Intraseasonal Variability of the Low-Level Jet Stream of the Asian Summer Monsoon

1APRIL 2004 JOSEPH AND SIJIKUMAR 1449 Intraseasonal Variability of the Low-Level Jet Stream of the Asian Summer Monsoon P. V. JOSEPH AND S. SIJIKUMAR Department of Atmospheric Sciences, Cochin University of Science and Technology, Kochi, India (Manuscript received 19 February 2003, in final form 16 August 2003) ABSTRACT The strong cross-equatorial low level jet stream (LLJ) with its core around 850 hpa of the Asian summer monsoon (June September) is found to have large intraseasonal variability. During the monsoon onset over Kerala, India, and during break monsoon periods, when the convective heating of the atmosphere is over the low latitudes of the Indian Ocean, the axis of the LLJ is oriented southeastward over the eastern Arabian Sea and it flows east between Sri Lanka and the equator and there is no LLJ through peninsular India. This affects the transport of moisture produced over the Indian Ocean to peninsular India and the Bay of Bengal. In contrast, during active monsoon periods when there is an east west band of strong convective heating in the latitudes 10 20 N from about longitude 70 to about 120 E, the LLJ axis passes from the central Arabian Sea eastward through peninsular India and it provides moisture for the increased convection in the Bay of Bengal and for the monsoon depressions forming there. The LLJ does not show splitting into two branches over the Arabian Sea. Splitting of the jet was first suggested by Findlater and has since found wide acceptance as seen from the literature. Findlater s findings were based on analysis of monthly mean winds. Such an analysis is likely to show the LLJ of active and break monsoons as occurring simultaneously, suggesting a split. Strengths of the convective heat source (OLR) over the Bay of Bengal and the strength of the LLJ (zonal component of wind) at 850 hpa over peninsular India and also the Bay of Bengal between latitudes 10 and 20 N have the highest linear correlation coefficient at a lag of 2 3 days, with OLR leading. The LLJ crossing the equator close to the coast of East Africa will pass through India only if there is active monsoon convection in the latitude belt 10 20 N over south Asia. The position in latitude of the LLJ axis between longitudes 70 and 100 E is decided by the south north movement of the east west convective cloud band of the monsoon in its 30 50-day oscillation. When there is little convection over south Asia in the latitude belt 10 20 N, the LLJ crossing the equator curves clockwise over the Arabian Sea under conservation of potential vorticity and bypassing India passes east close to the equator. It is speculated that the cyclonic vorticity associated with this low-latitude LLJ causes convergence in the boundary layer and consequent upward motion in the atmosphere resulting in the formation of a convective cloud band there that later moves into the Bay of Bengal as part of the monsoon s 30 50-day oscillation. Since LLJ is very important in monsoon dynamics, monsoon modelers should take adequate care to see that LLJ and the associated deep convection and their intraseasonal variability are properly simulated in their models. 1. Introduction A strong cross-equatorial low level jet stream (LLJ) with a core around 850 hpa exists over the Indian Ocean and south Asia during the boreal summer monsoon season June September. Bunker (1965) using aircraft observations of wind over the Arabian Sea during the International Indian Ocean Expedition (IIOE) traced an LLJ with large vertical wind shears off Somalia and across the central parts of the Arabian Sea. He showed that monsoon winds attained a speed of 50 kt in the southwestern parts of the Arabian Sea at the top of a 1000-m layer of air cooled by contact with the upwelled water off the Somali Arabia coasts. Analyzing the wind Corresponding author address: Dr. P. V. Joseph, Department of Atmospheric Sciences, Cochin University of Science and Technology, Fine Arts Avenue, Kochi 682 016, India. E-mail: porathur@md4.vsnl.net.in data of 5 yr collected by the radiosonde/radio wind network of India, Joseph and Raman (1966) established the existence of a westerly low-level jet stream over peninsular India with strong vertical and horizontal wind shears. This LLJ is seen over peninsular India on many days in the typical monsoon month of July with a core at about 1.5 km above mean sea level and core wind speeds of the order of 40 60 kt. Findlater (1969a,b) found that the Asian summer monsoon LLJ has its origin in the trade wind easterlies of the south Indian Ocean, it crosses the equator in a narrow longitudinal belt close to the East African coast as a southerly current with speeds at times even as high as 100 kt, turns into a westerly current over the Arabian Sea, and passes through India. This jet according to their computations accounts for nearly half the interhemispheric transport of air in the lower troposphere around the globe. Findlater s LLJ is a combination of the LLJs found by Bunker (1965) and Joseph and Raman (1966) 2004 American Meteorological Society

1450 JOURNAL OF CLIMATE VOLUME 17 FIG. 1. Wind field at 1 km for Aug over the Indian Ocean from Findlater (1971). Thick lines marked are the LLJ axes. Isotachs in m s 1 are shown as broken lines. and the LLJ crossing the equator off the East African coast discovered by him (Findlater 1966, 1967). Using monthly mean winds Findlater (1971) showed that the LLJ splits into two branches over the Arabian Sea, one branch passing southeastward toward Sri Lanka and the other eastward through peninsular India. Please see Fig. 1 taken from their paper, which shows the suggested splitting of the LLJ. We have examined whether the LLJ is really splitting over the Arabian Sea, using the daily National Centers for Environmental Prediction National Center for Atmospheric Research (NCEP NCAR) wind data. Using a one-level primitive equation model with a detailed bottom topography and a 1 latitude grid size, Krishnamurti et al. (1976) showed that many of the observed features of the cross-equatorial LLJ can be numerically simulated by including 1) the East African and Madagascar mountains, 2) the beta effect, and 3) a lateral forcing at 75 E due to land ocean contrast heating in the meridional direction, essentially following Murakami et al. (1970). They simulated an intense LLJ off the Somali coast. The split in the jet over the Arabian Sea was attributed to barotropic instability. The effect of a baroclinic boundary layer on the LLJ was investigated by Krishnamurti and Wong (1979) and Krishnamurti et al. (1983). A time-dependent primitive equation model with specified zonal flow, mountains and diabatic heating was used to study the LLJ by Hoskins and Rodwell (1995) and Rodwell and Hoskins (1995). The East African highlands and a land sea contrast in surface friction are shown to be essential for the concentration of the crossequatorial low-level flow into a LLJ. They found that surface friction and diabatic heating provided mechanisms for material modification of potential vorticity (PV) of the flow and both were found important for the maintenance of the LLJ. The study identified the strong sensitivity of the LLJ to changes in convective heating over the Indian ocean. When there is very little modification of the PV, the LLJ turns anticyclonically over the Arabian Sea and the flow tends to avoid India according to them.

1APRIL 2004 JOSEPH AND SIJIKUMAR 1451 During the last decade a number of studies have appeared in the literature regarding the interannual and intraseasonal variability of the lower-tropospheric (e.g., 850 hpa) monsoon circulation (e.g., Webster et al. 1998; Annamalai et al. 1999; Sperber et al. 2000; Krishnamurthy and Shukla 2000; Goswami and Ajaya Mohan 2001). They have discussed the convection and circulation associated with the active break cycle of the monsoon. In our study we have focused on the LLJ and its intraseasonal variability and the relation between LLJ and atmospheric convection. Details regarding the data used for this study are given in section 2. Section 3 describes the 30-yr (1961 90) climatology of LLJ. The intraseasonal variability of the LLJ is presented in section 4, analyzing 12-yr (1979 90) composites of the monsoon onset phase and the composites of active and break spells. Composites are made of an 850-hPa wind field and of OLR as a proxy for the convective heating of the atmosphere. The relation between the convective heating of the atmosphere and the strength of the LLJ is discussed in section 5. This is followed in section 6 by a detailed case study of the active break cycle of the First Global Atmospheric Research Programme (GARP) Global Experiment Monsoon Experiment (FGGE MONEX) year 1979 monsoon. Summary and conclusions are given in section 7. 2. Data Recently global datasets on a twice daily time scale were generated as part of the NCEP NCAR reanalysis project (Kalnay et al. 1996). We have used the NCEP NCAR daily wind data at standard pressure levels, on a 2.5 latitude longitude grid, to study the characteristics of the LLJ. The NCEP NCAR data output has been classified into four categories, depending on the relative influence of the observational data and the model used on the gridded variable. Wind data is under category A, which indicates that this variable is strongly influenced by observed data and hence it is in the most reliable class (Kalnay et al. 1996). However substantial difference exists between NCEP and the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis products, particularly around the LLJ region (Annamalai et al. 1999). For the strength of the convective heating of the atmosphere we have used National Oceanic and Atmospheric Administration (NOAA) interpolated outgoing longwave radiation (OLR) data. [The data have been taken from the Interpolated OLR Data provided by the NOAA CIRES (Cooperative Institute for Research in Environmental Sciences) Climate Diagnostics Center, Boulder, Colorado, from their Web site at http://www.cdc.noaa.gov; Gruber and Krueger (1984).] Monsoon onset dates are taken from India Meteorological Department. Joseph et al. (1994) has given a critique of the methods for determination of the date of TABLE 1. Date of monsoon onset over Kerala, and duration of active and break monsoon spells during 1979 90. Dates of monsoon onset as given by IMD, also by Ananthakrishnan and Soman/Soman and Krishnakumar (AS/SK). Active spells are during Jun Aug (as defined in text) and break spells during Jul and Aug (De et al. 1998), both of the duration of 3 or more days. Year 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 Monsoon onset over Kerala (IMD) (AS/SK) Active monsoon Break monsoon 13 Jun 30 May 13 Jun 31 May 28 May 4 Jun 2 Jun 26 May 3 Jun 19 May 1 31 May 29 May 12 Jun 24 May 13 Jun 2 Jun 17 May 23 Jun 2 Jul 31 Jul 12 Aug 2 7 Jul 5 8 Aug 11 16 Aug 11 15 Aug 14 18 Jun 19 27 Jun 16 21 Jul 6 11 Aug 15 20 Jul 31 Jul 2 Aug 21 26 Jul 22 Jun 8 Jul 17 23 Jul 15 31 Aug 17 20 Jul 26 30 Jul 23 27 Aug 22 25 Aug 20 24 Jul 22 25 Aug 23 26 Aug 29 31 Aug 28 Jul 1 Aug 5 8 Jul 13 15 Aug 10 12 Jul 29 31 Jul 8 10 Jul 27 31 Jul Total days 113 84 Monsoon Onset over Kerala (MOK). The long-term mean date of MOK is e with a standard deviation of about 8 days and extreme dates of 11 May (earliest onset) and 18 June (most delayed onset). Dates of MOK for the years 1979 90 are given in Table 1. For comparison dates of monsoon onset over south and north Kerala by an objective method using only daily rainfall data by Ananthakrishnan and Soman (1988) and Soman and Krishnakumar (1993) are also given in Table 1. Breaks in monsoon during July and August have been identified by De et al. (1998) for the period 1968 96. The main criteria used are a Monsoon trough running close to the foot hills of the Himalayas and absence of easterly winds over the northern parts of India up to 1.5 km above sea level. For this study we have taken the break spells in July and August lasting 3 days or more of the 12-yr period 1979 90. We have thus 17 break spells of a total duration of 84 days. These break spells are listed in Table 1. Data on the dates of MOK and break monsoon spells are available in the literature for more than 100 yr (De et al. 1998; Ramamurthy 1969). But similar long period data on active monsoon spells are not available. V. Magna and P. J. Webster (1996, personal communication) and Webster et al. (1998) have used indices based particularly on the strengths of 850-hPa wind flow and convection in the latitude belt 10 20 N over south Asia to define active monsoon spells. Goswami and Ajaya Mohan (2001) have used a similar wind criteria for defining active monsoon spells. An active monsoon is gen-

1452 JOURNAL OF CLIMATE VOLUME 17 FIG. 2. The 5-day moving avg of daily mean zonal wind at 850 hpa in the lat lon box (10 20 N, 70 80 E) for the period 31 Aug 1979 in m s 1. erally understood as the period when strong LLJ passes through the 10 20 N latitude belt accompanied by active convection (rainfall) in the same belt over south Asia, formation of monsoon lows and depressions in the head of the Bay of Bengal, etc. (Rao 1976). We define an active monsoon spell arbitrarily as one in which for each day of the spell the area-averaged zonal wind at 850 hpa in the latitude longitude box 10 20 N and 70 80 E in a 5-day period centered on that day is 15ms 1 or more. Figure 2 gives the variation of the daily zonal wind at 850 hpa derived in this manner for the period e 31 August 1979 showing two active spells of monsoon (as given in Table 1). This figure also shows the break spells given in the table for 1979 as the weak wind portion of the wind variation chart. During the June August months of the 12-yr period 1979 90 we could thus get 14 active monsoon spells of a total duration of 113 days as shown in Table 1. 3. Climatology of the LLJ The mean vertical profile of the zonal component (U) of the monsoon flow through the central Arabian Sea and peninsular India and the trade winds of the Southern Hemisphere and the jet streams of the upper troposphere during July and August of 1961 90 are shown in Fig. 3. Averaging is done over the longitudes 62.5 67.5 E for the central Arabian Sea and 75 80 E for peninsular India from latitudes 30 S 50 N. Monsoon westerlies extend from the surface to about the 400-hPa level between latitudes 5 S 25 N over India. The westerly mean monsoon current is strongest close to 850 hpa. While the LLJ of the monsoon westerlies has only one core at longitude 65 E it has two cores (wind maxima) at longitude 77.5 E, one at about 8 N, and the other at about 17 N in agreement with Findlater (1971). Trade wind (easterly) maximum of the Southern Hemisphere is at about 12 S latitude close to the 925-hPa level. The figure FIG. 3. Vertical profile of the mean zonal component of wind of Jul and Aug, averaged over the lon (a) 62.5 67.5 E representative of lon 65 E and (b) 75 80 E representative of lon 77.5 E as averages for the 30-yr period 1961 90. The contour interval is 1 m s 1. Data of eight vertical levels from 1000 to 300 hpa as given in NCEP NCAR reanalysis are used. also shows the subtropical westerly jet streams of both hemispheres. The subtropical jet stream is much stronger in the Southern Hemisphere. The easterly wind seen above the monsoon westerlies is the bottom part of the tropical easterly jet stream. 4. Intraseasonal variability of the LLJ 1979 90 Detailed examination of the daily NCEP NCAR reanalysis 850-hPa wind and OLR data over the Indian subcontinent and the adjacent regions of 12 monsoons 1979 90 provided clear insight into the characteristic features of the LLJ on the intraseasonal scale. The monsoon has two main phases, the active and break and the most important intraseasonal variability of the monsoon is the active break cycle described in detail in Rao (1976). In addition it has an onset phase. In the following paragraphs details are given about LLJ and the areas of active monsoon convection during composite onset phases and the composites of spells of active and break monsoons during the period 1979 90.

1APRIL 2004 JOSEPH AND SIJIKUMAR 1453 FIG. 4. Composites for the onset pentad ( 2 to 2 days around the day of monsoon onset over Kerala) of 12 yr 1979 90 in (a) OLR (isolines in W m 2 : 220 and lower at intervals of 10 W m 2 ), (b) 850-hPa wind vectors and isolines of the magnitude of wind in m s 1 : 6 and more at 2 m s 1 interval. FIG. 5. Composites for active monsoon days in Jun Aug of 1979 90. (a) OLR (isolines in W m 2 : 220 and lower at intervals of 10 W m 2 ), (b) 850-hPa wind vectors. Isolines of magnitude of the wind inms 1 : 6 and more at 2 m s 1 interval. The date of MOK is taken as 0 and the days before and after onset are taken as negative ( ) and positive ( ), respectively. The 12-yr composite of the onset pentad corresponding to 2 to 2 days of MOK of OLR (Fig. 4a) shows a large area of low OLR or high convection in the low latitudes of the north Indian Ocean. The 850-hPa wind composite of the corresponding pentads shows a strong LLJ beginning from the south Indian Ocean, crossing the equator passing close to the East African coast, and turning east off the Somalia coast and moving farther east (Fig. 4b). A well-marked LLJ maximum is present over the Indian longitudes between the equator and latitude 10 N. The onset phase is characterized by a single LLJ core with maximum wind speeds over south Asia and the Indian Ocean between the equator and latitude 10 N. Monsoon westerlies through India are weak. Figures 5a,b give the composite mean of the OLR and the 850-hPa zonal wind in the study area for active monsoon spells of June August as defined in section 2. The dates on which active monsoon conditions prevailed are given in Table 1. The areas of maximum wind (LLJ) and the maximum convection (lowest OLR) are in the latitude belt 10 20 N. The composite LLJ has only one axis and LLJ shows no splitting. A convection maximum is seen in the Bay of Bengal. Composite analysis of OLR and wind at 850 hpa for break monsoon spells during July and August are presented in Figs. 6a,b. Details of the break monsoon spells of July and August of 1979 90 are given in section 2 and in Table 1. The OLR minimum area of low latitudes at the time of MOK has moved northward to the central Bay of Bengal during active spells and in break spells it lies over northeast India and its neighborhood. A fresh area of OLR minimum has formed over the equatorial Indian Ocean. [Please see Sikka and Gadgil (1980) for the northward movement and regeneration of maximum cloud zones.] The composite anomaly chart of break monsoon spells prepared by Ramamurthy (1969) using rain gauge data from the Indian land area also show positive rainfall anomalies over the extreme south of peninsular India and also over northeast India supporting the OLR anomalies in the break composite in this paper [see Fig. 4 of Ramamurthy (1969)]. The LLJ axis of break monsoon passes south of the Indian peninsula between the equator and latitude 10 N to the convectively active area there. In the break composite a weak LLJ axis can also be seen passing through north India toward the convectively active region over northeast

1454 JOURNAL OF CLIMATE VOLUME 17 FIG. 6. Composites for break monsoon days in Jul and Aug of 1979 90. (a) OLR (isolines in W m 2 : 220 and lower at intervals of 10Wm 2 ), (b) 850-hPa wind vectors. Isolines of magnitude of the wind in m s 1 : 6 and more at 2 m s 1 interval. India. A composite vertical cross section through longitude 77.5 E made using the data of break monsoon days listed in Table 1 (Fig. 7a) shows the two LLJ axes clearly, one around 5 and the other around 25 N. The composite for the active monsoon spells (Fig. 7b) shows only one strong LLJ axis at latitude 16 N. The present study based on daily NCEP NCAR reanalysis data did not support the splitting of LLJ over the Arabian Sea as suggested by Findlater (1971). Study of the 12-yr composites of the monsoon onset and active and break spells and the examination of individual days confirmed that there is no splitting of LLJ over the Arabian Sea as described in Findlater (1971) and frequently referred to in the literature. The jet shows a single axis except during the break monsoon when the northern LLJ branch passes through the latitude of about 25 and not 17 N as shown in Findlater (1971). Findlater s observation of the splitting of the LLJ is based on monthly mean wind data. Monthly mean data can show two branches of LLJ, one corresponding to the break monsoon and the other associated with the active monsoon. The two branch structure (split) of the LLJ as suggested by Findlater (1971) was not observed in the analysis of 1995 and 1997 monsoons by Halpern et al. (1998) and Halpern and Woiceshyn (1999). FIG. 7. Vertical profile of the mean zonal component of wind of (a) break composite and (b) active composite, averaged over the lon 75 80 E representative of lon 77.5 E for the 12-yr period 1979 90. The contour interval is 1 m s 1. Data of eight vertical levels from 1000 to 300 hpa as given in NCEP NCAR reanalysis are used. 5. Relationship between convective heating and the LLJ We have examined the relationship between the convective heating of the atmosphere over south Asia and the LLJ. As a measure of the strength of the LLJ we used the daily U index, which is the area-averaged zonal component of the wind at 850 hpa in (i) the peninsular box bounded by latitudes 10 20 N and longitudes 70 80 E and (ii) the Bay of Bengal box bounded by latitudes 10 20 N and longitudes 80 100 E as averages of 0000 and 1200 UTC observations. For the strength of the convective heating we have used the daily OLR index, which is the area-averaged OLR in the latitude longitude box 10 20 N and 80 100 E (the Bay of Bengal box). The OLR index chosen is representative of the large-scale convection in the monsoon. According to Sikka and Gadgil (1980) a maximum cloud zone (MCZ) of deep convective clouds form in the low-latitude regions south of India and moves north to the Himalayas and this process is repeated with a periodicity of 30 50 days during the monsoon season June Sep-

1APRIL 2004 JOSEPH AND SIJIKUMAR 1455 FIG. 8. LCC between the daily OLR index for the Bay of Bengal box (area 10 20 N, 80 100 E) and the daily U index (a) for Bay of Bengal (box area same as for OLR) and (b) for peninsular India box (area 10 20 N, 70 80 E) for lags of 5 to 5 days. Max negative LCC is for lags of 2 3 days, OLR leading. Line (a) is with triangles and line (b) is with dots. tember. It is observed that LLJ is strong through peninsular India when the MCZ passes through the Bay of Bengal (active monsoon). The linear correlation coefficient (LCC) between the daily OLR index and the daily U index for lags of 5 to 5 days is given in Fig. 8. LCC increases with lag for both boxes of wind and reaches a maximum (magnitude) and then decreases. Maximum LCC is 0.51 between the daily OLR index and the daily U index of the peninsular box (for 744 pairs of the indices during July and August of 1979 90) for OLR index leading U index by 2 and 3 days. The LCC for significance at levels 99% and 99.9% for 744 pairs of data are 0.08 and 0.115, respectively, according to the Student s t test. For the U index of the Bay of Bengal box the maximum LCC is 0.41 at lags of 1 and 2 days. Thus, atmospheric heating by convection is able to accelerate the LLJ flow through peninsular India in about 2 3 days. When this heating between 10 and 20 N is weak the cross-equatorial LLJ moves to the central Arabian Sea and then moves southeastward to areas south of India as shown in the modeling studies by Hoskins and Rodwell (1995) and Rodwell and Hoskins (1995). It was seen in section 4 that in break monsoon spells when the active monsoon convection has moved to northeast India from the Bay of Bengal there is a branch of LLJ through north India (latitude about 25 N) that can carry moisture to this area from the Indian Ocean. We may infer that the MCZ of Sikka and Gadgil (1980) is closely associated with the cross-equatorial LLJ over south Asia. Thus, while the LLJ crosses the equator in a geographically fixed and narrow longitude band, the latitude of the core of the LLJ over peninsular India longitudes, moves from low latitudes FIG. 9. Hovmöller diagram showing evolution of (a) convection (OLR) and (b) 850-hPa zonal wind speed from to 31 Aug 1979. Averaging is done for the lon band 80 100 E and a 5-day moving avg is applied as a smoother. OLR contours at 220 and lower at intervals of 10 W m 2 and wind speed contours at 6 and more at intervals of 2ms 1. to almost 25 N along with the northward movement of the MCZ in its 30 50-day cycle. 6. Case study of ISO of the LLJ in the monsoon of 1979 The monsoon of FGGE MONEX year 1979 had strong intraseasonal oscillation and pronounced active break cycles (Krishnamurti 1985). Figure 9a shows the Hovmüller diagram of the mean OLR between longitudes 80 100 E and latitudes 10 S 30 N of the period e 31 August 1979 smoothed by a 5-day moving average. After an active monsoon spell in the second half of June, convection in the 10 20 N belt weakens. By mid-july two zones of convection are found over the 80 100 E zone, one around latitude 25 N and the other around the equator. A second active monsoon spell is observed during the first half of August when convection is again active in the 10 20 N latitude belt. This is followed by a long break monsoon spell (please see Table 1) when the main area of convection is around the equator. Figure 9b shows the Hovmöller diagram of the 850-hPa zonal wind (U) averaged over the longitudes 80 100 E from latitudes 10 S to30 N of the period e 31 August, smoothed as in the case of the OLR by a 5-day moving average. The two active

1456 JOURNAL OF CLIMATE VOLUME 17 FIG. 10. Hovmöller diagram showing evolution of 850-hPa zonal wind speed from to 31 Aug 1979. Averaging is done (a) for the lon band 62.5 67.5 E and (b) for the lon band 77.5 82.5 E and a 5-day moving avg is applied as a smoother. Wind speed contours at 6 and more at intervals of 2 m s 1. spells are seen as maxima of zonal wind in the 10 20 N region. These wind maxima are found to lag in time behind the OLR maxima by a few days in agreement with the findings of section 5. The zone of maximum convection is on the cyclonic U-shear vorticity zone of the LLJ where the frictional convergence in the boundary layer produces upward motion to generate cumulonimbus cloud heating in the conditionally unstable tropical atmosphere. It is speculated that the dynamics (cyclonic vorticity and the consequent frictional convergence producing Ekman pumping of the moist boundary layer air) and the thermodynamics (the convective heating of the atmosphere and the consequent lowering of atmospheric pressure below) cooperate to increase convection and strengthen the LLJ, increment by increment. Because for continuity the whole LLJ has to strengthen, intensification of LLJ has to lag behind the convection by a few days. This is a kind of instability in which a planetary-scale system (the LLJ) cooperates with the convection in a synoptic-scale cloud cluster over the Bay of Bengal and both intensify. This phenomenon is similar to the conditional instability of the second kind (CISK) in the case of a synoptic-scale system like the tropical cyclone. Figure 10a shows the Hovmöller diagram of zonal wind (U) of 850 hpa averaged over the longitude band 62.5 67.5 E and smoothed by a 5-day moving average for the period e 31 August 1979. Active monsoon spells are characterized by strong cores of U, but whether it is active or break monsoon, the strongest U is at one latitude only, about 15 N. The intraseasonal oscillation at this longitude (65 E) is the weakening and strengthening of the LLJ core without north south movement. In contrast is a similar section through longitude 77.5 (75 80 E) shown in Fig. 10b. After the active monsoon spell of the second half of June, LLJ appears as two axes, one moving to latitude 25 N and the other toward the equator. The movement to 25 N is in response to the northward movement of the area of active convection. The movement of the other axis equatorward is likely to be by the mechanism suggested by Rodwell and Hoskins (1995) that in the absence of heat sources in the 10 20 N latitude belt, LLJ moves southeastward from the central Arabian Sea conserving its potential vorticity. It is speculated that when the axis of LLJ reaches near the equator, a zone of strong U shear with cyclonic vorticity forms around the equator, which leads to frictional convergence in the boundary layer and the generation of an east west band of convection there. It may be noted that Ekman pumping is very effective in low latitudes, more so in equatorial latitudes [e.g., Eq. (5.38) in Holton (1992)]. This convective band will then strengthen the LLJ and a zone of strong cyclonic shear vorticity appears close to the equator, which generates an east west band of convection that then moves north. Wind and OLR data of this period support this speculation regarding the genesis of the equatorial east west convective band. Figures 11a 11d give the mean OLR and 850-hPa wind fields corresponding to the first active and break spells of the monsoon in 1979. In this year the active and break spells are very long as can be seen in Table 1. As mentioned earlier it was a year of very pronounced intraseasonal oscillation in the monsoon and a year in which we have the best datasets thanks to the FGGE MONEX experiment. In the active spell the LLJ through India is very strong and it has only one axis at about 15 N latitude. In the break monsoon spell there are two active areas of convection, one from 10 S to10 N and the other around northeast India. There is very little convection in the latitude belt 10 20 N over south Asia. Monsoon models should at least be able to simulate the broad features of the active break cycle described in this paper, particularly of the LLJ and the associated convection. 7. Summary and conclusions Using the wind data from NCEP NCAR reanalysis and the NOAA OLR data, a detailed study has been made of the LLJ of the Asian summer monsoon. Vertical structure of the LLJ, the intraseasonal variability of the LLJ, particularly its relation to the active break cycle of the monsoon and the relation between convective

1APRIL 2004 JOSEPH AND SIJIKUMAR 1457 heating of the atmosphere over the Bay of Bengal and the LLJ over south Asia have been investigated. The following are the important conclusions. 1) The core of the cross equatorial LLJ crosses the equator in a geographically fixed narrow longitudinal belt close to the East African coast as a southerly current and it crosses India as a westerly current at latitudes varying from the equator to 25 N. In active monsoon conditions, the core of the LLJ passes through peninsular India around latitude 15 N. In break monsoon conditions the LLJ from the central Arabian Sea moves southeastward and passes eastward close to Sri Lanka in the latitude belt from the equator to 10 N. There is often seen at this time a weaker LLJ axis through north India around latitude 25 N. 2) LLJ does not show splitting into two branches over the Arabian sea as suggested by Findlater (1971). His suggestion, which is widely accepted since then, is based on the analysis of monthly mean winds. Such an analysis is likely to show the LLJ of active and break monsoons as occurring at the same time, suggesting a split of the LLJ over the Arabian Sea. Two branches of LLJ through India are however seen during break monsoon spells, but the northern branch is at around latitude 25 and not at about 17 N as found by Findlater. 3) Convective heating of the atmosphere over the Bay of Bengal has a high and significant linear correlation coefficient with the zonal component of the wind at 850 hpa over peninsular India (70 80 E) and the Bay of Bengal (80 100 E) all between latitudes 10 and 20 N. The correlation is maximum for a lag of 2 3 days, convection leading. It is speculated that active convection occurring over the Bay of Bengal between latitudes 10 and 20 N accelerates the whole interhemispheric LLJ and takes the monsoon to an active spell. LLJ should thus have a prominent place in numerical modeling studies of the monsoon. We should be able to simulate the LLJ correctly in each phase of the monsoon in order that models simulate realistic monsoon rainfall and its intraseasonal variability. Acknowledgments. The authors thank the Department of Atmospheric Sciences, Cochin University of Science and Technology for providing them data, computer facilities, and other support to do this research. They also thank the two anonymous referees and the editor for their valuable comments and suggestions. FIG. 11. Avg of (a) OLR and (b) 850-hPa zonal wind averaged for the first active monsoon spell of 1979 (23 Jun 2 Jul). Isolines of OLR at 220 and less at intervals of 10 W m 2 and of zonal wind at 6 and more at intervals of 2 m s 1. Avg of (c) OLR and (d) 850- hpa zonal wind averaged for the first break monsoon spell of 1979 (17 23 Jul). Isolines of OLR at 220 and less at intervals of 10 W m 2 and of zonal wind at 6 and more at intervals of 2 m s 1. REFERENCES Ananthakrishnan, R., and M. K. Soman, 1988: The onset of the southwest monsoon over Kerala 1901 1980. J. Climatol., 8, 283 296. Annamalai, H., J. M. Slingo, K. R. Sperber, and K. Hodges, 1999: The mean evolution and variability of the Asian summer mon-

1458 JOURNAL OF CLIMATE VOLUME 17 soon: Comparison of ECMWF and NCEP/NCAR reanalyses. Mon. Wea. Rev., 127, 1157 1186. Bunker, A. F., 1965: Interaction of the summer monsoon air with the Arabian Sea (preliminary analysis). Proc. Symp. Meteorological Results, Bombay, India, International Indian Ocean Expedition, 3 16. De, U. S., R. R. Lele, and J. C. Natu, 1998: Breaks in south west monsoon. Prepublished Scientific Rep. 1998/3, India Meteorological Department, Pune, India, 24 pp. Findlater, J., 1966: Cross-equatorail jet streams at low levels over Kenya. Meteor. Mag., 95, 353 364., 1967: Some further evidence of cross-equatorial jet streams at low levels over Kenya. Meteor. Mag., 96, 216 219., 1969a: Inter-hemispheric transport of air in the lower troposphere over the western Indian Ocean. Quart. J. Roy. Meteor. Soc., 95, 400 403., 1969b: A major low level current near the Indian Ocean during northern summer. Quart. J. Roy. Meteor. Soc., 95, 362 380., 1971: Mean monthly airflow at low levels over the western Indian Ocean. Geophys. Memo., 16, 1 53. Goswami, B. N., and R. S. Ajaya Mohan, 2001: Intraseasonal oscillations and interannual variability of the Indian summer monsoon. J. Climate, 14, 1180 1198. Gruber, A., and A. F. Krueger, 1984: The status of the NOAA outgoing longwave radiation data set. Bull. Amer. Meteor. Soc., 65, 958 962. Halpern, D., and P. M. Woiceshyn, 1999: Onset of the Somali jet in the Arabian Sea during June 1997. J. Geophys. Res., 104, 18 041 18 046., M. H. Freeilich, and R. A. Weller, 1998: Arabian Sea surface winds and ocean transports determined from ERS-1 scatterometer. J. Geophys. Res., 103, 7799 7805. Holton, J. R., 1992: An Introduction to Dynamic Meterology. Academic Press, 391 pp. Hoskins, B. J., and M. J. Rodwell, 1995: A model of the Asian summer monsoon. Part I: The global scale. J. Climate, 8, 1 12. Joseph, P. V., and P. L. Raman, 1966: Existence of low level westerly jet-stream over peninsular India during July. India J. Meteor. Geophys., 17, 407 410., J. K. Eischeid, and R. J. Pyle, 1994: Interannual variability of the onset of Indian summer monsoon and its association with atmospheric features, El Niño, and sea surface temperature anomalies. J. Climate, 7, 81 105. Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437 471. Krishnamurti, T. N., 1985: Summer monsoon experiment A review. Mon. Wea. Rev., 113, 1590 1629., and V. Wong, 1979: A simulation of cross equatorial flow over the Arabian Sea. J. Atmos. Sci., 36, 1895 1907., and J. Shukla, 2000: Intraseasonal and interannual variability of rainfall over India. J. Climate, 13, 4366 4377., J. Molinari, and H. L. Pan, 1976: Numerical simulation of the Somali jet. J. Atmos. Sci., 33, 2350 2362., V. Wong, H. L. Pan, R. Pasch, J. Molinari, and P. Ardanuy, 1983: A three-dimensional planetary boundary layer model for the Somali jet. J. Atmos. Sci., 40, 894 908. Murakami, T., R. Godbole, and R. R. Kelkar, 1970: Numerical simulation of the monsoon along 80 E. Proc. Conf. on the Summer Monsoon of South East Asia, Norfolk, VA, Navy Weather Research Facility, 39 51. Ramamurthy, K., 1969: Monsoon of India: Some aspects of break in the Indian south west monsoon during July and August. Forecasting Manual, Part IV. 18.3, India Meteorological Department, New Delhi, India, 13 pp. Rao, Y. P., 1976: Southwest Monsoon. Meteor. Monogr. Synoptic Meteor., No. 1/1976, India Meteorological Department, 376 pp. Rodwell, M. J., and B. J. Hoskins, 1995: A model of the Asian Summer Monsoon. Part II: Cross-equatorial flow and PV behavior. J. Atmos. Sci., 52, 1341 1356. Sikka, D. R., and S. Gadgil, 1980: On the maximum cloud zone and the ITCZ over Indian longitudes during the southwest monsoon. Mon. Wea. Rev., 108, 1840 1853. Soman, M. K., and K. Krishnakumar, 1993: Space time evolution of meteorological features associated with onset of Indian summer monsoon. Mon. Wea. Rev., 121, 1171 1194. Sperber, K. R., J. M. Slingo, and H. Annamalai, 2000: Predictability and the relationship between subseasonal and interannual variability during the Asian summer monsoon. Quart. J. Roy. Meteor. Soc., 126, 2545 2574. Webster, P. J., V. O. Magana, T. N. Palmer, J. Shukla, R. T. Tomas, M. Yanai, and T. Yasunari, 1998: Monsoons: Processes, predictability, and the prospects of prediction. J. Geophys. Res., 103 (C7), 14 451 14 510.