June 1989 T. Nitta and S. Yamada 375. Recent Warming of Tropical Sea Surface Temperature and Its. Relationship to the Northern Hemisphere Circulation

June 1989 T. Nitta and S. Yamada 375 Recent Warming of Tropical Sea Surface Temperature and Its Relationship to the Northern Hemisphere Circulation By Tsuyoshi Nitta and Shingo Yamada Long-Range Forecast Division, Japan Meteorological Agency, Tokyo 113, Japan (Manuscript received 6 October 1988, in revised form 10 March 1989 Abstract Analyses of the global sea surface temperature (SST) were performed to examine year to decadescale variations of SST. It is found that the tropical SST, especially in the central and eastern Pacific and in the Indian Ocean, has being increasing since the late 1970's. SST averaged in the whole tropics between 20*N and 20*S was warmer by about 0.3**.0.4* in the 1980's than in the 1970's. Convective activity in the tropics detected by the satellite-measured outgoing longwave radiation becomes more enhanced corresponding to the recent warming of the tropical SST. The Southern Oscillation Index, which represents a strength of the tropical Walker circulation, accordingly tends to become negative after the late 1970's. The Pacific/North American teleconnection pattern with significant lowering of the geopotential height at 500 mb in the North Pacific has become dominant and persistent during winter seasons in the 1980's. It is suggested that these large atmospheric anomalies may be associated with tropical heat sources enhanced by the recent warming of the tropical SST. 1. Introduction In recent years, much attention has been paid to long-term variations of global atmosphere and ocean with time scales ranging from several months to decades under the World Climate Research Program (WCRP). Especially, variations with time-scales of about several years have been extensively studied in connection with the El Nino and Southern Oscillation (ENSO). Variations with time-scales of one month to several months have also been studied for the understanding of the tropical 30-60 day oscillation and for the development of long-range weather forecasts. Decade-scale variations of the global surface air temperature have been extensively examined by many investigators, e.g., Jones et al. (1986) and Yamamoto and Hoshiai (1980). Quite recently, Hansen and Lebedeff (1988) called attention to the fact that the global surface air temperatures in the 1980's are the warmest in the history of instrumental records. They suggested that this recent warming may be due to the greenhouse effect caused by increasing carbon dioxide. Folland et al. (1984) examined interannual fluctuations of worldwide sea surface temperature and near-surface marine air temperature for the period 1856-1981 and found several types of inter-decadal 1989, Meteorological Society of Japan variations. Folland el at. (1986) further demonstrated that wet and dry periods in the Sahel region of Africa are closely related to contrasting pattern of SST anomalies on a near-global scale. Hsiung and Newell (1983) has made, for the first time, an empirical-orthogonal-function analysis of the global sea surface temperature (SST) for the period 1949-79 and obtained the El Nino mode as the most important non-seasonal pattern. The second most important pattern shows evidence of decreasing SST trends in the North Atlantic and North Pacific in the 1965-79 period. Kashiwabara (1987) pointed out that the 500-mb height in the North Pacific was extremely low during 1977-1986. He suggested that tropical SST may give an impact on those decade-scale atmospheric anomalies, although the SST analysis was not performed. Quite recently, Namias el al. (1988) has examined the persistence of North Pacific SST and atmospheric flow pattern. He found that there exists a low-frequency variability with periods from several years to decades in North Pacific SST, the atmospheric Pacific/North American (PNA) index and the Southern Oscillation Index (SOI) which seem to be coupled with each other, but their SST analysis is limited in the North Pacific region north of 20*N. In this paper we make a global analysis of the sea surface temperature by using SST data for 1950-1987 and put a main focus on rapid warming of

376 Journal of the Meteorological Society of Japan Vol. 67, No. 3 tropical SST occurring from the late 1970's to the mid-1980's. Variations of tropical convective activity and the atmospheric anomalies in the Northern Hemisphere corresponding to the recent SST warming are also examined. 2. Data and method of analysis Monthly mean global sea surface temperature data during 1950-1987 at 2**2* grid points between 60*N and 40*S edited by Long-range Forecast Division, Japan Meteorological Agency (JMA) are used for analysis. These SST data are based on Comprehensive Ocean-Atmosphere Data Set (LOADS) for 1950-1969, NOAA dataset for 1970-1984 and those by Oceanographical Division of JMA for 1985-1987. Folland of al. (1984) discussed instrumental problems of SST related to changes in measuring techniques and concluded that corrections of SST after World War 2 are likely to be small. Since COADS data have better data coverage after the 1950's than before, data quality of SST for the analyzed period (1950-87) seems to be generally homogeneous, although quality may have gradually improved with time. Monthly mean outgoing longwave radiation (OLR) data at 2.5*2.5* grids during July 1974- March 1987 except for January 1978-March 1979 are used to analyze variations of tropical convection. Although there have been several changes in satellites and analysis schemes since the beginning of the OLR dataset in 1974, the OLR data used in this study have been corrected by National Environmental Satellite Data and Information Service (NESDIS) based on the correction algorithms as described in Gruber and Krueger (1984). Further, we have checked the global averages of OLR, but no systematic bias could be found except for January- March 1979 that corresponds to early periods of the TIROS N observations. We exclude these periods in addition to the whole of 1978 in which the OLR data were missing. The OLR data thus corrected and checked may be used for the long-period climatological studies. The 500-mb geopotential height and surface pressure (reduced to mean sea level) data at 10**10* grids in the Northern Hemisphere (north of 20*N) edited by JMA are analyzed to obtain circulation anomalies corresponding to the SST variations. Empirical orthogonal function (EOF) analysis was applied to SST and OLR data to detect dominant spatial patterns and their time variations. The EOFs are non-rotated and based on a correlation matrix. The EOFs are also computed based on a covariance matrix but general patterns for EOF1-3 are nearly the same as those by correlation matrix. Composite maps of SST, 500-mb geopotential height and the surface pressure in the Northern Hemisphere before and after 1977 are constructed to clarify the anomalies corresponding to the the decade-scale variations. 3. Decade-scale variations of global sea surface temperature EOF analysis is performed by using global SST data (60*N-40*S) for 38 years from 1950 to 1987 to determine dominant spatial patterns and their time variations. 8* longitude * 4* latitude grid points are used for the EOF computations. 38-year averaged monthly mean values are subtracted from individual monthly values and 3-month running means of these deviations are used for the EOF analysis. Fig. 1 shows the spatial patterns and time coefficients for the first two principal modes. The estimate of the sampling errors associated with each eigenvalue using the formula recommended by North of al. (1982) indicated that the first four functions all have sampling errors smaller than the spacing between their neighboring values. The first mode accounts for 17.2 % of the total variance. This mode has a large amplitude in the tropical eastern Pacific extending from the central equatorial region toward both northeast-eastward and southeast-eastward directions. Large positive anomalies are also found both in the tropical Indian and Atlantic Oceans. On the other hand, negative deviations are found in the extratropical regions both of the North and South Pacific. Time variations of the amplitude of the first mode (Fig. 1b) indicate that this mode was amplified corresponding to the El Nino years such as 1951-52, 1953-54, 1957-58, 1963-64, 1965-66, 1969-70, 1972-73, 1977-78, 1982-83 and 1986-87. This is consistent with the observational results that SST in the equatorial central and eastern Pacific becomes warmer than normal during E1 Nino years. In addition to these El Nino-related variations, we can see decade-scale variations. Especially, a trend increasing from the mid'70 to the mid'80 is noticeable. The first mode has a negative amplitude during the early to middle 70's (1970-76) except in 1972-73 (El Nino years), but tends to become positive after 1977. The results of the first mode obtained in this study generally correspond to those of the first eigenvector computed by Hsiung and Newell (1983) for 1949-79. The second mode which accounts for 6.3 % of the total variance has a dominant amplitude with opposite sign in the North and South Atlantic. Large amplitudes are also found in the Indian Ocean and Pacific. Time variations of the second mode show a downward downward linear trend changing sign around 1970, in addition to several years variations. Spatial and temporal variations of this mode are similar to those of the second eigenvector obtained by Hsiung and Newell (1983). This mode may also correspond to near-global SST variations which are strongly related to the Sahel Rainfall as obtained by Folland et al. (1986).

June 1989 T. Nitta and S. Yamada 377 Fig. 1. Distributions of the first two principal components of the EOF analysis of the global SST (a) and their time coefficients (b). The contour values of the EOFs are multiplied by 100. Contour intervals are 20 and negative contours are dashed. Thin lines and thick lines in (b) denote original time coefficients and their 10-month running means, respectively. Since the first mode of SST changed its sign before and after 1977, SST differences between 1977-1986 and 1967-1976 and their significance (above 95 %) according to a t-test are obtained to identify recent trends of SST variations (Fig. 2). Similar anomaly patterns patterns to those for the first mode (Fig. la) are obtained in general. The largest positive SST difference with an amplitude of about 0.8* occurs in the Southeast Pacific around 10*S, 120*W and also in the Northeast Pacific around 20*N, 120*W. A positive SST difference with an amplitude of about 0.4* is also found in the equatorial central Pacific near the dateline. SST difference in the tropical western Pacific is quite small and has little significance. Positive SST difference with about 0.4* is found in the whole Indian Ocean. SST in the North Atlantic shows a significant positive trend but the SST difference in the South Atlantic is small. In contrast to the general warming in the whole tropical belt between 20*N and 20*S, significant cooling takes place in the subtropical-middle latitudes both in the North and South Pacific centered around 160*W. Warming can be found along the west coast of North America extending from low latitudes to higher latitudes. SST anomalies in the North Pacific north of 20*N are quite similar to

378 Journal of the Meteorological Society of Japan Vol. 67, No. 3 Fig. 2. Differences between the 10-year mean SST for 1977-1986 and that for 1967-1976. Contour interval is 0.2* and shaded regions denote the 95 % level of significance according to a t-test. the first mode of North Pacific SST computed by Namias et al. (1988). They pointed that there is a significant upward linear trend in the north Pacific SST EOF1 especially during recent two decades. Our results indicated that the recent linear trend in SST is not limited to the North Pacific but dominant also in the whole tropics. The results in Fig. 2 indicate that SST in the tropical belt tends to increase in these 10 years. We then examined interannual variations of SST averaged in the whole tropics together with those averaged in the Indian Ocean (20*N-20*S, 40*E-100*E) and in the tropical eastern Pacific (20*N-20*S, 180*-80*W). Fig. 3a shows time variations of yearly mean SST averaged over the whole tropical belt (20*N-20*S). 5-year running means (dashed lines) are also plotted. There exist short-period variations with periods of about 3-7 years which may correspond to ENSO cycles. In addition to the short-period fluctuations, decade-scale variations can be noted. Negative SST peaks appear in the 5-year running means in the mid-1950's. The tropical SST is nearly constant around the mean value from the late 1950's to the early 1970's. The SST anomalies dropped to the minimum value of about -0.2* in the mid-1970's and then increased quite rapidly to about +0.2* in the mid-1980's. In the 5-year running means the tropical SST was increased by about 0.4* during the recent 10 years from the mid-1970's to the mid- 1980's. Time variations of SST in the Indian Ocean (Fig. 3b) are generally similar to those averaged in the whole tropics, but the amplitude of the former variations is slightly larger than the latter. The warming trend in the Indian Ocean can be seen not only during the recent two decades but also for the whole analysis period. The SST anomalies in the tropical eastern Pacific (Fig. 3c) exhibit large short-period fluctuations with time scales of several years mostly due to the ENSO cycles, but the decade-scale warming trend from the mid-1970's to the 1980's can be also noted. SST in the North and South Pacific (20 N-40 N, Fig. 3. Yearly mean SST averaged in the tropics between 20*N and 20 S (a), SST averaged in the Indian Ocean (b) and SST averaged in the tropical eastern Pacific (c). Dashed lines denote 5-year running means. Units are *. 160*E-140*W; 20*S-40*S, 160*E-140*W) generally show variations out of phase with that in the tropical eastern Pacific (not shown), indicating that fluctuations in SST in all regions of the Pacific including tropics and extratropics are strongly related. 4. Convective activity in the tropics It is suggested from the results of the SST analysis that the tropical convection may become enhanced corresponding to the recent warming of the tropical

June 1989 T. Nitta and S. Yamada 379 Fig. 4. Distributions of the first two principal components of the EOF analysis of OLR (a) and their time coefficients (b). The contour values of the EOFs are multiplied by 100. Contour intervals are 20 and negative contours are dashed. SST. We applied the EOF analysis to the OLR data at 5**5* grids between 25*N and 25*S for July 1974-March 1987 (January 1978-March 1979 is not included) to detect spatial and temporal variations of tropical convection. The estimate of the sampling errors using the formula by North et al. (1982) shows that the first three modes are distinguished from the neighboring modes. Fig. 4 shows the first two principal components of OLR and their time coefficients. The first component (Fig. 4a, upper) which accounts for 15.7 % of the total variance exhibits a east-west dipole pattern in the equatorial Pacific region. This component becomes positive in the 1976/77,1982/83 and 1986/87 El Nino events suggesting suggesting that this component is associated with the ENSO cycle. There are opposite variations in North Brazil and the equatorial Atlantic to those in the equatorial eastern Pacific. These results are consistent with the El Nino composite of precipitation obtained by Ropelewski and Halpert (1987), indicating that below normal precipitation occurs during the El Nino years. The second component (Fig. 4a, lower) accounts for 11.4 % of the total variance and has large amplitudes in the Indian Ocean and in the equatorial central Pacific. Positive anomalies occur in the Atlantic, while negative anomalies are found in the Southeast Pacific offshore of South America. Time coefficients of this mode (Fig. 4b, lower) show that this mode has a long-term trend changing its sign from the late 1970's to the early 1980's. These results indicate that the tropical convection especially in the Indian Ocean and in the central Pacific tends to have been more active in the 1980's than in the 1970's. This recent enhancement of convective activity in the tropics may correspond to recent warming of the tropical SST discussed in the previous section. In order to see enhancement of tropical connective activity corresponding to the recent warming of SST, 5-month running means of OLR anomalies averaged in the whole tropical belt (10*N-10*S) are

380 Journal of the Meteorological Society of Japan Vol. 67, No. 3 plotted in Fig. 5. The tropical OLR anomalies exhibit a significant downward linear trend from the 1970's to the 1980's, showing that the tropical convective activity becomes stronger in the 1980's. These results of the EOF analysis and time variations of tropical OLR clearly indicate that the convection in the global tropics was more active in the 1980's than in the middle to late 1970's. These are quite consistent with the recent warming of the tropical SST. Especially, convective activity in the tropical Indian Ocean and in the tropical central and eastern Pacific of the Southern Hemisphere is much enhanced. The latter may correspond to the northeast-eastward shift of the South Pacific Convergence Zone (SPCZ). Fig. 5. OLR anomalies averaged in the tropics between 10*N and 10*S. Units are W m-2. 5. Atmospheric circulation anomalies in the Northern Hemisphere Quite recently, Kashiwabara (1987) has pointed that the geopotential height at 500mb during Northern winter in the North Pacific had lowered considerably in the 10 years since 1977. It is probable that this large anomaly in the North Pacific may have resulted from the enhanced convective activity in the tropics due to the warming of the tropical SST demonstrated in the previous sections. Since Kashiwabara (1987) described only anomaly distributions of 500-mb geopotential height over the North Pacific, we computed similar distributions but in the whole Northern Hemisphere north of 20*N. Fig. 6 shows a map of the 500-mb geopotential height difference for winter (December-February) between a 10-year average for 1977-86 and that for 1967-76 (former minus latter). The trough in the North Pacific around 40*N, 170*E has deepened substantially in these 10 years (KashiwaJara, 1987). There exist positive anomalies along the west coast of North America and in the subtropical easte Pacicific. These anomaly patterns extending from the Pacific region to North America are quite similar to the so-called Pacific/North American (PNA) patterns found by many investigators, e. g., Wallace and Gatzler (1981) and Barnston and Livezey (1987). The PNA pattern is thought to be enhanced by the tropical forcing associated with the El Nino events Fig. 6. Differences of the 500-mb geopotential height during winter between for 1977-86 and for 1967-76. Contour intervals are 10 gpm and negative contours are dashed. The 95% level of significance is shaded.

June 1989 T. Nitta and S. Yamada 381 Fig. 7. As in Fig. 6 but for the surface pressure. Contour intervals are 1 mb. Fig. 8. Southern Oscillation Index (Tahiti-Darwin). Unit is normalized by the standard deviation. (lore! and Wallace, 1981). It is likely that the enhancement of the PNA pattern in these 10 years may be largely due to the recent intensification of tropical convection especially in the Pacific ocean. The trend toward stronger PNA patterns during the recent decades has been also noted by Namias et al. (1988). There exists a weak positive anomaly with a significance level above 95 % over the North Africa, but further analyses may be needed to understand how this anomaly is generated. Fig. 7 shows the same anomaly map as Fig. 6 except for the surface pressure. The region with a high significance level is limited to the North Pacific where the mean surface pressure in recent 10 years has dropped by about 5mb as compared with that in previous 10 years. These regions generally correspond to those where the 500-mb height is significantly dropped. It is suggested that the SST warming and its related enhancement of tropical convection may change not only atmospheric circulations in the extratropical regions as demonstrated above, but also large-scale circulations in the tropics. The Southern Oscillation Index (SOl) is known to be an index representing the tropical Walker (east-west) circulations. Fig. 8 shows time series of 501 during 1950-1987. In addition to time variations with periods of several years corresponding to the ENSO cycles, there exists a long-term trend which tends to have negative values after 1977. This has been already pointed out by Kashiwabara (1987) and Namias et al. (1988). This tendency of negative 501 seems to be consistent with spatial distributions and time variations of OLR EOF2 as shown in Fig. 4. The results of OLR EOF2 shows that the convection around Tahiti

382 Journal of the Meteorological Society of Japan Vol. 67, No. 3 (Southeast Pacific) became active in the recent 10 years resulting in a lowering of the surface pressure; the convection around Darwin (North Australia), however, has not changed. The changes of the convective activity in the tropical Pacific may produce a negative tendency of SOI. It can be tentatively concluded that the ENSOlike phenomena have tended to occur more frequently during the recent ten years than during the previous years because of enhanced convection in the tropical central and eastern Pacific corresponding to the SST warming in these regions. The PNA anomaly patterns may become intensified by these tropical forcing 6. Conclusions Analyses of global SST showed that the tropical SST especially in the Indian and central and eastern Pacific Oceans tends to have been warmer in the 1980's than in the 1970's. SST in the Atlantic also increased. The mean SST in the tropical belt between 20*N and 20*S was warmer by about 0.3* 0.4* in the 1980's than in the 1970's. SST in the subtropical to mid-latitude regions in the Pacific Ocean of the both hemispheres decreased conversely. The mean tropical SST as shown in Fig. 3a appears to have decade-scale variations including the recent warming trend. It is generally thought that the time scales of the global ocean circulations including the deep layers range from decades to thousand years and the SST may be largely affected by these circulations. However, the structure and time variations of global ocean circulations have not been fully understood. The World Ocean Circulation Experiment (WOCE) is planned to be carried out internationally in the 1990's and its outcomes will make significant contributions to the knowledge of the global ocean circulations. It has become clear that the coupling between the atmosphere and ocean may be quite important for the long-term variations of the atmosphere and ocean. Bryan et al. (1982) has studied the transient climate response to increasing atmospheric carbon dioxide by means of a detailed three-dimensional atmosphere-ocean model. They determined that the initial response to a sudden increase of atmospheric CO2 is a rapid rise of SST in the tropics. Recently, Hansen and Lebedeff (1988) have shown observationally that the global surface air temperature has substantially increased in the 1980's. Current air temperature trends are of special interest because of the expectation that increases in infrared-absorbing trace gases such as carbon dioxide will cause a global warming. These results of the warming of the global surface air temperature in the 1980's are consistent with those of the recent warming trend of the tropical SST obtained in this study. In addition to the recent warming, both the global air temperature Fig. 9. 5-year running means of the global mean surface air temperature (Ta) (after Hansen Lebedeff, 1988) and those of SST averaged in the Indian Ocean (Ts). Units are *. and the tropical SST, especially the Indian Ocean SST, show similar time variations after about 1960 as shown in Fig. 9 suggesting a close relationship between air temperature and SST. However, the detailed coupling processes of the decade-scale variations between the air and SST are not fully understood and need further study. The OLR analyses showed that the convection in the tropics, especially over the Indian Ocean and the central and eastern Pacific, was more active in the 1980's than in the 1970's. This enhanced convective activity may have resulted from the warming of the tropical SST. Atmospheric circulations seem to be disturbed by the warming of SST and the enhancement of convective activity in the tropics. SOI, which indicates the strength of east-west circulations in the Pacific Ocean, tends to become more negative after the late 1970's. This result is consistent with that of enhanced convection in the central and eastern Pacific corresponding to the SST warming in these regions. Large anomaly circulations can be found during the Northern winter corresponding to the recent SST warming. Substantial lowering of the 500- mb geopotential height and the surface pressure in the North Pacific was observed during 1977-1986. The PNA pattern becomes more dominant during the 1980's. It seems that the PNA pattern may be largely forced by the tropical heating. Negative SST anomalies in North Pacific may also contribute to these patterns to some extent as suggested by the numerical experiments by Pitcher et al. (1988). Recent trends in decreasing SST in the North Pacific, decreasing SOI index and intensification of the PNA patterns were also discussed by Namias et al. (1988).

June 1989 T. Nitta and S. Yamada 383 The ocean-atmosphere conditions in the tropical Pacific has fallen into a strong anti-el Nino mode since the spring of 1988. The SOI index in late 1988 had an extremely large positive value which is comparable to that of 1975 according to Monthly Monitoring Report by J apan Meteorological Agency (1989). This large positive SOI in 1988 is opposed to the recent negative trend discussed in this study. It is of interest to monitor whether this large opposite swing in the tropical Pacific is a signal of a new trend or not. Acknowledgement The authors wish to extend their thanks to Mr. K. Koizumi and Mr. S. Matsubayashi for providing geopotential height and SST data. References Barnston, A.G. and RE. Livezey, 1987: Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Mon. Wea. Rev., 115, 1083-1126. Bryan, K., F.G. Komro, S. Manabe and M.J. Spelman, 1982: Transient climate response to increasing atmospheric carbon dioxide. Science, 215, 56-58. Folland, C.K., D.E. Parker and F.E. Kates, 1984: Worldwide marine temperature fluctuations 1856-1981. Nature, 310, 670-673. Folland, TN. Palmer and D.E. Parker, 1986: Sahel rainfall and worldwide sea temperature, 1902-85. Nature, 320, 602-607. Gruber, A. and A.F. Krueger, 1984: The status of the NOAA outgoing longwave radiation data set. Bull. Amer. Meteor. Soc., 65, 958-962. Hansen, J. and S. Lebedeff, 1988: Global surface air temperatures: Update through 1987. Geophy. Res. Letters, 15, 323-326. Horel, J.D. and J.M. Wallace,1981: Planetary-scale atmospheric phenomena associated with the Southern Oscillation. Mon. Wea. Rev., 109, 813-829. Hsiung, J. and RE. Newell, 1983: The principal nonseasonal modes of variation of global sea surface temperature. J. Phys. Oceanogr., 13, 1957-1967. Japan Meteorological Agency, 1989: Monthly Report on Climate System, December 1988. 50pp. Jones, P.D., S.C.B. Raper, R.S. Bradley, H.F. Diaz, P.M. Kelly and T.M.L. Wigley, 1986: Northern Hemisphere surface air temperature variations: 1851-1984. J. Climate Appl. Meteor., 25, 161-179. Kashiwabara, T., 1987: On the recent winter cooling in the North Pacific. Tenki, 34, 777-781 (in Japanese). Namias, J., X. Yuan and DR. Cayan, 1988: Persistence of North Pacific sea surface temperature and atmospheric flow patterns. J. Climate, 1, 682-703. North, G.R., T.L. Bell, R.F. Cahalar and F.J. l\toeng, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev., 110, 699-706. Pitcher, E.J., ML. Blackmon, G.T. Bates and S. Munoz,1988: The effect of North Pacific sea surface temperature anomalies on the January climate of a general circulation model. J. Atrnos. Sci., 45, 173-188. Ropelewski, C.F. and M.S. Halpert, 1987: Global and regional scale precipitation patterns associated with the El Nino/ Southern Oscillation. Mon. Wea. Rev., 115, 1606-1626. Wallace, J.M. and D.S. Gutzler, 1981: Teleconnections in the geopotential height during the Northern Hemisphere winter. Mon. Wea. Rev., 109, 784-812. Yamamoto, R. and M. Hoshiai, 1980: Fluctuations of the Northern Hemisphere mean surface air temperature during recent 100 years estimated by optimum interpolation. J. Meteor. Soc. Japan, 58, 187-193.