation of
cyclones, the growth rate is slowed down and prolonged in FILT. This is supportive of the longer lifetime
of cyclones in FILT shown in Figure 2d. All the above results point to the fact that the presence of oceanic
eddies supports stronger storm growth rate and generates intensified cyclones.

3.3. Moisture Related Response to Oceanic Eddies

We next explore the dynamics connecting mesoscale oceanic eddies and extratropical cyclogenesis. Previous
studies have suggested moist baroclinicity associated with diabatic heating release is a likely candidate to
explain the storm track response to mesoscale oceanic eddies (Foussard et al., 2019; Ma et al., 2017;
Willison et al., 2013). To gain further understanding on this mechanism, we compare moisture related pro-
cesses between CTRL and FILT. Figures 5a and 5c show 850‐hPa water vapor mixing ratio (Q) and air tem-
perature (T) anomalies at 18‐hr during cyclogenesis. Warm and moist air is transported in front of cyclone
along warm conveyer belt and eventually converged to cyclone's center, giving rise to a comma structure
and eastward shift of maximum Q and T, consistent with previous findings (Foussard et al., 2019; Hirata
et al., 2015). We then compare the vertical structure of Q and T where their maxima occur between the
two simulations (Figures 5b and 5d). Large amount of water vapor is extracted from the ocean and trapped
below 500 hPa in cyclones during their genesis. The amount of water vapor supply in CTRL is about twice of
that in FILT as shown in Figure 5b. In contrast, the temperature change is less than 10% and does not pass

ZHANG ET AL. 6416

 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

the significance test (Figure 5d). It is surprising that mesoscale oceanic

eddies can bring such tremendous moisture change during cyclogenesis,
which is not noted by previous studies.
Composite of surface turbulent heat fluxes demonstrates the maximum
heat fluxes out of the ocean locates behind the cyclone center due to the
cold and dry air carried by the cold branch of cyclones. The presence of
mesoscale oceanic eddies leads to remarkable increase of turbulent heat
fluxes with ~120 W/m2 turbulent heat flux release during cyclogenesis
in CTRL compared to ~60 W/m2 of that in FILT (Figure 6). Further
decomposition of turbulent heat fluxes into sensible and latent heat flux
reveals that the difference mainly comes from latent heat flux (figure
not shown), consistent with the water vapor change. We also notify that
Figure 4. Time series of sea level pressure anomalies (SLPa) deepening rate heat flux response in the presence of mesoscale SST forcing in the compo-
(hPa/6 hr) in CTRL (blue) and FILT (red). SLPa deepening rate is
computed as the difference of minimum SLPa at cyclone centers between n
site analysis is stronger than previous study (Foussard et al., 2019). This
and n + 1 time step based on SLPa composite. raises the concern that if there is contribution from large‐scale SST asso-
ciated with cyclones chosen for the composite analysis. To verify this,
the difference of SST composite between CTRL and FILT is shown in Figure 7. SST difference pattern is basi-
cally noisy and weak with no significant large‐scale SST structure especially during cyclone initial stages.
There is even SST cooling in CTRL at T = 36 hr (though insignificant), suggesting a possible negative

Figure 5. Composite of geopotential height (contours, with interval of 10 m), wind (vector, m/s) and water vapor mixing ratio Q (shading, kg/kg) anomalies at 850
hPa in CTRL at T = 18 h (a). Composite of vertical section of Q in CTRL (contours) and difference of that between CTRL and FILT (shading) at T = 18 hr (b). (c, d) As
in (a) and (b) but for air temperature (T, K) anomalies accompanying with cyclones. The vertical profile is computed by averaging 4° zonal band centered along
longitude = 2 on each isobaric layer. Longitude = 2 is where maximum Q/T occurs as shown in Figure 3a/3c. Strongest storm growth occurs at T = 18 hr as shown in
Figure 4. Differences significant at 90% confidence level based on T test are shaded by black dots.

ZHANG ET AL. 6417

 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

Figure 6. Composite of surface turbulent heat fluxes (shading) and sea level pressure anomalies (contours) in CTRL at T = 0, 12, 24, 36 hr (a). Difference of tur-
bulent heat fluxes composite between CTRL and FILT at T = 0, 12, 24, 36 hr (b). Composite region is extended to include complete turbulent heat fluxes structure.
Differences significant at 90% confidence level based on T test are shaded by black dots.

contribution to turbulent heat flux change in Figure 6. The results confirm that the increased turbulent heat
fluxes in CTRL should primarily come from mesoscale SST anomalies associated with oceanic eddies. We
speculate the stronger heat flux is a combined effect of mesoscale SST anomalies and cyclone intensity
change. The presence of mesoscale SST anomalies first supports stronger storm growth through enhanced
water vapor and heat flux release, the intensified cyclones then in turn increase heat flux release, indicating
a positive feedback between heat flux release and cyclone growth. We also make an effort to estimate the
heat flux and water vapor change induced by cyclone intensity difference (see details in the supporting infor-
mation). Although a positive relationship between heat flux and cyclone intensity is confirmed, as the effect

Figure 7. Sea surface temperature composite (contours) and difference (shading) of that between CTRL and FILT at T = 0, 12, 24, 36 hr. Differences significant at
90% confidence level based on T test are shaded by black dots.

ZHANG ET AL. 6418

 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

Figure 8. Composite of vertically integrated (from 1,000 to 300 hPa) diabatic heating (shading, hPa·K/s) and geopotential height anomalies at 850 hPa
(contours, with interval of 10 m) in CTRL at T = 0, 12, 24, 36 hr (a). Difference of vertically integrated diabatic heating composite between CTRL and FILT at T
= 0, 12, 24, 36 hr (b). A 4° × 4° boxcar filter is applied for clarity. Composite region is extended to include complete diabatic heating structure. Differences significant
at 90% confidence level based on T test are shaded by black dots.

of mesoscale SST forcing and cyclone intensity are highly connected, the enhancement of cyclone intensity is
essentially motivated by mesoscale eddies.
The trapped water vapor in cyclones is then transported upward. The rising motion of moist air will lead to
intense diabatic heating release in the troposphere, which is crucial to cyclogenesis according to moist bar-
oclinicity theory (Emanuel et al., 1987; Lapeyre & Held, 2004). Figure 8 shows the vertical integrated diabatic
heating composite at different time steps during cyclogenesis. Similarly to water vapor distribution, the max-
imum diabatic heating also occurs east of cyclone center due to the moist air and rising motion inside the
warm branch of cyclones. The presence of mesoscale oceanic eddies results in significant increase along
the warm branch of cyclones. In line with the enhanced diabatic heating release, the upward water vapor
transport (w′Q′) by cyclones is significantly stronger in CTRL (Figure 9a). The increased diabatic heating
further disturbs the stability and enhances vertical motion (Figure 9b), leading to more energy conversion
from eddy potential energy (EPE) to eddy kinetic energy (EKE). Figure 9c displays the w′T′ composite which
can be used as the estimation of EPE‐to‐EKE conversion according to temperature variance equation
(Lorenz, 1955; Marshall & Shutts, 1981). While the maximum w′Q′ is confined below 500 hPa, the maximum
w′T′ can extend deeper into the upper troposphere attributed to the rising motion inside cyclones.
Comparison of w′T′ between CTRL and FILT shows ~15% increase of EPE‐to‐EKE conversion in CTRL, cor-
responding to the enhanced water vapor transport and vertical motion. The increased energy conversion
finally contributes to intensified cyclones with stronger EKE as estimated by v′v′ (Figure 9d). Collectively,
results above indicate that mesoscale oceanic eddies induce remarkable modification of water vapor inside
cyclones, which is the source for further cyclogenesis. The tracking and composite analyses provide explicit
evidence for moist diabatic mechanism by exhibiting water vapor, diabatic heating, and energy conversion
change directly over cyclones.

ZHANG ET AL. 6419

 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

Figure 9. Composite of vertical section of w′Q′ (a), w′w′ (b), w′T′ (c), and v′v′ (d) in CTRL (contours) and difference of that between CTRL and FILT (shading) at T
= 18 hr. The vertical profile is constructed in the following way: Take w′Q′ as an example, a zonal average (4°) centered along maximum w′Q′ at each isobaric
layer is first calculated and then the averaged lines are aligned with increasing height. The x axis is oriented from south to north. Differences significant at 90%
confidence level based on T test are shaded by black dots.

4. Discussion and Conclusions

Using regional high‐resolution atmospheric model simulations with and without mesoscale SST forcing, we
identified and tracked cyclones in the Kuroshio extension region and found that the presence of mesoscale
SST anomalies associated with oceanic eddies tends to sustain stronger storm growth rate and intensified
cyclones with shorter life cycle. The presence of mesoscale SST anomalies results in remarkable water vapor
increase, which stimulates enhanced diabatic heating and eddy energy conversion associated with the
upward transport of moisture, eventually supporting stronger storm growth rate and intensified cyclones.
This study highlights that moisture processes play a crucial role for understanding of interaction between
oceanic eddies and cyclogenesis in the western boundary current regions in the midlatitude. Mesoscale ocea-
nic eddies can remarkably modify moisture process associated with cyclogenesis and thus resolving mesos-
cale oceanic eddies may have important implication for improving extratropical cyclone prediction in
climate models. Although our study focuses in Kuroshio extension region, the findings may be applied to
other frontal regions like Gulf Stream where both oceanic eddies and storm track are active.
It is also worth pointing out that since there is no significant large‐scale background SST difference between
CTRL and FILT in this study, the net increase of turbulent heat flux and water vapor during cyclogenesis in
the presence of oceanic eddies suggests a stronger influence from warm eddies than cold eddies. The rectified
effect of warm and cold eddies on turbulent heat fluxes is also notified by a recent study (Foussard et al.,
2019). However, the heat fluxes response in their study (~10 W/m2) is much weaker compared to that
revealed in our study (~60 W/m2), possibly because they computed zonal averaged heat flux change to
oceanic eddies while we focus on heat fluxes change following cyclones. In addition, the larger heat flux
and water vapor response during cyclogenesis in our study suggests a positive feedback. The presence of
mesoscale oceanic eddies supports stronger cyclone growth through enhanced water vapor and heat flux
release; the intensified cyclones in turn can further increase water vapor and heat flux release. But note that
cyclone changes are fundamentally attributed to mesoscale SST anomalies associated with oceanic eddies.

ZHANG ET AL. 6420

 Journal of Geophysical Research: Atmospheres
Acknowledgments
We thank Ning Zhao and the other
anonymous reviewer who helped
greatly improve the manuscript. This
research is supported by National Key
R&D Program of China
(2017YFC1404100 and
2017YFC1404101) and National
Natural Science Foundation of China
(41776013). We thank the Texas A&M
Supercomputing Facility and the Texas
Advanced Computing Center (TACC)
at the University of Texas at Austin and
Center for High Performance
Computing and System Simulation at
Qingdao Pilot National Laboratory for
Marine Science and Technology for
providing high performance computing
resources that contributed to the
research results reported in this paper.
The reanalysis and SST data used in this
study can be obtained from NCEP
(https://cfs.ncep.noaa.gov/cfsr/down-
loads/) and the Remote Sensing
Systems website (http://www.remss.
com/measurements/sea‐surface‐tem-
perature). Data sets used for producing
all figures are publicly available
(https://doi.pangaea.de/10.1594/
PANGAEA.901018).
ZHANG ET AL.
10.1029/2019JD030595
References
Brayshaw, D. J., Hoskins, B., & Blackburn, M. (2008). The storm‐track response to idealized SST perturbations in an aquaplanet GCM.
Journal of the Atmospheric Sciences, 65(9), 2842–2860. https://doi.org/10.1175/2008JAS2657.1
Bryan, F. O., Tomas, R., Dennis, J. M., Chelton, D. B., Loeb, N. G., & McClean, J. L. (2010). Frontal scale air–sea interaction in high‐
resolution coupled climate models. Journal of Climate, 23(23), 6277–6291. https://doi.org/10.1175/2010JCLI3665.1
Chang, E. K., Lee, S., & Swanson, K. L. (2002). Storm track dynamics. Journal of Climate, 15(16), 2163–2183. https://doi.org/10.1175/1520‐
0442(2002)015<02163:STD>2.0.CO;2
Chelton, D. B., Schlax, M. G., Freilich, M. H., & Milliff, R. F. (2004). Satellite measurements reveal persistent small‐scale features in ocean
winds. Science, 303(5660), 978–983. https://doi.org/10.1126/science.1091901
Chelton, D. B., & Xie, S.‐P. (2010). Coupled ocean‐atmosphere interaction at oceanic mesoscales. Oceanography, 23(4), 52–69. https://doi.
org/10.5670/oceanog.2010.05
Emanuel, K. A., Fantini, M., & Thorpe, A. J. (1987). Baroclinic instability in an environment of small stability to slantwise moist convection.
Part I: Two‐dimensional models. Journal of the Atmospheric Sciences, 44(12), 1559–1573. https://doi.org/10.1175/1520‐
0469(1987)044<1559:BIIAEO>2.0.CO;2
Foussard, A., Lapeyre, G., & Plougonven, R. (2019). Storm tracks response to oceanic eddies in idealized atmospheric simulations. Journal
of Climate, 32(2), 445–463. https://doi.org/10.1175/JCLI‐D‐18‐0415.1
Frenger, I., Gruber, N., Knutti, R., & Münnich, M. (2013). Imprint of Southern Ocean eddies on winds, clouds and rainfall. Nature
Geoscience, 6(8), 608–612. https://doi.org/10.1038/ngeo1863
Gentemann, C. L., Meissner, T., & Wentz, F. J. (2010). Accuracy of satellite sea surface temperatures at 7 and 11 GHz. IEEE Transactions on
Geoscience and Remote Sensing, 48(3), 1009–1018. https://doi.org/10.1109/TGRS.2009.2030322
Gentemann, C. L., Wentz, F. J., Mears, C. A., & Smith, D. K. (2004). In situ validation of Tropical Rainfall Measuring Mission microwave
sea surface temperatures. Journal of Geophysical Research, 109, C04021. https://doi.org/10.1029/2003JC002092
Hirata, H., Kawamura, R., Kato, M., & Shinoda, T. (2015). Influential role of moisture supply from the Kuroshio/Kuroshio Extension in the
rapid development of an extratropical cyclone. Monthly Weather Review, 143(10), 4126–4144. https://doi.org/10.1175/MWR‐D‐15‐0016.1
Hoskins, B., & Pedder, M. (1980). The diagnosis of middle latitude synoptic development. Quarterly Journal of the Royal Meteorological
Society, 106(450), 707–719. https://doi.org/10.1002/qj.49710645004
Hoskins, B. J., McIntyre, M. E., & Robertson, A. W. (1985). On the use and significance of isentropic potential vorticity maps. Quarterly
Journal of the Royal Meteorological Society, 111(470), 877–946. https://doi.org/10.1002/qj.49711147002
Joyce, T. M., Kwon, Y. O., & Yu, L. (2009). On the relationship between synoptic wintertime atmospheric variability and path shifts in the
Gulf Stream and the Kuroshio Extension. Journal of Climate, 22(12), 3177–3192. https://doi.org/10.1175/2008JCLI2690.1
Kuwano‐Yoshida, A., & Minobe, S. (2017). Storm‐track response to SST fronts in the northwestern Pacific region in an AGCM. Journal of
Climate, 30(3), 1081–1102. https://doi.org/10.1175/JCLI‐D‐16‐0331.1
Lapeyre, G., & Held, I. (2004). The role of moisture in the dynamics and energetics of turbulent baroclinic eddies. Journal of the Atmospheric
Sciences, 61(14), 1693–1710. https://doi.org/10.1175/1520‐0469(2004)061<1693:TROMIT>2.0.CO;2
Lorenz, E. N. (1955). Available potential energy and the maintenance of the general circulation. Tellus, 7, 157–167.
Ma, X., Chang, P., Saravanan, R., Montuoro, R., Hsieh, J.‐S., Wu, D., et al. (2015). Distant influence of Kuroshio eddies on North Pacific
weather patterns? Scientific reports, 5(1), 17,785. https://doi.org/10.1038/srep17785
Ma, X., Chang, P., Saravanan, R., Montuoro, R., Nakamura, H., Wu, D., et al. (2017). Importance of resolving Kuroshio front and eddy
influence in simulating the North Pacific storm track. Journal of Climate, 30(5), 1861–1880. https://doi.org/10.1175/JCLI‐D‐16‐0154.1
Marshall, J., & Shutts, G. (1981). A note on rotational and divergent eddy fluxes. Journal of Physical Oceanography, 11(12), 1677–1680.
https://doi.org/10.1175/1520‐0485(1981)011<1677:ANORAD>2.0.CO;2
Minobe, S., Kuwano‐Yoshida, A., Komori, N., Xie, S.‐P., & Small, R. J. (2008). Influence of the Gulf Stream on the troposphere. Nature,
452(7184), 206–209. https://doi.org/10.1038/nature06690
Nakamura, H., & Shimpo, A. (2004). Seasonal variations in the Southern Hemisphere storm tracks and jet streams as revealed in a rea-
nalysis dataset. Journal of Climate, 17(9), 1828–1844. https://doi.org/10.1175/1520‐0442(2004)017<1828:SVITSH>2.0.CO;2
Neu, U., Akperov, M. G., Bellenbaum, N., Benestad, R., Blender, R., Caballero, R., et al. (2013). IMILAST: A community effort to inter-
compare extratropical cyclone detection and tracking algorithms. Bulletin of the American Meteorological Society, 94(4), 529–547. https://
doi.org/10.1175/BAMS‐D‐11‐00154.1
O'Reilly, C. H., & Czaja, A. (2015). The response of the Pacific storm track and atmospheric circulation to Kuroshio Extension variability.
Quarterly Journal of the Royal Meteorological Society, 141(686), 52–66. https://doi.org/10.1002/qj.2334
Piazza, M., Terray, L., Boé, J., Maisonnave, E., & Sanchez‐Gomez, E. (2016). Influence of small‐scale North Atlantic sea surface tempera-
ture patterns on the marine boundary layer and free troposphere: A study using the atmospheric ARPEGE model. Climate Dynamics,
46(5‐6), 1699–1717. https://doi.org/10.1007/s00382‐015‐2669‐z
Putrasahan, D. A., Miller, A. J., & Seo, H. (2013). Isolating mesoscale coupled ocean–atmosphere interactions in the Kuroshio Extension
region. Dynamics of Atmospheres and Oceans, 63, 60–78. https://doi.org/10.1016/j.dynatmoce.2013.04.001
Révelard, A., Frankignoul, C., Sennéchael, N., Kwon, Y. O., & Qiu, B. (2016). Influence of the decadal variability of the Kuroshio Extension
on the atmospheric circulation in the cold season. Journal of Climate, 29(6), 2123–2144. https://doi.org/10.1175/JCLI‐D‐15‐0511.1
Saha, S., Moorthi, S., Pan, H.‐L., Wu, X., Wang, J., Nadiga, S., et al. (2010). The NCEP climate forecast system reanalysis. Bulletin of the
American Meteorological Society, 91(8), 1015–1058. https://doi.org/10.1175/2010BAMS3001.1
Sheldon, L., Czaja, A., Vannière, B., Morcrette, C., Sohet, B., Casado, M., & Smith, D. (2017). A ‘warm path’ for Gulf Stream‐troposphere
interactions. Tellus A: Dynamic Meteorology and Oceanography, 69(1). https://doi.org/10.1080/16000870.2017.1299397
Skamarock, W., Klemp, J., Dudhia, J., Gill, D., Barker, D., Duda, M., et al., (2008). A description of the advanced research WRF Version 3,
NCAR technical note, Mesoscale and Microscale Meteorology Division. National Center for Atmospheric Research, Boulder, Colorado,
USA. doi:https://doi.org/10.5065/D68S4MVH.
Small, R. J., deSzoeke, S. P., Xie, S.‐P., O'Neill, L., Seo, H., Song, Q., et al. (2008). Air‐sea interaction over the ocean fronts and eddies.
Dynamics of Atmospheres and Oceans, 45(3‐4), 274–3319. https://doi.org/10.1016/j.dynatmoce.2008.01.001
Smirnov, D., Newman, M., Alexander, M. A., Kwon, Y.‐O., & Frankignoul, C. (2015). Investigating the local atmospheric response to a
realistic shift in the Oyashio sea surface temperature front. Journal of Climate, 28(3), 1126–1147. https://doi.org/10.1175/JCLI‐D‐14‐
00285.1
Wernli, H., & Schwierz, C. (2006). Surface cyclones in the ERA‐40 dataset (1958–2001). Part I: Novel identification method and global
climatology. Journal of the Atmospheric Sciences, 63(10), 2486–2507. https://doi.org/10.1175/JAS3766.1
6421
 Journal of Geophysical Research: Atmospheres 10.1029/2019JD030595

Willison, J., Robinson, W. A., & Lackmann, G. M. (2013). The importance of resolving mesoscale latent heating in the North Atlantic storm
track. Journal of the Atmospheric Sciences, 70(7), 2234–2250. https://doi.org/10.1175/JAS‐D‐12‐0226.1
Yoshiike, S., & Kawamura, R. (2009). Influence of wintertime large‐scale circulation on the explosively developing cyclones over the
western North Pacific and their downstream effects. Journal of Geophysical Research, 114, D13110. https://doi.org/10.1029/
2009JD011820

ZHANG ET AL. 6422