Abnormal Meridional Temperature Gradient and its Relation to the Okhotsk High

Transcription

1 Journal of the Meteorological Society of Japan, Vol. 82, No. 5, pp , Abnormal Meridional Temperature Gradient and its Relation to the Okhotsk High Yoshihiro TACHIBANA Institute of Observational Research for Global Change, JAMSTEC, Yokohama, Japan Liberal Arts Education Center, Tokai University, Hiratsuka, Japan Takuya IWAMOTO Liberal Arts Education Center, Tokai University, Hiratsuka, Japan Masayo OGI Frontier Research Center for Global Change, JAMSTEC, Yokohama, Japan and Yohei WATANABE Graduate School of Environmental Earth Science, Hokkaido University, Sapporo, Japan (Manuscript received 15 January 2004, in final form 5 July 2004) Abstract The climatological meridional atmospheric temperature structure in the Okhotsk region in summer is characterized by a poleward increase of the surface air temperature. Under this anomalous temperature gradient, the stationary anticyclone, referred to as the Okhotsk high, occasionally occurs. The relationship of the interannual variation of the meridional temperature gradient anomaly is statistically investigated, with those of the Okhotsk high, the sea surface temperature, and the global atmosphere mainly using NCEP reanalysis data. The correlation between the anomaly and the Okhotsk high is quite high. That is, in years when the meridional temperature gradient is positive, the Okhotsk high appears more than normal. The anomaly is composed of two independent factors, the warmness of eastern Siberia and the coldness of the northwestern North Pacific. The vertical structures of the anticyclone, related to the Siberian warmness, are different from those related to the Pacific coldness. The former anticyclone has deep structure, whereas the latter has shallow structure. Therefore, there are two types of the Okhotsk High. The warmness of Siberia is connected to the Rossby wave propagating along the northern coast of the Eurasian continent. The coldness of the North Pacific is, on the other hand, influenced by the variation of the tropical Pacific. Consequently, the interannual variation of the Okhotsk high is influenced by completely different remote sources; one is the tropical Pacific, and the other is the highlatitude areas facing the Arctic Ocean. To understand the interannual variation of the Asian summer monsoon, which is closely related to the occurrence of the Okhotsk high, both the Arctic and the tropics should be considered. Corresponding author: Yoshihiro Tachibana, Liberal Arts Education Center, Tokai University, Hiratsuka, Japan. tachi@rh.u-tokai.ac.jp ( 2004, Meteorological Society of Japan 1. Introduction A poleward decline in atmospheric temperature, and the existence of associated westerlies

2 1400 Journal of the Meteorological Society of Japan Vol. 82, No. 5 Fig. 1. The climatological meridional 1000 hpa temperature gradient in July. Areas of positive values, which are drawn by shading, represent the areas of poleward increase of the temperature. The unit is Kelvin degree 1. Square-drawn areas around the Okhotsk Sea represent averaged areas for indexes (see text). in mid-latitudes, are basic features of the atmosphere. In some high latitude areas during the summer months, the temperature inversely becomes warmer in the poleward direction. Figure 1 shows the climatological meridional 1000 hpa temperature gradient in the northern hemisphere in July, by averaging the NCEP (National Center for Environmental Prediction) reanalysis dataset from 1958 through In most of the areas, the temperature becomes colder with increasing latitude, while in eastern Siberia and Alaska, a positive temperature gradient area is widespread. This extraordinary atmospheric meridional structure in this region is brought about mainly by the difference of the thermal inertia between the southern cold oceans, i.e., the Okhotsk Sea and the Bering Sea, and the northern warm continents. Under this large thermal contrast, a stationary anticyclone, referred to as the Okhotsk high, occasionally develops in the Okhotsk and the western Bering Sea regions. Although the variation of the Okhotsk high seems to be a regional climatological issue, it is

3 October 2004 Y. TACHIBANA, T. IWAMOTO, M. OGI and Y. WATANABE 1401 still important for global climate. For example, Okawa (1973) reported that the Okhotsk high is usually associated with the dense marine fog around the Okhotsk Sea and the northwestern North Pacific. Norris (1998) and Norris et al. (1998) suggested that variations of the marine fog, the solar insolation, and the sea level pressure field in the North Pacific in summer, are synchronized with each other. Based on these reasons, the variation of the Okhotsk high may influence the summertime oceanic surface condition in the North Pacific. From a synoptic scale meteorology point of view, it has been pointed out that the Okhotsk high influences the Asian summer monsoon, and the associated Baiu/ Meiyu stationary front stretching from south China to the central North Pacific in early summer (Suda and Asakura 1955; Ninomiya and Mizuno 1985a; Ninomiya and Mizuno 1987; Wang 1992; Kodama 1997). For example, the Okhotsk high brings about abnormally cold and rainy summers in Japan (e.g., Ninomiya and Mizuno 1985b). Thus, understanding the cause of the interannual variation of the Okhotsk high is important, not only for regional interests, but also for gaining insight into large-scale climatic issues. Nevertheless, few scientific studies directly targeting the interannual variation of the Okhotsk high have been undertaken (e.g., Wang and Yasunari 1994). Because the study executed by Wang and Yasunari (1994) was only based on six years of data, a full scientific understanding of the long-term variation has not been achieved yet. Very recently, Ogi et al. (2003) and Ogi et al. (2004) found that summertime large-scale atmospheric circulations are connected to the North Atlantic Oscillation (NAO) of the previous winter. They pointed out that for years in which the winter NAO was in a positive phase, the summertime 500 hpa geopotential height field in the mid latitudes was positive overall, especially over the Okhotsk Sea, suggesting the NAO regulates the interannual variation of the Okhotsk high. The Okhotsk high is thought to occur as a result of an atmospheric blocking phenomenon (e.g., Wang 1992; Wang and Yasunari 1994; Nakamura and Fukamachi 2004). According to the climatology of Lejenas and Okland (1983), there is a blocking maximum over the central and western part of the Pacific in summer. This blocking maximum may correspond to the occurrence of the Okhotsk high. Nakamura and Fukamachi (2004) very recently pointed out that the abnormal meridional temperature structure, as shown in Fig. 1, plays an important role of the blocking occurring in this region based on case studies. Their composite analyses based on some prominent blocking events showed that weak westerlies due to the large anomalous meridional temperature gradient over the Okhotsk Sea are not favorable for the eastward propagation of the Rossby wave, and this unfavorable condition for the eastward propagation strengthens the blocking anticyclones over the Okhotsk Sea. However, in the interannul time scale, no studies on the relationship between the strength of the abnormal temperature structure and the blocking in this region have been executed. The aim of this study is to determine whether the poleward temperature increase in this region is related to the Okhotsk high in the interannual time scale. The common features of the blocking consist of a north-south dipole geopotential height pattern associated with the weak westerlies, and the split of the westerlies into the northern and southern branches (e.g., Rex 1950; Nakamura et al. 1997). The interanual variation of the blocking occurring in the Okhotsk region in this season might be due to the interannual variation of the abnormal meridional temperature structure. The interannual variation of the components of the meridional thermal contrast, such as continental grand hydrological and oceanic processes, may weaken the westerlies in the interannual time scale and bring the interanual variation of the blocking phenomenon. Pelly and Hoskins (2003) proposed a new perspective on blocking, based on the meridional reversal of the potential temperature gradient on a potential vorticity surface. They compared their new blocking formulation with those of a conventional height field index presented by Tibaldi and Molteni (1990). They furthermore conclude that the method based on the meridional potential temperature structure is better able to detect W blockings, in which the westerly jet meanders in the shape of the Greek character W, than conventional indexes. Based on an idea similar to the blocking formulation by Pelly and Hoskins (2003), In this

4 1402 Journal of the Meteorological Society of Japan Vol. 82, No. 5 study the relationship between the strength of the meridional temperature gradient and occurrence of the Okhotsk high in the interannual time scale is examined, mainly using objective re-analysis. The remote connection to the interannual variation of the Okhotsk high is determined. In addition, the relations of the Okhotsk high with the anomalous cold summer in Japan are examined. Owing to the fact that the anomalous meridional temperature as shown in Fig. 1 is prominent in June and July, Our analyses are focused on June and July. Because the results of June was quite similar to those of July, in this study we only show the case of July. 2. Data The atmospheric data used in this study is from the National Center for Environmental Prediction/National Center for Atmospheric Research (NCEP/ NCAR) monthly mean reanalysis dataset with a 2:5 2:5 regular latitude-longitude grid from 1958 to 1998 (Kalnay et al. 1996). The monthly mean SST data, issued by the Japan Meteorological Agency (JMA), with a 2 2 resolution from 1958 to 1998 are also used. Using these monthly mean data sets, statistical analyses are applied to clarify the relationship between the meridional temperature gradient anomaly, and large-scale atmospheric and oceanic conditions. The indexes representing the strength of the Okhotsk high and the anomaly in the meridional temperature gradient are defined as follows: The Okhotsk high index (OH Index): the areaaveraged monthly mean 850 hpa geopotential height within the Okhotsk Sea. The averaging area is from 55 N 57.5 N through E E. The usage of the sea level pressure (SLP) for the OH index might be better than that of the 850 hpa height. However, as mentioned above, because the Okhotsk Sea is usually covered by dense fog in summer, fog with cold boundary-layer air mass, which may deform the SLP field over the Okhotsk Sea, may not be reproduced by the reanalysis datasets. Because of this reason, in this study we avoid adopting the SLP as the OH index. However, results by using the SLP as the OH index were quite similar to those using the 850 hpa geopotential height. The index of the thickness difference between Siberian and the Pacific regions (SP index): the area-averaged monthly mean thickness between 1000 hpa and 500 hpa over the eastern Siberia area, minus the area-averaged monthly mean thickness between 1000 hpa and 500 hpa over the northwestern part of the North Pacific. The averaging areas are, respectively, from 60 N 65 N through 140 E 150 E and 40 N 45 N through 150 E 160 E. This index represents the strength of the anomalousness of the meridional temperature gradient in the lower troposphere around the Okhotsk Sea. The Siberian thickness index (SB index) and the Pacific thickness index (PC index) are also used. The averaging area for each index is the same as that for each area of the SP index. The standardized values of these four indexes are used for linear regression analyses. In Fig. 1, the three squared-drawn areas around the Okhotsk Sea represent the averaged areas for indexes. By using the above-mentioned NCEP dataset, the climatological meridional temperature gradient shown in Fig. 1 was drawn. 3. Results 3.1 Abnormal meridional temperature gradient and the Okhotsk high Figure 2 shows the correlation coefficient of the thickness field with the Okhotsk high index in July. The thickness defined here is between the 1000 hpa and 500 hpa geopotential heights. Two definite positive and negative correlation areas are located over eastern Siberia and the northwestern part of the North Pacific. This pattern reflects the fact that in the years when the atmospheric vertical averaged temperature over eastern Siberia was higher than normal and when the atmosphere over the northwestern part of the North Pacific was colder than normal, the Okhotsk high tended to develop. To confirm the importance of the negativity of the meridional temperature gradient, the correlation and the regression coefficients of the 1000 hpa field with the SP index in July are shown in Fig. 3. The areas of the SP index referring to the positive maximum, and, the negative minimum regression areas in Fig. 2 are defined. In the center of the Okhotsk Sea, a high correlation area is evident as shown in Fig. 3. The regressed geopotential field at

5 October 2004 Y. TACHIBANA, T. IWAMOTO, M. OGI and Y. WATANABE 1403 Fig. 2. The correlation and regression coefficients of the thickness between 1000 hpa and 500 hpa in July with the OH index. The contours show the regression coefficient, and the gradations of the shaded areas indicate the correlation coefficient exceeding 90%, 95%, 99%, and 99.9% statistical confidence levels, which are determined by t-test assuming the sample data are independent. The unit of the regression is one meter. 850 hpa shows a similar pattern (not shown). Therefore, the SP index well represents the variation of the Okhotsk high. Figure 4 shows the year-to-year variations of the OH index and the SP index. The variations of the two lines are almost the same, and the correlation coefficient is This means about half of the interannual variability of the Okhotsk high in the lower troposphere can be expressed by the variation of the SP index. We therefore use this index as representative of the Okhotsk high in the following analyses. The same analyses are applied to the regions around the Bering Sea and Alaska, in which the climatological meridional temperature gradient is also positive (See Fig. 1). We, however, were not able to find any statistically significant correlations of the geopotential field over the Bering Sea, with the Fig. 3. The same as Fig. 2 except for the correlation and regression coefficients of the 1000 hpa geopotential height in July with the SP index. thickness fields over Alaska and over the North Pacific. These results show that the close connection to the abnormity of the meridional temperature to the stationary anticyclone is one of the unique features of the Okhotsk region. Figure 5 shows the correlation and regression coefficient map of the 500 hpa geopotential height field with the SP index. A striking wavelike pattern appearing along the northern coast of the Eurasian continent and spreading toward the northwestern North Pacific is evident. A similar pattern in which the amplitude is weaker, can also be seen in the other atmospheric regressed fields (see Figs. 2 and 3). Therefore, we expect that the variation of the Okhotsk high is related to a wide range of areas, from the subtropical Pacific through northern Eurasia. 3.2 Warmness of Siberia The meridional temperature abnormality is composed of the coldness of the northwestern North Pacific, and the warmness of the eastern Siberian continent. The warmness of the Siberian continent is represented by the SB index, whereas the coldness of the northwestern

6 1404 Journal of the Meteorological Society of Japan Vol. 82, No. 5 Fig. 4. Interannual variation of the Okhotsk high (OH index) and the meridional temperature abnormality (SP index). Bold and thin lines respectively, show the OH index and SP index. North Pacific is represented by the PC index. If the correlation between the SB index and the PC index is high, it is difficult to distinguish between them. However, the correlation coefficient between the SB index and PC index is Fig. 5. The same as Fig. 2 except for the correlation and regression coefficients of the 500 hpa height in July with the SP index. The unit of the regression is one meter. small, with a value of only For this negligibly small correlation coefficient, the interannual variation of these two factors defining the strength of the anomalous temperature gradient can be regarded as independent. This independence indicates that the causes of the variation of one index are different from those of another. This suggests the existence of two different types of the Okhotsk high. To distinguish the connection to the Siberian warmness from the connection to the Pacific coldness, the correlation analyses for each component of the meridional temperature gradient anomaly are executed. Figure 6-a shows the correlation and regression coefficients of the 1000 hpa geopotential height with the SB index. A high positive correlation area is still evident in the center of the Okhotsk Sea. Because the positive regression area is shaped as a concentric configuration around the Okhotsk Sea, the SB index represents not a ridge stretching from somewhere, but the Okhotsk high itself. Along with the high positive correlation area in the Okhotsk Sea, a wave-like pattern from the western Eurasian continent to the Okhotsk Sea is more obvious than that in Fig. 3, and the path of the wave-like pattern is the same as that in Fig. 3. In contrast to Fig. 3, no statistically significant correlation in the northwestern North Pacific can be seen. The correlation, and the regression coefficients of the 500 hpa geopotential height

7 October 2004 Y. TACHIBANA, T. IWAMOTO, M. OGI and Y. WATANABE 1405 with the SB index are shown in Fig. 6-b. Similarly, a clearer and more statistically significant wave train is detected along the northern coast of Eurasia, than when using the SP index (cf. Fig. 5). In agreement with the surface atmosphere, there are no correlations with the Pacific region. An additional investigation was carried out based on a wave-activity flux formulated by Takaya and Nakamura (1997, 2001), which is parallel to the local group velocity of stationary Rossby waves embedded on the climatological-mean flow. In the calculation of the wave-activity flux, Climatological-mean fields are defined as the mean fields, and regression-coefficient fields with the SB index as the deviation fields. The procedure using climatic mean fields and regressed monthly fields is commonly used in order to diagnose the energy propagation of Rossby waves (e.g., Honda et al. 1999; Honda et al. 2001). The arrows in Fig. 6-b clearly exhibit large eastward flux reaching to the north of the Okhotsk Sea from central Siberia, and this eastward propagation can be traced to the Barents region. This diagnostic analysis supports the fact that the pattern of the Siberian wave train represents the stationary Rossby wave. Figure 6-c shows the correlation, and the regression coefficients of the 1000 hpa temperature with the SB index. Wave-like positivenegative correlation areas along the northern coast of the Siberian continent are also obvious. One particular positive correlation area, centered in the western Barents Sea, is strikingly high. On the other hand, no significant connec- Fig. 6. The same as Fig. 2 except for the correlation and regression coefficients of (a: upper) the 1000 hpa geopotential height, (b: middle) the 500 hpa geopotential height, and (c: lower) the 1000 hpa temperature in July with the SB index. Arrows in (b) indicate the horizontal component of the phaseindependent wave-activity flux (m 2 s 1 ) formulated by Takaya and Nakamura (1997, 2001), which is parallel to the local group velocity of the stationary Rossby waves embedded in the climatological-mean flow.

8 1406 Journal of the Meteorological Society of Japan Vol. 82, No. 5 tion can be seen in the Pacific region. Correlation patterns of the upper tropospheric temperature, with the SB index, are similar to that of 100 hpa temperature (Figures not shown). Therefore, the anomalous high temperature from the surface to the upper troposphere, over the eastern part of Siberia, is closely related to the interannual variation of the Okhotsk high, and this anomalous temperature is probably due to the eastward propagating Rossby wave along the northern coast of the Siberian continent. 3.3 Coldness of the Pacific Ocean The other independent factor defining the meridional temperature abnormity is the coldness of the northwestern North Pacific. In this subsection, The relationship of the coldness of this area to the global atmosphere and the ocean is described. Figure 7-a shows the correlation and regression coefficients of the 1000 hpa geopotential height with the PC index. In the Okhotsk Sea, there is a high negative correlation area, and the center of the maximum correlation area is the same as that shown in Fig. 6-a. The correlation, and the regression coefficients with the PC index in the Okhotsk region, are also almost the same as those of the SB index shown in Fig. 6-a. Therefore, the degrees of importance of these indexes for the Okhotsk high are comparable to each other. Because the positive regression area is shaped as a concentric configuration around the Okhotsk Sea, the PC index represents not a ridge intruding from the North Pacific, but the Okhotsk high itself. In the Pacific region there are two significantly correlated areas; one is the area to the south of Japan, and the other is in the central part of the North Pacific (Fig. 7-a). Figure 7-b shows the correlation and regression coefficients of the 500 hpa geopotential height field with the PC index. A high positive correlation area, covering the northwestern North Pacific toward Japan, is obvious. This high correlation basically represents the area of the PC index itself. In addition, the connection with the tropics is more distinct than it is in the surface atmosphere, and the Fig. 7. The same as Fig. 6 except for the correlation and regression coefficients of (a: upper) the 1000 hpa geopotential height, (b: middle) the 500 hpa geopotential height, and (c: lower) the 1000 hpa temperature in July with the PC index.

9 October 2004 Y. TACHIBANA, T. IWAMOTO, M. OGI and Y. WATANABE 1407 tropical-midlatitude connection is more evident than it is in the correlation maps using the SP index (cf. Fig. 5). Contrary to Fig. 6, no significant west-east wave-like pattern along the northern coast of Eurasia, as shown in Fig. 6, can be seen in Fig. 7. Because no obviously systematic wave activity flux, associated with stationary Rossby waves, can be seen by the wave activity analysis as was executed in Fig. 6b, the wave activity flux is not superimposed in this figure. Figure 7-c shows the correlation and regression coefficients of the 1000 hpa atmospheric temperature with the PC index. Similar to Fig. 7-b, the statistically significant correlation over the eastern part of the Siberian continent is small, whereas there is a zonally spreading positive high correlation area along the 40 N latitude, where the Baiu/Meiyu front usually occurs during this season. This positive anomaly agrees with the abnormally cold weather in Japan, when the Okhotsk high occurs (e.g., Suda and Asakura 1955; Kato 1985; Ninomiya and Mizuno 1985a; Wang 1992; Wang and Yasunari 1994). Figure 8 shows the correlation coefficients of the SST in June, July and August, with the PC index in July. A notable feature shown in this figure is that the SST in July and August in the northwestern part of the North Pacific, whose area is almost the same as the PC index area defined in this study, is highly positively correlated with the PC index. This indicates that in the years when the Okhotsk high occurred more frequently, the SST in the PC index region was lower than normal. Moreover, it should be noted that a negative correlation area can be seen in the eastern tropical Pacific, and this tendency continues to August. On the other hand, no significant correlation of the SST in June, with the PC index in July, can be seen. The relation of the SB index with the SST field in the Pacific, was also analyzed. No statistically significant relation can be seen, however, in the Pacific SST (Figure not shown). Fig. 8. The same as Fig. 2 except for the lag correlation and regression coefficients of the SST field in June, July, and August with the PC index in July. 3.4 Vertical structure and heat flux There are two different types of Okhotsk highs, one is related to the high latitudes, and the other is related to the tropics. In this subsection, differences are determined in the vertical structures, the surface heat flux, and the horizontal advection. Figures 9-a and -b respectively show that the vertical-meridional structures of the geopotential height, regressed with the SB index, and the PC index, along the 150 E longitude. For the convenience of comparison, the signs of the regression and the correlation coefficients in the PC index panel are reversed, since a positive value of the PC index, corresponds to the weakness of the Okhotsk high. In the lower altitude at 55 N in each figure, the maximum geopotential anomaly, with respect to the latitude, can be

10 1408 Journal of the Meteorological Society of Japan Vol. 82, No. 5 Fig. 9. The same as Fig. 2 except for the vertical-meridional structures of the geopotential height regressed with the SB index (a: upper) and the PC index (b: lower) along the 150 E longitude. For the convenience of comparison, the signs of the regression and the correlation coefficients in the PC index panel are reversed, since the PC index itself corresponds to the weakness of the Okhotsk high. seen. This geopotential maximum in the low altitudes corresponds to the center of the Okhotsk high in the lower troposphere. The maximum tilts northward with an increase in height. This indicates that the vertical axis of the center of the Okhotsk high tilts northward, and that the tilting angle of the PC index case is larger than that of the SB index. The geopotential anomaly field in the upper troposphere at 55 N, at which the surface center of the Okhotsk high is located, is negative in the PC index case, while in the SB index case, the regressed geopotential in the upper troposphere at 55 N is still positive. Moreover, the largest difference of these figures is in the upper troposphere. In the geopotential height field regressed with the SB index, the maximal amplitude on the axis of the Okhotsk high is in the upper troposphere, whereas in the case regressed with the PC index, the significance in the upper troposphere on the axis of the Okhotsk high, is less than that in the lower troposphere. Therefore, the Okhotsk high that is related to the Pacific coldness has a shallow structure, whereas the Okhotsk high that is related to the Siberian warmness, has a deep anticyclone, although the surface pressure fields of the Okhotsk high in both cases appear almost the same, as shown in Figs. 6-a and 7-a. Hereafter, for the convenience of the description, the Okhotsk high related to the south and the north is referred to as the shallow Okhotsk high and the deep Okhotsk high, respectively. Another notable feature in the shallow Okhotsk high, is the existence of a dipole pattern in the middle and upper troposphere centered at about 45 N, and about 20 N. This south-north dipole pattern can also be seen in Fig. 7-b. Next, the thermal structures of the PC and the SB index regions are examined. Table 1 shows each component of the area-averaged surface heat fluxes issued by the NCEP/NCAR re-analysis on the SB index and PC index regions. It may not be highly reliable to quantitatively evaluate the surface heat flux data from reanalysis data, because there might be errors in the numerical model output and observation data. Therefore, the following heat flux analyses should be used, not for quantifying air-surface interactions, but for roughly estimating the tendency or at least the direction, of the heat fluxes related to the Okhotsk high. The values listed in Table 1 represent the areaaveraged regression coefficients, with the SB index and the PC index. The signs of the regression and correlation coefficients with the PC index are reversed as shown in Fig. 9. Since the SB index and PC index are normalized, the regression coefficients indicate the anomalies of the heat flux when the value of the index is at the standard deviation. The anomaly and the correlation coefficient of the net heat flux over the Pacific region during the occurrence of the shallow Okhotsk high, is larger (11.1 W m 2 )

11 October 2004 Y. TACHIBANA, T. IWAMOTO, M. OGI and Y. WATANABE 1409 Table 1. Correlation and regression coefficients of the area-averaged values of the components of the surface heat flux with the SB and PC indexes. The averaged areas are the SB and the PC index regions. For example, in the PC region, the SB index vs. the short wave indicates the regression coefficient of the area averaged short wave heat flux in the PC region with the SB index. Values in brackets show the correlation coefficients. The sign of the PC index is reversed. The positive values indicate upward heat flux. The unit of the regressed heat flux is Wm 2. Horizontal advection to the SB, and the PC index regions are added to this heat flux list. Heat flux Siberian (SB) region Pacific (PC) region SB index (deep high) PC index (shallow high) SB index (deep high) PC index (shallow high) short wave 1.2 (0.12) 3.2 ( 0.33) 0.0 (0.00) 3.3 ( 0.41) long wave 0.6 (0.17) 1.1 (0.31) 0.5 ( 0.11) 4.5 (0.95) latent heat 0.9 ( 0.13) 1.0 (0.15) 2.4 ( 0.26) 7.0 (0.75) sensible heat 2.0 (0.24) 0.4 (0.05) 1.5 ( 0.32) 3.0 (0.64) net 2.9 (0.35) 0.7 (0.08) 4.5 ( 0.25) 11.1 (0.63) horizontal adv than it is during the occurrence of the deep Okhotsk high. The largest value in the component of the net heat flux is in the latent heat flux. Therefore, the atmosphere gains more heat, and the ocean loses more heat in the Pacific region during the occurrence of the shallow Okhotsk high, than during the occurrence of the deep Okhotsk high. As for the surface heat flux over eastern Siberia, no statistically significant correlation with each component of the surface heat flux was found. Therefore, the land-air interaction over eastern Siberia is not important for the occurrence of the deep Okhotsk high. Vertically integrated horizontal heat advection in the PC index, and the SB index regions, is also listed in the bottom line of Table 1. The averaged areas are the same as the regions defined as the PC index and the SB index regions, and the vertical integration is from 1000 hpa to 500 hpa. The anomaly of the horizontal advection was estimated by the climatological mean temperature advection by the wind anomaly plus the anomaly temperature advection, by the climatological mean wind, i.e., V 0 T þ V T 0 ; ð1þ where and 0 are the climatological mean value and the regression coefficient, respectively. If the regression coefficients with the SB index are, for example, chosen as perturbation fields, Eq. (1) represents the monthly mean horizontal advection in the year when the SB index is at one standard deviation. In addition, we calculated V 0 T 0. However, this term was negligibly small compared to that estimated by Eq. (1). In the Siberian region, the horizontal heat advections during the occurrence of both the deep and the shallow Okhotsk high are small. On the other hand, in the Pacific index region for both cases, the advection is negative and larger than that in the Siberian region. This cold air advection in the PC index region is brought about by the northeasterly wind circulating around the Okhotsk high. These results indicate that regardless of the type of Okhotsk high, the occurrence of the Okhotsk high brings about horizontal cold air advection to the Pacific region, but relatively little horizontal heat advection to eastern Siberia. 4. Discussion The hypothesis for the relationship between the Okhotsk high and the meridional temperature gradient anomaly described in the introduction, is strongly supported by the present

12 1410 Journal of the Meteorological Society of Japan Vol. 82, No. 5 analyses. The Okhotsk high is closely related to the northern warmness and the southern coldness in the interannual time scale. In this section, we discuss what brings about the interannual variation of the anomalous temperature gradient. We also found that the northern warmness and the southern coldness are independent of each other. This independence shows that the anomaly itself is not physically but statistically important. Understanding the formation processes of coldness and warmness respectively is more important than finding those of the anomaly itself. As shown in the previous section (see Fig. 9), the vertical structure of the Okhotsk high related to the southern coldness is shallower and more baroclinic with a large northward tilt, whereas the vertical structure of the Okhotsk high, related to the northern warmness, is deeper and less baroclinic with a small northward tilt. Because the structure is different, the formation process of each is likely to be different. The existence of these two types disclosed by the present objective analyses is in agreement with a subjective classification executed by Okawa (1973), in which the Okhotsk high was classified into a deep Okhotsk high related to the blocking and a shallow one related to the air-sea interaction. First, the Siberian warmness is described. As shown in Fig. 6, the warmness is related to the wave propagation of the stationary Rossby wave along the Siberian continent. Because the surface heat flux in the SB region is small (Table 1), the formation of the upper anticyclone over the SB index region cannot be brought about by local diabatic heating due to the surface heat flux. In addition, the horizontal heat advection toward eastern Siberia is small, as shown in Table 1. In order for the SB index region to remain warm, the air mass must therefore gain the heat from somewhere. One possibility for this is the vertical heat advection due to the downward wind. The requirement of the downward vertical motion is consistent with the anticyclonic phase of the Rossby wave in the SB region. Hence, the variation of the SB index is mainly explained by the stationary Rossby wave propagation from western Siberia. According to Wang (1992) and Wang and Yasunari (1994), eastward Rossby wave propagation from the west of East Siberia, might be one of the possible causes for the formation of the Okhotsk high. The wave pattern shown above partially agrees with their suggestion. However, the path of the Rossby wave suggested by Wang (1992) is to the south of our figures. On the other hand, case studies on the Rossby wave propagation associated with the Okhotsk high executed by Nakamura and Fukamachi (2004) agree with our results. These results suggest that there are some dominant propagation paths related to the Okhotsk high. However, our results show that at least in the interannual time scale, the most dominant path of the wave connected to the Okhotsk high in July is along the northern coast of the Eurasian continent. A similar westeast wave train can be seen in Fukutomi et al. (2003), who illustrated a summertime atmospheric pattern correlated with summertime precipitation over major Siberian river basins. It is also interesting to note that Ogi et al. (2003) and Ogi et al. (2004) identified a summertime atmospheric pattern in association with the NAO in the previous winter. The wavy patterns over Siberia in Fig. 2e of Ogi et al. (2003), and in Fig. 7 of Fukutomi et al. (2003), are quite similar to our Fig. 6. Although the targets shown by these recent studies are different from those of the present study, the similarity of the wave pattern over Siberia suggests that the Siberian wave train is a dominant summertime atmospheric pattern. It should be noted that in Fig. 1 a maximum negative poleward temperature gradient area, zonally located along the northern coast of the Eurasian continent, where the Siberian wave train lies, can be seen. This corresponds to the Arctic frontal zone, which occurs under the conditions of the cold icy Arctic and the ice-free warm continent, represented by Serreze et al. (2001). Although determining whether the Siberian wave train is a summertime dominant mode or not is an interesting issue, and it is beyond our scope. Further analyses should be undertaken as a next step of this study. The convergence of the wave activity flux over the Okhotsk Sea (see arrows in Fig. 6-b) suggests that weak westerlies, due to the large anomalous meridional temperature gradient over the Okhotsk Sea, are not favorable for the eastward propagation of the Rossby wave, and this unfavorable condition for the eastward propagation strengthens the blocking

13 October 2004 Y. TACHIBANA, T. IWAMOTO, M. OGI and Y. WATANABE 1411 anticyclones over the Okhotsk Sea. The convergence of the wave activity flux agrees with some numerical, and observational, studies about the reflection and the breaking of the Rossby wave under the condition of weak winds (e.g., Naoe et al. 1997). This result is in agreement with Nakamura and Fukamachi (2004), but from our study, the stationary Siberian wave-train pattern in the interannual time scale is a physically meaningful pattern, associated with Rossby wave propagation along the northern coast of the Eurasian continent. A pair of large meridional temperature gradients, one is negative facing to the Arctic and the other positive in the Okhotsk Sea, (See Fig. 1) may be a key factor to the large variability of the summertime atmosphere there. Without eastern Siberia facing both the Arctic and the Okhotsk Sea, the Okhotsk high would not appear. Once again, looking at Fig. 6, one can see that the correlation map of the 1000 hpa temperature, with the SB index, has a large positive correlation area centered at the coastal area of the Barents Sea. Moreover, the wave activity flux shown in Fig. 6 suggests that around the coastal region of the Barents Sea, large emanation of the wave activity flux can be seen. This suggests that some variations in this region, such as the variations of the soil moisture, snow cover, sea ice, and the SST, might be sources of the excitement of the Rossby wave. Such surface influences on the summertime atmospheric circulations have also been illustrated in Ogi et al. (2003) and Ogi et al. (2004). Seeking the source regions of the Rossby wave and explaining the dynamical processes in detail should be undertaken in the future. Next, the coldness of the northwestern North Pacific is described. In the PC index region, both the air column and the SST are cold when the shallow Okhotsk high occurs, whereas for the deep Okhotsk high, neither the air temperature nor the SST in the PC index region has any correlation with the Okhotsk high. In the occurrence of the shallow Okhotsk highs, the horizontal heat advection to the PC index region is negative and larger than that of the deep Okhotsk high. This anomalous cold air advection in the PC index region can be a cause of the coldness of Japan. Another reason for the coldness around Japan during the occurrence of the shallow Okhotsk high might be the cold ocean. The direction of the net surface heat flux, regressed with the PC index is, albeit, upward (Table 1), i.e., the SST is cooled by the air. The lag correlation analyses shown in Fig. 8 also illustrate that the anomaly of the SST does not lead to the anomaly of the air temperature above. For these reasons, it is expected that the cold SST in the PC region is not the cause of the cold atmosphere, but rather a response to it. By the oceanic positive heat flux, the cold advection is partially canceled. This indicates that the coldness in the PC index region cannot be explained only by the large horizontal cold air advection. An additional cooling process is long wave radiation cooling occurring at the top of the fog, which appears in association with the Okhotsk high (e.g., Okawa 1973; Kato 1985; Kodama 1997). This fog top cooling enables the atmospheric mixed layer to remain cold. During the deep Okhotsk high, on the other hand, there is no signature of the atmospheric temperature anomaly in and around Japan. This represents the fact that the Okhotsk high with deep structure does not always influence the summer climate in Japan. We further discuss this contradictive result to previous studies, in which the Okhotsk high brings anomalous cold weather in Japan (e.g., Ninomiya and Mizuno 1985a). One reason is the weakness of the cold air advection as described above. The horizontal cold air advection related to the deep Okhotsk high is only a half value of that related to the shallow one (Table 1). In addition, the meridional wind component, which is the main cause of the horizontal temperature advection, is not statistically significantly correlated with SP index in the PC index region (Figure not shown). Let us look again at Figs. 6a and 7a. The horizontal gradient of the regressed 1000 hpa in the PC index region in Fig. 7a is larger than that in Fig. 6a. Because of the difference of the horizontal gradient of the geopotential height, the wind anomaly, which brings the cold advection, is weak and insignificant during the deep Okhotsk high. According to Kodama (1997), some Okhotsk highs bring cold advection to northern Japan, and some bring warm air advection. His back trajectory analyses suggest that this difference is brought by the difference in the wind direction, depending on the location

14 1412 Journal of the Meteorological Society of Japan Vol. 82, No. 5 of the ridge. Our result that the meridional wind component has no significant relation to the coldness of the PC region during deep anticyclone is in agreement with Kodama (1997). The finding that the Okhotsk high with deep structure does not always bring a cold summer in Japan, is also in agreement with a recent study (Ogi et al. 2003). Figure 2 of Ogi et al. (2003) showed the summertime atmospheric patterns regressed with the wintertime NAO. In Figs. 2(d) and 2(e) of their study, a significant anticyclonic signal over the Okhotsk Sea is obvious. However, no temperature anomaly over Japan can be seen in Fig. 2(f ) of Ogi et al. (2003). During the shallow Okhotsk high, the cyclonic anomaly is located in the northern North Pacific, while during the deep Okhotsk high (Fig. 6a), no significant anomalies can be seen there. By the existence of this cyclonic anomaly, the horizontal cold air advection is strengthened only during the shallow anticyclone. These results suggest that key processes for the coldness in Japan are not only the existence of the Okhotsk high itself, but also the existence of the cyclonic anomaly in the northern North Pacific. Because during the deep Okhotsk high there are no significant anomalies in the Pacific, the weak and insignificant cold advection associated with only the Okhotsk high is not sufficient to bring significant a cold anomaly around Japan. The remote influences of ENSO on the summer climate in Japan and eastern Asia have been examined by many studies. During La Niña years, summer weather in Japan becomes, over all, anomalously warm (e.g., Nitta 1987; Kurihara and Tsuyuki 1987; Ueda et al. 1995; Kawamura et al. 1996). They pointed out that when tropical convective activity in the western part of the tropical Pacific is strong, i.e., in a La Nina year, the convective forcing excites the barotropic Rossby wave propagating toward a northeastward direction. This wave train is called the Pacific-Japan (PJ) pattern (Nitta 1987), in which the northwestern Pacific is covered by an anticyclonic anomaly in response to the Rossby wave propagation in La Niña years. This anticyclone anomaly over the northwestern North Pacific brings about abnormally warm and dry summers in Japan. In the present study, when the shallow Okhotsk high appears, the cyclonic anomaly appears in the northwestern part of the Pacific. This cyclonic anomaly and the subtropical-midlatitude dipole patterns shown in Fig. 7b and Fig. 9b, can be the antiphase of the PJ pattern. Wang et al. (2000) pointed out that the anomalous Philippine Sea anticyclone, which is closely influenced by El Niño, is a key phenomenon for the variation of the Asian summer monsoon. The anomaly fields in the geopotential fields over the Philippine Sea, shown in Figs. 7b and 9b, are in agreement with these studies as well as in agreement with the SST anomaly field in the Philippine Sea (See Fig. 8). Therefore, the relation to the SST in the eastern tropical Pacific, shown in Fig. 8, is consistent with the previous studies on the Asian summer monsoon. This study adds the possiblilty of remote influence of ENSO further extends toward the Okhotsk high. Such remote influence on the Okhotsk high has not been clarified yet by previous studies, probably because in the previous studies classifications of the Okhotsk high were not properly executed to isolate only the shallow Okhotsk high. From the viewpoint of the influence of the Okhotsk high on the summer climate of Japan, the influences of the deep and shallow Okhotsk high are different. As shown in Fig. 6-c, deep Okhotsk highs do not have any significant correlations with the temperature field over Japan, whereas, in a shallow Okhotsk high, Japan becomes anomalously cold (See Figs. 7-c and 8). This finding suggests that whenever the Okhotsk high appears, Japan does not always have anomalous cold summers, and that the distinction of the type of the Okhotsk high is crucial for summertime weather forecasts in Japan. 5. Conclusions We have found that both the northern Siberian warmness, and the southern Pacific coldness, are key factors involved in the interannual variation of the Okhotsk high. In addition, the interannual variation of the Okhotsk high is influenced by different independent remote sources; the remote source of the Siberian warmness is in northwestern Eurasia facing the Arctic, and the other remote source is in the tropical Pacific. According to Nakamura and Fukamachi (2004), there are

15 October 2004 Y. TACHIBANA, T. IWAMOTO, M. OGI and Y. WATANABE 1413 Table 2. Summarized characteristics of the both types of Okhotsk highs. type temperature remote vertical structure surface interaction horizontal advection climate of Japan deep warm Siberia Arctic barotropic small small no relation shallow cold Pacific tropics baroclinic large large cold two types of the blocking Okhotsk high. One usually occurs in July, and it is related to the Rossby wave over Siberia. This type is in agreement with the deep anticyclone in our study. The other type, which is related to the Pacific anticyclone, usually occurs in May (Nakamura and Fukamachi 2004). This May type is different from the shallow anticyclone found in our analysis, because as shown in Fig. 7, the shallow anticyclone is related to cyclonic circulation over the Pacific. The anticyclones classified by Nakamura and Fukamachi (2004) are related to the seasonal march, while the anticyclones classified by our analyses are the phenomena occurring in the same month. The summarized characteristics of both Okhotsk highs distinguished in this study are listed in Table 2. The relation to both the Arctic and the tropics suggests that for long-term weather forecasting of the Asian summer monsoon, both the Arctic and the tropics should be equally watched, although many previous studies on the Asian monsoon have tended to observe the tropics only. However, in the present study, we were not able to clarify why the surface Okhotsk high is enhanced by the upper north anticyclone located over northern Siberia, nor were we able to clarify why the surface Okhotsk high is enhanced by the upper southern cyclone located in the northwestern North Pacific. We can speculate that the development of the surface Okhotsk high might be influenced by such unusual occurrences as the appearance of marine fog or of low level clouds, both of which can dramatically change the surface heat budgets in this region, as was pointed out by Norris (2000). Nakamura and Fukamachi (2004) also pointed out that horizontal advection of the cold air, which may be related to the marine fog, plays an important role in the development of the Okhotsk high. The formation of the cold marine mixed layer, due to the longwave radiation cooling at the top of the marine fog, may raise the SLP. Solving this issue will be the central topic in the next step of the study. Acknowledgements In preparing this paper, the authors have had many discussions with their colleagues, including Prof. M.J. Wallace of Washington University and Dr. Nakamura of the University of Tokyo. Many suggestive advices of Prof. Yamazaki of Hokkaido University and anonymous reviewers are also highly fruitful. The authors also express gratitude to Mr. Yamamoto, Mr. Azuma and Mr. Kamata of Tokai University, for devoting themselves to illustrating the figures. This work is partially funded by the Core Research for Evolutional Science and Technology of Japan. The Grid Analysis and Display System (GrADS) was used for drawing the figures. References Fukutomi, Y., H. Igarashi, K. Masuda and T. Yasunari, 2003: Interannual variability of summer water balance components in three major river basins of northern Eurasia. J. Hydrometeor. 4, Honda, M., K. Yamazaki, H. Nakamura and K. Takeuchi, 1999: Dynamic and thermodynamic characteristics of atmospheric response to anomalous sea-ice extent in the Sea of Okhotsk. J. Climate, 12, , H. Nakamura, J. Ukita, I. Kousaka and K. Takeuchi, 2001: Interannual seesaw between the Aleutian and Icelandic lows. Part 1: Seasonal dependence and life cycle. J. Climate, 14, Kato, K., 1985: Heat budget in the atmosphere and the variation of the air temperature in the lower layer over the Okhotsk Sea (A case study during the Baiu Season in 1979). Tenki, 32, (in Japanese). Kalnay, E., M. Kanamitsu and co-authors, 1996: The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc., 77,