Inertial instability flow in the troposphere over Suriname during the

South American Monsoon

J. Paul F. Fortuin, Hennie M. Kelder,1 Michael Sigmond,1 and Radchis OemrawRoyal Netherlands Meteorological Institute, The Netherlands

Cor R. BeckerMeteorological Service of Suriname, Suriname

Received 11 December 2002; revised 25 March 2003; accepted 4 April 2003; published 10 May 2003.

[1] Weekly sonde observations in Suriname, supported byECMWF analyses and a linear stability analysis, are used toanalyze the recurrence of inert ial instability as the ITCZmigrates over land during the South American Monsoon. Alayer of cool moist air from the Atlantic Ocean is thenadvected southward over Suriname in the shape of a coldfront, displacing the warmer air over the continent. The returnflow northward, by the upper branch of the Hadley cell, is aregion where inertial instability pervades due to cross-equatorial advection o f anticyclonic vorticity and theproximity of the subtropical jet as it migrates closer to theEquator. This unstable region evidently leads to the episodicalformation of a meridional sub-cell below the tropopause,where the damping is calculated to be strong enough tostabilize flow at smaller vertical scales, and yet weak enoughto allow the observed cell recurrence - at approximately theinertial frequency of the underlying latitudes. This instabilityshould also contribu te to the Hadley cell formation thr oughnorthward acceleration in the upper branch. The moistsaturated conditions in the lower troposphere do allowinertial instability here, but the high damping values withinthe boundary layer suggests that the observed southwardacceleration in the lower branch of the Hadley cell has a causeother than inertial instability. INDEX TERMS: 3374

Meteorology and Atmospheric Dynamics: Tropical meteorology;

3319 General circulation; 3364 Synoptic-scale meteorology; 0368

Atmospheric Composition and Structure: Troposphere—constituent

transport and chemistry. Citation: Fortuin, J. P. F., H. M. Kelder,

M. Sigmond, R. Oemraw, and C. R. Becker, Inertial instability flow

in the troposphere over Suriname during the South American

Monsoon, Geophys. Res. Lett., 30(9), 1482, doi:10.1029/

2002GL016754, 2003.

1. Introduction

[2] Suriname, one of the Guyana countries at the northerncoast of South America, lies approximately in the middle ofthe annual migration range of the inter-tropical convergencezone (ITCZ) in this region, and hence experiences two wetand two dry seasons. Weekly balloon sonde releases atParamaribo station (5.8�N 55.2�W) started in September1999, measuring morning (around 13h00 GMT) profiles ofozone, temperature, water vapor and wind. These sonde

observations are used to study the structure of the ITCZ, andits associated large-scale circulation, as it migrates over landduring the South American Monsoon. Of particular interestis the regular occurrence of inertial instability in the uppertroposphere above Suriname during this period. Inertialinstability in the northern hemisphere (NH) occurs whenthe potential vorticity (PV) of air becomes negative, suchthat a disturbance experiences a meridional poleward orequatorward acceleration in the unstable area. A study byStevens [1983] demonstrates that the north-south symmetryobserved in zonal flow in the tropical atmosphere andoceans is not accidental, as a horizontal wind shear nearthe Equator will induce a negative PV that results in inertialinstability and a flow that counteracts the wind shear. Thisflow will be one of mass overturning through verticallystacked meridional circulation cells, if the conditions forzonal symmetric instability are met, as first indicated fromtheory by Dunkerton [1981], hence called D81. Apart fromevidence of such flow occurring in the extratropical tropo-sphere [Bennetts and Hoskins, 1979; Emanuel, 1983], it hasalso been identified near the Equator by Hitchman et al.[1987], as a so-called ‘‘pancake structure’’ of alternatingtemperature extrema in the lower mesosphere. The currentstudy presents evidence of inertially unstable flow occurringin the tropical troposphere over Suriname during the mon-soon period, when the subtropical jet migrates closer to theEquator in NH winter.

2. Structure of the ITCZ During theMonsoon Period

[3] The chronology of wet and dry seasons experiencedin Suriname can be traced in the recorded water vapor sondeprofiles versus time in Figure 1. The onset of the short wetseason in December occurs almost simultaneously with thetransition of the trade winds from southeasterly to north-easterly flow (Figure 2, indicated with an ‘‘A’’). In contrast,the onset of the long wet season around April happens wellin advance of the reverse transition in meridional windaround July. When the ITCZ comes from the north and firstreaches the northern coast of South America around Nov–Dec, there is an initial rapid outflow of moist air over theAmazon basin to the south as the locus of precipitationsuddenly shifts to the thermal low created over land - as istypical for a monsoon system. Convection over Paramaribois suppressed by subsidence on the equatorward flank of theNorth Atlantic high during the short dry season from Feb–Mar [Hastenrath, 2000], when the ITCZ reaches its south-

GEOPHYSICAL RESEARCH LETTERS, VOL. 30, NO. 9, 1482, doi:10.1029/2002GL016754, 2003

1Also at the Technical University of Eindhoven, The Netherlands.

Copyright 2003 by the American Geophysical Union.0094-8276/03/2002GL016754$05.00

35 -- 1

ern-most position around the Equator. From this position,the rainy band then slowly retreats back northward alongwith the ITCZ over the Atlantic Ocean, resulting in the longwet season from Apr–Jul. The northeasterly flow in theboundary layer during the monsoon period consists ofcooler air, which, coming from the Atlantic Ocean, is richin moisture and produces abundant rain over the warmercontinent. One would expect the cooler air to have a wedgedshape as it displaces the warmer air over the continent.Assuming a gradual northward migration after the ITCZ hasreached its southernmost point, a pole-ward slope of thiswedge (from Figure 2, marked ‘‘B’’) can be inferred of�1:500, which agrees well with the slope inferred from theECMWF case study presented later on (cf. Figure 5c). Theinduced deep convection often penetrates through the boun-

dary layer, up to maximal heights of around 12 km. At thisheight, a distinct northward flow is apparent (Figure 2,within the ‘‘C’’ area) corresponding with the upper branchof the Hadley cell.

3. Inertial Instability During the MonsoonPeriod

[4] Since the start of the Paramaribo observation pro-gram, collocated ECMWF analysis profiles were calculatedeach day. In general, the ECMWF wind profiles comparevery well with the Paramaribo observations. Of particularinterest is the fact that the northward Hadley branch is mostoften flanked, directly above and below, by southward flow(Figure 2a, square area marked with a ‘‘C’’), and that thispersistently coincides with negative PV values (Figure 3).This raises the question whether the return flow could be amanifest of the vertically stacked cellular flow predicted byD81 for symmetrical inertial instability, as is schematicallydepicted in Figure 4. This issue is investigated further withECMWF operational analyses over a 10-day period, whenthe conditions for secondary flow resulting from symmetricinertial stability seem to be clearly met: from February 23 toMarch 3, 2000. During this period, there is a clear anti-correlation between meridional and zonal wind in the uppertroposphere (Figure 2, at the ‘‘C’’ mark), a feature also used

Figure 1. Time-height cross section of relative humidityfrom sonde observations over Paramaribo; the inferredtropopause height is in bold black. The markings (in yellow)are described in the text. The first day of each month ismarked on the x-axis.

Figure 2. a) Observed zonal wind (top), and b) meridionalwind (bottom) over Paramaribo, similar to Figure 1. Themarkings (in yellow) are described in the text.

C

Figure 3. Time-height cross section of ECMWF potentialvorticity over Paramaribo. The markings (in yellow) aredescribed in the text.

Figure 4. Cartoon of the ITCZ and the induced meridionalcirculation cells resulting from negative PV (blue shaded) inthe upper troposphere during the South American Monsoon.The pink arrows indicate zonal wind (westerly pointeddownward), the Equator is at y = 0 and Paramaribo at Yo.

35 - 2 FORTUIN ET AL.: INERTIAL INSTABILITY FLOW OVER SURINAME

as an indicator by Hayashi et al. [2002] to mark theoccurrence of symmetric inertial flow around the strato-pause. Similar to their analysis, we will study the averagesover a 10-day period in order to filter out transient dis-turbances on shorter timescales.[5] A latitude-height cross-section of the meridional and

zonal wind field at the Paramaribo longitude (55�W) isshown in Figures 5a and 5b. Figure 5b also shows the windvectors composed of the meridional wind and an enhanced(10�) vertical wind velocity. It is evident that the strongestnorthward winds of the upper Hadley branch coincide withthe region of maximal zonal wind shear caused by thesubtropical jet at the same altitude to the north. Above theHadley cell, a sub-cell can be distinguished directly under-neath the tropopause, with a latitudinal span between theEquator and 12�N and a maximal southward velocityaround 7�N. The location of the meridional wind maximumsouthward (at 150 hPa) and northward (250 hPa) of thissub-cell approximately coincides with the zero PV line.Viewing the days of the 10-day episode separately, it isfound that on 5 of the 10 days the sub-cellular structure canbe identified, and on 3 occasions the entire sub-cell liessubmerged in the negative PV region, with local meridionalwind maxima about halfway between Equator and the outeredge of the unstable layer. This is conform the analyticaltwo-dimensional solution of D81 for symmetric inertialinstability, predicting stacked cellular flow contained withinthe unstable area and maximal meridional velocities at thelatitudinal middle, shifting towards the Equator as thevertical extent of the unstable domain increases [Stevens,1983]. However, the linear stability analysis of D81 alsopredicts the highest growth rates for infinite small verticalscales. Stacked flow with a finite vertical extent onlybecomes possible when one assumes that viscosity or eddydiffusivity locally stabilizes flow at smaller vertical scales,as suggested by D81. Assuming a Prandtl number anddiffusivity equal to unity, and then subsituting the ECMWFparameter values of this case study in equation 2.6 of D81,gives an almost uniform vertical wavelength at whichmaximal growth occurs: between 2.2 to 3 km for the entireunstable region (not shown). As a second indication, onecan also calculate the maximum vertical wavelength atwhich growth of the unstable flow is still theoreticallypossible:

lvert >p�u2y2N b

; ð1Þ

as can be inferred from the dispersion relation in D81 forzonal symmetric disturbances of inviscid flow on a b-plane.Calculating equation (1) from ECMWF values for our casestudy, Figure 5a (bold dashed contours within the negativePV region) yields a vertical wavelength range of 2–10 kmwithin the unstable area. This shows that, in principle,symmetric instability can grow on the vertical scaleobserved for the upper tropospheric sub-cell, given thatviscosity or eddy diffusivity stabilizes the flow on smallervertical scales. For the months Jan–Mar, when the deepestconvection occurs near the Equator, it is found that the sub-cell reappears about each 5 days or so and lasts for 2–3days. This is consistent with the inertial frequency (ffif, theCoriolis parameter) at this latitude.

[6] The sub-cell is mirrored below by a much bigger cell,the Hadley cell, which is confined to approximately thesame latitude band and lies in a vertically split domainwhere either the PV (upper troposphere) or the PVeq (lowertroposphere) is less than zero. Bennetts and Hoskins [1979],and later Emanuel [1983], showed that symmetric inertialinstability is likely to occur in a moist baroclinic atmosphereunder the same conditions as for a dry atmosphere, exceptthat a PVeq (instead of PV) less than zero is now thecriterion for the occurrence of inertial instability. Figure 5shows (PVeq, bold dotted) that the lower troposphere nowalso becomes unstable, which raises the question to whichextent inertial instability plays a role in the formation of theHadley cell. The equator-ward position of the southwardvelocity maximum in the lower branch of the Hadley cell isin agreement with the larger vertical extent of the Hadleycell, as mentioned earlier [Stevens, 1983]. However, thestudy by Tomas et al. [1999] argues against an accelerationby inertial instability within the boundary layer due tostrong damping here, indicating that the observed south-ward acceleration has another cause which clearly needsfurther investigation. Calculations for the upper tropospheredo however suggest that the upper branch of the Hadley cellwill feel the northward acceleration by inertial instabilitythat also feeds the upper sub-cell.[7] Symmetric unstable flow as discussed above is valid

for the specific case where the disturbance has no longi-tudinal dependence and results in a uniform vertical massoverturning over the entire unstable zonal band. In practice,a zonally asymmetric disturbance will often happen, whichfor the same background conditions as discussed beforeleads to a non-linear solution first presented by Dunkerton[1983]. Well-behaved solutions were found to exist only for

Figure 5. ECMWF pressure (hPa) - latitude cross sectionsat 55�W, averaged over the period 23/2–3/3/2003, for: (a)meridional wind (colored contours), the zero line for PV(bold solid) and PVeq (bold dotted), maximum verticalwavelength (see text, bold dashed in km); (b) zonal wind(colored contours), vectors (green) composed of meridionaland (10�) vertical wind, minimal horizontal wavelength(see text, bold dashed in �longitude); (c) temperaturedifference with the meridional mean; (d) percentage relativehumidity. Paramaribo station is indicated with a light bluevertical line.

FORTUIN ET AL.: INERTIAL INSTABILITY FLOW OVER SURINAME 35 - 3

zonal wave numbers below a certain shortwave cutoff valuecorresponding with lhor <

2pb�uy. For discrete wavenumbers

with respect to the spherical earth, a maximum growth ratebetween the symmetric value (k = 0) and the cutoff value isthen possible. This minimal horizontal wavelength is givenby the bold dashed contours within the negative PV regionof Figure 5b, showing a range of 20 to 40 degrees longitudewithin the unstable zone. In this and other cases studied sofar, the zonal extent of the unstable region is narrower thanthese cutoff wavelength values, indicating that a symmetricinertial flow will be enforced within the unstable region.[8] Figure 5c shows the temperature difference from the

meridional mean (25�S–25�N) for each pressure layer. Thelatitudinal slope (1:500) of the cooler boundary layerinferred earlier for the monsoon period can be viewed asbeing approximately 3 km depth over a 1500 km latitudespan. It is clear that the cooler air coming from the AtlanticOcean is advected over the warmer continent to just south ofthe Equator, where deep convection occurs and the upwardbranch of the Hadley cell is formed. These features arereflected in the ECMWF humidity field of Figure 5d.Interestingly enough, a dryer layer can be distinguished atthe location of the sub-cell, with a similar meridional extentand depth. If water vapor is regarded as a passive tracer inthe upper troposphere, this dry cell could be a signature ofthe meridional sub-cell as moisture is first advected north-ward by the upper Hadley branch, and then upward andsouthward just below the tropopause, leaving the stagnantair in the middle of the cell to dry. Unfortunately, humiditymeasurements of radiosondes become unreliable at thesehigh altitudes, so that this signal cannot be confirmed bysonde observations.

4. Conclusions

[9] Paramaribo observations, supported by ECMWFanalyses, show the systematic recurrence of an inertiallyunstable domain in the upper troposphere over Surinameduring the South American Monsoon, reaching from theEquator towards the subtropical jet. This unstable field isthe result of the subtropical jet migrating towards theEquator during NH winter (raising local vorticity abovelocal Coriolis values), or comes from cross-equatorial north-ward transport of anticyclonic vorticity (from the SH upper-air anticyclone related to deep convection in the Amazonebasin below). A linear stability analysis points out that therecurrence of a meridional sub-cell positioned on top of theHadley cell can be due to symmetrical inertial instability, asmaximum growth rates can occur on the vertical scaleobserved for typical damping values of the upper tropo-sphere. Inertial instability may also play an important role inthe formation of the Hadley cell, which is approximatelyconfined over the same latitude band during this period,through a northward acceleration of the upper Hadley

branch at the height of the subtropical jet. The moistsaturated conditions in the lower troposphere do allowinertial instability here, but the high damping values withinthe boundary layer suggests that the observed southwardacceleration in the lower branch of the Hadley cell has acause other than inertial instability. During the monsoonperiod, it is found that the cool moist air advected over theAmazon basin by the northeasterly trade winds has thewedge-like shape of a cold front sloping poleward (about1:500), as it replaces the warmer continental air and inducesdeep convective events that pierce through the boundarylayer up to about the height of the upper Hadley branch at12 km. This study presents first evidence, not conclusiveproof, that inertial instability plays an important role in theformation of meridional circulation cells in the tropicaltroposphere. More extensive analyses and modeling areclearly needed to substantiate this and infer its validity onbroader temporal and geographical scales.

[10] Acknowledgments. We thank the Suriname operator team fortheir dedication and good work. Paramaribo station has benefited from itsmembership to the SHADOZ network. The first author wishes to acknowl-edge the Radio Science Center for Space and Atmosphere (RASC) at KyotoUniversity, Japan, where most of this work was done during his stay asvisiting scientist. The referees are acknowledged for the valuable sugges-tions that were incorporated in the revision. This study is part of the projectRADCHIS (Research on Atmospheric Dynamics and Chemistry in Sur-iname), financed by the Netherlands Organization for Scientific Research(NWO-ALW).

ReferencesBennetts, D. A., and B. J. Hoskins, Conditional symmetric instability - Apossible explanation for frontal rainbands, Quart. J. Roy. Meteor. Soc.,105, 945–962, 1979.

Dunkerton, T. J., On the inertial instability of the Equatorial middle atmo-sphere, J. Atmos. Sci., 38, 2354–2364, 1981.

Dunkerton, T. J., A symmetric Equatorial inertial instability, J. Atmos. Sci.,40, 807–813, 1983.

Emanuel, K. A., The Lagrangian parcel dynamics of moist symmetric in-stability, J. Atmos. Sci., 40, 2368–2376, 1983.

Hastenrath, S., Interannual and long-term variability of upper air circulationin the Northeast Brazil-tropical Atlantic sector, J. Geophys. Res., 105,7327–7335, 2000.

Hayashi, H., M. Shiotani, and J. C. Gille, Horizontal wind disturbancesinduced by inertial instability in the Equatorial middle atmosphere as seenin rocketsonde observations, J. Geophys. Res., 107(D14), 4228, doi:10.1029/2001JD000922, 2002.

Hitchman, M. H., C. B. Leovy, J. C. Gille, and P. B. Bailey, Quasi-sta-tionary zonally asymmetric circulations in the Equatorial lower meso-sphere, J. Atmos. Sci., 44, 2219–2236, 1987.

Stevens, D. E., On symmetric stability and instability of zonal mean flowsnear the Equator, J. Atmos. Sci., 40, 882–893, 1983.

Tomas, R. A., J. R. Holton, and P. J. Webster, The influence of cross-equatorial pressure gradients on the location of near-equatorial convec-tion, Q. J. R. Meteorol. Soc., 125, 1107–1127, 1999.

�����������������������J. P. F. Fortuin, H. M. Kelder, M. Sigmond, and R. Oemraw, Royal

Netherlands Meteorological Insitute, P.O. Box 201, 3730 AE De Bilt, TheNetherlands. (fortuin@knmi.nl; kelder@knmi.nl)C. R. Becker, Meteorological Service of Suriname, Magnesiumweg 41,

Paramaribo, Suriname.

35 - 4 FORTUIN ET AL.: INERTIAL INSTABILITY FLOW OVER SURINAME