The impact of latent heating on the location, strength and structure of the Tropical Easterly Jet in the Community Atmosphere Model, version 3.1: Aqua-planet simulations
aa r X i v : . [ phy s i c s . a o - ph ] A ug The impact of latent heating on the location, strengthand structure of the Tropical Easterly Jet in theCommunity Atmosphere Model, version 3.1:Aqua-planet simulations
S. Rao , Department of Mechanical EngineeringIndian Institute of ScienceBangalore 560012, India Engineering Mechanics UnitJawaharlal Centre for Advanced Scientific ResearchBangalore 560064, IndiaEmail: [email protected], [email protected]: +919916675221July 2, 2018
Abstract
The Tropical Easterly Jet (TEJ) is a prominent atmospheric circulation featureobserved during the Asian Summer Monsoon (ASM). The simulation of TEJ by theCommunity Atmosphere Model, version 3.1 (CAM-3.1) has been discussed in detail.Although the simulated TEJ replicates many observed features of the jet, the jetmaximum is located too far to the west when compared to observation. Orographyhas minimal impact on the simulated TEJ hence indicating that latent heating is thecrucial parameter. A series of aqua-planet experiments with increasing complexitywas undertaken to understand the reasons for the extreme westward shift of theTEJ.The aqua-planet simulations show that a single heat source in the deep tropicsis inadequate to explain the structure of the observed TEJ. Equatorial heating isnecessary to impart a baroclinic structure and a realistic meridional structure. Jetzonal wind speeds are directly related to the magnitude of deep tropical heating.The location of peak zonal wind is influenced by off-equatorial heating which is losest to it. Hence the presence of excess rainfall in Saudi Arabia has been shownto be the primary reason for the extreme westward shift of the TEJ maximum. Keywords:
Tropical Easterly Jet Indian Summer Monsoon CAM-3.1 orographyaqua-planet precipitation
The Tropical Easterly Jet (TEJ) is one of the most defining features of the Indian SummerMonsoon (ISM) which itself is a part of the Asian summer monsoon (ASM). The jetis observed mostly during the ISM that is in the months of June to September. Ithas a maximum between 50 ◦ E-80 ◦ E, Equator-15 ◦ N and at 150hPa. The TEJ was firstdiscovered by Koteswaram (1958). It has a great influence on the rainfall in Africa(Hulme and Tosdevin, 1989). The correct simulation of the TEJ is important for accurateseasonal predictions and weather forecasting.The Tropical Easterly Jet is believed to be influenced by the Tibetan Plateau. Previousstudies (Flohn, 1968, Krishnamurti, 1971) have highlighted the presence of a huge uppertropospheric anticyclone above the Tibetan Plateau in summer. The origin of the Ti-betan anticyclone itself has been attributed to the summertime insolation on the TibetanPlateau. This sensible heating is widely regarded as the primary reason for this region toact as an elevated heat source. Flohn (1965) first suggested that in summer, southern andsoutheastern Tibet, i.e. south of 34 ◦ N-35 ◦ N, act as an elevated heat source which changesthe meridional temperature and pressure gradients and contributes significantly to thereversal of high tropospheric flow during early June. Flohn (1968) showed that sensibleheat source over the Tibetan Plateau as well as latent heat release due to monsoonal rainsover central and eastern Himalayas and their southern approaches generates a warm coreanticyclone in the upper troposphere at around 30 ◦ N. This was a primary mechanismfor establishing the south Asian monsoonal circulation over south Asia. According toKoteswaram these winds are a part of the Tibetan Anticyclone which forms during thesummer monsoon over South Asia. He believed that the southern flank of the anticy-clone preserved its angular momentum and became an easterly at approximately 15 ◦ N.Contradicting Koteswaram, Raghavan (1973) opined that the upper tropospheric zonalcomponent of the equatorward outflow from the Tibetan anticyclone does not agree withthe law of conservation of angular momentum. However the thermal gradient balancewas applicable and this was responsible for the origin of the TEJ.The TEJ is not simply a passive atmospheric phenomena. Hulme and Tosdevin (1989)studied the impact of the TEJ on Sudan rainfall. El Ni˜no events were suggested as a di-rect control over Sahelian rainfall via the TEJ. The decelerating limb of the TEJ showedinterannual variations in location in relation to Eastern Pacific warming. Fluctuationsin the jet were shown to be responsible for altering precipitation patterns in Sudan.2amberlin (1995) also reported on the significant linkages between interannual varia-tions of summer rainfall in Ethiopia-Sudan region and strength and latitudinal extentof upper-tropospheric easterlies. More recently Nicholson et al. (2007) showed that thewave activity of the TEJ influences the African easterly jet (AEJ). The AEJ in turninfluences the weather patterns over the Atlantic coast of Africa. Bordoni and Schneider(2008) discussed the role of TEJ in modifying the monsoonal circulation. Upper-leveleasterlies shield the lower-level cross-equatorial monsoonal flow (which later becomes theSomali jet) from extratropical eddies. This made the angular momentum conservationprinciple applicable to overturning Hadley cell dynamics. In the larger picture this im-plies a strengthening of the monsoonal regime. This shows that the TEJ actively shapesthe climate and weather pattern in regions not under the direct influence of the IndianSummer Monsoon.In fact the TEJ also shows intraseasonal and interdecadal variations in its location.Sathiyamoorthy et al. (2007) reported that the axis of the jet undergoes meridional move-ment in response to active and break phases of the Indian Monsoon. Sathiyamoorthy(2005) also observed a reduction in the spatial extent of the TEJ between 1960-1990. Ac-cording to him the TEJ almost disappeared over the Atlantic and African regions. Thisreduction coincided with the 4-5 decade prolonged drought conditions over the Sahel re-gion. He speculated that these two phenomena were associated with each other. Thisexplanation also implies that latent heating significantly influences the characteristics ofthe TEJ.Ye (1981) did laboratory experiments to simulate the heating effect of elevated land.He introduced heating in an ellipsoidal block resulting in vertical circulation, an anticy-clone in the upper layer and a cyclonic flow in the lower layer. The pattern was found tobe qualitatively similar to summer time atmospheric circulation in south Asia. Raghavan(1973) discussed the importance of Tibetan Plateau and explained that the jet owedits existence the the temperature gradients that was partly influenced by the Tibetanhigh. According to Wang (2006) the elevated heat source of the Tibetan Plateau wasinstrumental in providing an anchor to locate the Tibetan high. On the other handHoskins and Rodwell (1995) and Liu et al. (2007) have argued that orography plays asecondary role in determining the position of the summertime upper tropospheric anticy-clone. Jingxi and Yihui (1989) studied the TEJ at 200 hPa and found that precipitationchanges in the west coast of India led to changes in the jet structure. Chen and van Loon(1987) observed that at 200 hPa the jet was weaker during El Ni˜no and Indian droughtevents. The TEJ was weaker in the drought years of 1979, 1983 and 1987 and strongerin the excess monsoon years 1985 and 1988.According to Zhang et al. (2002), the South Asian High had a bimodal structure. Itstwo modes – one the Tibetan mode and the other the Iranian mode, both of which werefairly regular in their occurrence, were mostly influenced by heating effects. The former3wed its existence to diabatic heating of the Tibetan Plateau, while the latter occurreddue to adiabatic heating in the free atmosphere and diabatic heating near the surface.Thus it can be seen that while there have been attempts in the past to explain someaspects of the TEJ and its effects, the issue about the importance of orography on theTEJ has not yet been settled. In the previous discussions variations in the jet were inthe presence of the Himalayas and Tibetan Plateau. Although the importance of latentheating on the TEJ is quite clear now, it is necessary to confirm the importance oforography in deciding the location of the TEJ. It is necessary to study the jet in a GCMand understand the relative roles played by latent heating and orography on the TEJ.An Atmospheric General Circulation Model (AGCM), CAM-3.1, has been used to studythe impact of orography and monsoonal heat sources on the TEJ. The jet location andits response to have been studied by:(i) Modifying the deep-convective scheme and thereby changing the monsoonal heatingpattern.(ii) Removing orography.A series of aqua-planet experiments have been conducted determine the factors influencingthe location, structure and strength of the TEJ in CAM-3.1.
For wind data, NCEP (Kalnay et al., 1996) and ERA40 (Uppala et al., 2005) data havebeen used, and GPCP (Adler et al., 2003) and CMAP (Xie and Arkin, 1997) data forprecipitation. Unless otherwise mentioned the time period chosen spans years 1980 to2002. The maximum zonal winds occur at 150hPa. From Fig. 1 it is seen that the TEJpeaks in July to August. Hence, in present work the focus is on the TEJ during themonth of July when it first reaches its maximum value.Figure 1: Time series of 150 hPa zonal wind averaged between 50 ◦ E-80 ◦ E, Equator-15 ◦ Nshowing TEJ peaking in July-August (NCEP: 1980-2002 avg)4he magnitude of peak zonal wind and its location are listed in Table 1. Due to thegood agreement between zonal winds of NCEP and ERA40, only shown the former isshown. In Fig. 2a, the zonal wind and location of peak zonal wind, henceforth U max ,for NCEP at 150hPa, is shown. The ‘cross-diamond’ indicates the horizontal location of U max . The location of peak zonal wind of each individual year from 1980-2002 was foundand then the means were calculated. The standard deviation for the zonal winds forNCEP and ERA40 was found to be 2 m s − and 1.7 m s − respectively. The meridionalsection of the zonal wind is shown in Fig. 3a. This vertical cross-section corresponds tothe longitude where the zonal wind is maximum. The meridional structure shows the jetpeak lying between equator and ∼ ◦ N. A vertical equator to pole tilt towards higherpressure levels can be observed. The mean height of maximum easterly zonal wind speedsare higher at higher latitudes. In Fig. 4a the velocity vectors and geopotential high isshown. The peak is not over the Tibetan Plateau but to the west of it.Table 1: Magnitude ( m s − ) and location of peak zonal wind, and centroid of meanprecipitation reanalysis and observation Case Zonal wind PrecipitationPeak Lon Lat Press Lon LatNCEP 34.48 65.1 ◦ E 10.3 ◦ N 150ERA40 33.61 68.7 ◦ E 10.4 ◦ N 150GPCP 96.0 ◦ E 11.9 ◦ NCMAP 95.8 ◦ E 10.0 ◦ N CAM-3.1 simulations
Case Zonal wind PrecipitationPeak Lon Lat Press Lon LatCtrl 47.19 42.5 ◦ E 8.8 ◦ N 125 76.7 ◦ E 9.7 ◦ NnoGlOrog 43.45 43.5 ◦ E 7.6 ◦ N 125 77.0 ◦ E 7.4 ◦ NCtrl: default orographynoGlOrog: no orography
The TEJ is not a stationary entity. Variations in its position in different years can alsobe observed. This is most strikingly observed in July 1988 and 2002. The location of5 a) NCEP (150 hPa)(b) Ctrl (125 hPa)(c) noGlOrog (125 hPa)
Figure 2: July horizontal zonal wind profile at pressure level where maximum zonal windis attained, ‘cross-diamond’ is location of peak zonal wind: NCEP, Ctrl and noGlOrog.6 a) NCEP (b) NCEP(c) Ctrl (d) Ctrl(e) noGlOrog (f) noGlOrog
Figure 3: July cross-section of zonal wind profile at the longitude (left panel) and latitude(right panel) where U max is attained (see Table 1 for details): NCEP, Ctrl and noGlOrog.7 a) NCEP (b) Ctrl Figure 4: July velocity vectors and geopotential height ( × m) at 150 hPa. ‘Cross-diamond’ is location of peak geopotential height. Contour for NCEP and Ctrl is 14.36and 14.43 respectively which is 99.5% of individual peak.8he TEJ in during these two years is radically different. This is seen in Fig. 5a wherethe 30 m s − zonal wind contour and U max locations show significant differences. In factthe location of U max is shifted eastwards by ∼ ◦ in 2002. These differences are seen inERA40 data as well. The vertical structure in these two years is similar to Fig. 3a.It is well known that in India 2002 was a drought year (Sikka, 2003, Bhat, 2006)while 1988 was an excess monsoon year. The relationship between precipitation and thejet location in the month of July for these two years is now clarified. Here in order tobe self-consistent,both precipitation and zonal winds from NCEP data have been used.This is because reanalysis data is self-consistent and that the dynamic response of theatmosphere is entirely on account of forcing from the same data. If the TEJ is influencedlargely by latent heating due to Indian Summer Monsoon then the July 2002 shift shouldbe due to the mean rainfall being eastward shifted. This is clearly seen in Fig. 5b whereprecipitation differences between July 1988 and 2002 have been plotted. The differencesare quite striking. In July 2002 there is a clear eastward shift in rainfall with the maximumbeing in the Pacific warm pool and relatively little in the Indian region. This is in contrastwith July 1988 where Indian region received high amounts of precipitation. The jet in1988 peaks at 60 ◦ E-65 ◦ E while in 2002 the maximum is in the southern Indian peninsula,implying a ∼ ◦ westward shift. The same behaviour was also found for ERA40 dataand hence is not shown. The AGCM that has been used for the present work is the Community AtmosphereModel, version 3.1 (CAM-3.1). The finite-volume dynamical core using the recommended2 ◦ × ◦ grid resolution has been used for all simulations. The time step is 30 minutes and26 levels in the vertical are used. Deep convection is the Zhang and McFarlane (1995)scheme while shallow convection is the Hack (1994) scheme. Stratiform processes em-ploys the Rasch and Kristj´ansson (1998) scheme updated by Zhang et al. (2003). Cloudfraction is computed using a generalization of the scheme introduced by Slingo (1989).The shortwave radiation scheme employed is described in Briegleb (1992). The longwaveradiation scheme is from Ramanathan and Downey (1986). Land surface fluxes of mo-mentum, sensible heat, and latent heat are calculated from Monin-Obukhov similaritytheory applied to the surface. Climatological mean SST was specified as the boundarycondition. Sea surface temperatures are the blended products that combine the globalHadley Centre Sea Ice and Sea Surface Temperature (HadISST) dataset (Rayner et al.,2003) for years up to 1981 and Reynolds et al. (2002) dataset after 1981.The model was run in its default configuration for a five year period. This simulation9 a) NCEP zonal wind, 1988 (red), 2002 (blue)(b) NCEP precipitation difference: 1988–2002 Figure 5: (a) 30 m s − zonal wind contour (dashed) in 1988 and 2002 showing shift inTEJ in response to heating, cross-sections are at U max pressure level (150hPa), ‘cross-diamond’ shows the location of maximum zonal wind; (b) precipitation difference ( mmday − ), year 2002 subtracted from 1988. All data from NCEP.10s referred to as the control (Ctrl) simulation. Additionally another simulation (referredto as noGlOrog) has been conducted to check the influence of orography on the TEJ. Thishas also been run for five years with same boundary conditions but with orography allover the globe removed. This latter simulation is used to investigate the direct influenceof topography on the TEJ. All the simulation results presented in this paper are basedon five year means. The 5 year average of maximum zonal wind and its corresponding location have beencomputed. The horizontal (Figs. 2b,c) and meridional (Figs. 3c,e) profiles of the TEJ forCtrl and noGlOrog simulations are shown. As with reanalysis, the vertical cross-sectionis at the location of zonal wind maximum. The existence of a Tropical Easterly Jet canbe observed. In both the simulations the first noticeable feature is a ∼ ◦ westward shiftof the simulated jet. This shift in the default Ctrl was also documented by Hurrell et al.(2006) where they analyzed the 200hPa JJA zonal wind fields. The location of the peakzonal wind is virtually the same for Ctrl and noGlOrog simulations while the jet is weakerin the absence of orography. The zonal cross-section also shows that the zero line at500 hPa is sandwiched between the peaks in the TEJ and low-level Somali jet. Onemajor difference appears in the Ctrl and noGlOrog simulation - the maximum in theSomali jet is greater and restricted east of ∼ ◦ E in the former. This was explained byChakraborty et al. (2008). The absence of orography in the latter caused the low-levelwind to spread out and thereby reduce the intensity without compromising on the totalflow. This is also seen in NCEP data (Fig. 3b) where the Somali jet is also maximum tothe east of ∼ ◦ E. The meridional cross-section is very similar to real-planet and showsthe familiar vertical equator to pole tilt with the maximum lying between equator and ∼ ◦ N. The depth is also maximum on the poleward side.The impact of orography in determining the location and spatial structure of the jetin CAM-3.1 is thus minimal. The location of the peak zonal wind is virtually the samefor Ctrl and noGlOrog simulations while the jet is weaker in the absence of orography.Although CAM-3.1 does show reasonable fidelity in determining the spatial features of theTEJ, the discrepancies in location and magnitude of the maximum velocity of the TEJneed to be understood. With the insight gained from section 2.1 the spatial distributionof simulated and observed rainfall is now studied.
The westward shift of the simulated TEJ indicates that sensible heating from the TibetanPlateau may not play a major role in the existence and location of TEJ. During the11onsoon season the major source of heating in the tropics is latent heating and hence itis necessary to look at the role of latent heating. (a) Precipitation: GPCP (b) Precipitation: CMAP(c) Precipitation: Ctrl (d) Precipitation difference
Figure 6: (a), (b), (c) July precipitation ( mm day − ). (d) July precipitation difference( mm day − ) (Ctrl–(GPCP+CMAP)/2).Figs. 6a-c show the precipitation in the month of July of GPCP, CMAP and Ctrl.The difference between Ctrl and mean of GPCP and CMAP is shown in Fig. 6d. Theobservational data have been averaged since the precipitation patterns are very similar.The contrast between reanalysis and model simulations is quite striking. Most noticeablediscrepancies in model simulations are (i) significantly reduced precipitation in northernBay of Bengal, East Asia, western Pacific warm pool (ii) a significant precipitation tonguejust south of the equator between 50 ◦ E-100 ◦ E and (iii) Spurious precipitation in the SaudiArabian region which is quite prominent and equal in magnitude to the precipitationpeaks in central Arabian Sea and south-western Bay of Bengal. This implies a majorrealignment in the local heating pattern. This unrealistic precipitation has been discussedby Hurrell et al. (2006).Thus from Table 1 and Fig. 6 it appears that westward shift in the peak of the sim-ulated precipitation is responsible for the TEJ to be centered east Africa. The strongestimpact seems to be from the significantly high anomalous precipitation in the Saudi Ara-bian region which could play a part in the extreme westward shift of the TEJ.As in Kucharski et al. (2009) and Davis et al. (2012), precipitation is used as a proxyfor latent heating. The centroid of precipitation ( P c ) has been computed for in a regionthat is spatially quite significant and covers the monsoon region. Different regions for12eanalysis and CAM-3.1 simulations have been chosen since the precipitation pattern isspatially different. For reanalysis the west Pacific warm pool is also considered, while itis ignored for model simulations. The region chosen for reanalysis is 60 ◦ E-130 ◦ E, 16 ◦ S-36 ◦ N and for the simulations it is 40 ◦ E-110 ◦ E, 16 ◦ S-36 ◦ N (region demarcated by redboxes in Figs. 6a-c). From Table 1 it is seen that the mean precipitation in simulationsand reanalysis are in reasonable agreement although Ctrl experiment overestimated theprecipitation by ∼ P c , is computed using the following equation: x c = P i P i x i P i P i , y c = P i P i y i P i P i (1)where, x c and y c are the zonal and meridional coordinates of the P c , P i is the precipitation at each grid point, x i and y i are the zonal and meridional distances from a fixed coordinate system, in eachcase the grid point where peak precipitation occurs.The distances are measured from a coordinate system centered at a point where theprecipitation is maximum in the chosen region. There is a clear ∼ ◦ westward shiftin the precipitation centroid in Ctrl with respect to reanalysis. Precipitation pattern ofnoGlOrog is similar to Ctrl and hence even in this case the shift is ∼ ◦ westward. Thelatitudinal differences are not significant. Choosing slightly different averaging regionsdoes not significantly distort this relationship.Consequent to the westward shift in the precipitation centroid and TEJ, the peakgeopotential height at 150hPa in the simulations has also shifted westwards by ∼ ◦ .This is clearly seen in Fig. 4 where the geopotential contour corresponds to 99.5% ofthe peak, the location of which is indicated by the cross-diamond. The velocity vectorsclearly show the anti-cyclonic flow around the geopotential high.Though this offers evidence of the primacy of latent heating in determining the lo-cation of the jet, the validity of this argument is further demonstrated by showing theprecipitation patterns of the Ctrl τ d and noGlOrog τ d simulations. The mean precipi-tation and centroid are shown in Table 1. Since the precipitation pattern is now similarto real-planet (not shown), the same region as in real-planet has been used to computethe location of the precipitation centroid. Figs. 7a,b show the precipitation and TEJ inthese two simulations. The location of the TEJ agrees well with real-planet. Althoughthe mean precipitation is more in comparison to the previous simulations, the peak zonalwind speed is almost the same. The geopotential high (Fig. 7c) in the Ctrl τ d simula-tion is now almost same as that in NCEP (Fig. 4a). The spatial separation between theprecipitation centroid and U max is included in Fig. 8. The zonal separation is also incloser agreement to real-planet and this is may be attributed to the increased similarityin spatial pattern of precipitation in real-planet and these two simulations. The similarity13n the precipitation patterns and TEJ structures in both these simulations as well as thepreviously discussed Ctrl and noGlOrog simulations suggests that it is the location andstrength of the heat source and not orography that controls the TEJ. The only differencein precipitation between the Ctrl τ d and noGlOrog τ d simulations is that the formerhas additional heating to the north of 20 ◦ N around 90 ◦ E longitude. The absence of thispeak in the latter further suggests that the Tibetan Plateau is not the primary reason toset up the temperature gradients that lead to the formation of the TEJ. Sensible heatingdue to the presence of Tibetan Plateau may be less important when dominated by latentheat release due to convective processes.Thus it has been demonstrated that there is a close correspondence between the sourceof heating, spatial location of TEJ and the geopotential high. The negligible influenceof orography and strong effect of heating on upper level wind patterns has also beendiscussed by Liu et al. (2007). It is also worthwhile to note that the Somali jet in NCEPand Ctrl simulation is roughly in the same location and hence the lower and upper-levelwinds (Figs. 3b,d) are not merely reflections of each other, not just in magnitude but alsoin location, as simple though insightful models, e.g. Gill (1980) suggest. However thespatial location of the low-level jet and upper-level TEJ is more closely correlated in thenoGlOrog simulation and this is due to the absence of orography as will also be shown inthe subsequent discussion on aqua-planet simulations.However this mean pattern does not present the full picture. Between 70 ◦ E-100 ◦ Ethe precipitation peak in real-planet is more northwards in comparison to CAM-3.1 whilethe opposite is true between 30 ◦ E-60 ◦ E. In fact real-planet hardly shows rainfall in thelatter region. Hence it is not clear how the spatial structure of the precipitation influencesthe positioning of the TEJ. This aspect will need to be clarified before further analysisand this will be the focus of the next section where a major simplification is adopted byrunning CAM-3.1 in the aqua-planet (AP) mode.
Since orography has hardly any impact on the TEJ in the real-planet simulations it isinstructive to study the atmospheric response only due to heating. The multiplicity ofheat sources both in reanalysis and Ctrl preclude any easy interpretation of the influencethat each heat source has on the TEJ. A major simplification is conceivable if one removesorography, land, sea-ice totally and further remove seasonal cycles, and yet retain all theimportant physics that the AGCM offers. This also implies that one needs to prescribeSSTs as a boundary condition. All this is possible in the aqua-planet configuration ofCAM-3.1. The solar insolation is perpetually fixed at 21 st March which is March Equinox.In order to understand the role played by heating in determining the structure andlocation, CAM-3.1 has been run in aqua-planet configuration. The basic state consists14 a) Ctrl τ d (125 hPa)(b) noGlOrog τ d (125 hPa)(c) Ctrl τ d (150 hPa) Figure 7: (a), (b) July precipitation ( mm day − , shaded) and horizontal zonal wind ( ms − ) profile at pressure level where maximum zonal wind is attained; ‘cross-diamond’ islocation of peak zonal wind, (c) velocity vectors and geopotential height ( × m) at 150hPa; ‘cross-diamond’ is location of peak geopotential height; contour value is 14.46 whichis 99.5% of peak: Ctrl τ d and noGlOrog τ d .15igure 8: Separation ( P c – U l ) between location of precipitation centroid and maximumzonal wind. ‘Cross-diamond’ denotes aqua-planet simulations, ‘plus’ denotes real-planetsimulations and reanalysis.of a uniform background SST on which additional heat sources are imposed. The heatsources are indirectly specified by setting an SST perturbation on this uniform SST back-ground. Precipitation induced on account of this SST perturbation is representative oftotal atmospheric heating. The implicit assumption is that latent heating is the dominanteffect in the region of precipitation.The rationale for imposing heat sources on a uniform SST background in contrast to azonally symmetric and meridionally varying SST profile is now explained. Rajendran et al.(2013) showed that equatorial easterlies will be simulated even if a zonally symmetric butmeridionally varying SST profile, symmetric about the equator, is used. The existenceof an equatorial jet also depends on the presence of twin or single Inter-Tropical Conver-gence Zone (Rajendran et al. (2013), Neale and Hoskins (2001)). In such a meridionallyvarying SST profile it is more difficult to determine the role played by weak heat sourcesin the formation of a jet in aqua-planet.The aqua-planet simulations with just uniform background SSTs of 20 ◦ C and 25 ◦ Chad a weak equatorial easterly of about 10-15 m s − peaking at ∼ ◦ N/ ◦ S do not significantlyinfluence equatorial dynamics. This was also suggested by Hoskins and Rodwell (1995)when they conducted a series of experiments to understand the Asian summer monsoon.Thus a uniform background SST is deemed to be the simplest basic state on whichadditional heating may be imposed to study the influence of heating on the TEJ.The reasons for the existence of these easterlies is as follows: these simulations hada band of weak precipitation near the equator. This would naturally imply that moistparcels arise aloft implying the presence of a weak Hadley Cell. This means that air ismixed latitudinally. Therefore there must be westerly motion at higher latitudes andeasterly motion at lower latitudes owing to angular momentum conservation. The time-averaged torque on the whole atmosphere due to surface friction must be zero, which16equires that there be both easterly and westerly winds. Thus easterlies must prevailnear the equator. Drag on surface easterlies also transfers angular momentum from thesurface to the atmosphere (Schneider (2006)). The issue of the existence of equatorialupper tropospheric easterlies has also been discussed by Lee (1999). According to her,in the deep tropics the horizontal transient eddy momentum flux accelerates the zonalmean zonal wind. Transient eddies of intraannual and interannual timescales were im-portant determining factors. Hadley cell dynamics was also important. According toBordoni and Schneider (2008) the Hadley cell approaches the angular momentum conser-vation limit because of these upper-level easterlies (upper-level easterlies strengthen whenadditional off-equatorial heating is present).The details of all the aqua-planet simulations are in Tables 2 and 3. In Fig. 9, thedifferent locations where SSTs are imposed are shown. The names of the aqua-planetsimulations start with ‘AP’. The magnitude of peak zonal wind and mean precipitationare listed in Table 3. The important point to note is that the SSTs have been chosensuch that mean of precipitation exceeding 5 mm day − compares well with reanalysisand real-planet simulations. Fig. 8 shows the zonal and meridional distance between P c and U max . Equation (1) has been used to compute the centroid. The zonal windfor each month has not been individually computed and then averaged; the zonal findfields have been added and then averaged for the six month period. Hence these locationscorrespond to grid point values of the model. This method is acceptable since the heatsource is stationary and the response too will average out over a six month time scale.The coarse model resolution implies that minor fluctuations in location will hardly distortthe main observations and inferences. The vertical sections at the longitude and latitudewhere U max is attained. Some representative cases are discussed.Figure 9: Schematic showing locations of imposed SSTs for aqua-planet simulations.Refer Table 2 for further details. (Not drawn to scale).17able 2: Location and nomenclature of SST profiles for aqua-planet simulations. Seetable 3 for full nomenclature. Figure 9 shows the schematic of the locations.Region Shape of SST profile Location of SST peakS Circle (C) 20 ◦ diameter 90 ◦ E,20 ◦ NOval (Oa) 50 ◦ major axis16 ◦ minor axisOval (Ob) 90 ◦ major axis16 ◦ minor axisE Rectangle 10 ◦ slope 62 ◦ E,4 ◦ S to 94 ◦ E,2 ◦ SCircle 20 ◦ diameter 60 ◦ E,4 ◦ NB Circle 20 ◦ diameter 50 ◦ E,20 ◦ NC 70 ◦ E,14 ◦ ND 85 ◦ E,10 ◦ NP Oval As in (Ob) above 120 ◦ E,20 ◦ N150 ◦ E,10 ◦ N130 ◦ E,2 ◦ S For single heat source simulations, the SST profiles imposed have circular and oval shapesall centered at 90 ◦ E,20 ◦ N which mimic the off-equatorial monsoonal heat source in thenorthern Bay of Bengal. The circular profiles (AP S1 and AP S2) have a diameter of20 ◦ while the oval shaped profiles have major axes of either 50 ◦ (AP S3 and AP S4) or90 ◦ (AP S5 and AP S6). All oval profiles have a minor axis of 16 ◦ . The simulationswith oval-shaped SSTs serve to demonstrate the effect of the shape of the heating regionon the jet. The peak SSTs are at 90 ◦ E,20 ◦ N. They go linearly down to the backgroundtemperature. However the profile with major axes of 90 ◦ (AP S5 and AP S6) has a non-linear SST gradient which is used to study if the gradient has any qualitative change onthe jet. The background temperature is 25 ◦ C. In each set, the first simulation in each sethas 29 ◦ C peak SST while the second has 32 ◦ C peak SST. For example, AP S3 has 29 ◦ Cand AP S4 has 32 ◦ C SST peak on a uniform background temperature of 25 ◦ C.The meridional and zonal cross-sections (Figs. 10a,b) are shown for AP S6 simulation.Meridionally, the jet is not symmetric. The major difference with reanalysis and real-planet simulations lies in the vertical shape of the jet. With single off-equatorial heatsource the jet has a pole to equator tilt towards higher pressure levels. This is oppositeto reanalysis and real-planet (Figs. 3a,c). Below 500 hPa, there is also a region of weaklow-level jet which is the aqua-planet counterpart of the Somali jet (not shown). Further,18able 3: Magnitude ( m s − ) and location of peak zonal wind, and magnitude ( mmday − ) and centroid of mean precipitation for aqua-planet simulations. Refer Fig. 9 andtable 2 for shape and location of SST profiles. Single heat source (at region S) simulations
Case SST Zonal wind PrecipitationPeak Shape Peak Lon Lat Press Mean Lon LatAP S1 29 ◦ C Circle (C) 23.64 70 ◦ E 6 ◦ N 175 8.81 90.2 ◦ E 21.9 ◦ NAP S2 32 ◦ C 33.97 80 ◦ E 10 ◦ N 150 12.41 90.5 ◦ E 21.6 ◦ NAP S3 29 ◦ C Oval (Oa) 30.75 65 ◦ E 6 ◦ N 175 9.24 88.8 ◦ E 20.7 ◦ NAP S4 32 ◦ C 37.98 67.5 ◦ E 6 ◦ N 150 14.74 89.4 ◦ E 21.4 ◦ NAP S5 29 ◦ C Oval (Ob) 35.58 45 ◦ E 4 ◦ N 175 10.17 86.3 ◦ E 20.5 ◦ NAP S6 32 ◦ C 42.56 40 ◦ E 4 ◦ N 150 15.06 88.6 ◦ E 20.1 ◦ NUniform background temperature: 25 ◦ C Multiple heat source simulations
Case SST region Zonal wind PrecipitationPeak Lon Lat Press Mean Lon LatAP E1 E 15.54 52.5 ◦ E 2 ◦ N 225 11.52 74.2 ◦ E 3.6 ◦ SAP E2 17.58 37.5 ◦ E 6 ◦ N 225 11.49 72.5 ◦ E 2.4 ◦ SAP N1 B,C,D 33.08 37.5 ◦ E 10 ◦ N 150 9.02 67.7 ◦ E 16.7 ◦ NAP N2 30.41 57.5 ◦ E 6 ◦ N 175 9.27 72.3 ◦ E 16.2 ◦ NAP NE1 B,E 26.73 37.5 ◦ E 8 ◦ N 175 10.37 67.0 ◦ E 2.7 ◦ NAP NE2 23.80 42.5 ◦ E 14 ◦ N 150 7.51 67.3 ◦ E 3.1 ◦ NAP M1 C,D,E 30.89 55 ◦ E 6 ◦ N 125 11.40 70.8 ◦ E 5.4 ◦ NAP M2 24.91 52.5 ◦ E 4 ◦ N 150 11.62 71.6 ◦ E 2.6 ◦ NAP M3 B,C,D,E 24.31 40 ◦ E 10 ◦ N 150 9.03 71.9 ◦ E 2.3 ◦ NAP M4 36.02 35 ◦ E 8 ◦ N 150 9.03 68.7 ◦ E 5.4 ◦ NAP M5 34.43 32.5 ◦ E 6 ◦ N 150 10.86 69.2 ◦ E 4.9 ◦ NAP M6 B,C,D,E,P 35.70 32.5 ◦ E 8 ◦ N 150 10.22 68.8 ◦ E 3.1 ◦ NUniform background temperature: 22 ◦ C Peak temperatures: ≤ ◦ C19he low-level westerly was observed to be either very weak or almost non-existent for thosesimulations with reduced spatial extent of heating and/or lesser overall heating rates, forexample AP S2 and AP S5.To the east of the jet, there is an upper-level westerly intrusion. A hint of this westerlyis also observed in the reanalysis and real-planet simulations (Fig. 3, right panel). In Fig.10c, the precipitation region is demarcated by the 5 mm day − and above shaded contours.It can be seen that precipitation (also refer Table 3) for the AP S6 case has very high rateswhich is unrealistic. For comparatively low heating rates, a baroclinic structure similarto reanalysis and real-planet simulations does not exist. This only suggests that for suchcases, the easterlies developed are more of an add-on to the equatorial easterlies developedwith uniform background SSTs. However even low heating more than doubles the peakzonal winds generated in comparison to only uniform SST everywhere. This shows thateven relatively higher zonal easterly wind speeds may not create vertical structures thathave any resemblance with reanalysis and real-planet cases. So one cannot understand thestructure of the TEJ with a single heat source. In all the single heat-source simulationsthe easterlies were present around the entire tropics. In each of the circular and oval sets,increased heating due to increased SSTs also increases the zonal wind speeds. Points Aand B in Fig. 8 are for AP S2 and AP S6.The first thing to note is that compared to the change in mean precipitation mag-nitudes, the zonal separation differs significantly. The precipitation centroid also variesby less than ∼ ◦ . The maximum separation is for the 90 ◦ major axis oval-shaped SST(AP S6) while the least is for the circular SST profile with 32 ◦ C peak (AP S2). Thisdemonstrates that more zonally constricted heating reduces the zonal separation between P c and U max . When the spatial extent of heating is less, the zonal wind has relativelyrapid zonal acceleration and a slower zonal deceleration. Intense convergence in a rela-tively small region and subsequent south-easterly and then almost easterly movement ofthe air mass will cause zonally rapid acceleration and slow deceleration. Mishra (1987)attributed the southward movement of the jet to the spherical geometry of the earth.Non-linearity was also responsible for causing the jet to shift towards the equator (Mishra(1993)). Increased spatial extent of heating causes increased acceleration length as is alsoevident from Table 3. It appears that spatially extensive heating imparts greater energyto the mean tropical flows, since higher zonal winds exist over longer zonal lengths. Thisis an important point and will be stressed upon in the discussion where multiple heatsources are introduced. The jet is thus to the south-west of the peak heating which is adeparture from reanalysis and real-planet simulations.The location of the geopotential contour is very near to the region of maximum precip-itation. This is clearly seen in Fig. 10d. Between 70 ◦ E-110 ◦ E the meridional velocitiesare also significant. This is a major departure from real-planet simulations where thecontour is very much to the west of the region of precipitation.20 a) Meridional cross-section (b) Zonal cross-section(c) Horizontal cross-section: (150 hPa) (d) Velocity vectors and geopotential height Figure 10: Zonal wind ( m s − ) profile. Vertical sections are at the (a) longitude and(b) latitude where U max is attained; (c) horizontal cross-section at pressure level where U max is attained, ‘cross-diamond’ is location of U max , ‘star’ is location of P c , precipitationcontours ( mm day − ) are shaded; (d) velocity vectors and geopotential height ( × m)at 150 hPa; ‘cross-diamond’ is location of peak geopotential height; contour value is 14.09which is 99.5% of peak: AP S6 (Aqua-planet with single off-equatorial heat source inregion S shown in Fig. 9). 21dditional simulations were also done with SST peaks at 90 ◦ E,10 ◦ N. The uniformbackground temperatures were 20 ◦ C and 25 ◦ C. The magnitudes of the SST peaks rangedfrom 23 ◦ C to 37 ◦ C. For heating at 10 ◦ N, zonal wind speeds remained approximately ∼
20 m s − and did not show any increase even with higher heating rates. In contrast,for heating at 20 ◦ N, zonal winds showed a monotonic increase with heating. The windspeeds increased from ∼
20 m s − to ∼
50 m s − as heating increased. Thus heating inlower latitudes does not result in increase in jet zonal velocities. The major finding of the previous section is that a single off-equatorial heat source isinadequate in explaining the structure of either the reanalysis TEJ or the full AGCMTEJ. Hence it is necessary to study the impact of combinations of heat sources. Fig. 9shows the regions where precipitation exists in Ctrl simulation. An additional advantagethat Ctrl simulation has more precipitating regions than reanalysis in the region of interestand so it is more beneficial to understand these heat sources in determining the locationand strength of the TEJ. If these aqua-planet simulations have a jet structure resemblingthe TEJ in Ctrl or noGlOrog simulation in location as well as in magnitude, then it is animportant step forward. This is because it can be then claimed that the location of theTEJ as largely a construct of the location of heating and not on any complicated land-ocean interactions. This section is an attempt to understand the interactions betweenthe various heating and how they interact to produce jet-like structures.The names AP N1 to AP M6 are for these cases. The locations and peak SSTsare given in Tables 2, 3 (also refer Fig. 9). As explained before, in regions of overlapthe maximum SSTs are chosen which removes any kinks in the profiles. The uniformbackground temperature is 22 ◦ C. Though it would have been more realistic to incorporatea 25 ◦ C background temperature, it was thought that 22 ◦ C would enable us to clearlydecipher the role the different zones play as elsewhere precipitation would be less whileat the same time 22 ◦ C background temperature would allow for minimal convection tooccur. The peak SSTs again do not exceed 29 ◦ C and in most cases are less than that.In all cases, SSTs linearly increase to the peak value. Although only a few simulationsof each type have been listed, more such simulations have been conducted with differentSST peaks in the same locations in order to check the validity of the assertions made.They are not presented for want of space.
These cases are AP N1 and AP N2 which have heating in regions B, C and D as shown inFig. 9. This set has no equatorial heating. Since the heating is mostly off-equatorial, the22eridional and zonal structures are similar to those with single heat source at 90 ◦ E,20 ◦ Nand hence are not shown. The increased heating caused a low-level westerly in bothcases. The interesting point to note from Table 3 and Fig. 11 is the ∼ ◦ westward shiftof U max in the 1 st simulation (Fig. 11a) in comparison to the 2 nd simulation (Fig. 11a).The zonal separation between P c and U max (Fig. 8) also shows a similar difference. Theprecipitation region in these two simulations shows that in the second case, precipitationregion B is less than 10 mm day − . This shows the importance of the heating that isclosest to the peak zonal wind location. When the westernmost heating is below a certainthreshold, the jet is influenced by the next closest heating, in this case the one in regionC in Fig. 9. Simulations with heating only in regions C and D have also been conducted(not shown). In these cases, the zonal separation between P c and U max is about 15 ◦ less compared to AP N1 and AP N2. In other words, additional heating at locationB increases this separation. This could be one of the reasons why Ctrl and noGlOrogsimulations show a westward shift in the TEJ relative to reanalysis. In the real-planetsimulations, there is significant precipitation in the Saudi Arabian region. (a) AP N1 (150 hPa) (b) AP N2 (175 hPa) Figure 11: Shift in location of U max in response to locations of heating. Zonal wind (m s − ) profile at pressure level where maximum zonal wind is attained. Precipitationcontours ( mm day − ) are shaded, ‘cross-diamond’ indicates location of U max , ‘star’ islocation of P c . Aqua-planet with multiple off-equatorial heat sources in regions B, C andD shown in Fig. 9. Referring to Fig. 6c one can see that in the Ctrl simulation, during July, there is a broadprecipitation zone just south of the Equator between 50 ◦ E-100 ◦ E and a zone just north ofthis zonal tongue centered at 60 ◦ E. Our goal is to induce similar precipitation patterns byimposing SST peaks in this region. Two such simulations have been conducted which arelabelled as AP E1 and AP E2. The details are given in Tables 2, 3 and Fig. 9. Heatingis present only in region E. Whenever there is an overlap, the maximum temperature is23sed which ensures that there are no kinks in the SST profiles. However the backgroundSST is now 22 ◦ C. The rationale for this choice will be explained in the next subsectionwhere multiple-heat source simulations are discussed. Both rectangular and circular SSTprofiles have been imposed. The peak SSTs do not exceed 29 ◦ C and are made to increaselinearly to the peak value. (a) AP E2: Meridional section (b) AP E2: Horizontal section: (225 hPa)
Figure 12: Zonal wind ( m s − ) profile. (a) meridional section is at the longitude where U max is attained; (b) horizontal cross-section at pressure level where U max is attained,‘cross-diamond’ is location of U max , ‘star’ is location of P c , precipitation contours ( mmday − ) are shaded: Aqua-planet with multiple heat sources in region E shown in Fig. 9.From Table 3 it can be seen that the zonal wind speeds barely qualify as a jet. Thespeeds are hardly more than one-and-a-half times the easterly obtained for a uniformbackground simulation. The zonal wind structure is shown in Fig. 12. The location ofpeak zonal wind is at a lower height (below 200 hPa). The most important observation isthe stark contrast in the meridional structure vis-a-vis the off-equatorial heating. Fromthe meridional section (Fig. 12a), it is seen that the equator to pole tilt is towards higherpressure levels as in Ctrl and reanalysis. A low-level westerly of comparable strengthextending below 500 hPa was also observed. Thus with just equatorial heating similar toreal-planet, the vertical structure becomes baroclinic. Compared to reanalysis and real-planet there is a significant westerly to the east of the upper-level easterly. This signifiesthe reduced zonal extent of the easterly flows in comparison to single heat-source cases.But the horizontal structure (Fig. 12b) immediately shows that the zonal wind struc-ture is actually nowhere near the real-planet TEJ. The peak zonal wind was observedto be to the west of 90 ◦ W and hence very far from P c . This far-off value has not beentabulated since the region chosen for locating the peak zonal wind is 0-90 ◦ E. In spite ofthe maximum heating being in the south of the equator, the easterly is in the northernhemisphere. All these show that near-equatorial heating is necessary for imparting a baro-clinic structure somewhat resembling reality. But stand-alone near-equatorial heating isby itself insufficient to generate a jet structure that resembles the TEJ.24 a) AP NE1: Meridional section (b) AP NE1: Horizontal section: (175 hPa)(c) AP NE2: Meridional section (d) AP NE2: Horizontal section: (150 hPa)
Figure 13: Zonal wind ( m s − ) profile. (a),(c) meridional sections are at the longitudewhere U max is attained; (b),(d) horizontal cross-section at pressure level where U max is attained, ‘cross-diamond’ is location of U max , ‘star’ is location of P c ; precipitationcontours ( mm day − ) are shaded: Aqua-planet with multiple heat sources in regions Band E shown in Fig. 9. 25 .2.3 Interplay between heating in near-Equator and Saudi Arabian regions It is interesting to study the effect of heating in region B in addition to region E discussedabove. These are the AP NE1 and AP NE2 simulations. It can be seen from Tables 1and 3, that in these simulations the location of U max is closer to Ctrl simulation.The meridional structure (Fig. 13a) is similar to the equatorial heating case. Thereis also the equator to pole tilt that is observed in real-planet and real-planet cases. Themeridional structures show the familiar equator to pole tilt (as in reanalysis) till theprecipitation in region B (Saudi Arabia) does not increase beyond a threshold. Thisincrease can be inferred from Fig. 13c. Even then the depth is moderated by the equatorialheating effects. The change in tilt is due to precipitation being more in region B as canbe seen from Fig. 13d in comparison to Fig. 13b. Here the equatorial heating in thelatter is quite low and yet heating in B is not enough to increase the tilt in comparisonto similar heating rates in the single heat source at 20 ◦ N).This shows that heating near the equator is essential in creating a meridional structureresembling reanalysis and real-planet.
The discussion on the influence of and near-equatorial and off-equatorial heating can nowbe used to combine all of them and study how close the net effect is on the Ctrl simulation.These are the AP M1 to AP M5 simulations (refer Table 3). In this effort, once againthe role of the easternmost heating on the jet structure and location is studied. The onlydifference between AP M1 and AP M2, and AP M3 to AP M5 simulations is the presenceof heating in region B in the latter set. Heating in region D is expected to have minimalrole to play as it is not the easternmost source.The first and foremost difference, as observed from Table 3, is in the zonal location of U max . Precipitation in region B, as expected, causes a 15 ◦ -20 ◦ westward shift in the jetlocation while the location of P c is hardly changed. This is clearly seen from Fig. 8 wherepoints G and H are for AP M1 and AP M5 simulations. The zonal separation withoutheating in region B in the former causes the location of U max to be much closer to P c .The jet zonal velocities are generally greater and this is because of heating in regions B,Cand D (see section 5.2.3).In all cases, the other zonal wind structures are similar in all cases (except for thedifference in location of U max when heating in region B is present). Hence in Fig. 14only AP M5 simulation is shown. The upper-level meridional structure of the jet (Fig.14a) resembles real-planet simulations (Fig. 3, left panel). The equator to pole tilt isalways towards higher pressure levels. This demonstrates that in aqua-planet configura-tion, precipitation patterns similar to real-planet simulations result in meridional shapessimilar to full AGCM simulations. The zonal jet lengths are still less than real-planet26imulations.The eastward extent of the upper-level westerly intrusion is also unrealistic.This discrepancy will be addressed in section 5.2.5 below.The relatively weaker and more westward location of peak zonal wind speed in AP M3simulation compared to AP M5 is because the precipitation in region B in AP M3 wasmuch lower in comparison to AP M5 (not shown). This once again underscores theinfluence of heating in region B (Saudi Arabia) in distorting the location of the jet in theCtrl simulation. In spite of lower heating in region B in AP M3 simulation, the meridionalstructure has the familiar equator to pole tilt (not shown). Except for the westerly tothe east of 90 ◦ E, the overall TEJ simulated has a structure similar to Ctrl (Figs. 2b and3c,d).The location of the geopotential high (Fig. 14d) is now bears resemblance to Ctrlsimulation (Fig. 4b). Although this peak in AP M5 is shifted more eastwards in com-parison, from Tables 1 and 3 it is seen that the location of U max as well as P c is alsowestwards in comparison. On closer inspection, from Fig. 8 (points J and H) it can beobserved that the zonal and meridional separation between P c and U max almost the same.Thus in the aqua-planet simulations if all the SST profiles had been shifted westwards by ∼ ◦ , then the mosaic of precipitation, zonal wind and geopotential heights would haveshifted westwards by about the same amount. Then there would be more similarity withreal-planet simulations.It is also interesting to observe that the jet velocities increase by only ∼ − whencompared with simulations with just an additional source at region B. In the previoussingle heat source simulations, the jet velocities increased when the intensity of heatingincreased both spatially and magnitude wise. Here, additional the heat source at 20 ◦ Ndoes not significantly increase the jet strength. This shows that when multiple, well-distributed heat sources resembling reality are present, it is the intensity of heating andnot the number of heating zones that determines the zonal wind strength.
In all the above multiple heat source simulations, the greatest mismatch in zonal windstructure with the real-planet simulations is the westerly that was always present to thewest of 90 ◦ E. This westerly was prominently reflected in all the zonal cross-sections.From Fig. 6 it is observed that in reanalysis there is significant precipitation occurringeast of 100 ◦ E. Although in this region, in both Ctrl and noGlOrog, the precipitation isless than observation, there was significant precipitation in excess of 5 mm day − in thePacific warm pool which had not been incorporated in the aqua-planet configurations.Hence an additionale experiment, AP M6, incorporating additional heating in region Pas shown in Fig. 9, was conducted. The ratio of mean precipitation between 110 ◦ E-180 ◦ and 20 ◦ S-30 ◦ N for AP M6 and Ctrl is ∼ a) Meridional section (b) Zonal section(c) Horizontal section (150 hPa)
111 1 1 1 1 (d) Velocity vectors and geopotential height (150hPa)
Figure 14: Zonal wind ( m s − ) profile. Vertical sections are at the (a) longitude and (a)latitude where U max is attained; (c) horizontal cross-section at pressure level where U max is attained, ‘cross-diamond’ is location of peak zonal wind, ‘star’ is location of P c , pre-cipitation contours ( mm day − ) are shaded; (d) velocity vectors and geopotential height( × m) at 150 hPa, ‘cross-diamond’ is location of peak geopotential height, contourvalue is 13.79 which is 99.5% of peak. AP M5 (Aqua-planet with multiple heat sourcesin regions B, C, D and E shown in Fig. 9).28ow shows the extension of the easterly beyond 90 ◦ E and this is also reflected in thezonal section (Fig. 15a). Thus the acceleration length of the TEJ is now closer to real-planet and reanalysis. The meridional section is similar to AP M5 simulation and henceis not shown. The location and magnitudes of peak zonal wind in AP M5 and AP M6are almost the same. Thus heating in the Pacific warm pool is necessary for the eastwardextent of the TEJ. (a) Zonal cross-section(b) Horizontal cross-section (150 hPa)
Figure 15: Zonal wind ( m s − ) profile. (a) zonal wind at the latitude where U max isattained, (b) horizontal cross-section at pressure level where U max is attained, ‘cross-diamond’ is location of U max , ‘star’ is location of P c , precipitation contours ( mm day − )are shaded. AP M6 (Aquaplanet with multiple heat sources in all regions, except S,shown in Fig. 9). The July climatological structure of the Tropical Easterly Jet in observations has beenstudied and compared with the simulations by an AGCM. The TEJ in reanalysis in July1998 and 2002 had significant zonal shifts. Reduced precipitation in the Indian region in2002 made the jet have its maxima in the southern Indian peninsula as compared to 1988when high rainfall in the Indian region resulted in the TEJ having its maxima over theArabian sea region. 29he TEJ in the Ctrl simulation has errors in the spatial location; otherwise the sim-ulated TEJ bears similarities with observations. Removing orography left the spatiallocation and structure is practically unchanged. This leads us to conclude that the TEJis not directly influenced by orography. The primary reason for the shift in the simulatedTEJ was because the location of precipitation in both Ctrl and noGlOrog is westwardswhen compared to reanalysis.Additional experiments were conducted to check if the TEJ is primarily influencedby latent heating. Changing the deep-convective relaxation time scale both in Ctrl andnoGlOrog simulations confirmed this. In these new simulations the precipitation is moreaccurately simulated and most importantly anomalous precipitation in Saudi Arabia nolonger occurred. The jet followed the shift in precipitation and relocated to the correct cli-matological position. The absence of orography once again had no impact on the locationof the jet. This conclusively proves that the TEJ is independent of orography. Changingthe default value of deep convective time-scale also demonstrated the secondary role oforography. Both Ctrl τ d and noGlOrog τ d have very similar precipitation patterns andhence the TEJ in both is correctly simulated.To understand why the TEJ was shifted westward in the AGCM, aqua-planet exper-iments were conducted.1 The simulation of TEJ in an aqua-planet configuration of the AGCM shows that orog-raphy and land-sea interactions are not as important as latent heat release.2 The total acceleration length is cirrelated to the zonal extent of the heating.3 A heat source at 20 ◦ N appears more to be robust in generating wind speeds that maybe referred to as a jet while equatorial heating alone does not generate TEJ.4 Equatorial heating is necessary to generate a strong low-level westerly that imparts thevertical baroclinic structure to the TEJ. However it is insufficient in generating trueTEJ horizontal structure.5 Equatorial heating is essential to create meridional structures seen in observations andfull AGCM simulations. Greater poleward depth is possible only if equatorial heatingis present.6 The longitudinal location of peak zonal wind is influenced by the off-equatorial heatingthat is closest to it. It has been demonstrated that rainfall in Saudi Arabia causesthe extreme westward shift of the TEJ in full AGCM simulations in comparison toobservations.7 Heating in the Pacific warm pool is essential to cause eastward extension and increasedacceleration length of the TEJ. 30hen all the important heat sources are incorporated in the aqua-planet configuration,many observed features of the TEJ were simulated. Thus aqua-planet simulations play animportant role in understanding the role of heat sources in the absence of any influenceof land and orography.
References
R. F. Adler, G. J. Huffman, A. Chang, R. Ferraro, P.-P. Xie, J. Janowiak, B. Rudolf,U. Schneider, S. Curtis, D. Bolvin, A. Gruber, J. Susskind, P. Arkin, and E. Nelkin.The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitationanalysis (1979-present).
Journal of Hydrometeorology , 4:1147–1167, 2003.G. S. Bhat. The Indian drought of 2002 – a sub-seasonal phenomenon?
Quarterly Journalof the Royal Meteorological Society , 132(621):2853–2602, 2006. doi: 10.1256/qj.05.13.S. Bordoni and T. Schneider. Monsoons as eddy-mediated regime transitions of thetropical overturning circulation.
Nature Geosciences , 1(8):515–519, 2008. doi: 10.1038/ngeo248.B. P. Briegleb. Delta-Eddington approximation for solar radiation in the NCAR Com-munity Climate Model.
Journal of Geophysical Research , 97(D7):7603–7612, 1992. doi:10.1029/92JD00291.P. Camberlin. June-september rainfall in north-eastern Africa and atmospheric signalsover the tropics.
International Journal of Climatology , 15(7):773–783, 1995. doi: 10.1002/joc.3370150705.A. Chakraborty, R. S. Nanjundiah, and J. Srinivasan. Impact of African orography andthe Indian summer monsoon on the low-level Somali jet.
International Journal ofClimatology , 29(7):983–992, 2008.T.-C. Chen and H. van Loon. Interannual variation of the Tropical Easterly Jet.
MonthlyWeather Review , 115:1739–1759, 1987. doi: 10.1175/1520-0493(1987)115 h i Atmos. Chem. Phys. , 12:1865–1879, 2012. doi:10.5194/acp-12-1865-2012.H. Flohn. Thermal effects of the Tibetan Plateau during the Asian monsoon season.Correspondence, University of Bonn, 1965.31. Flohn. Contributions to a meteorology of the Tibetan highland. Atmospheric SciencePaper 130, Colorado State University, Colorado State University, Fort Collins, 1968.A. E. Gill. Some simple solutions for heat-induced tropical circulation.
Quarterly Journalof the Royal Meteorological Society , 106(449):447–462, 1980.J. J. Hack. Parameterization of moist convection in the National Center for AtmosphericResearch Community Climate Model (CCM2).
Journal of Geophysical Research , 99(D3):5551–5568, 1994. doi: 10.1029/93JD03478.B. J. Hoskins and M. J. Rodwell. A model of the Asian summer monsoon. Part 1: Theglobal scale.
Journal of the Atmospheric Sciences , 52:1329–1340, 1995. doi: 10.1175/1520-0469(1995)052 h i Theoretical and Applied Climatology , 39(4):179–187, 1989. doi: 10.1007/BF00867945.J. W. Hurrell, J. J. Hack, A. S. Phillips, J. Caron, and J. Yin. The dynamical simulationof the Community Atmosphere Model version 3 (CAM3).
Journal of Climate , 19:2162–2183, 2006. doi: 10.1175/JCLI3762.1.L. Jingxi and D. Yihui. Climatic study on the summer Tropical Easterly Jet at 200 hPa.
Advances in Atmospheric Sciences , 6(2):215–226, 1989. doi: 10.1007/BF02658017.E. Kalnay, M. Kanamitsu, R. Kistler, W. Collins, D. Deaven, L. Gandin, M. Iredell,S. Saha, G. White, J. Woollen, Y. Zhu, M. Chelliah, W. Ebisuzaki, W. Higgins,J. Janowiak, K. C. Mo, C. Ropelewski, J. Wang, A. Leetma, R. Reynolds, R. Jenne,and D. Joseph. The NCEP/NCAR 40-year reanalysis project.
Bulletin of the Ameri-can Meteorological Society , 77:437–471, 1996. doi: 10.1175/1520-0477(1996)077 h i Tellus , 10(1):43–57, 1958. doi:DOI:10.1111/j.2153-3490.1958.tb01984.x.T. N. Krishnamurti. Observational study of the tropical upper tropospheric motion fieldduring the northern hemisphere summer.
Journal of Applied Meteorology , 10:1066–1096, 1971.F. Kucharski, A. Bracco, J. H. Yoo, A. M. Tompkins, L. Feudale, P. Rutic, and Dell’Aquila. A Gill-Matsuno-type mechanism explains the tropical Atlantic influence onAfrican and Indian monsoon rainfall.
Quarterly Journal of the Royal MeteorologicalSociety , 135(640):569–579, 2009. doi: 10.1002/qj.406.S. Lee. Why are the climatological zonal winds easterly in the equatorial upper tropo-sphere?
Journal of the Atmospheric Sciences , 56:1353–1363, 1999.32. Liu, B. J. Hoskins, and M. Blackburn. Impact of Tibetan orography and heating onthe summer flow over Asia.
Journal of the Meteorological Society of Japan , 85B:1–19,2007. doi: 10.2151/jmsj.85B.1.S. K. Mishra. Linear barotropic instability of the Tropical Easterly Jet on a sphere.
Journal of the Atmospheric Sciences , 44(2):373–383, 1987.S. K. Mishra. Nonlinear barotropic instability of the upper-tropospheric Tropical EasterlyJet on the sphere.
Journal of the Atmospheric Sciences , 50(21):3541–3552, 1993.R. B. Neale and B. J. Hoskins. A standard test for AGCMs including their physicalparametrizations. II: Results for The Met Office Model. 1, 2001. doi: 10.1006/asle.2000.0020.S. E. Nicholson, A. I. Barcilon, M. Challa, and J. Baum. Wave activity on the TropicalEasterly Jet.
Journal of the Atmospheric Sciences , 64:2756–2763, 2007. doi: 10.1175/JAS3946.1.K. Raghavan. Tibetan anticyclone and Tropical Easterly Jet.
Pure and Appied Geo-physics , 110(1):2130–2142, 1973. doi: 10.1007/BF00876576.K. Rajendran, A. Kitoh, and J. Srinivasan. Effect of SST variation on ITCZ in APEsimulations.
Journal of the Meteorological Society of Japan , 91A:195–215, 2013.V. Ramanathan and P. Downey. A nonisothermal emissivity and absorptivity formulationfor water vapor.
Journal of Geophysical Research , 91:8649–8666, 1986. doi: 10.1029/JD091iD08p08649.P. J. Rasch and J. E. Kristj´ansson. A comparison of the CCM3 model climate usingdiagnosed and predicted condensate parameterizations.
Journal of Climate , 11:1587–1614, 1998. doi: 10.1175/1520-0442(1998)011 h i Journal of Geophysical Research , 108(D14):4407, 2003. doi: 10.1029/2002JD002670.R. W. Reynolds, N. A. Rayner, T. M. Smith, D. C. Stokes, and W. Wang. An improvedin situ and satellite SST analysis for climate.
Journal of Climate , 15:1609–1625, 2002.doi: 10.1175/1520-0442(2002)015 h i Geo-physical Research Letters , 32(14):1–4, 2005. doi: 10.1029/2005GL022956.33. Sathiyamoorthy, P. K. Pal, and P. C. Joshi. Intraseasonal variability of the TropicalEasterly Jet.
Meteorology and Atmospheric Physics , 96(3-4):305–316, 2007. doi: 10.1007/s00703-006-0214-7.T. Schneider. The general circulation of the atmosphere.
Annu. Rev. Earth Planet. Sci. ,34:655–688, 2006.D. R. Sikka. Evaluation of monitoring and forecasting of summer monsoon over Indiaand a review of monsoon drought of 2002.
Proceedings of the Indian National ScienceAcademy , 69, A(5):479–504, 2003.A. Slingo. A GCM parameterization for the shortwave radiative properties of clouds.
Jour-nal of the Atmospheric Sciences , 46:1419–1427, 1989. doi: 10.1175/1520-0469(1989)046 h i Quarterly Journal of the RoyalMeteorological Society , 131(612):2961–3012, 2005. doi: 10.1256/qj.04.176.B. Wang.
The Asian Monsoon . Springer, 2006.P. Xie and P. A. Arkin. Global precipitation: A 17-year monthly analysis based on gaugeobservations, satellite estimates, and numerical model outputs.
Bulletin of the Amer-ican Meteorological Society , 78(11):2539–2558, 1997. doi: 10.1175/1520-0477(1997)078 h i Bulletin of the American Meteorological Society , 62:14–19, 1981. doi: 10.1175/1520-0477(1981)062 h i Atmosphere-Ocean , 33(3):407–446, 1995. doi: 10.1080/07055900.1995.9649539.M. Zhang, C. S. Bretherton, J. J. Hack, and P. J. Rasch. A modified formulation offractional stratiform condensation rate in the NCAR Community Atmospheric ModelCAM2.
Journal of Geophysical Research , 108(D1), 2003. doi: 10.1029/2002JD002523.34. Zhang, G. Wu, and Y. Qian. The bimodality of the 100 hPa south asia high and itsrelationship to the climate anomaly over east asia in summer.