1 INTRODUCTION

Tropical cyclone (TC), as one of the most destructive natural disasters, has been studied extensively. To understand the impacts of general circulation on TCs is very important, and is very useful for forecasting TC genesis. Gray (1968, 1979) suggested that sea surface temperature (SST), vertical wind shear, low level vorticity, relative humidity at low and mid levels, and stratified stability are key environmental factors for TC genesis. Emanuel and Nolan (2004) developed a genesis potential index (GPI), motivated by the work of Gray (1979). Four factors were used in this index, including absolute vorticity, vertical wind shear, relative humidity and potential intensity. This index had a strong climatological relationship with TC genesis. Camargo et al. (2007) diagnosed impacts of the El Niño and La Niña on TCs using this index. The composite results showed qualitative replication of interannual variation of TC genesis location at global scale. It was also used to study seasonal and intraseasonal variation of TC cyclogenesis (Huang et al., 2011; Jiang et al., 2012).

Based on the location of SST anomaly center, it is well accepted that El Niño can be divided into two types, an eastern Pacific El Niño (EP-El Niño) and central Pacific El Niño (CP-El Niño) (Ashok et al., 2007; Kao and Yu, 2009). The GPI has been widely used in previous studies to study impacts of these two types of El Niño on TC activity. For example, Kim et al. (2011) investigated the influence of the two types of El Niño and La Niña on TC activity in the North Pacific, and the GPI was used to address the contribution of environmental factors to cyclogenesis. The results showed that with westward translation of sea surface heating during a CP-El Niño, TC activity also shifted westward, and circulation condition anomalies favored TC landfall in coastal areas of East Asia. Wang et al. (2013) studied the relationship between TCs in the western north Pacific (WNP) and two types of El Niño and La Niña in the early season (April-June), peak season (July-September) and late season (October-December), and the GPI was used as a proxy to study the variation of TC genesis. The results showed that atmospheric and oceanic conditions during the early and late season favored more TC activities in the south, whereas in peak season, favorable conditions appeared in the north. Yang et al. (2015c) investigated the influences of decay phases of the two types of El Niño on TC activities in the WNP during summer with the GPI. They demonstrated that compared with EP-El Niño, during CP-El Niño, a suppressed TC genesis center in the central WNP is further north, associated with stronger vertical shear and weaker relative vorticity. Xu and Huang (2015) studied the effects of two types of El Niño on TCs in the entire Pacific via the GPI. They indicated that during January-February-March of the decaying years of warming events, TC frequency and intensity both had positive anomalies over the South Pacific. The anomalies in EP-El Niño years had larger amplitude and wider spatial distribution than those in CP-El Niño years. Lowlevel vorticity, 700 hPa relative humidity, and vertical wind shear played key roles.

As a useful index, the GPI has been used to study the types of El Niño in several basins (Kim et al., 2011; Wang et al., 2013; Xu and Huang, 2015; Yang et al., 2015c), but it is still not clear whether this index is capable of reproducing the variation of global TC genesis during the developing and decaying phases of these two types of El Niño events. Bruyère et al. (2012) suggested that GPI did not have skill in describing interannual variation and trends of TC genesis in the entire Atlantic Basin. In the study of Kim et al. (2011), the GPI anomalies during CP-El Niño cannot reproduce the negative genesis anomalies over the eastern Pacific. Therefore, it is useful to investigate the relationship between GPI and TC genesis location during the two types of El Niño, from their developing to decaying phases. Considering the key role of descending motion in suppressed TC genesis over the eastern Pacific during CP-El Niño events (Kim et al., 2011), we chose a modified GPI (MGPI) version with the addition of omega at 500 hPa (Murakami et al., 2011) to do the following analysis.

This paper is organized as follows. Section 2 describes the data used and identifies the two types of El Niño. Section 3 demonstrates the skill of the modified GPI for variation of global TC genesis during the developing and decaying phases of the two types. Section 4 presents the contributions of environmental factors to cyclogenesis variability at global scale. Section 5 provides a summary and discussion.

2 DATA AND METHOD

Genesis information for the WNP, North Indian Ocean and Southern Hemisphere was derived from best-track data from the Joint Typhoon Warning Center (JTWC). Information for the North Atlantic and eastern Pacific were acquired from the National Oceanographic and Atmospheric Administration’s National Hurricane Center. From a global standpoint, the best-track data are of high quality after 1975, so the study covered the period 1975-2013. The region where TCs were first detected was defined as the genesis location. The best-track data were re-gridded into 2.5°×2.5° grid data and smoothed by a nine-point weighted smoother (Wang et al., 2010) to study spatial patterns of TC genesis. Environmental factor data such as winds, relative humidity and omega were used to calculate the modified GPI, which were extracted from National Centers for Environmental Prediction (NCEP) -National Center for Atmospheric Research reanalysis data (www.cdc.noaa.gov/cdc/reanalysis) with horizontal resolution 2.5°×2.5°. SST data for our study period were from version 3b of the Extended Reconstructed SST analyses, with horizontal resolution 2°×2°.

The modified version of the Emanuel and Nolan (2004) GPI, adding the influence of omega at 500 hPa (Murakami et al., 2011), was used to examine how various environmental factors contributed to the effect of the two El Niño types on TC genesis. The MGPI is defined as

Here, η is absolute vorticity at 850 hPa (/s), H is relative humidity (%) at 700 hPa, V_pot is potential intensity (m/s), V_shear is the magnitude of vertical wind shear between 850 and 200 hPa (m/s), and ω is the vertical wind velocity (Pa/s) at 500 hPa.

The two types of El Niño were identified by the El Niño indices proposed by Ren and Jin (2011). Definitions of the indices are from the Niño3 (SST anomalies averaged over 5°N-5°S, 150°-90°W) and Niño4 (those anomalies averaged over 5°N-5°S, 160°E-150°W) indices, as follows.

N_EP and N_CP are indices for EP-El Niño and CP-El Niño, respectively. Using these indices, five EP-El Niños (1976/1977, 1982/1983, 1986/1987, 1991/1992, and 1997/1998) and five CP-El Niños (1990/1991, 1994/1995, 2002/2003, 2004/2005, and 2009/2010) were defined after 1975. The developing (the first year of the defined El Niños events) and decaying (the second year of the defined El Niños events) phases of El Niños were designated by 0 and 1, respectively. The following addresses MGPI variation in developing summer (July-September, JAS0), developing fall (October-December, OND0), phaselocked season (January-March, JFM1), decaying summer (April-June, AMJ1) and decaying summer (July-September, JAS1). Here, the developing summer and fall are these two season in developing year of the EP-(1976, 1982, 1986, 1991, and 1997) and CP-(1990, 1994, 2002, 2004, and 2009) El Niños, the phase-locked season, decaying spring and summer are in the decaying year of EP-(1977, 1983, 1987, 1992, and 1998) and CP-(1991, 1995, 2003, 2005, and 2010) El Niños. In addition, the TC genesis anomaly was the difference between TC genesis of each season and the climatology mean of 39-yrs (1975-2013) for that particular season.

3 VARIATION OF TC GENESIS AND MGPI AS INFLUENCED BY TWO TYPES OF EL NIÑOS

It is well known that the TC formation is significantly influenced by the large-scale environmental conditions. Variation in thermodynamic and dynamic factors can influence the frequency and location of genesis of TC. The heating pattern and life cycle of CP-El Niño and EP-El Niño are quite different, which cause differential variation of the environmental factors from its developing phase to decaying phase, as a result, different impacts on TC genesis. The five variables used to define MGPI have significant contribution on formation of TC (Gray, 1968, 1979) and also are sensitive to the SST anomaly and its spatial pattern. The anomalies of MGPI associated with the two types of El Niño are likely to capture the characteristic of TC anomalies caused by El Niño. Therefore, at first variation of TC genesis due to the two types of El Niño is analyzed and then the ability of the MGPI for representation of TC genesis anomalies associated with El Niño in the active basins is investigated.

TC activity varies with basins and for most of basins, the active season is in summer; e.g., the peak season of TC activity in the North Pacific and Atlantic is JAS, and that in the Southern Hemisphere is JFM (Austral summer). But because of stronger vertical wind shear caused by Indian monsoon in summer, active seasons of TC in the north Indian Ocean are observed in the pre-monsoon season (AMJ) and late monsoon season (OND) (Wang et al., 2010). Based on characteristics of TC variation of different basins, we focus on TC genesis in those basins during different evolution periods of the two types of El Niño. We investigate TC genesis variation in the North Pacific and Atlantic during JAS0 and JAS1. In JFM1, we pay more attention to TC variation in the southern Hemisphere. In AMJ and OND, TC genesis in the north Indian Ocean will be emphasized. TC genesis variation in the WNP during OND and AMJ is also addressed, because of higher frequency of TC genesis in this basin during these two seasons.

Characteristics of TC geneses in WNP, north Atlantic, north Indian Ocean, Southern Hemisphere during developing and decaying phases of the two types of El Niño are shown in Fig. 1. Consistent with previous results, the anomalies of TC genesis during EP-El Niño JAS0 are positive in the southeastern portion of the WNP and negative in the northwestern portion (Wang and Chan, 2002; Ho et al., 2004; Camargo and Sobel, 2005; Chen et al., 2006; Camargo et al., 2007). In contrast to EP-El Niño, in the CP-El Niño, the positive TC genesis anomaly shifted westward significantly in the North Pacific (Kim et al., 2011). In the north Atlantic, TC activity was significantly suppressed during EP-El Niño (Goldenberg and Shapiro, 1996; Shaman et al., 2009), but the influence of CP-El Niño on TC in the north Atlantic is still unclear. Kim et al. (2009) showed greater-than-average TC frequency and increased TC landfalls on the Gulf of Mexico coast and Central America. However, Lee et al. (2010) suggested that greater TC frequency in 1969 and 2006 were associated with the variation of local circulations, without a remote influence from tropical Pacific. Larson et al. (2012) pointed out that the CP-El Niño had insubstantial impacts on Atlantic TC activity. In our study, the anomaly of cyclogenesis was negative, except in the southern north Atlantic in JAS0 (Fig. 1f).

As we know, OND was the peak season of the TC activity in north Indian Ocean (Fig. 1b and g). In this basin, most TCs formed in the Bay of Bengal and eastern Arabian Sea, there were about 4 TCs per year in the Bay of Bengal and the frequencies of TC in this region were 3-4 times of that in Arabian Sea (Sadhuram et al., 2006). During both types of El Niño, TC genesis decreased in the Bay of Bengal (Singh et al., 2000; Ng and Chan, 2012) and increased in the southern Arabian Sea, while they were more prominent during CP-El Niños. This dipole pattern of TC anomalies between Bay of Bengal and Arabian Sea in this season was related to the feature of TC genesis in this basin that November storms formed in either the Arabian Sea or the Bay of Bengal but not in both during the same year (Evan and Camargo, 2011). Because OND was the second major season of TC formation in the WNP, it is also focused on in this season of developing years of El Niños (Fig. 1b and g). In the WNP, TC genesis anomalies are mainly negative in the western and central WNP; positive anomalies only appeared in its southeastern part during EP-El Niños (Wang et al., 2013). However, in the CP-El Niño, anomalies of TC genesis had an eastwest dipole pattern, and the positive anomaly area was larger than that during EP-El Niños (Wang et al., 2013).

During boreal winter, TC genesis was active only in the southern Hemisphere. For the south Indian Ocean, there are few papers distinguishing the impacts of the two types of El Niño on TC activity. The negative anomalies almost covered the entire south Indian Ocean during EP-El Niño except in an area around 70°E where positive anomalies can be found (Fig. 1c). In contrast with EP-El Niño, during CP-El Niño, the TC genesis anomaly revealed a north-south dipole pattern, with a positive center near the equator (Fig. 1h). In the south Pacific, there was enhanced TC formation in its northeastern part during EP-El Niño (Basher and Zheng, 1995; Chand and Walsh, 2009; Vincent et al., 2011), while during CP-El Niño, positive anomalies only appeared around 180°E (Fig. 1h).

During AMJ1, TCs were active in the north Indian Ocean and WNP. In the north Indian Ocean, the distribution of the TC anomalies were similar to those in OND0, except that the positive anomaly in the Arabian Sea was further north and there was a slight positive anomaly in the Bay of Bengal (Fig. 1i). Gray (1968) noted that in the Arabian Sea, TC formed northward in the spring and southward in the fall associated with withdraw of the monsoon trough. That might be the main reason of the different centers of TC anomalies between the two seasons. In the WNP, TC formation was substantially suppressed over the entire basin during EP-El Niño (Fig. 1d), whereas it was reduced in the western part and enhanced in the southeastern part during CP-El Niño (Fig. 1i).

During JAS1, there were TC active centers in the North Pacific and Atlantic. Yang et al. (2015c) investigated the various impacts of the two types of El Niño on TC activity in the WNP. There are still no published works for the eastern Pacific or north Atlantic. In the WNP during EP-El Niño, a positive anomaly mainly appeared in the southwest (Yang et al., 2015c). However, during CP-El Niño, the TC genesis anomaly in the WNP had a southwest-northeast dipole pattern with a positive center in the southwest (Fig. 1g). In the central and eastern Pacific, there was a positive genesis anomaly during EP-El Niños, while during CP-El Niños they were almost all negative. In the Atlantic, the difference between two types was very significant; EP-El Niños suppressed TC formation over the entire basin, except for the Gulf of Mexico. CP-El Niños enhanced TC genesis in the entire basin.

Since there was such contrasting variation for the two types of El Niño from developing to decaying phases, it is necessary to determine the cause of such contrast. Therefore, we analyzed whether MGPI variation could capture this contrast of TC genesis caused by different phases of the two types of El Niño (Fig. 2). First, during JAS0 of EP-El Niño, the MGPI anomalies were positive in the southeast and negative in the northwest, consistent with the genesis variation (Fig. 2a). A westward shift of the positive TC genesis anomaly in the WNP was clearly found by the variation of MGPI during CP-El Niño (Fig. 2f). In the eastern Pacific, MGPI anomalies were markedly positive during both EP-and CP-El Niños, so the index could capture most enhanced TC cyclogenesis. Suppressed TC genesis in the northeastern part of eastern Pacific (eastern part) during EP-El Niño (CPEl Niño) could not be captured by the MGPI. An obvious decrease of TC genesis number in the north Atlantic during both types of El Niños was well captured, but a slight positive anomaly over the southern north Atlantic was insufficient to match strongly enhanced TC formation in this region during CP-El Niños (Fig. 2a, f).

During OND0, a strong difference between EPand CP-El Niños in the WNP was well demonstrated. In the north Indian Ocean, clearly suppressed TC formation in the Bay of Bengal during both types of El Niño and increased TC number in the Arabian Sea during CP-El Niños were reflected, but a positive MGPI anomaly in the southern Bay of Bengal overestimated TC variation there (Fig. 2b, g).

During JFM1, MGPI anomalies in the south Indian Ocean were negative in most areas, and only around 70°E was there a positive anomaly during EP-El Niños, consistent with TC genesis variation. During CP-El Niños, the MGPI anomalies also had a northsouth dipole pattern, consistent with anomalies of TC genesis. In the South Pacific, the distribution of MGPI anomalies could explain genesis variation during EP El Niño. However, during CP-El Niño, the positive anomaly area was larger (Fig. 2c, h).

During AMJ1, the MGPI was able to capture most variation of TC genesis anomalies in the WNP and Bay of Bengal during both El Niño types, but positive anomalies in the Arabian Sea were not reproduced (Fig. 2d, i).

During JAS1, MGPI anomalies were also negative in the WNP, except for the southwest part. They were positive in the central Pacific during EP-El Niños and had a northeast-southwest dipole pattern in the WNP during CP-El Niños. In the eastern Pacific, negative MGPI during CP-El Niños agrees well with the anomaly of cyclogenesis, but increased TC genesis over the entire basin during EP-El Niños was not exhibited well. In the north Atlantic, the entire basin had negative MGPI anomalies during EP-El Niños and positive ones during CP-El Niños, consistent with genesis variation (Fig. 2e, j). Since the MGPI anomalies could capture the principal characteristics of TC genesis anomalies as influenced by the two types of El Niño from developing to decaying phases, we now address environmental factor contributions to MGPI variation.

The MGPI could represent main characteristic of TC genesis anomalies related to the two types El Niños in main basins except for the Arabian Sea. Comparing them with previous studies of normal GPI with the two types El Niños, the MGPI can significantly improve the reproduction of some aspects. For example, in the results of Kim et al. (2011), the positive GPI anomaly in the WNP cannot reach westward enough comparing with genesis anomaly due to CP-El Niños. However, using MGPI, it could improve the reproducing of westward shift of the positive anomalies in the central and western north Pacific in the CP-El Niños in JAS0. In the south Pacific, the dipole pattern of MGPI anomalies agreed better with TC genesis anomalies than that of normal GPI during EP-El Niño events. Although the MGPI overestimated the positive anomaly of TC genesis related to the CP-El Niño, it was still better than the normal GPI’s representation (Xu and Huang, 2015). Because of complex oceanic and atmospheric environment in north Indian Ocean, the cyclogenesis are influenced by several air-sea interact systems such as El Niño, Indian Ocean dipole and Madden-Julian oscillation (Sumesh and Ramesh Kumar, 2013). The reproducing of TC genesis’s variation in this basin is more difficult. Due to modified omega to GPI, the MGPI reproduced TC genesis anomaly in Bay of Bengal was much better than GPI, but in the Arabian Sea, even MGPI still showed bad representation. Therefore, a new GPI in Arabian Sea is needed.

4 FACTORS INFLUENCING THE EFFECTS OF EL NIÑO TYPES ON MGPI

In this section, the relative importance of the five variables used in the MGPI in determining anomalies of the two types of El Niño is evaluated. Because of strong MGPI nonlinearity, its variation cannot be represented by simply summing the five factors. Therefore, the MGPI was recalculated by setting climatological values for four factors and varying the other one interannually. Through comparing the recalculated MGPI with the MGPI with variation of the five factors, we could determine the contribution of the varying factor to the recalculated MGPI. Repeating this process for each MGPI factor and compositing them from developing to decaying phases of the two types of El Niño, we could ascertain the main environmental factors involved in El Niño effects on the MGPI (Camargo et al., 2007; Wang et al., 2013; Yang et al., 2015c).

Figure 3 shows the recalculated MGPI anomalies of EP-(Fig. 3a, b, c, d and e) and CP-El Niños (Fig. 3f, g, h, i and j) for Northern Hemisphere during JAS0. This figure shows cases of varying vertical shear (Fig. 3a and f), vorticity (Fig. 3b and g), omega (Fig. 3c and h), relative humidity (Fig. 3d and i) and potential intensity (Fig. 3e and j), in each case with the other four fixed at their long-term means. For the WNP, negative MGPI anomalies caused by vertical shear and potential intensity during both EP-and CP-El Niños. Positive anomalies during the two types of El Niños are mainly associated with variation of vertical shear, relative vorticity at 850 hPa, mid-level omega, and 700 hPa relative humidity. Compared with EP-El Niños, the westward shift of positive anomalies for CP-El Niños is mainly related to vertical shear, midlevel omega, and 700 hPa relative humidity. For the eastern Pacific, a substantially enhanced TC genesis in this basin was caused by vertical shear, whereas the lack of a negative anomaly may be attributed to underestimated vorticity. For the north Atlantic, clear negative anomalies during EP-El Niños are associated with vertical shear and potential intensity. During CPEl Niños, suppressed TC genesis is related to vertical shear and relative humidity, and enhancement is associated with vertical shear.

Figure 4 is similar to Fig. 3, but for OND0. During OND0, vertical shear is also very important to the dipole pattern of MGPI anomalies in the WNP during the two types of El Niños. The westward extension of a positive anomaly center in the WNP during CP-El Niños is mainly caused by vertical shear and relative vorticity. In the north Indian Ocean, relative vorticity, mid-level omega, and relative humidity are more important in TC genesis variation. As discussed above, there was contrasting variation in the Bay of Bengal and Arabian Sea. For the substantially suppressed TC genesis in the Bay of Bengal for both types of El Niño, relative vorticity, mid-level omega and relative humidity are all important. The larger magnitude of a negative anomaly in the Bay of Bengal during CP-El Niño is associated with relative vorticity and 500 hPa omega. A positive anomaly in the southern Bay of Bengal was caused by vertical shear. In the Arabian Sea, there is a weak positive MGPI anomaly during EP-El Niños, which is related to vertical shear and relative humidity.

JFM1 is the peak season of TC activity in the southern Hemisphere. Anomalies of MGPI there during the two types of El Niños are shown in Fig. 5 (similar to Figs. 3 and 4, but for JFM1). In the south Indian Ocean, a reduced MGPI is the main characteristic of this basin during EP-El Niños. Such variation is clearly caused by the factors including relative vorticity, omega, relative humidity, and potential intensity. A positive anomaly around 70°E is related to omega and relative humidity. In contrast with EP-El Niños, during CP-El Niños there was a dipole pattern of the MGPI anomaly. The northern positive anomaly is strongly influenced by vertical shear, and the southern negative one is associated with relative vorticity and relative humidity. In the South Pacific, the positive anomalies of the northwestsoutheast anomaly dipole have a close relationship with relative vorticity and humidity, whereas the negative anomalies are related to vertical shear and potential intensity. Although the contribution of 500 hPa omega does not appear strong, it shows a dipole pattern similar to that of the MGPI with five varying factors. During CP-El Niños, negative MGPI anomalies were still mainly influenced by vertical shear, whereas the overestimated positive anomaly over the South Pacific is caused by relative vorticity. The positive anomaly of MGPI-varying relative vorticity is from 170°E to 170°W, whereas the genesis anomaly in the basin was only found east of 180°. For the positive anomaly center east of 180°, mid-level omega and relative humidity were key. The negative center is associated with variation of vertical shear.

Figure 6 (similar to Figs. 3, 4 and 5, but for AMJ1) shows recalculated MGPI anomalies of EP-and CPEl Niños during their decaying springs. In the WNP, reduced MGPIs during EP-El Niños are associated with all the five variables. During CP-El Niños, relative humidity contributed to the dipole pattern of MGPI anomalies. In the Bay of Bengal, substantial variation of the MGPI is still negative, and relative vorticity, mid-level omega and relative humidity may be responsible for MGPI variation during both EPand CP-El Niños.

TC variation in the Northern hemisphere during summer following El Niño events is also important. As Yang et al. (2015c) showed, there is clear difference between the two types of El Niño during their decaying summers. For the mainly negative MGPI anomalies in the WNP during EP-El Niños, the five factors are all important, especially relative humidity. For the weak positive anomaly in the southwestern WNP, vertical shear is the dominant factor. However, during CP-El Niños, the distribution of TC genesis anomalies shows a northeast-southwest dipole. The negative center in the central WNP is further north, and positive anomalies cover a larger area (Yang et al., 2015c). In this dipole pattern, vertical shear may be the key. For the northward shift of the negative center in the central WNP, relative vorticity, 500-hPa omega, and relative humidity all had important contributions (Fig. 7; same to Figs. 3-6, but for JAS1)). In the decaying summer, the positive MGPI anomaly in the central Pacific during EP-El Niños is mainly caused by the variation of relative humidity. In the eastern Pacific, there is even inverse variation between the two types of El Niño, which has not been addressed by earlier studies. Unfortunately, the MGPI could not describe TC genesis variation in this basin during EPEl Niños, which had a substantial negative anomaly. However, during CP-El Niños, the MGPI could capture suppressed TC genesis in the eastern Pacific, and vertical shear controlled the variation of the index. Over the North Atlantic, there was nearly opposite variation between EP-and CP-El Niños during JAS1. A reduced MGPI during EP-El Niños is mainly related to variation of the vertical shear, relative humidity, and potential intensity. The opposite variation during CP-El Niños was produced by inverse variation of these three factors.

5 SUMMARY AND DISCUSSION

In this paper, a modified GPI version was examined from a global perspective during the two types of El Niño. The results showed that the index could capture the variation of TC genesis in most basins during the developing to decaying phases of EP-and CP-El Niños. The crucial environmental factors varied substantially by basin.

Over the WNP, we focused on impacts of the two types of El Niño on TC genesis using the MGPI for JAS0, OND0, AMJ1, and JAS1. The results showed that the five environmental factors were all important for TC genesis variation during both types of El Niños, especially in their peak seasons (JAS0 and JAS1). Substantially different variation between EPand CP-El Niños during their developing and decaying summers were caused mainly by contrasting variations of vertical shear, 500-hPa omega, and relative humidity. In the second major season (SON) of TCs, the difference of cyclogenesis influenced by the two El Niño types were caused by vertical shear, relative vorticity, and humidity. For the decaying spring (AMJ1) of El Niño events, the critical environmental factor was relative humidity.

In the eastern Pacific, TC genesis was enhanced in the central and southeast parts and suppressed in the northeast during developing summers (JAS0) of EPEl Niños. TCs were active (inactive) west (east) of 110°W during CP-Niño developing summers (Kim et al., 2011). EP-El Niños enhanced cyclogenesis over nearly the entire basin, whereas CP-El Niños suppressed TC genesis nearly basin-wide during their decaying summers (JAS1). Vertical shear played the key role in TC genesis variation for the two types of El Niño during their developing and decaying summers (JAS0 and JAS1). Relative vorticity was another important factor. A weakening of the contribution of relative vorticity to MGPI caused a lack of negative MGPI anomalies in the eastern Pacific during both EP-and CP-El Niño developing summers (JAS0). Overestimation of a negative anomaly along the South American coast during decaying EP-Niño summers was associated with vertical shear and relative vorticity.

In the north Atlantic, cyclogenesis was greatly reduced over the entire basin during EP-El Niños in JAS0, and was also suppressed during CP-El Niños except for south part of the basin. There was still no published work on the substantially different variation of TC genesis between EP-and CP-El Niños in their decaying summers (JAS1). That revealed an almost inverse variation between EP-and CP-El Niños in the Atlantic Basin. The MGPI was able to capture the main pattern of TC genesis during the two types of El Niños in developing and decaying summers. The crucial environmental factors impacting the cyclogenesis during these two types El Niños were vertical shear, relative humidity, and potential intensity.

In the Bay of Bengal, TC genesis was reduced during the developing fall (OND0) of the two types of El Niño, especially for CP-El Niño. During AMJ1, cyclogenesis decreased in the north and slightly increased in the south for EP-and CP-El Niños, and was more obvious during the latter. During OND0, the MGPI generally replicated TC genesis very well, except that it failed to capture the positive anomaly in the southern Bay of Bengal during CP-El Niños. Relative vorticity, humidity, and omega were the crucial factors for strongly decreased MGPI anomalies. For the simulation error in the southern Bay of Bengal, vertical shear played the key role. During AMJ1, relative vorticity, mid-level omega, and relative humidity were the important factors for MGPI variation in that basin. In the Arabian Sea, genesis anomalies caused by El Niños were not substantial during OND0 and AMJ1. There were positive anomalies in both CP-and EP-El Niños. The MGPI generated poor simulations for the Arabian Sea. The reason might be that in the Arabian Sea, factors such as the Indian dipole mode together with El Niños more greatly aff ect TC activity than El Niño events alone (Sumesh and Ramesh Kumar, 2013). Considering the complex conditions in the Arabian Sea, there are likely more factors important to TC genesis than the five MGPI variables. Therefore, to develop a new GPI which can show good simulation in the Arabian Sea may be a very useful work and need to be done in the future.

In the Southern Hemisphere, there were clearly different variations of TC genesis between the two types of El Niño during their peak season (JFM1). Cyclogenesis was obviously reduced in most of the South Indian Ocean, except for the region around 70°E during EP-El Niños. However, CP-El Niños suppressed TC formation in the south and enhanced it in the north. In contrast with the South Indian Ocean, in the South Pacific, the cyclogenesis anomaly during EP-El Niños had a southwest-northeast dipole pattern. During CP-El Niños the negative anomalies were obviously and the positive one only over east of 180° and north of 20°S. The MGPI for the Southern Hemisphere could capture the main characteristics of TC genesis variation except overestimate the positive anomaly over south Pacific during CP-El Niños, and the five environmental factors were all important.

In addition, because there are different scenarios of TC formation in different basins, the main factors controlling the TC genesis are also different. In the WNP, the TC formation is primarily associated with monsoon trough. Besides ENSO, the quasi-biennial oscillation (QBO), intraseasonal oscillation (ISO), and Antarctic Oscillation have important impacts on TC in the WNP on interannual timescale (Chan, 1995; Yang et al., 2015a, b; Ho et al., 2005). However, the relationship between QBO and TC cannot be held during ENSO events. In the decaying phases of the two types El Niños, the Indian basin mode can influence on TC in WNP through maintaining the anomalous anti-cyclone over the WNP (Du et al., 2011; Yang et al., 2015c). In the Atlantic, TCs are related to African easterly waves (Thorncroft and Hodges, 2001). And the African easterly waves are controlled by the western Sahel rainfall which has good relationship with ENSO (Goldenberg and Shapiro, 1996; Tang and Neelin, 2004). The changes of the vertical shear play the key role in the ENSO-TC relationship (Gray and Sheaffer, 1991). Moreover, the North Atlantic Oscillation can enhance the suppressing of El Niños on TC genesis during their negative phase (Larson et al., 2005). ENSO plays a dominant role on interannual variability of TC genesis in the eastern Pacific (Wu and Lau 1992). The TC formation in the south Pacific, especially east of the dateline, is related to the variation of monsoon trough and the south Pacific convergence zone (Vincent et al., 2011). Therefore, the ENSO plays the key role in the variation of TC genesis in south hemisphere (Terry, 2007; Dowdy et al., 2012). However, the negative Indian dipole mode events can enhance the impacts of La Niña on TC formation in the eastern Australia region (Liu and Chan, 2012). In the North Indian Ocean, the TC formation is closely related to the Monsoon Depression (Goswami et al., 2003). Due to the TC developing location is near to land, it is so difficult for the evolution from the monsoon depression to TC. Meanwhile, there are complex airsea systems in the north Indian Ocean. The TC activity in the north Indian Ocean is influenced not only El Niños, but also the Indian dipole mode (Sumesh and Ramesh Kumar, 2013; Mahala et al., 2015). So the MGPI shows bad representation in the basin, especially in the Arabian Sea.

In this paper, we evaluated the ability of the MGPI to reproduce the different variation of TC genesis due to the two types of El Niño from developing to decaying phases and confirmed the specific environmental factors which were most important in the influence of El Niños on TCs. As we know, the forecast of TC activity is very important for the reducing of the economic loss. The MGPI can provide a way to forecast TC number through forecasting the large-scale environmental factor. Since there is a better presentation in TC genesis anomaly associated with the two types of El Niño, the MGPI may be used to do interannual prediction of TC formation.