The Pacific Meridional Mode and the Occurrence of Tropical 1 Cyclones in the Western North Pacific 2

3 4

W. Zhang 1,2,*, G. A. Vecchi1,2, H. Murakami1,2, G. Villarini3, 5

L. Jia1,2 6

7 1National Oceanic and Atmospheric Administration/Geophysical Fluid 8

Dynamics Laboratory, Princeton, NJ, USA 9 2Atmospheric and Oceanic Sciences Program, Princeton University, 10

Princeton, NJ, USA 11 3 IIHR-Hydroscience & Engineering, The University of Iowa, Iowa City, 12

Iowa 13 14 15 16 17 18

*Corresponding author: 19 Wei Zhang 20 21 National Oceanic and Atmospheric Administration/Geophysical Fluid 22 Dynamics Laboratory, Princeton, NJ, USA, 08540 23 Phone: 609-452-5817 24 Email: wei.zhang@noaa.gov 25

26 27 28

29

1

Abstract 30

This study investigates the association between the Pacific Meridional Mode (PMM) 31

and tropical cyclone (TC) activity in the western North Pacific (WNP). It is found that 32

the positive PMM phase favors the occurrence of TCs in the WNP while the negative 33

PMM phase inhibits the occurrence of TCs there. Observed relationships are 34

consistent with those from a long-term preindustrial control experiment (1000 years) 35

of a high-resolution TC-resolving Geophysical Fluid Dynamics Laboratory (GFDL) 36

Forecast-oriented Low Ocean Resolution (FLOR) coupled climate model. The 37

diagnostic relationship between the PMM and TCs in observations and the model is 38

further supported by sensitivity experiments with FLOR. The modulation of TC 39

genesis by the PMM is primarily through the anomalous zonal vertical wind shear 40

(ZVWS) changes in the WNP, especially in the southeastern WNP. The anomalous 41

ZVWS can be attributed to the responses of the atmosphere to the anomalous 42

warming in the northwestern part of the PMM in the positive PMM phase, which 43

resembles a classic Matsuno-Gill pattern. Such influences on TC genesis are 44

strengthened by a cyclonic flow over the WNP. The significant relationship between 45

TCs and the PMM identified here may provide a useful reference for seasonal 46

forecasting of TCs and interpreting changes in TC activity in the WNP. 47

48

49

50

51 2

1. Introduction 52

Tropical cyclones (TC) are among the most destructive and costly natural disasters on 53

the earth (e.g., Rappaport 2000; Pielke Jr et al., 2008; Zhang et al., 2009). The 54

occurrence of TCs has long been the subject of scientific inquiry(e.g., Mitchell, 1932; 55

Gray, 1968; Henderson-Sellers et al., 1998; Simpson et al., 1998; Vitart and Stockdale, 56

2001; Liu and Chan, 2003; Klotzback, 2007; Chan, 2008; Sippel and Zhang, 2008; 57

Lin et al., 2014; Vecchi et al., 2014). Large-scale climate variations in various basins 58

affect the occurrence of TCs in the WNP by both local and remote forcing. In general, 59

variations in TC occurrence in the WNP are closely associated with the sea surface 60

temperature (SST) patterns such as the El Niño Southern Oscillation (ENSO) (Chan, 61

1985; Wu and Lau, 1992; Chan, 2000; Wang and Chan, 2002; Camargo and Sobel, 62

2005; Zhang et al., 2012), the Indian ocean warming (Du et al., 2010; Zhan et al., 63

2010; Zhan et al., 2014), the North Pacific Gyre Oscillation (Zhang et al., 2013) and 64

the Pacific Decadal Oscillation (PDO; Lee et al., 2012; Liu and Chan, 2012; 65

Girishkumar et al., 2014). Recently, it was suggested that the SST in the North 66

Atlantic also modulates TC activity in the WNP (Li et al., 2013; Huo et al., 2015). 67

The Pacific Meridional Mode (PMM) is here defined as the first Maximum 68

Covariance Analysis (MCA) Mode of SST and 10-m wind vectors in the Pacific, and 69

describes meridional variations in SST, winds and convection in the tropical Pacific 70

Ocean (Chiang and Vimont, 2004). The PMM bears marked resemblance to the 71

Atlantic Meridional Mode (AMM) in terms of its principal mechanisms, impacts and 72

3

interactions with extra-tropical climate systems (Chiang and Vimont, 2004). For 73

example, the North Pacific Oscillation (NPO) has been found to force the PMM, 74

while the North Atlantic Oscillation (NAO) acts to force the AMM (Vimont et al., 75

2003). 76

The PMM is closely associated with ENSO (Chang et al., 2007; Zhang et al., 2009; 77

Larson and Kirtman, 2013; Lin et al., 2014), with the PMM peaking around April and 78

May whereas ENSO usually reaches its peak in boreal winter (Chang et al., 2007; 79

Zhang et al., 2009; Larson and Kirtman, 2013; Lin et al., 2014). Previous studies have 80

indicated that the PMM may set the stage for the seasonal peak of ENSO (Larson and 81

Kirtman, 2014). However, the PMM can also exist independently of ENSO, with the 82

ENSO effect routinely removed through linear regression onto the cold tongue index 83

(CTI) when calculating the PMM index (Chiang and Vimont, 2004; Zhang et al., 2009; 84

Larson and Kirtman, 2014). 85

Previous research has found that the seasonal occurrence and accumulated energy of 86

hurricanes in the North Atlantic are strongly influenced by the AMM (Vimont and 87

Kossin, 2007; Smirnov and Vimont, 2010). More recently, Patricola et al., (2014) 88

examined the combined influence of the PMM and ENSO on hurricane activity in the 89

North Atlantic. Basically, the PMM consists of two parts: a northwestern part and a 90

southeastern part in the Eastern Pacific (Chiang and Vimont, 2004). During the 91

positive PMM phase, positive SST anomalies prevail in the northwestern part while 92

negative SST anomalies prevail in the southeastern part (Chiang and Vimont, 2004). 93

4

Such SST anomalies can be hypothesized to affect TC activity in the WNP by 94

modulating the large-scale circulation, much like the PMM has been found to 95

modulate weather and climate in East Asia (Li et al., 2010; Li and Ma, 2011). 96

Therefore, it is of great interest to examine whether the PMM modulates the 97

occurrence of TCs in the WNP. 98

The objectives of this study are thus: i) to assess whether and to what extent the PMM 99

modulates TC activity in the WNP, and ii) to dissect the mechanisms underpinning the 100

modulation based on both observations and model simulation. This study is expected 101

to enhance the understanding of TC activity in the WNP and to provide valuable 102

insights for seasonal forecasting of the occurrence of TCs over that region. Moreover, 103

since the PMM is linked to extra-tropical climate systems such as NPO (Rogers, 1981) 104

and NPGO (Di Lorenzo et al., 2010) which is the second principal mode of SST 105

anomalies in the North Pacific (Vimont et al., 2001), this study can play a role in 106

bridging the association of extra-tropical and tropical interactions. 107

The remainder of this paper is organized as follows. Section 2 presents the data and 108

methodology, while Section 3 discusses the results based on observations and 109

simulations. Section 4 includes the discussion and summarizes the main conclusions. 110

2. Data and Methodology 111 2.1 Data 112

The TC data are obtained from International Best Track Archive for Climate 113

Stewardship (IBTrACS; Knapp et al., 2010). The study period is 1961 - 2013 because 114

TCs after the 1960s are more reliable (Chan, 2008). The key meteorological variables 115

5

such as zonal and meridional wind fields, geopotential height, vorticity and relative 116

humidity are from the Japan Meteorology Agency (JMA) 55 reanalysis data (JRA-55). 117

To substantiate and reinforce the results, the National Centers for Environmental 118

Prediction and the National Center for Atmospheric Research (NCEP/NCAR; Kalnay 119

et al,. 1996) reanalysis data and the Modern Era Retrospective-analysis for Research 120

and Applications (MERRA) reanalysis are also employed. The JRA-55 is available 121

since 1958 on a global basis with a spatial resolution of 1.25° × 1.25°. It is based on a 122

new data assimilation system that reduces many of the problems reported in the first 123

JMA reanalysis (Ebita et al., 2011; Kobayashi et al., 2015). MERRA is produced by 124

the National Aeronautics and Space Administration (NASA) for the satellite era, using 125

a major new version of the Goddard Earth Observing System Data Assimilation 126

System Version 5 (GEOS-5) (Rienecker et al., 2011). MERRA data is only available 127

from 1979. The zonal vertical wind shear (|du/dz| or |uz|) is defined as the magnitude 128

of the differences in zonal wind between 200 hPa and 850 hPa level. The years used 129

in the above three reanalysis data are consistent with TC data. 130

Monthly estimates of SST are taken from the Met Office Hadley Centre 131

HadSST3.1.1.0 (Kennedy et al., 2011). The PMM index is calculated following the 132

method described in Chiang and Vimont (2004). In calculating it, the seasonal cycle 133

and the linear trend are first removed by applying a three-month running mean to the 134

data, and then subtracting the linear fit to the cold tongue index (CTI, Deser and 135

Wallace, 1987) from every single spatial point to remove correlations with El Niño 136

6

(Chiang and Vimont, 2004). The CTI is defined as the average SST anomaly over 137

6 °N - 6 °S, 180 - 90 °W minus the global mean SST (Deser and Wallace, 1987). 138

The peak season for TCs is defined as July-October. The results are consistent if we 139

slightly change the definition of the TC peak season from June to November or from 140

July to November (not shown). 141

2.2 Model 142

This study employs a newly developed high-resolution coupled climate model for 143

experiments, the Geophysical Fluid Dynamics Laboratory (GFDL) Forecast-oriented 144

Low Ocean Resolution Version of CM2.5 or FLOR (e.g., Vecchi et al., 2014; Jia et al., 145

2015;Yang et al., 2015). The aim in the development of FLOR was to represent 146

regional scales and extreme weather and climate events (e.g., TCs) keeping 147

computational costs relatively low. For these reasons, this model is characterized by 148

high-resolution land and atmosphere components but relatively low-resolution ocean 149

component (Vecchi et al., 2014). The atmosphere and land are the same as those of the 150

GFDL Climate Model version (CM) 2.5 (CM2.5; Delworth et al., 2012) with a spatial 151

resolution of 50 km × 50 km. The ocean and sea ice components of FLOR are similar 152

to those in the CM2.1 (Delworth et al., 2006) with a spatial resolution of 1×1 degree. 153

The relatively low spatial resolution of the ocean and sea ice components in FLOR 154

enables large ensembles and better efficiency for seasonal forecasting. This study 155

utilizes the flux-adjusted version of FLOR (FLOR-FA) for sensitivity experiments in 156

which climatological adjustments are applied to the model's momentum, enthalpy and 157

freshwater flux from the atmosphere to the ocean to give the model's SST and surface 158 7

wind stress a closer climatology to the observations over 1979 - 2012 (Vecchi et al., 159

2014). For more details about the FLOR model, please refer to Vecchi et al. (2014), 160

Jia et al. (2015) and references therein. 161

2.3 Control simulation 162

A 2500-year long control simulation was performed with the FLOR-FA by prescribing 163

radiative forcing and land use representative of 1860. These experiments are referred 164

to as "Pre-industrial" experiments with FLOR-FA. The control experiments are run for 165

a total of 2500 years, but here we focus on the first 1000 years for the analysis of 166

PMM and TCs for the sake of computing efficiency. The 500-year control simulation 167

by prescribing radiative forcing and land use representative of 1990 with FLOR-FA is 168

also analyzed on PMM-TC association. Because the PMM-TC association based on 169

FLOR-FA 1990 is consistent with that on FLOR-FA 1860, this study only shows 170

results based on FLOR-FA 1860. 171

2.4 Tracking algorithm 172

TCs in FLOR are tracked from 6-hourly model output by using the tracker developed 173

by Harris et al. (in prep) as implemented in Murakami et al. (2015, submitted to 174

Journal of Climate). The temperature anomaly averaged vertically over 300 and 500 175

hPa (ta) and sea level pressure (SLP) are key factors of this tracker. The tracking 176

procedures are as follows. 177

(1) Local minima of the smoothed SLP field are found. The location of the center is 178

properly adjusted by fitting a biquadratic to the SLP and locating the center at the 179

minimum. 180

8

(2) Closed contours is in an interval of 2 hPa (dp) around every single SLP low center. 181

The Nth contour is marked as the contiguous region surrounding a low central pressure 182

P with pressures lower than dp × N + P, as detected by a “flood fill” algorithm. It is 183

noted that the contours are not required to be circular and a maximum radius of 3,000 184

km is searched from each candidate low center. 185

(3) If the distances of contours are close enough, the low is counted as a TC center. In 186

this way, the tracker attempts to find all closed contours within a certain distance of 187

the low center and without entering contours belonging to another low. The maximum 188

10-m wind inside the set of closed contours is taken as the maximum wind speed at 189

that time for the storm. 190

(4) Warm cores are detected via similar processes: closed 1K contours for FLOR are 191

found surrounding the maximum ta within a TC's identified contours, no more than 1 192

degree from the low center. This contour must have a radius less than 3 degrees in 193

distance. If such a core is not found, it should not be marked as a warm-core low 194

center and the center is rejected. 195

(5) TC centers are combined into a track by taking a low center at time T – dt, 196

extrapolating its motion forward dt, and then seeking for storms within 750 km. It is 197

noted that a deeper low center has higher priority of tracking. 198

(6) The following criteria are required to pick up the final TCs. 199

a. A lifespan of at least 72 hours. 200

b. At least 48 cumulative (not necessarily consecutive) hours with a warm core. 201

9

c. At least 36 consecutive hours of a warm core with winds greater than 17.5 202

m s–1. 203

TC genesis is confined equatorward of 40 °N. TC density in the WNP is binned into 204

5° × 5° grid boxes at a 6-hour interval. 205

3. Results 206

The analysis results of the PMM-TC association are derived from observations, a 207

1000 year control experiment, and sensitivity experiments based on FLOR-FA. 208

3.1 Observations 209

The regressions of SST and 10-m wind fields onto the PMM index in observations are 210

shown in Figure 1a. There is a characteristic pattern to the PMM in the central and 211

eastern Pacific, with warming in the northwestern part (subtropical region) and 212

cooling in the southeastern part (tropical) (Figure 1a). The warming in the central 213

Pacific is in the vicinity of the western North Pacific and may affect TC activity there. 214

The regressions of TC track density onto the PMM index for 1961-2013 are largely 215

positive in the WNP in observations, indicating that TC activity is enhanced 216

(suppressed) in the WNP in the positive (negative) PMM phase (Figure 1b). In 217

addition, we calculate the correlation between the time series of the PMM index and 218

the TC frequency in the WNP for the period of 1961 - 2013. For each year, the PMM 219

index is calculated by averaging the monthly PMM index over the peak TC season 220

(July-October). The correlation coefficient between the two indices is 0.46, significant 221

at the 0.01 level (Figure 2). Although previous studies have argued that the occurrence 222

10

of the PMM is independent of ENSO (Zhang et al., 2009; Larson and Kirtman, 2014), 223

and the linear ENSO effect has been removed when calculating the PMM index, 224

partial correlation is calculated by controlling the Niño 3 index to further exclude the 225

linear influence of ENSO on the PMM-TC association. The partial correlation 226

coefficient is slightly more significant (0.48) after controlling for the ENSO effect 227

(not shown). 228

To analyze the observed relationship between WNP TC genesis and PMM, the WNP 229

is divided into five sub-regions following the definition of sub-regions in the WNP in 230

Wang et al. (2013): the South China Sea (SCS), northwestern part (NW), northeastern 231

part (NE), southwestern part (SW) and southeastern part (SE) (Figure 3). The SE part 232

has a relatively high density of TC genesis (Figure 3). The PMM index has a 233

significant positive correlation with TC frequency in the SE, NE and NW parts 234

(Figure 3). 235

To further assess which large-scale atmospheric conditions may be acting to modulate 236

the TC occurrence, a suite of variables associated with TC genesis (Chan, 2000; Wang 237

and Chan, 2002) based on JRA-55 are regressed onto the PMM index (Figure 4). 238

Chan and Liu (2004) found that the increase in local SST has no significant influence 239

on TC activity in the WNP. Indeed, the modulation of local large-scale circulation by 240

remote SST anomalies (e.g., ENSO) largely leads to changes in TC activity there 241

(Chan, 2000; Wang and Chan, 2002; Chan and Liu, 2004; Camargo and Sobel, 2005). 242

Therefore, instead of analyzing SST and maximum potential intensity, we analyze 243

11

600-hPa relative humidity, 850-hPa relative vorticity and zonal vertical wind shear 244

(ZVWS) for TC genesis. In general, the regressions of the PMM index on JRA-55, 245

NCEP/NCAR, and MERRA reanalysis data are consistent. Thus, we only show the 246

results from JRA-55 reanalysis data. The regression of ZVWS is characterized by 247

strong negative anomalies in a large portion of the WNP (0 - 20 °N, 120 °E - 180 °E), 248

which would act to enhance TC genesis (Figure 4). Further analysis indicates that the 249

changes in ZVWS arise from anomalous westerly (easterly) in 850 hPa (200 hPa) 250

level (Figure 4) although the anomalous westerly in 850 hPa is mainly located in the 251

western part of the WNP. Such anomalous winds in the lower and upper levels can 252

weaken the prevalent westerlies in the upper level (200 hPa) and the prevalent 253

easterlies (850 hPa) in the lower level over the WNP. We hypothesize that the large-254

scale circulation responses to the SST patterns of the PMM alter the Walker 255

Circulation, which is responsible for the changes in the lower and upper level wind 256

flow. The mechanisms underlying how the vertical wind shear is modulated by the 257

large-scale circulation are discussed in detail in Subsection 3.3. The anomalous 258

positive relative humidity in the SE part also supports higher TC genesis in the WNP 259

in MERRA. Although the MERRA reanalysis is only available from 1979, the 260

satellite-based data may be more reliable (not shown). The AMM pattern (the Atlantic 261

counterpart of PMM) also has a significant positive correlation with hurricane 262

frequency in the North Atlantic because of increased subtropical SST, convergence 263

12

and low sea level pressure and decreased vertical wind shear (Vimont and Kossin, 264

2007; Smirnov and Vimont, 2010). 265

266 3.2 Results from the long-term control run 267

The 1000-year control simulation of FLOR-FA with 1860 radiative forcing is 268

analyzed to further understand the PMM-TC association. The main purpose is to test 269

whether such PMM-TC association also holds for the long-term control simulation of 270

FLOR-FA, which would provide a dynamical framework in which to understand the 271

causes of this connection. The PMM index in the control experiment is also calculated 272

based on the method proposed in Chiang and Vimont (2004). The PMM in FLOR-FA 273

control simulation (Figure 5a) captures the fundamental features of the observed 274

PMM (Figure 1a), indicating that the FLOR-FA model is capable of simulating the 275

basic structure of the PMM in a fully-coupled numerical climate system. Despite 276

these encouraging comparison, the control experiment of FLOR-FA produces a 277

weaker cooling and 10-m wind fields in the southeastern part of PMM, and stronger 278

cooling northwest of the positive anomalies compared to observation. In addition, the 279

squared covariance fraction of the PMM in observation is 53% (Chiang and Vimont, 280

2004) whereas it is 30% in the control experiment, suggesting a weaker role of PMM 281

in the control experiment. 282

The regressions of TC track density on the PMM index are generally positive in the 283

WNP in control experiment (1000-year, Figure 5b), consistent with observations 284

(Figure 1b). However, the positive regressions are largely confined in the southeastern 285

13

WNP (Figure 5b). The correlation coefficient between the simulated PMM and WNP 286

TC frequency is 0.37 for the 1000-year simulation, significant at 0.01 level of 287

significance (Figure 6a). Figure 6b shows the histogram of the correlation coefficients 288

in moving 50-year chunks and all are positive. The correlation coefficient from the 289

observations is within the model spread. This suggests that the observed TC-PMM 290

association may be robust, as it is significant even in a long-term climate control 291

simulation. 292

The model shows negative anomalies in ZVWS in the southeastern part (0 – 20 °N, 293

150 °E–180 °E) of the WNP (Figure 7), similar to those based on observations 294

(Figures 4). There are positive anomalies of ZVWS westward of 150°E in the control 295

experiment, suggesting unfavorable condition for TC genesis there. This is 296

responsible for different spatial patterns for TC genesis between observation and 297

control simulation (Figure 3). Observations show that TC genesis is promoted in 298

almost all the sub-regions in observations (Figure 3). In contrast, the control 299

simulation shows that TC genesis is enhanced in the southeastern part while 300

suppressed in the southwestern sub-region of the WNP (Figure 3). It is noted that we 301

have also analyzed the 500-year FLOR-FA control with 1990 radiative forcing and the 302

results derived from both control runs are consistent in terms of the PMM pattern and 303

TC-PMM association. A possible reason responsible for discrepancy in TC genesis 304

between the control experiment and observations is the pre-industrial radiative forcing 305

in the control experiment. 306

14

In summary, the 1000-year control experiment produces quantitatively consistent 307

results with observation, with positive phases of the PMM conducive to WNP TC 308

genesis, and anomalous reductions of ZVWS during the positive PMM phases. 309

However, TC genesis exhibits different regional relationships between observations 310

and the control experiment, which may be explained by different anomalous ZVWS 311

patterns in model and observations. Interestingly, the regressions of relative humidity 312

in this control experiment is consistent with those in MERRA reanalysis (figure not 313

shown) in which positive anomalies are observed eastward of 140°E. In order to 314

further investigate the mechanisms underpinning how PMM modulates TC genesis in 315

the WNP, a set of sensitivity experiments have been performed using FLOR-FA and 316

will be discussed in next subsection. 317

3.3 Results from sensitivity experiments 318

To isolate the role of the PMM, a set of perturbation experiments were performed 319

with FLOR-FA using 1860 radiative forcing. In the first experiment, SST is restored 320

to a repeating annual cycle in the PMM region with a 5-day restoring timescale, 321

which is taken as the control run (CTRL) (Figure 8). The second experiment is set up 322

by prescribing the sum of the annual cycle of SST and the temporally constant SST 323

anomalies associated with positive PMM patterns (denoted as PPMM) in the PMM 324

region (Figure 8). After an initial 100-year spinup, both experiments are integrated for 325

100 years. The mask of the restoring SST is shown in Figure 8, with the coefficient 326

varying from 1 to 0 when moving from the center to the boundary of the masking 327

15

region. By subtracting the CTRL experiment from the PPMM experiment, one can 328

assess the net influence of the positive PMM phase on the climate system. 329

The WNP mean TC frequency during the peak TC season (JASO, July to October) in 330

the 100-year PPMM experiment (19.4) is significantly higher than the control 331

experiment (17.5) at 0.01 significance level (Table 1). The difference in WNP TC 332

frequency between the PPMM experiment (26.7) and CTRL (24.7) is also significant 333

for June-November (JJASON) (Table 1). To better represent the differences between 334

the two experiments, TC density is computed by binning TC centers onto each 5°×5° 335

grid box in the WNP. Positive anomalies of TC density are observed in the eastern 336

part of the WNP (Figure 9). However, the positive anomalies have higher magnitude 337

than the negative ones. These positive anomalies arise from an enhanced TC genesis 338

in the PPMM experiment (Figure 10). The occurrence of TC genesis in the PPMM 339

experiment is higher than that in the control run in the eastern part of the WNP 340

(Figure 10). These experiments corroborate that the positive PMM phases promote 341

TC genesis in the WNP, especially in the eastern portion of the WNP. A number of 342

studies have proposed mechanisms on how remote SST patterns affect TC genesis in 343

the WNP, including the Matsuno-Gill pattern (Matsuno, 1966; Gill, 1980), Pacific-344

East Asia teleconnections (Wang et al., 2000; Wang and Chan, 2002), the Philippine 345

anticyclone induced by Indian ocean warming (Xie et al., 2009; Ha et al., 2014) and 346

SST gradients (Zhan et al., 2013; Li and Zhou, 2014). These studies have emphasized 347

16

the importance of the Philippines anticyclone and monsoon trough in modulating TC 348

genesis and how SST anomalies modulate the large-scale circulation in remote areas. 349

To explain why TC genesis in the PPMM experiment is higher in the eastern part of 350

the WNP, the large-scale atmospheric factors related to TC genesis are analyzed. The 351

differences between the PPMM and control run resemble the regression of those 352

variables onto the PMM index in both control experiment and observations (Figure 353

11). This suggests that the SST pattern in the positive PMM phase and within that 354

particular box drives the modeled and observed TC genesis in the WNP, especially in 355

its eastern part. Specifically, negative anomalies of ZVWS are induced by the PPMM, 356

arising from westerly anomalies at the850 hPa level and easterly anomalies at the 200 357

hPa (Figure 11). These changes can be described as a change in the Walker 358

Circulation. The question is then related to the mechanisms that lead to the changes in 359

the Walker Circulation. It is known that the positive PMM pattern features a warming 360

in the northwestern part and a cooling in the southeastern part in the tropical eastern 361

Pacific (Figures 1, 5, 12). A cyclonic 850 hPa flow is forced by the anomalous 362

warming in the tropical central Pacific through the Matsuno-Gill response (Matsuno, 363

1966; Gill, 1980). The responses to anomalous positive PMM warming are quite 364

similar to the Matsuno-Gill pattern under asymmetric heating (Gill, 1980), showing 365

that a cyclonic pattern is located northwestward of the heating (Figure 12a). In 366

addition, the diabatic heating center can also be identified by negative outgoing 367

longwave radiation (OLR) anomalies (Figure 12a). The anomalous cyclonic flow 368

17

pattern in the WNP is more likely to cause a weaker Philippine anticyclone, a weaker 369

subtropical high and elongated monsoon trough, which are favorable conditions for 370

TC genesis in the WNP (Wang et al., 2000; Wang and Chan, 2002). Intensified low-371

level westerly and easterly flows move towards the PPMM heating center from east 372

and west, respectively (Figure. 12a). Such patterns are also pronounced in the profile 373

of zonal and vertical winds (Figure 12b). An updraft of the Walker circulation shifts to 374

the eastern part of the WNP (around 180°). Notice that the zonal wind is averaged 375

over 0° to 20°N when calculating vertical profile for the zonal and vertical wind 376

(Figure 12b). Anomalous easterlies (westerlies) in upper (lower) levels is elongated 377

from 140°E to 180°E (Figure 12b). Such tropospheric flow patterns weaken the 378

westerlies in the upper levels (e.g., 200 hPa) and easterlies in the lower levels (e.g., 379

850 hPa). The combined effect of changes in the lower and upper level winds is 380

responsible for the weakened zonal vertical wind shear in the eastern part of the WNP 381

(Figure 11), which leads to increased TC genesis there. Therefore, enhanced TC 382

genesis in the WNP can be attributed to weaker ZVWS, which may arise from the 383

responses of the atmosphere to the anomalous warming which is part of the positive 384

PMM pattern by the Matsuno-Gill mechanisms. 385

386 4. Discussion and conclusion 387

The Pacific Meridional Mode (PMM) has been the subject of heightened interest by 388

the scientific community because of its role in bridging across the extra-tropical and 389

tropical climate systems. It is thus of high scientific relevance whether the PMM is 390

18

linked to TCs in the WNP given that previous studies claimed that its counterpart 391

AMM is intimately associated with hurricane activity in the North Atlantic (Vimont 392

and Kossin, 2007; Smirnov and Vimont, 2010; Patricola et al., 2014). 393

This study has analyzed the observed and modeled association between the PMM and 394

TC activity in the WNP. Our results show that the positive PMM phase promotes the 395

occurrence of TCs in the WNP. There is a general consistency in the diagnostic TC-396

PMM relationship in observations and in a long-term control experiment (1000 years) 397

of FLOR-FA. Sensitivity experiments using the model show causality in the PMM/TC 398

relation. The modulation TC genesis by the PMM originates mainly from the 399

abnormal ZVWS in the WNP, especially in its southeastern portion as shown in the 400

control experiment and sensitivity experiments. The anomalous ZVWS can be 401

attributed to the responses to the atmosphere to the anomalous warming in the 402

positive PMM pattern through Matsuno-Gill mechanisms. Such results are supported 403

by the cyclonic flow in the Philippine Sea and relevant flow patterns over the Pacific 404

in the sensitivity experiments. 405

The PMM is linked to the climate modes in the North Pacific such as NPO and NPGO 406

(Chiang and Vimont, 2004). For example, NPO, which is the atmospheric component 407

of NPGO, tends to force the PMM (Chiang and Vimont, 2004). Therefore, PMM 408

bridges the extra-tropical and tropical climate systems. As shown in Zhang et al. 409

(2013), NPGO/NPO modulates the occurrence of TCs in the WNP and it plays an 410

even stronger role than ENSO and PDO based on observations. In addition, Chen et al. 411

19

(2015) reported that the Spring NPO can modulate TC activity in the WNP in the 412

following peak TC season. However, detailed mechanisms are yet to be unraveled. 413

This study may provide some hints to explain the impacts of NPGO or NPO on TCs 414

in the WNP. It is known that the PMM pattern is obtained by removing the effect of 415

ENSO via linearly removing of the cold tongue index (Deser and Wallace, 1990). TCs 416

in the WNP are affected by various climate modes such as ENSO, the Indian ocean 417

warming, NPGO, PDO, and the SST in the North Atlantic Ocean. This study has also 418

found that the PMM significantly modulates the occurrence of TCs in the WNP. An 419

on-going comprehensive study involving all of these SST patterns is analyzing their 420

relative roles in influencing TC genesis in the WNP. Given the important role of the 421

PMM in modulating the climate, the linkage between PMM and TC activity in other 422

ocean basins will be analyzed in our future study. Moreover, the results of this work 423

indicate that the predictability of the PMM is potentially of great value for the 424

prediction of TC activity; future studies will examine the predictability of the PMM 425

and TCs, by leveraging seasonal forecasting experiments with FLOR-FA. 426

427

Acknowledgments 428

The authors thank Xiaosong Yang and Andrew Wittenberg for their valuable 429

comments on an earlier version of this paper. WZ benefits from discussion with 430

Lakshmi Krishnamurthy on this study. The authors thank Seth Underwood and 431

Fanrong Zeng for their helpful assistance in experiments. 432

20

433

21

Reference 434 Camargo, S. J. and A. H. Sobel, 2005: Western North Pacific tropical cyclone 435

intensity and ENSO. Journal of Climate, 18, 2996-3006. 436 Chan, J. C. L., 1985: Tropical Cyclone Activity in the Northwest Pacific in Relation to 437

the El Niño Southern Oscillation Phenomenon. Monthly Weather Review, 113, 438 599-606. 439

Chan, J. C. L., 2000: Tropical cyclone activity over the western North Pacific 440 associated with El Niño and La Nina events. Journal of Climate, 13, 2960-2972. 441

Chan, J. C. L. and K. S. Liu, 2004: Global Warming and Western North Pacific 442 Typhoon Activity from an Observational Perspective. Journal of Climate, 17, 443 4590-4602. 444

——, 2008: A Simple Seasonal Forecast Update of Tropical Cyclone Activity. 445 Weather and Forecasting, 23, 1016-1021. 446

Chan, J.C.L., 2008: Decadal variations of intense typhoon occurrence in the western 447 North Pacific. Proceedings of the Royal Society A: Mathematical, Physical and 448 Engineering Science, 464, 249-272. 449

Chang, P., L. Zhang, R. Saravanan, D. J. Vimont, J. C. H. Chiang, L. Ji, H. Seidel, and 450 M. K. Tippett, 2007: Pacific meridional mode and El Niño—Southern Oscillation. 451 Geophysical Research Letters, 34, L16608. 452

Chen, D., Wang, H., Liu, J. and Li, G. , 2015: Why the spring North Pacific 453 Oscillation is a predictor of typhoon activity over the Western North Pacific. Int. J. 454 Climatol, in press. 455

Chiang, J. C. H. and D. J. Vimont, 2004: Analogous Pacific and Atlantic Meridional 456 Modes of Tropical Atmosphere–Ocean Variability. Journal of Climate, 17, 4143-457 4158. 458

Delworth, T. L., A. Rosati, W. Anderson, A. J. Adcroft, V. Balaji, R. Benson, K. Dixon, 459 S. M. Griffies, H.-C. Lee, R. C. Pacanowski, G. A. Vecchi, A. T. Wittenberg, F. 460 Zeng, and R. Zhang, 2012: Simulated Climate and Climate Change in the GFDL 461 CM2.5 High-Resolution Coupled Climate Model. Journal of Climate, 25, 2755-462 2781. 463

Delworth, T. L., A. J. Broccoli, A. Rosati, R. J. Stouffer, V. Balaji, J. A. Beesley, W. F. 464 Cooke, K. W. Dixon, J. Dunne, K. A. Dunne, J. W. Durachta, K. L. Findell, P. 465 Ginoux, A. Gnanadesikan, C. T. Gordon, S. M. Griffies, R. Gudgel, M. J. Harrison, 466 I. M. Held, R. S. Hemler, L. W. Horowitz, S. A. Klein, T. R. Knutson, P. J. 467 Kushner, A. R. Langenhorst, H.-C. Lee, S.-J. Lin, J. Lu, S. L. Malyshev, P. C. D. 468 Milly, V. Ramaswamy, J. Russell, M. D. Schwarzkopf, E. Shevliakova, J. J. Sirutis, 469 M. J. Spelman, W. F. Stern, M. Winton, A. T. Wittenberg, B. Wyman, F. Zeng, and 470 R. Zhang, 2006: GFDL’s CM2 Global Coupled Climate Models. Part I: 471 Formulation and Simulation Characteristics. Journal of Climate, 19, 643-674. 472

Deser, C. and J. M. Wallace, 1987: El Niño events and their relation to the Southern 473 Oscillation: 1925–1986. Journal of Geophysical Research: Oceans, 92, 14189-474 14196. 475

22

Deser, C. and J. M. Wallace, 1990: Large-Scale Atmospheric Circulation Features of 476 Warm and Cold Episodes in the Tropical Pacific. Journal of Climate, 3, 1254-477 1281. 478

Di Lorenzo, E., K. Cobb, J. Furtado, N. Schneider, B. Anderson, A. Bracco, M. 479 Alexander, and D. Vimont, 2010: Central Pacific El Nino and decadal climate 480 change in the North Pacific Ocean. Nature Geoscience, 3, 762-765. 481

Du, Y., L. Yang, and S.-P. Xie, 2010: Tropical Indian Ocean Influence on Northwest 482 Pacific Tropical Cyclones in Summer following Strong El Niño. Journal of 483 Climate, 24, 315-322. 484

Ebita, A., S. Kobayashi, Y. Ota, M. Moriya, R. Kumabe, K. Onogi, Y. Harada, S. 485 Yasui, K. Miyaoka, K. Takahashi, H. Kamahori, C. Kobayashi, H. Endo, M. Soma, 486 Y. Oikawa, and T. Ishimizu, 2011: The Japanese 55-year Reanalysis "JRA-55": an 487 interim report, SOLA, 7, 149-152. 488

Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. 489 Quarterly Journal of the Royal Meteorological Society, 106, 447-462. 490

Girishkumar, M. S., V. P. Thanga Prakash, and M. Ravichandran, 2014: Influence of 491 Pacific Decadal Oscillation on the relationship between ENSO and tropical 492 cyclone activity in the Bay of Bengal during October–December. Climate 493 Dynamics, 1-11. 494

Gray, W. M., 1968: Global view of the origin of tropical disturbances and storms. 495 Monthly Weather Review, 96, 669-700. 496

Ha, Y., Z. Zhong, X. Yang, and Y. Sun, 2014: Contribution of East Indian Ocean 497 SSTA to Western North Pacific tropical cyclone activity under El Niño/La Niña 498 conditions. International Journal of Climatology,35,506-519. 499

Henderson-Sellers, A., H. Zhang, G. Berz, K. Emanuel, W. Gray, C. Landsea, G. 500 Holland, J. Lighthill, S. L. Shieh, P. Webster, and K. McGuffie, 1998: Tropical 501 cyclones and global climate change: A post-IPCC assessment. Bulletin of the 502 American Meteorological Society, 79, 19-38. 503

Huo, L., P. Guo, S. N. Hameed, and D. Jin, 2015: The role of tropical Atlantic SST 504 anomalies in modulating western North Pacific tropical cyclone genesis. 505 Geophysical Research Letters, 2015GL063184. 506

Jia, L., X. Yang, G. A. Vecchi, R. G. Gudgel, T. L. Delworth, A. Rosati, W. F. Stern, A. 507 T. Wittenberg, L. Krishnamurthy, S. Zhang, R. Msadek, S. Kapnick, S. 508 Underwood, F. Zeng, W. G. Anderson, V. Balaji, and K. Dixon, 2014: Improved 509 Seasonal Prediction of Temperature and Precipitation over Land in a High-510 Resolution GFDL Climate Model. Journal of Climate, 28, 2044-2062. 511

Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. 512 Amer. Meteor. Soc., 77, 437–471. 513

Kennedy, J. J., N. A. Rayner, R. O. Smith, D. E. Parker, and M. Saunby, 2011: 514 Reassessing biases and other uncertainties in sea surface temperature observations 515 measured in situ since 1850: 2. Biases and homogenization. Journal of 516 Geophysical Research: Atmospheres, 116, D14104. 517

23

Klotzback, P. J., 2007: Recent developments in statistical prediction of seasonal 518 Atlantic basin tropical cyclone activity. Tellus Series a-Dynamic Meteorology and 519 Oceanography, 59, 511-518. 520

Knapp, K. R., M. C. Kruk, D. H. Levinson, H. J. Diamond, and C. J. Neumann, 2010: 521 The International Best Track Archive for Climate Stewardship (IBTrACS). 522 Bulletin of the American Meteorological Society, 91, 363-376. 523

Larson, S. and B. Kirtman, 2013: The Pacific Meridional Mode as a trigger for ENSO 524 in a high-resolution coupled model. Geophysical Research Letters, 40, 3189-3194. 525

Larson, S. M. and B. P. Kirtman, 2014: The Pacific Meridional Mode as an ENSO 526 Precursor and Predictor in the North American Multimodel Ensemble. Journal of 527 Climate, 27, 7018-7032. 528

Lee, H. S., T. Yamashita, and T. Mishima, 2012: Multi-decadal variations of ENSO, 529 the Pacific Decadal Oscillation and tropical cyclones in the western North Pacific. 530 Progress in Oceanography, 105, 67-80. 531

Li, C. and H. Ma, 2011: Coupled modes of rainfall over China and the pacific sea 532 surface temperature in boreal summertime. Advances in Atmospheric Sciences, 28, 533 1201-1214. 534

Li, C., L. Wu, and P. Chang, 2010: A Far-Reaching Footprint of the Tropical Pacific 535 Meridional Mode on the Summer Rainfall over the Yellow River Loop Valley. 536 Journal of Climate, 24, 2585-2598. 537

Li, R. C. Y. and W. Zhou, 2014: Interdecadal Change in South China Sea Tropical 538 Cyclone Frequency in Association with Zonal Sea Surface Temperature Gradient. 539 Journal of Climate, 27, 5468-5480. 540

Li, X., S. Yang, H. Wang, X. Jia, and A. Kumar, 2013: A dynamical-statistical forecast 541 model for the annual frequency of western Pacific tropical cyclones based on the 542 NCEP Climate Forecast System version 2. Journal of Geophysical Research: 543 Atmospheres, 118, 12,061-12,074. 544

Lin, C.-Y., J.-Y. Yu, and H.-H. Hsu, 2014: CMIP5 model simulations of the Pacific 545 meridional mode and its connection to the two types of ENSO. International 546 Journal of Climatology, in press. 547

Lin, I. I., I.-F. Pun, and C.-C. Lien, 2014: 'Category-6' Supertyphoon Haiyan in Global 548 Warming Hiatus: Contribution from Subsurface Ocean Warming. Geophysical 549 Research Letters, 2014GL061281. 550

Liu, K. S. and J. C. L. Chan, 2003: Climatological characteristics and seasonal 551 forecasting of tropical cyclones making landfall along the South China coast. 552 Monthly Weather Review, 131, 1650-1662. 553

Liu, K. S. and J. C. L. Chan, 2012: Inactive Period of Western North Pacific Tropical 554 Cyclone Activity in 1998–2011. Journal of Climate, 26, 2614-2630. 555

Matsuno, T., 1966: Quasi-geostrophic motions in the equatorial area. J. Meteor. Soc. 556 Japan, 44, 25-43. 557

Mitchell, C. L., 1932: West Indian Hurricanes and Other Tropical Cyclones of the 558 North Atlantic of the North Atlantic Ocean. Monthly Weather Review, 60, 253-253. 559

24

Patricola, C. M., R. Saravanan, and P. Chang, 2014: The Impact of the El Niño–560 Southern Oscillation and Atlantic Meridional Mode on Seasonal Atlantic Tropical 561 Cyclone Activity. Journal of Climate, 27, 5311-5328. 562

Pielke Jr, R., J. Gratz, C. Landsea, D. Collins, M. Saunders, and R. Musulin, 2008: 563 Normalized hurricane damage in the United States: 1900–2005. Natural Hazards 564 Review, 9, 29-42. 565

Rappaport, E. N., 2000: Loss of life in the United States associatedwith recent 566 Atlantic tropical cyclones. Bull. Amer. Meteor.Soc., 81, 2065–2073. 567

Rienecker, M. M., M. J. Suarez, R. Gelaro, R. Todling, J. Bacmeister, E. Liu, M. G. 568 Bosilovich, S. D. Schubert, L. Takacs, G.-K. Kim, S. Bloom, J. Chen, D. Collins, 569 A. Conaty, A. da Silva, W. Gu, J. Joiner, R. D. Koster, R. Lucchesi, A. Molod, T. 570 Owens, S. Pawson, P. Pegion, C. R. Redder, R. Reichle, F. R. Robertson, A. G. 571 Ruddick, M. Sienkiewicz, and J. Woollen, 2011: MERRA: NASA’s Modern-Era 572 Retrospective Analysis for Research and Applications. Journal of Climate, 24, 573 3624-3648. 574

Rogers, J. C., 1981: The north Pacific oscillation. International Journal of Climatology, 575 1, 39-57. 576

Simpson, J., J. B. Halverson, B. S. Ferrier, W. A. Petersen, R. H. Simpson, R. 577 Blakeslee, and S. L. Durden, 1998: On the role of “hot towers” in tropical cyclone 578 formation. Meteorology and Atmospheric Physics, 67, 15-35. 579

Sippel, J. A. and F. Zhang, 2008: A Probabilistic Analysis of the Dynamics and 580 Predictability of Tropical Cyclogenesis. Journal of the Atmospheric Sciences, 65, 581 3440-3459. 582

Smirnov, D. and D. J. Vimont, 2010: Variability of the Atlantic Meridional Mode 583 during the Atlantic Hurricane Season. Journal of Climate, 24, 1409-1424. 584

Kobayashi, S., Y. Ota, Y. Harada, A. Ebita, M. Moriya, H. Onoda, K. Onogi, H. 585 Kamahori, C. Kobayashi, H. Endo, K. Miyaoka, and K. Takahashi , 2015: The 586 JRA-55 Reanalysis: General Specifications and Basic Characteristics. J. Meteor. 587 Soc. Japan, 93, 5-48, doi:10.2151/jmsj.2015-001. 588

Vecchi, G. A., T. Delworth, R. Gudgel, S. Kapnick, A. Rosati, A. T. Wittenberg, F. 589 Zeng, W. Anderson, V. Balaji, K. Dixon, L. Jia, H. S. Kim, L. Krishnamurthy, R. 590 Msadek, W. F. Stern, S. D. Underwood, G. Villarini, X. Yang, and S. Zhang, 2014: 591 On the Seasonal Forecasting of Regional Tropical Cyclone Activity. Journal of 592 Climate, 27, 7994-8016. 593

Vimont, D. J. and J. P. Kossin, 2007: The Atlantic Meridional Mode and hurricane 594 activity. Geophysical Research Letters, 34, L07709. 595

Vimont, D. J., D. S. Battisti, and A. C. Hirst, 2001: Footprinting: A seasonal 596 connection between the tropics and mid-latitudes. Geophysical Research Letters, 597 28, 3923-3926. 598

Vimont, D. J., J. M. Wallace, and D. S. Battisti, 2003: The Seasonal Footprinting 599 Mechanism in the Pacific: Implications for ENSO*. Journal of Climate, 16, 2668-600 2675. 601

25

Vitart, F. and T. N. Stockdale, 2001: Seasonal forecasting of tropical storms using 602 coupled GCM integrations. Monthly Weather Review, 129, 2521-2537. 603

Wang, B. and J. C. L. Chan, 2002: How strong ENSO events affect tropical storm 604 activity over the Western North Pacific. Journal of Climate, 15, 1643-1658. 605

Wang, B., R. Wu, and X. Fu, 2000: Pacific-East Asian Teleconnection: How Does 606 ENSO Affect East Asian Climate? Journal of Climate, 13, 1517-1536. 607

Wang, C., C. Li, M. Mu, and W. Duan, 2013: Seasonal modulations of different 608 impacts of two types of ENSO events on tropical cyclone activity in the western 609 North Pacific. Climate Dynamics, 40, 2887-2902. 610

Wu, G. X. and N. C. Lau, 1992: A Gcm Simulation of the Relationship between 611 Tropical-Storm Formation and Enso. Monthly Weather Review, 120, 958-977. 612

Xie, S.-P., K. Hu, J. Hafner, H. Tokinaga, Y. Du, G. Huang, and T. Sampe, 2009: 613 Indian Ocean Capacitor Effect on Indo–Western Pacific Climate during the 614 Summer following El Niño. Journal of Climate, 22, 730-747. 615

Yang, X., G. A. Vecchi, R. G. Gudgel, T. L. Delworth, S. Zhang, A. Rosati, L. Jia, W. F. 616 Stern, A. T. Wittenberg, S. Kapnick, R. Msadek, S. Underwood, F. Zeng, W. 617 Anderson, V. Balaji, 2015: Seasonal predictability of extratropical storm tracks in 618 GFDL’s high-resolution climate prediction model, Journal of Climate, in press. 619

Zhan, R., Y. Wang, and X. Lei, 2010: Contributions of ENSO and East Indian Ocean 620 SSTA to the Interannual Variability of Northwest Pacific Tropical Cyclone 621 Frequency. Journal of Climate, 24, 509-521. 622

Zhan, R., Y. Wang, and M. Wen, 2013: The SST Gradient between the Southwestern 623 Pacific and the Western Pacific Warm Pool: A New Factor Controlling the 624 Northwestern Pacific Tropical Cyclone Genesis Frequency. Journal of Climate, 26, 625 2408-2415. 626

Zhan, R., Y. Wang, and L. Tao, 2014: Intensified Impact of East Indian Ocean SST 627 Anomaly on Tropical Cyclone Genesis Frequency over the Western North Pacific. 628 Journal of Climate, 27, 8724-8739. 629

Zhang, L., P. Chang, and L. Ji, 2009: Linking the Pacific Meridional Mode to ENSO: 630 Coupled Model Analysis. Journal of Climate, 22, 3488-3505. 631

Zhang, Q., Q. Liu, and L. Wu, 2009: Tropical Cyclone Damages in China 1983–2006. 632 Bulletin of the American Meteorological Society, 90, 489-495. 633

Zhang, W., Y. Leung, and J. Min, 2013: North Pacific Gyre Oscillation and the 634 occurrence of western North Pacific tropical cyclones. Geophysical Research 635 Letters, 2013GL057691. 636

Zhang, W., H. F. Graf, Y. Leung, and M. Herzog, 2012: Different El Niño Types and 637 Tropical Cyclone Landfall in East Asia. Journal of Climate, 25, 6510-6523. 638

639

26

List of Tables 640

Table 1. TC frequencies in JASO and JJASON for PPMM and control experiment 641

respectively. Boldface numbers represent the differences that are statistically 642

significant and * represents 0.01 level of significance. 643

644

645

646

647

648

649

650

651

652

653

654

655

656

657

658

659

27

660

List of Figures 661

Figure 1. The regressions of (a) SST (unit: °C) and 10 m wind fields (unit: ms-1) and 662

(b) tropical cyclone track density onto the standardized PMM index. The spatial 663

domain in (a) is consistent with that in Chiang and Vimont (2004). 664

Figure 2. (a) The standardized time series of the yearly PMM index and WNP TC 665

frequency for the period 1961-2013 and (b) scatterplot of PMM vs WNP TC 666

frequency (dot) and fitted line (black line). The yearly PMM index and WNP TC 667

frequency are calculated by averaging over July-October . 668

Figure 3. The five sub-regions of TC genesis in the WNP. Blue dots represent TC 669

genesis locations in the period of 1961 - 2013. The text "obs" denotes the correlation 670

between PMM and TC frequency in observations whereas "ctrl" denotes that in 671

control run. The numbers following "obs" or "ctrl" denote correlation coefficients in 672

each subregion. "WNP(obs)" represents the correlation coefficient between PMM and 673

TC frequency in the WNP in observations while "WNP(ctrl)" denotes that in control 674

run (1000 years). *,**, and *** represent the significance level 0.1, 0.05 and 0.01, 675

respectively. 676

Figure 4. The regressions of 600 hPa relative humidity, 850 hPa relative vorticity 677

(×10-6), zonal vertical wind shear (ms-1), u850 (ms-1) and u200 (ms-1) onto the 678

PMM index based on JRA-55 reanalysis. The thick red lines are used to divide the 679

five subregions in the WNP. 680

28

681

Figure 5. Regression of (a) SST anomalies (°C) and 10 m wind fields (ms-1) and (b) 682

TC density onto the standardized PMM index in a 1000-yr control simulation of the 683

FLOR-FA. 684

Figure 6. (a) The scatterplot of PMM vs WNP TC frequency (blue dots) and fitted 685

line (black) in the control experiment, and (b) the histogram of correlation between 686

WNP TC frequency and PMM for every 50 years (the correlation between WNP TC 687

frequency and PMM based on observations is shown in the thin red bar). 688

Figure 7. The regressions of 600 hPa relative humidity, 850 hPa vorticity (×10-6), 689

zonal vertical wind shear (ms-1), u850 (ms-1) and u200 (ms-1) onto the PMM index 690

in the control experiment of FLOR-FA. The thick red lines are used to divide the five 691

subregions in the WNP. 692

Figure 8. The mask for the PMM region and the SST anomalies (°C) in positive PMM 693

phase added to the seasonal cycle in the experiment. 694

Figure 9. TC track density in the PPMM experiment (top panel), control experiment 695

(CTRL; middle panel) and the difference PPMM - CTRL (bottom panel). 696

Figure 10. As in Fig.9, but for TC genesis. 697

Figure 11. The differences in 600 hPa relative humidity, 850 hPa relative vorticity 698

(×10-6), ZVWS (200 - 850 hPa, unit: ms-1), u850 (ms-1) and u200 (ms-1) between 699

PPMM and control experiments. The thick red lines are used to divide the five 700

subregions in the WNP. 701

29

Figure 12. (a) The differences in 850 hPa wind fields (vector, unit: ms-1) and OLR 702

(contour, unit: W·m-2) between PPMM and CTRL experiments with FLOR-FA, and 703

the SST pattern of the positive PMM pattern (shading, unit: °C) and (b) the vertical 704

profile of zonal wind (ms-1) from 1000 hPa to 100 hPa level (the zonal wind is 705

averaged over 0° and 20°N) between PPMM and CTRL experiments. The vertical 706

velocity is multiplied by -100 in (b). 707

708

709 710

30

Table 1. TC frequencies in JASO and JJASON for PPMM and control experiment 711 (CTRL) respectively. Boldface numbers represent the differences that are statistically 712 significant, with *** representing results significant at the 0.01 level based on t-test. 713 Frequency JJASON JASO PPMM 26.7 19.4 CTRL 24.7 17.5 PPMM - CTRL 2.0*** 1.9***

714

31

Figures 715

Figure 1. The regressions of (a) SST (unit: °C) and 10 m wind fields (unit: ms-1) and 716

(b) tropical cyclone track density onto the standardized PMM index. The spatial 717

domain in (a) is consistent with that in Chiang and Vimont (2004). 718

719

32

720

721 Figure 2. (a) The standardized time series of the yearly PMM index and WNP TC 722 frequency for the period 1961-2013 and (b) scatterplot of PMM vs WNP TC 723

33

frequency (dot) and fitted line (black line). The yearly PMM index and WNP TC 724 frequency are calculated by averaging over July-October . 725 726

34

727

Figure 3. The five sub-regions of TC genesis in the WNP. Blue dots represent TC 728

genesis locations in the period of 1961 - 2013. The text "obs" denotes the correlation 729

between PMM and TC frequency in observations whereas "ctrl" denotes that in 730

control run. The numbers following "obs" or "ctrl" denote correlation coefficients in 731

each subregion. "WNP(obs)" represents the correlation coefficient between PMM and 732

TC frequency in the WNP in observations while "WNP(ctrl)" denotes that in control 733

run (1000 years). *,**, and *** represent the significance level 0.1, 0.05 and 0.01, 734

respectively. 735

35

736

737 Figure 4. The regressions of 600 hPa relative humidity, 850 hPa relative vorticity 738 (×10-6), zonal vertical wind shear (ms-1), u850 (ms-1) and u200 (ms-1) onto the PMM 739 index based on JRA-55 reanalysis. The thick red lines are used to divide the five 740 subregions in the WNP. 741

36

742

743 Figure 5. Regression of (a) SST anomalies (°C) and 10 m wind fields (ms-1) and (b) 744 TC density onto the standardized PMM index in a 1000-yr control simulation of the 745 FLOR-FA. 746 747 748 749 750

37

751 Figure 6. (a) The scatterplot of PMM vs WNP TC frequency (blue dots) and fitted line 752 (black), and (b) the histogram of correlation between WNP TC frequency and PMM 753 for every 50 years (the correlation between WNP TC frequency and PMM based on 754 observations is shown in the thin red bar) in the control experiment of FLOR-FA. 755 756

38

757 Figure 7. The regressions of 600 hPa relative humidity, 850 hPa vorticity (×10-6), 758 zonal vertical wind shear (ms-1), u850 (ms-1) and u200 (ms-1) onto the PMM index in 759 the control experiment of FLOR-FA. The thick red lines are used to divide the five 760 subregions in the WNP. 761 762

39

763

Figure 8. The mask for the PMM region and the SST anomalies (°C) in positive PMM 764 phase added to the seasonal cycle in the experiment. 765 766 767 768

40

769 Figure 9. TC track density in the PPMM experiment (top panel), control experiment 770 (CTRL; middle panel) and the difference PPMM - CTRL (bottom panel). 771 772 773

41

774 Figure 10. As in Fig.9, but for TC genesis. 775

776

42

777 Figure 11. The differences in 600 hPa relative humidity, 850 hPa relative vorticity 778 (×10-6), ZVWS (200 - 850 hPa, unit: ms-1), u850 (ms-1) and u200 (ms-1) between 779 PPMM and control experiments. The thick red lines are used to divide the five 780 subregions in the WNP. 781 782

43

783 Figure 12. (a) The differences in 850 hPa wind fields (vector, unit: ms-1) and OLR 784 (contour, unit: W·m-2) between PPMM and CTRL experiments with FLOR-FA, and 785 the SST pattern of the positive PMM pattern (shading, unit: °C) and (b) the vertical 786 profile of zonal wind (ms-1) from 1000 hPa to 100 hPa level (the zonal wind is 787 averaged over 0° and 20°N) between PPMM and CTRL experiments. The vertical 788 velocity is multiplied by -100 in (b). 789 790 791

44