Climate change trend and causes of tropical cyclones affecting the South China Sea during the past 50 years

BO Xiang,XINNING Dong and YONGHUA Li

Chongqing Climate Center,Chongqing,China

ABSTRACT

KEYWORDS Tropical cyclones affecting the South China Sea;climatic characteristics;cyclone activity index;cause analysis

1. Introduction

The influence of tropical cyclones(TCs)on China is serious every year. Previous studies show that, on average,approximately seven tropical cyclones hit China per year, causing severe disaster hazards to coastal provinces(Chen and Ding 1979;Li,Chen,and Zhang 2004).In particular, the TCs in the South China Sea (SCS) have always been a focus of concern. This is because the SCS has a special geographic environment, i.e., it is surrounded by the mainland of China, the Indochina Peninsula, the Philippine Islands, and the Indonesian archipelago.Therefore,the TCs formed in the SCS usually make land soon after their formation.Meanwhile,most of the surrounding coastal areas are densely populated by rapid economic growth, posing an urgent need for improved accuracy in TC forecasting. Therefore, an indepth understanding of the various climatic characteristics of TCs in the SCS is necessary.

Wu, Wang, and Geng (2005) found that under the background of global warming, there is significant change in the prevailing tracks of TCs in the western North Pacific(WNP).This change has an important influence on the climatic characteristics of SCS TCs.Previous studies have pointed out that there are remarkable seasonal and intra-seasonal variation in the formation and track distribution of SCS TCs(Li,Liu,and Cheng 2007;Li et al.2010;Yang et al.2015).It is suggested by Yang et al.(2012) that the changes in large-scale environmental dynamical and thermo-dynamical factors lead to the interdecadal variation of SCS TC frequency during 1979-2008. Some studies have found that the variation of the sea surface temperature (SST) is an important factor affecting the formation and activity of SCS TCs(Chen 2011).This is because a change in SST can usually stimulate an abnormal secondary circulation in the lowlatitude atmosphere (Li and Zhou 2014). Additionally,the El Ni?o/Southern Oscillation (ENSO) events also influence the TC activity. Chan (2000) found that the formation frequency of the SCS TCs was higher in September-October of the year prior to El Ni?o,whereas it was lower in the year of El Ni?o.Goh and Chan(2010)found that during 1965-2005, the decreasing trend of the SCS TC frequency is directly related to ENSO.Overall,deeper and further understanding of the SCS TC activity are still in need especially when the TCs that are formed in the WNP and then entered and affected the SCS.

While TC activity is receiving more and more attention, the frequent TC activity in the SCS in recent years has caused serious disaster and loss in the southern provinces of China. To investigate this issue, the post-1965 TC cases affecting the SCS (including the formed and entering ones)are selected in this study to preliminarily analyse the variation in their main climatic characteristics and the possible causes,providing a reference for the studies of TCs in the SCS.

2. Data

In this study, we adopt two TC best track datasets: one from the Shanghai Tropical cyclone Institute of China Meteorological Administration (CMA-STI) and the other from the Joint Tropical cyclone Warning Center of the United States(JTWC).The two datasets both contain the name and number,the location(including longitude and latitude) at every six hours, the maximum surface wind,and the central minimum pressure of TCs over the WNP.In particular,the JTWC dataset records the maximum wind speed radius for the TCs between 2001 and 2013.Studies have shown that the two datasets are slightly different in terms of TC intensity in the WNP, but the difference in their track data is relatively little(Wu and Zhao 2012).

Since the mid-1960s, satellite observation technology has been gradually applied to the field of TC monitoring and forecasting.In addition,Dvorak(1975)proposed the TC intensity estimation technique based on the satellite cloud picture in 1975. Therefore, in this study, we select the TC data between 1965-2017, from which we screen out the cases that affected the SCS with maximum intensities above tropical storm intensity(with maximum wind speed near the centre ≥17.2 m s?1).In this study,the TCs affecting the SCS are defined as the cases formed in the SCS basin as well as those formed originally in the WNP that later entered the SCS which is similar to the definition as suggested by Goh and Chan (2010).The formation of a TC is defined as the time when it reaches the tropical storm intensity for the first time during its lifespan.

This study also adopts the monthly average data from the National Centers for Environmental Prediction/National Center for Atmospheric Research(NCEP/NCAR)global reanalysis to analyse the large-scale environmental conditions.The data are from 1965-2017 with a spatial resolution of 2.5°×2.5°and 17 layers in the vertical direction.

3. Variation of the climatic characteristics of tropical cyclones affecting the South China Sea

The results show that,since 1965,there have been a total of 535 TCs that affected the SCS, with an annual average of approximately 10. There is a remarkable seasonal difference in the monthly distribution. In each year, TCs are mainly concentrated in the summer and the fall (June-November)(Figure 1(a)).During summer,the TCs affecting the SCS gradually increased in number,reaching a peak in July, which had a total of 92 TC cases. The frequencies months August to October are greater than 80 as well.

Figure 1(b) indicates the time series of the annual frequency for the above selected cases. The data show an annual frequency distribution that is significantly different.There were 18 cases(the highest)in 1974 and only 5 cases(the lowest)in 1969,1997,and 2002,respectively.It is worth noting that,the annual frequency of TCs affecting the SCS exhibits a significant downward trend at the 95% confidence level.The fitted trend result shows that the annual average frequency decreased from 11.0 in 1965 to 9.0 in 2017, i.e., decreased by approximately 18.1%, indicating that the TC activity affecting the SCS and its influence may have gradually lessened in the past decades. This phenomenon has an important significance for the climate prediction and seasonal forecast of the SCS TCs.

The location of the formation is an important factor affecting TC activity, because it will directly affect the subsequent TC track and intensity. Therefore, we examined the variation of annual average formation longitude and latitude and the results are shown in Figure 2. The results indicate that the average formation latitude of TCs affecting the SCS is 14.8°N and it exhibits a significant uprising trend after 1965 at the 95% confidence level,from an average of 14.3°N in 1965 to 15.3°N in 2017,amounting to a variation amplitude of 1°. The average formation longitude of the selected TC cases is 123.9°E.Meanwhile,the annual average formation longitude gradually moved westward from an average of 124.8°E in 1965 to 123.1°E in 2017,amounting to a variation amplitude of 1.7°.The above results indicate that the formation location of TCs affecting the SCS during 1965-2017 had the tendency of shifting northward and westward.However, the average formation longitude and latitude were located outside the SCS (i.e., the ocean east of the Philippines).Therefore,the influence of tropical cyclones formed in the WNP and then entering the SCS should be paid more attention.

The TCs in the SCS are usually characterised by complex and changeable tracks.Results show that the activity region of TCs affecting the SCS exhibits a zonal distribution(105°-135°E,10°-25°N)covering from the SCS to the ocean east of the Philippines (not shown). The period of 1965-2017 is equally divided into two periods (1965-90 and 1991-2017), and the difference between the two periods are computed. It is found that, since the 1990s,the TC activity over most areas to the east and south of Hainan Island decreased remarkably (Figure 2(c)). This phenomenon may be directly related to the significant decrease in the count of TCs affecting the SCS,as shown in Figure 1.Moreover,there was an increase in TC activity in the regions of the Gulf of Tonkin and the Taiwan Strait.

Figure 1.(a)Monthly distribution and(b)annual frequency of the TCs affecting the SCS from 1965 to 2017.The black solid line is the annual distribution,and the black dashed line is the fitted trend line.

4. Cause analysis of climate change

TC movement is mainly controlled by large-scale environmental steering flows. Therefore, the variation of largescale flows in the WNP and the SCS will directly affect the climatic trend of TC activity in these regions. Both Wu,Wang,and Geng(2005)and Wu and Wang(2008)found that the large-scale environmental flows in the WNP basin varied significantly,causing a change in the prevailing TC track.In this study,the large-scale flows are calculated on each vertical level from 850 hPa to 300 hPa,respectively,during June-November of each year.Then an average is attained from the vertical layers. Further, we obtain the difference field of large-scale circulation (mean field of 1991-2017 minus that of 1965-90). The results show that there is a notable cyclonic circulation anomaly in the mid-to-low latitude region west of the WNP. This anomalous circulation centre is located in the north of southern China(Figure 3(a)).The westerlies in the southern periphery of this cyclone directly affects the northern region of the SCS (18°-20°N). It means that the westerly component of large-scale steering flows in this sea area gradually becomes stronger, or the easterly component gradually becomes weaker,which is consistent with previous studies(Wu,Wang,and Geng 2005;Wu and Wang 2008).This change is not conducive to a TC formed in the WNP moving westward into the SCS,nor to the westward movement of TCs formed in the SCS. Therefore, the frequency of TCs affecting the SCS decreased (Figure 1(b))and the TC activity is significantly reduced (Figure 2(c)).Subjected to the influence of the anomalous westerlies,it is easier for TCs to deflect toward the coast of southeastern and eastern China. Consequently, the TC activity increased near the Taiwan Strait as shown in Figure 2(c).The Gulf of Tonkin is located on the southwest side of this anomalous cyclone with remarkable abnormal northerlies.Thus,it is difficult for the TCs over this region to move northwestward or northward,and they may tend to slows down in the Gulf of Tonkin, which leading to increasing the TC activity(Figure 2(c)).

In addition to the large-scale flows that is not conducive to the TC entering the SCS,the changes in environmental conditions can also directly affect the TC activity. It is demonstrated that the changes in the vertical wind shear(the difference between the horizontal winds at 850 hPa and at 200 hPa)in the SCS was relatively weak during past decades,and therefore may not significantly influence on the SCS TC formation(not shown).However,the mid-and low-level relative humidity in most regions of the SCS notably decreased after the 1990s (Figure 3(b)). Gray(1979) noted that a TC can more readily be formed in a high humidity environment because it is conducive to stimulating more cumulus convection and higher precipitation efficiency. Therefore, the weakening humidity condition in the SCS is not beneficial to TC formation,and this is likely one of the important reasons for the frequency decrease and activity reduction of TCs affecting the SCS.

Figure 2.(a)Average latitude and(b)average longitude of annual formation of TCs affecting the SCS from 1965 to 2017 (units:degree). The black solid line is the annual distribution, and the black dashed line is the fitted trend line. (c) Differences in track occurrence for TCs affecting the SCS in 1965-2017.The difference field is the mean field of 1991-2017 minus that of 1965-90.

Figure 3. (a) The difference field of mean large-scale environmental field at 850-300 hPa(units:m s?1)and(b)the difference field of the average relative humidity at 950-600 hPa(units:%)for the tropical cyclones affecting the South China Sea during 1965-2017.The difference field is the mean field of 1991-2017 minus the mean field of 1965-90.

5. Activity variation of tropical cyclones affecting SCS

The intensity and scale are also important factors to be considered in the study of TC activity since larger and stronger TCs tend to have more severe influences.Recently, Haig, Nott, and Reichat (2014) defined a new cyclone activity index(CAI)includes TC intensity and scale as parameters and combines with the TC track to quantitatively calculate the comprehensive influence of the TC on a certain fixed point.This index is expressed as follows:

where=（Kt+Kt-1）n，Kt=max（t）/d（t）,Nis the total number of TCs in each year, andnis the total lifecycle length of a TC.Vmaxis the intensity(near surface maximum wind speed),Rmax(t) is the maximum wind speed radius andd(t)is the distance from the TC centre to a certain fixed site.Li,Wu,and Liu(2016)used the CAI to calculate the influence of TCs in the last 20 years on the main coastal cities of China. They found that it can quantify the influence of a TC on a given location,evaluate the climate change in TC influences.

Because the CMA-STI data lack records of the maximum wind speed radius of the TCs,Li,Wu,and Liu(2016)selected the maximum wind speed radius data from the JTWC data during the period 2001-13.They obtained the fitted quadratic formula for the relation between the maximum wind speed radius and TC intensity as follows:

In this study, we screen out the 535 TC cases analysed above from the JTWC data.Then the TC scale and the CAI are calculated based on the same dataset. Figure 4(a)shows the variation of the average CAI in some provincial capital cities in southern China after 1965.The results show that the CAI exhibits significant descending trends during 1965-2017 which is consistent with the previous studies(e.g.Li,Wu,and Liu 2016).Therefore,the reliability of the CAI analysis is verified and the result also is in line with the earlier phenomenon in this paper(Figures 1 and 2).

Figure 4. (a) Annual average CAI (solid line) and its trend line(dashed line) (units: m3 s?3) for main provincial capital cities in southern China during 1965-2017. Panel (a) is for Guangzhou(blue line) and Haikou (black line). (b) Average climatic spatial distribution (blue contour) of the CAI (units: m3 s?3) and its difference field(shaded area),and(c)time series of the maximum value region(red boxed region)of the CAI during 1965-2017.The difference field is the mean field of 1991-2017 minus the mean field of 1965-90.

We have further checked the CAI variation in a 1°×1°grid field associated with the selected TCs in the SCS during 1965-2017, as shown in Figure 4(b). The largevalue area of the regional spatial distribution of the CAI in the SCS is mainly located in the northern part from the east of Hainan Island to the west of the Bashi Channel.The coverage range is consistent with the influence region of the TC track indicating that the spatial distribution of the CAI is mainly affected by the TC track.The spatial average difference field of the CAI is calculated as the mean field of 1991-2017 minus that of 1965-90.It is found that the CAI in the entire South China Sea exhibits a decreasing trend.In Figure 2(c),there is an increase of TC activity in the Gulf of Tonkin and the Taiwan Strait,but the difference of the CAI is still negative, meaning that even though the frequency of the TC track increases in these regions, the influence of TC activity still declines. To validate the CAI variation of TC activity affecting the SCS, we select the region with maximum CAI distribution (15°-22°N,111°-120°E) and calculate the annual variation of the mean CAI in this region. The results also demonstrate a significant downward trend at the 0.05 level, and the descending amplitude of the fitted trend during 1965-2017 is 10.7%.The results of above analysis indicate that,after combining the factors of TC intensity and scale,the CAI values of the TCs affecting the SCS exhibit the climatic characteristics of significant decline during past decades.It indirectly validates the trend of overall activity reduction and influence decline for the TCs affecting the SCS, thereby providing a very valuable reference for future TC forecasts of the SCS.

6. Discussion and conclusions

The results of previous studies indicate that there is a notable difference in the climatic characteristics between the TCs affecting the SCS and the WNP.The variation of the climatic characteristics and the relevant causes of the SCS TCs need further investigation. Therefore, in the present study, we screen out the TC cases in 1965-2017 that affected the SCS and statistically analyse the variation of climatic features of frequency and activity for these cases.We further examine the possible causes for these variations,with the following conclusions:

(1) In 1965-2017, there were a total of 535 TCs formed in the SCS or moving from the WNP into the SCS. These TC activities are mainly concentrated in June to November every year. In addition, the annual TC frequency exhibited a significant decreasing trend,indicating that the TC activity affecting the SCS and its influences has likely weakened during past decades.

(2) The annual average formation location of TCs affecting the SCS demonstrates a significant trend shifting northward and slightly westward. Further analysis shows that the TC track occurrences in most regions of the SCS declines remarkably;nevertheless, TC activity increases in the Gulf of Tonkin and the Taiwan Strait.These characteristics have rarely been reported in previous studies.

(3) The variation in large-scale circulation forms anomalous westerlies in the northern part of the SCS, which is not conducive for TCs to move westward and to enter the SCS. Therefore, TCs deflected to the coastal areas of southeast and eastern China. It leads to a decrease in the TCs frequency and activity affecting the SCS.Meanwhile, the anomalous northerlies in the Gulf of Tonkin inhibits the TC from continuing to move northwestward or northward, therefore increasing the TC activity in this area.In addition,the decrease of mid-and low-level relative humidity is not beneficial to the formation of TCs and it may be one of the causes for the significant decrease of TC frequency and activity decline in the SCS.

(4) The results of the CAI analysis show that the influence of the selected TC cases on southern China declines significantly.The activity influence of these TCs in most parts of the SCS also weakens remarkably. These results validate the climatic characteristics of frequency decline and activity influence weakening for the TCs affecting the SCS.

It should be noted that the cases in this study include two types of TCs,i.e.those locally formed in the SCS,and those formed in the WNP and later moving into the SCS.The climatic characteristics of the two TC types and their different contributions to the phenomenon found in the present study will be further compared in future research.

Acknowledgments

The authors are grateful to two anonymous reviewers for their insightful suggestion that improved this paper. The authors are also grateful to Zhenzhen LI and others for selflessly providing the calculation program as well as writing guidance in helping to improve the paper. The authors are appreciated to all the researchers whose work is involved in this study. The help and the inspiration of their research enabled the completion of this study.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was jointly supported by General Project of Technological Innovation and Application Demonstration of Chongqing Municipality [cstc2018jscx-msybX0165]; Special Project for Development of Key Technology for Meteorological Forecast Service of China Meteorological Administration [YBGJXM (2018) 04-08]; National Natural Science Foundation of China [41875111]; and Innovation Team Project of Intelligent Meteorological Technology of Chongqing Meteorological Bureau[ZHCXTD-201804].