JP 3.18 A CLIMATOLOGICAL FEATURES OF MAKING LANDFALL OVER THE KOREAN PENINSULA

Baek-Jo Kim1 and Ki-Seon Choi2∗

1Policy Research Laboratory, National Institute of Meteorological Research, Meteorological Administration, Seoul 156-720, Republic of Korea 2Earthquake Observation Division, Korea Meteorological Administration, Seoul 151-747, Republic of Korea

1. INTRODUCTION affected the KP, Lee et al. (1992) classified the KP-affecting TCs into six-track type The tropical cyclones (TCs), which form during the period 1960-1990 and then over the western North Pacific (WNP), analyzed the large-scale circulation account for 30 percent of the global total. associated with each track type. Also, Park et Among these TCs, on average, three per al. (2006) investigated the KP-affecting TCs year affect the Korean Peninsula (KP), and in terms of the relationship between rainfall one lands over the KP with tremendous connected to TCs and TCs' tracks or kinetic damage (Typhoon White Book 1996). energy. Ho et al. (2004) reported that, According to National Emergency through analysis of summertime TC activity in Management Agency (NEMA 2005), TCs the WNP during the period 1951-2001, the explain approximately 65 percents of the passage frequency of TCs from the amounts of damage caused by a natural Philippine Sea to northeast China including disaster during the last 10 years. Also, it is and the western region of noted that damage by TCs may be much was reduced after the 1980s. It is also known more increased in case of considering a loss that TC motions are significantly controlled by of life by TCs. Therefore, in order to reduce regional atmospheric circulation rather than the damage brought by a TCs, the long-term the variability of the SST in the subtropics or variability of TC activity in each ocean basin tropics (Kim et al. 2005). has already been studied by many However, it is not easy to find the previous researchers. Chan and Shi (1996) showed study on KP-landfall TCs, but the study of that TCs activities in the western and central Kim et al. (2006) is just found. They defined North Pacific basins have increased since the TC landfall as an event where the TC center 1980s. In a study on the activity of TCs enters a circle with a radius of 5° from central affecting the KP, Park et al. (2006) showed Korea (128°E and 36°N). After that, they that the number of TC has rapidly increased analyzed the large-scale atmospheric since the 1990s. Actually, top 10 TCs with circulation associated with heavy rainfall the highest economic damage in South influenced only by the landfalling TCs. Korea have struck in recent years and made There have been many previous studies on landfall over the KP (KMA 2002). In addition, the climatological characteristics of TC in the a heavy rainfall influenced only by the landfall WNP(Chan 1985, Dong 1988, Wu and Lau TCs has increased since late 1970s (Kim et 1992, Lander 1994, Wang and Chan 2002, al. 2006). From this point of view, a study of and Yumoto and Matsuura 2001). They focus the long-term variability of the activity of KP- on the variability of TC genesis in the WNP, landfall TCs will be meaningful and valuable. which is mainly related to the El Niño- In studies of the climatological and Southern Oscillation (ENSO). statistical characteristics of TCs that have In the case of TC studies associated with the KP, the KP-affecting TCs are mainly ∗Corresponding author address: Ki-Seon made one of the objectives in this study. But, Choi, Earthquake Observation Division, as stated above, economic loses caused by Korea Meteorological Administration, Seoul TCs in South Korea are much more great by 156-720, Republic of Korea. E-mail those making landfall over the KP. address : [email protected] Additionally, TC activity related to the KP need to be studied separately from the 10 8 Total climate patterns of subtropical or tropical ) TD 8 regions in order to reduce the damage by TY TS, TY 6 . TCs making landfall over the KP. The present TS study aims to show the interannual and 6 , TD

4 ( interdecadal variabilities of the KP-landfall TC. y 4

2 uenc q 2. DATA (Total) Frequency 2

Fre In order to select the KP-landfall TC, the 0 0 present study employs the dataset of TC E1950 L1950 E1960 L1960 E1970 L1970 E1980 L1980 E1990 L1990 E2000 best- track for the WNP that has been Fig. 1. 5-year variation of the landfall archived by the Regional Specialized frequency of KP-landfall TCs. The bar and Meteorological Centers (RSMC)-Tokyo solid and dashed lines denote a 5-year total Typhoon Center during the period 1951-2004. frequency, a TD and extratropically This dataset includes 6-hourly latitude- transiotioned cyclone frequency, and a TS longitude positions and intensity such as and TY frequency, respectively. A capital E central pressure (hPa) and maximum and L denotes "early" and "late," respectively. sustained wind (MSW; kt) of TCs. Here, the MSW along the TC intensity is divided into 5 Until the 1970s, most TCs made landfall in grades such as tropical depression (TD; August, but, since the 1980s, the months of MSW < 34kt), tropical storm (TS; 34kt ≤ more frequent landfall has changed to June MSW ≤ 47kt), severe tropical storm (STS; and July. Generally, we can see that the 48kt ≤ MSW ≤ 63kt), typhoon (TY; MSW ≥ landfall frequency of KP-landfall TCs is the 64kt), and (ET). -1 highest in July and August, but the ratio of We also use the 6-hourly wind (m s ) and that in June to the total landfall frequency is geopotential heights (gpm) reanalyzed by the also not low. National Center for Environmental Prediction- Month the National Center for Atmospheric 56789105678910 Research (NCEP-NCAR) to characterize the 8 8 6 1950s 1990s 6 related atmospheric circulation patterns 4 4 (Kalnay et al. 1996). 2 2 0 0 8 8 6 1960s 2000s 6 3. INTERANNUAL VARIATION IN THE KP- 4 4 LANDFALL TCs 2 2

0 0 Frequency 8 30 TOTAL 6 1970s 20 Fig. 1 shows the 5-year variation of the Frequency 4 landfall frequency of the KP-landfall TCs. The 2 10 0 0 landfall frequency of the early 1960s and 8 10 )

Percentage % 6 1980s 8 ( early 2000s is quite high and that of the 6 4 4 period of the 1970s to the 1980s is relatively 2 2 0 0 low. However, the landfall frequency again Percent increases from the late 1980s onward, which 5678910 5678910 Month is consistent with Park et al. (2006), who Fig. 2. Decadal variation of the monthly analyzed the trend of TCs that affected the frequency of KP-landfall TCs. The units of the KP. The correlation coefficient between both light and dark gray bars are frequency and is 0.55, which is significant at the 95% percent, respectively. confidence level. The increase in the landfall frequency of KP-landfall TCs after the late The decadal variation of the landfalling 1980s is more remarkable in that of a TC with track patterns of KP-landfall TCs shows more intensity greater than the TS. Simultaneously, distinctive characteristics (Fig. 3). TCs in the the landfall frequency of the TD decreases. past mainly passed through the middle or This result reflects that the landfall frequency northern region of the KP, but, in recent of TCs with strong intensity has increased years, they have made landfall on the south more and more in recent years. coast and then passed over the east coast of Fig. 2 shows the decadal variation of the the KP. That is, this reflects that the main monthly landfall frequency of KP-landfall TCs. landfalling track of KP-landfall TCs tends to year mean of KP-landfall TCs. The region of shift south-eastward. This south-eastward the maximum recurving is in the vicinity of shift is more clear in the change of the mean Nanjing in Mainland China and the axis of regression track at each decade (thick lines relatively high frequency also is toward the in Fig. 3). southwestern sea of the KP. In the variation of the 10-year mean recurving location, we 1950s(8) 1960s(12) 1970s(6) can see the clockwise and simultaneously southward shift of the recurving location, which is the same direction as the landfalling track shift of KP-landfall TCs.

1980s(8) 1990s(9) 2000(7)

Fig. 3. Decadal variation of the landfalling track of KP-landfall TCs. In (b), thick lines denote regression mean tracks and the numbers in parentheses represent the landfall frequency of KP-landfall TCs during each decade.

This is because, in the past, many KP- Fig. 5. Decadal variations of the recurving landfall TCs tended to land at the KP after locations (dark circles) and the recurving recurving on Mainland China. However, in frequency (contour) of KP-landfall TCs recent years, the tracks of the TCs have changed to come not via Mainland China but 4. INTERDECADAL VARIATION IN THE KP- from the East China Sea (Fig.4). Therefore, LANDFALL TCs the increase in recent landfall frequency of KP-landfall TCs with strong intensity is due to In order to examine the characteristics of the recent increase of TCs passing over the interdecadal variation on the frequency of TC sea that can sufficiently supply the energy to landfalling the KP, the change-point analysis the TC. is applied. The indices associated with the KP-landfall TC used in the analysis are 1950s(29%) 1960s(29%) 1970s(5%) following :

ㆍ Frequency of TC (TDET) with intensity of tropical depression to extratropical transition of TC ㆍ Frequency of TC (TSTY) with intensity of 1980s(18%) 1990s(14%) 2000s(14%) tropical storm to typhoon ㆍ Accumulated cyclone energy (ACE) on TC with intensity of tropical storm to typhoon ㆍ Total moving distance (TMD) of TC for its life time.

Fig. 4. Decadal variation of the full tracks of Fig. 6 shows the interdecadal variations of KP-landfall TCs. The numbers in the TDET and TSTY, ACE and TMD. The 5-year parentheses denote the ratio of TCs that running mean, 54-year mean, period mean in passed through Mainland China to the total each index are given to find a significant number of TCs during each decade change-point. It can be found that there are

obvious change-points in mid-1960s and mid- The landfalling track change of KP-landfall 1980s. The exact years showing a significant TCs is likely to be related to that of the change-point are somewhat different among recurving location. Fig. 5 shows the recurving indices. For example, in case of TDET, frequency at each grid box and that of the 10- significant change-point years are 1967 and it might be mentioned that the intensity (ACE) 1984 at 95% confidence level and ACE are of the KP-landfall TC is largely dependent on 1964 and 1986. In particular, the change- the duration of TC with intensity greater than point of mid-1980s is consistent with the TS though its frequency is high. Therefore, in result of Choi and Kim (2007) that the this study, the frequency of the KP-landfall frequency of KP-landfall TC begins TC can be divided into three periods as increasing after late-1980s. Higher TDET and follows: TSTY and larger ACE than 54-year mean values are appeared until early 1960s and ㆍ High frequency period (H5165) : afterward late 1980s. TMD lies in phase with 1951~1965 (15 years) other indices. Undoubtedly, as the TDET and ㆍ Low frequency period (L6685) : TSTY with intensity greater than TS 1966~1985 (20 years) increases, the other indices will increase. ㆍ High frequency period (H8604) : However, if the duration of TC with intensity 1986~2004 (19 years) greater than TS is short even though its frequency is high, the ACE will become small And the significant change-points of mid (Goldenberg et al. 2001). In addition, as the 1950s and early 2000s obtained from the TC intensity is weak, the moving distance of analysis are not considered here to make a the TC is short (Wu et al. 2004). Eventually, comparative study of the intensity and track of TC and the associated atmospheric H 5165 L6685 H 8604 circulations in each period with 15~20 years. 4 2 Shown Fig. 7 are the frequency, central 3 TDET 2 0 pressure at landfall, and life time of the KP- 1 Frequency -2 landfall TC in each period. Landfalling 0 3 2 frequency is three times higher in H5165 and TSTY 2 0 H8604 than in L6685. TDET has the same

1 -2 number in both H5165 and H8604, while Frequency 0 -4 TSTY is much higher in H8604. That, is, most 4

40 ACE value t - of TCs (approximately 70 percentage) landed 2

-4 30

10 0 in H8604 has the intensity greater than TS. It

2 20

kt 10 -2 is found that in recent years the KP-landfall 0 -4 16 4 TC becomes stronger in the intensity as well 12 TM D 2 as higher in the frequency, which is in good -4 8 0

km agreement with the study of Choi and Kim 4 -2 0 -4 (2007). They also showed that the main 1950 1960 1970 1980 1990 2000 landfalling track of KP-landfall TCs tends to Year 5-Y ear R unning M ean shift southeastward from the climatological 54-Y ear M ean analysis of the track of KP-landfall TCs. This Period M ean t-v alu e finding could be confirmed in the analysis of the central pressure at landfall of TC. The Fig. 6. Decadal variation of Korean peninsula central pressure at landfall is the lowest in (KP) - landfall TC. The indices in order from H8604 with 978 hPa, meaning that the KP- an upper-most side is the frequency of total landfall TC is the strongest among each KP-landfall TC (TDET), the frequency of KP- period. The difference of central pressure at landfall TC with intensity greater than tropical landfall between H5165 (averaged central storm (TSTY), accumulated cyclone energy pressure : 988 hPa) and L6685 (averaged (ACE), and total movement distance (TMD) central pressure : 993 hPa) is only 5 hPa, for the TC life time, respectively. The periods showing the small variation of intensity of KP- of H5165, L6685, and H8604 indicate the first landfall TC during 35 years of 1951 to 1985. high frequency period 1951-1965, the low Recently, Wu et al. (2004) suggested that the frequency period 1966-1985, and the second stronger TC is, the longer TC life time is. high frequency period 1986-2004 calculated However, the life times of TDET showing from t-value (bold dashed line) of the change from tropical depression to transition of TC point analysis. The thick solid, dashed, and are very similar irrelevant of high or low dotted lines denote 5-year running mean, 54- frequency periods. On the contrary, the life year mean, and each period mean, time of TSTY with intensity greater than TS is respectively. the longest in H8604. Concretely speaking, (50%) of20 TCcasesmake landfall in the casesKP. 10 TC the southcoastof before sothat TCsmainly make landfall in in H8604 shiftsmore southeastward than Sobaeksanmaek. Themean regression track cases of 11 TC cases passed through the west coast oftheKP andthen 8 TC cases (63%) of 11TC cases make landfall in mean regression track. ForL6685, 7TC southward through thanthatinL5165 the know thatlandfalling tracks inL6685go more through Sobaeksanmaek incommon. We the KP.Inparticular, the cases do not pass pass throughand northcoast of themiddle landfall inthe west coast of theKPand then cases (80%) of20TCcases forH5165 make of KP-landfall TCsin each period. 16 TC during therecent 19years case ofTCwithintensity greater thanTS, intensity ofKP-landfall TC, particularly in the higher frequencyon andstronger Thus,muchattention should be paid L6685. than thatinH5165 andfour days than in in H8604is threedayslongerthe lifetime than TS. the best-track andduring theintensity greater barbs denote thetimes until disappearing in most andlower-most panels, dark and light (lower) ofthe KP-landfall TC.Inthe upper- pressure atlandfall(middle), andlifetime Fig. 7.Thefrequency (upper), central Fig. 8showsthelandfallingandfulltracks

Central pressure (hPa)

Life time (hour) 1000 970 980 990 100 200 300 400 Frequency 10 15 20 25 0 5 0 56 68 H8604 L6685 H5165 TSTY TDET TSTY TSTY TDET

adalT eoesrne n ogri landfall TCbecome stronger andlonger in result ofthat,intensityandlifetimeKP- further inland ofmainland China. With the region, those for L6685tend topenetrate into for H8604 only gopassbytheeastcoast TC cases pass through themainland China percent as that ofH6685. However, while the H8604 is 30%, whichis nearlythe same of 11TC cases).These kinds of TC cases in China rapidly reduce until 36%(4TC cases in 6685, TC cases pass throughthe mainland mainland China beforelanding attheKP,but (60%) of20 TC cases pass through the H5165,12TCcasesfull trackinFig.8.In We can confirm thechange ofKP-landfallTC this track tends to increase inrecent years intensity ofKP-landfallTC bythechangeof China Sea. Also, theysuggested that the not viaMainland China but fromthe East China but,in recentyears,change tocome the KPafterpassingthroughmainland past, many KP-landfall TCs tend toland at pass through there. cases make landfall inthe south coast do not through Sobaeksanmaek. However,10 TC west coast of theKPandthentheyallpass thick lines denote themean regression tracks. (b) L6685, and (c)H8604.panel, Intheleft of KP-landfall TCforthe periods of (a) H5165, Fig. 8.Landfalling(left)and fulltracks(right) Choi and Kim (2007) showed thatin the (c) H8604 (b) L6685 (a) H5165

recent years, respectively. There seems no REFERENCES distinctive feature except that the genesis of KP-landfall TC is located in farthest east in Chan, 1995: Tropical cyclone activity in the H8604. western North Pacific in relation to the stratospheric quasi-biennial 6. SUMMARY oscillation. Mon. Wea. Rev., 123, 2567-2571. In this study, we investigated the _____, J. C. L., 1985: Tropical cyclone climatological characteristics of tropical activity in the northwest Pacific in cyclones (TCs) that made landfall over the relation to the El Niño/Southern Korean Peninsula (KP) during the period Oscillation phenomenon. Mon. Wea. 1951-2004, which can be summarized as Rev., 113, 599-606. follows: _____, and J. Shi, 1996: Long-term trends 1) Landfall frequency has increased since the and interannual variability in tropical late 1980s. Especially, that of TCs with an cyclone activity over the western greater intensity than tropical storm (TS) North Pacific. Geophys. Res. Lett., has rapidly increased. 23, 2765-2571. 2) The pattern of landfall changed from the Choi and Kim, 2007: Climatological middle or northern region of the west coast characterisitcs of tropical cyclones of the KP to the south coast of the KP. making landfall over the Korean That is, the tracks of the landfalling TCs peninsula: J. Korean Meteor. Soc., tend to shift southeastward in recent years. 43(2), 97-109. 3) The change-point analysis showed the TC Chu, P. S., 2002: Large-Scale circulation frequency as three periods such as high features associated with decadal frequency period of 1951~1965 (H5165; 20 variations of tropical cyclone activity cases), low frequency period of over the central North Pacific. J. 1966~1985 (L6685; 11 cases), and high Climate, 15. 2678-2689. frequency period of 1986~2004 (H8604; 20 Collins, J. M. and I. M. Mason, 2000: Local cases). environmental conditions related to 4) TC intensity at landfall was much stronger seasonal tropical cyclone activity in in H8604 so that the duration of intensity the northeast Pacific basin. Geophys. greater than tropical storm (TS) was three Res. Lett., 27, 3881-3884. days longer in H8604 than in H5165. Dong, K., 1988: El Niño and tropical cyclone 5) Full track of the KP-landfall TC passed frequency in the Australian region through the mainland China in last years, and the northwest Pacific. Aust. but did not pass in recent years. This Meteor. Mag., 36, 219-225. became the reason of the increases of TC Goldenberg, S. B., C. W. Landsea, A. M. intensity in recent years. Mestas-Nunez, and W. M. Gray, 6) The Pacific-Japan (PJ) pattern in H8604 2001: The recent increase in Atlantic developed stronger than that in the other hurricane activity: Causes and periods, in particular the anomalous low implications, Science, 293, 474-479. (monsoon trough) around the Philippines Gong, D. -Y., and C. -H. Ho, 2002: Shift in and was much stronger. Also, the the summer rainfall over the Yangtze negative vertical wind shear (VWS) River valley in the late 1970s. anomaly was distinctive in the western Geophys. Res. Lett., 29, 1436, North Pacific. These synoptic conditions doi:10.1029/2001GL014523. became the appropriate environment that Ho, C. -H., J. -J. Baik, J. -H. Kim, and D. Y. TC in H8604 can develop more than those Gong, 2004: Interdecadal changes in in the other periods. summertime typhoon tracks. J. Climate, 17, 1767-1776. Acknowledgment: This research was Kalnay, E., and Coauthors, 1996: The carried out as a part of "A Study on Strategy NCEP/NCAR 40-Year reanalysis Development of Meteorological Technology project. Bull. Amer. Meteor. Soc., 77, (METRI-2008-B-10)" supported by the 437-471. National Institute of Meteorological Research. Kim, J. -H., C. -H. Ho, C. H. Sui, and S. -K, Park, 2005: Dipole structure of interannual variations in summertime tropical cyclone activity over East Wu, G., and N.-C. Lau, 1992: A GCM Asia. J. Climate, 18, 5344-5356. simulation of the relationship _____, C. -H. Ho, M. -H. Lee, J. -H. Jeong, between tropical-storm formation and and D. Chen, 2006: Large increase in ENSO. Mon. Wea. Rev., 120, 958- heavy rainfall associated with tropical 977. cyclone landfalls in Korea after the Wu, M. C., W. L. Chang, and W. M. Leung, late 1970s, Geophys. Res. Lett. 33. 2004: Impacts of El Niño-Southern L1876, doi:10.1029/2006GL027430. Oscillation Events on tropical cyclone Korean Meteorological Administration, 1996: landfalling activity in the western Typhoon White Book, 22 pp. North Pacific. J. Climate, 17, 1419- _____, 2002: Analysis of the characteristic of 1428. the 15th typhoon "RUSA", 121 pp. Yumoto, M., and T. Matsuura, 2001: Lander, M. A., 1994: An exploratory analysis Interdecadal variability of tropical of the relationship between tropical cyclone activity in the western North storm formation in the western North Pacific. J. Meteor. Soc. Japan, 79, Pacific and ENSO. Mon. Wea. Rev., 23-35. 122, 636-651. _____, M. A., 1993: Comments on ''A GCM simulation of the relationship between tropical storm formation and ENSO.'' Mon. Wea. Rev., 121, 2137- 2143. Landsea, C. W., N. Nicholls, W. M. Gray and L. A. Avilia, 1996: Quiet early 1990s continues trend of fewer intense Atlantic hurricanes. Geophys. Res. Letters, 23, 1697-1700. _____, C. W., 2000: El Niño-Southern Oscillation and the seasonal predictability of tropical cyclones. El Niño: Impacts of multiscale variability on natural ecosystems and society, H. F. Diaz and V. Markgraf, Eds., 149- 181. Lee, D. K., D. E. Jang, and T.K. Wee, 1992: typhoon approaching Korea, 1960- 1989 Part Ⅰ: Statistics and synoptic overview, J. Korean Meteor. Soc., 28, 133-147. Matsuura, T., M. Yumoto, and S. Iizuka, 2003: A mechanism of interdecadal variability of tropical cyclone activity over the western North Pacific. Climate Dyn, 21, 105-117. National Emergency Management Agency, 2005: 2005 disaster annual report, 619 pp. Park, J. K., B. S. Kim, W. S. Jung, E. B. Kim, and D. G. Lee, 2006: Change in statistical characteristics of typhoon affecting the Korean Peninsula. J. Korean Meteor. Soc., 16, 1-17. Wang, B, and J. C. L. Chan, 2002: How strong ENSO events affect tropical storm activity over the western North Pacific. J. Climate, 15, 1643-1658.