generation of low-level mesoscale convergence lines over …roger/... · 1998-03-11 · thermally...

JANUARY 1998 167S M I T H A N D N O O N A N

q 1998 American Meteorological Society

Generation of Low-Level Mesoscale Convergence Lines over Northeastern Australia

ROGER K. SMITH

Meteorological Institute, University of Munich, Munich, Germany

JULIE A. NOONAN

CSIRO Division of Atmospheric Research, Aspendale, Victoria, Australia

(Manuscript received 3 September 1996, in final form 8 May 1997)

ABSTRACT

Thermally forced atmospheric circulations over the Gulf of Carpentaria region of northeastern Australia areinvestigated using a mesoscale numerical model. The region is renown for the common occurrence of longwestward-moving convective- and wave-cloud lines, including the celebrated ‘‘morning glory’’ phenomenon.In the model, it is found that for uniform flows over the region ranging from northeasterly to southeasterly,westward-moving, low-level convergence lines develop over the gulf during the night and early morning. Theauthors suggest that similar convergence lines in the atmosphere are responsible for the initiation and maintenanceof the observed cloud lines. For northeasterly and easterly flow, the convergence lines show little day-to-dayvariation, despite the relatively long inertial period in the region, which is nearly two days. The calculations,which extend an earlier study by the same authors, lead to a new hypothesis to account for the observed longevityof morning glory bore waves. They provide also an explanation for the marked diurnal oscillation in the low-level easterly flow observed at Weipa during a field experiment to investigate the so-called north Australiancloud line.

1. Introduction

The Gulf of Carpentaria and Cape York Peninsularegion1 of northeastern Australia is distinguished for theregular occurrence there of a range of convective- andwave-cloud lines. These include the celebrated ‘‘morn-ing glory’’ phenomenon, a spectacular traveling wave-cloud system commonly observed over the southern partof the gulf and adjacent seaboard [see Smith (1988) andChristie (1992) for recent reviews], and the socallednorth Australian cloud line (NACL), a line of convectivecloud that is seen in satellite imagery to extend fre-quently across the entire gulf (Drosdowsky and Holland1987; Drosdowsky et al. 1989). Both types of distur-bance form on the western side of the peninsula, theformer typically in the late afternoon and the latter inthe late evening, and they both move with a westwardcomponent across the gulf. Moreover, both are respon-sible for significant weather events in the region. Themorning glory is accompanied by sudden wind squalls,

1 A map of the region with place names is shown in Fig. 1.

Corresponding author address: Prof. Roger K. Smith, Meteoro-logical Institute, University of Munich, Theresienstr. 37, 80333 Mu-nich, Germany.E-mail: [email protected]

intense low-altitude wind shear, and a marked verticaldisplacement of air parcels, sometimes sufficient to ini-tiate showers or thunderstorms in the wake of the dis-turbance. The NACL brings a wind change also, andwhile the depth of convection is sometimes limited byan inversion at a height of about 3 km, embedded show-ers and thunderstorms may occur at times of the yearwhen the air over the gulf is sufficiently moist. On oc-casion, the morning glory forms the southern extensionof the NACL (Smith and 985).

There is much evidence that the majority of both typesof disturbance have their origin in organized mesoscalecirculations that develop over the peninsula and adjacentgulf associated with sea breezes (Clarke et al. 1981;Clarke 1984; Noonan and Smith 1986, 1987; Dros-dowsky et al. 1989). However, some morning glory dis-turbances moving from the south are generated by fron-tal systems crossing central Australia (Smith et al. 1995;Smith et al. 1986).

Some insight into the precursors of cloud line for-mation was provided by the study of sea-breeze cir-culations over Cape York Peninsula by Noonan andSmith (1987; henceforth referred to as NS) using thethree-dimensional mesoscale numerical model devel-oped by Pielke and collaborators (McNider and Pielke1981). This model was initialized at sunrise (0600 EST2)

2 EST denotes Australian eastern standard time 5 UTC 1 10 h.

168 VOLUME 126M O N T H L Y W E A T H E R R E V I E W

with a uniform (5 m s21) easterly geostrophic airstreamwith a temperature sounding characteristic of the CoralSea in November. The integration, which covered a 20-hperiod, showed the evolution in the late afternoon andevening of a westward-moving convergence line overthe southern part of the peninsula and over the northerngulf. It was hypothesized that this line is the modelanalog of the convergence line that initiates the NACLand northeasterly morning glories. However, there weremany limitations of the calculation, necessitated at thattime by computational constraints. One was the rela-tively small size of the computational domain, whichincluded only the eastern half of the gulf and little ofthe southern gulf seaboard. Furthermore, the calculationlasted less than half an inertial period and did not con-sider the effects of orography. In the intervening years,computational resources have improved markedly, ashas the efficiency of numerical models, and the presentpaper revisits the problem, seeking to remove the fore-going limitations and to investigate additional questions.

Several outstanding questions that will be addressedhere are the following.

R What is the effect of orography on the circulationsover Cape York Peninsula?

R How are the mesoscale circulations over the peninsulaand over the gulf affected by changes in the geo-strophic wind direction over the region and whichdirections are favorable or unfavorable for the for-mation of the morning glory convergence line?

R How representative is the evolution of the mesoscalecirculations in the calculation by NS in view of thelong inertial period (about 46 h) in the gulf region?

R Do the sea-breeze circulations on the western side ofthe gulf have any significant effect on those on theeastern side and in particular on the formation of con-vergence lines over the gulf? For example, do thepersisting effects of inertial turning induced in thelow-level flow over the western side of the gulf haveany appreciable effect on the flow over the easternhalf? In the calculations by NS, the influence of thesea-breeze circulations on the western side of CapeYork Peninsula extended to the central gulf by theevening and inertial turning of the low-level windspersisted until the calculation terminated at 0200 ESTon the following morning (see also section 4a below).

A further question arose from a field experiment car-ried out in October 1986 to study the NACL (Hollandet al. 1986). From serial radiosonde soundings at Weipaduring the experiment it was found that there was aregular diurnal oscillation in the strength of the zonalflow when the flow over the peninsula was easterly. Theoscillation had a maximum amplitude at a height ofabout 900 mb and there was little corresponding signalin the meridional wind component. We explore here thereason for this oscillation and give a dynamical expla-nation for its occurrence. The existence of these oscil-lations calls into question the representativeness of ra-

diosonde and/or rawinsonde soundings made at Weipaand at other stations where such oscillations occur.

In essence we present the results of a series of nu-merical experiments encompassing a much larger do-main than that used by NS, covering the entire gulf,part of the ‘‘Top-End,’’ 3 and a substantial region to thesouth. The calculations include orography and extendfor a period of 84 h (nearly two inertial periods). Likethe previous model, the present one considers only dryconvection.

A brief description of the model is given in section2. In section 3 we present the results of a series of modelintegrations with and without orography and with a uni-form geostrophic wind from different directions. Thecalculations enable us to provide a rather comprehensivepicture of the diabatically forced mesoscale circulationsthat commonly occur in the region and of the low-levelconvergence lines to which they give rise.

2. The numerical model

The model used is LADM (the Lagrangian Atmo-spheric Dispersion Model), a hydrostatic primitive-equation model developed at the Commonwealth Sci-entific Industrial and Research Organization (CSIRO)Division of Atmospheric Research in Australia (Physicket al. 1994). It is fully compressible and formulated inan (x, y, s) coordinate system where s 5 p/ps, p beingthe pressure, and ps, the surface pressure. The momen-tum, heat, and moisture fluxes are parameterized usingthe scheme of Louis (1979). In the surface layer, theObukhov flux-profile relationships are specified in termsof numerically fitted functions of bulk Richardson num-ber. Above the surface layer, a gradient diffusion ap-proach is used to calculate the turbulent fluxes. Short-and longwave radiation parameterizations are used also.The sea surface temperature is kept constant at 300 K,and the surface temperature over land is diagnosed froma surface heat balance condition at each time step. Aheat diffusion equation is solved at six levels in the soilto compute the heat flux into or out of the ground. Eachgrid point can be representative of a grid square thatconsists of part land and part water. The horizontal ad-vection and geostrophic adjustment processes are eval-uated separately and a semi-Lagrangian method is usedto compute the horizontal advection. An advantage ofthis scheme is that it allows the advective terms to beevaluated using a larger time step than that determinedby the Courant–Friedrichs–Lewy criterion for an Eu-lerian-based scheme, hence improving the speed of themodel.

The plotting domain is shown in Fig. 1. The calcu-lation domain is larger than this in order to reduce the

3 The name given to the region west of the gulf including ArnhemLand and extending as far as Darwin.


FIG. 1. Topographic map of northeastern Australia with place namesmentioned in the text: Burketown (B), Delta Downs (D), Gove (G),Karumba (K), Macaroni Station (M), Mornington Island (MI), andWeipa (W). Topography contours have 100-m intervals. The figureindicates also the subdomain on which the model output fields areplotted and the corresponding grid numbers referred to in the text.The grid spacing is 20 km.

influence of boundary effects in the region of interest.It has 120 3 90 grid points in the horizontal with a gridspacing of 20 km. The model has 25 levels in the verticalwith finer resolution in the boundary layer. The ap-proximate heights of the sigma levels below 3 km arethe same as in the NS calculation [i.e., 2 m (s 5 0.9998),10, 50, 100, 300, 500, 700, 900, 1200, 1500, 2000, 2500,and 3000], as are the surface parameters, which aredetailed in Table 1 of NS.

3. The model experiments

Each 84-h model run starts at 0200 local solar time(LST) on day 1 and continues until 1400 LST on day4 to provide three realizations of the morning gloryconvergence line. As in the calculation of NS, the modelis initialized at each grid point with the November meanvertical profiles of temperature and humidity from WillisIsland (east of Cape York Peninsula) and a uniformgeostrophic wind speed. We present the results of fourcalculations.

In the control experiment, a uniform easterly windwith a speed of 8 m s21 is taken; this is 3 m s21 largerthan in the NS calculation (without orography) and waschosen because, as shown below, the orography tendsto delay the formation of nocturnal convergence lines.

In a second experiment, we repeat the control cal-culation but without orography. Finally, we report ontwo other experiments with orography and with a uni-form northeasterly or southeasterly geostrophic windagain with a speed of 8 m s21.

4. Easterly geostrophic flow

a. Evolution of the circulation

We study first the evolution of the thermally inducedcirculations in the region of interest for the control ex-periment. In many respects this is similar to that in theNS study, but the present discussion covers an extendedtime period and investigates other features, includingthe evolution in selected vertical cross sections.

Figure 2 shows vector plots of the surface wind forthis experiment at three selected times (1400, 2200, and0600 LST) starting on the afternoon of day 1 and endingon the morning of day 3. A prominent feature of theflow at 1400 LST on day 1 is the presence of an onshoresea-breeze circulation along the west coast of Cape YorkPeninsula with strong low-level convergence near thecoast. Along the east coast of the peninsula the flow isonshore and consequently no sea-breeze front developsthere, although later in the day a front develops farinland (see below). By 2200 LST, the sea breeze alongthe west coast has largely collapsed, although manifes-tations of the sea-breeze circulations over the peninsulaand the gulf remain, as is evidenced by the deviationsof the surface flow from its undisturbed state. Thesedeviations are associated with inertial motion of the low-level flow. At this time, what was previously the sea-breeze convergence line along the west coast of thepeninsula has moved westward over the gulf, while, asa continuation of this line over the southwestern part ofthe peninsula, the easterly flow has developed a line ofmarked convergence also. We suggest that it is the at-mospheric counterpart of this line that initiates themorning glory disturbance, while its northwestern ex-tension over the gulf is instrumental in the formationof the NACL (Drosdowsky et al. 1989). By 0600 LSTon day 2 the convergence line in the south of the gulfhas moved westward to a position that is typical ofmorning glory cloud lines at this time, whereas thenorthern part of the line has largely decayed. The latterresult may be an indication that moist convection is anecessary component in the maintenance of low-levelconvergence required to account for the observed lon-gevity of NACL. The convergence line in the southdecays in the late morning as the sea-breeze flow alongthe southern gulf coast becomes reestablished. A similarsequence of events occurs on the following two days(the last day is not shown), and while the details differslightly because of the different starting conditions atsunrise, the principal features described above are pres-ent. Some of the day-to-day changes are highlighted inFigs. 3–5, which help also to elucidate the dynamicalprocesses that are involved.

Longitude–height cross sections of the isentropic sur-faces below 3.5 km along the latitude of MorningtonIsland are shown in Fig. 3 at four hourly intervals be-tween 1800 LST on day 1 and 0600 LST on day 2, andFig. 4 shows the corresponding isotach cross sectionsof the zonal (u) and meridional (y) wind components


FIG. 2. Vector plots of the surface wind in the control experiment at three selected times (1400, 2200, and 0600LST) starting on the afternoon of day 1 and ending on the morning of day 3.


FIG. 3. Vertical cross sections of the isentropic surfaces below 3.5 km along the meridian that passesthrough Mornington Island in the south of the gulf at selected times: (a) 1800 LST on day 1, (b) 2200 LSTon day 1, (c) 0200 LST on day 2, and (d) 0600 LST on day 2.

for the same period. A conspicuous feature in the is-entropes at 1800 LST is the scarcity of contours below2 km over the western side of the peninsula and overland near the western edge of the domain, indicatingthe development of a deep convectively well-mixed lay-er there during the daytime. The region of nearly verticalcontours over the eastern side of the peninsula and overthe land immediately to the west of the gulf indicatethe inland development of the mixed layer with the meanpotential temperature increasing inland. Prominent fea-tures of the zonal wind field at 1800 LST are a regionof accelerated easterly flow over the ranges on the east-ern half of the peninsula together with a shallow regionof westerly flow on the eastern side of the gulf—thewest coast sea breeze on the peninsula (Fig. 4a). Thereturn branch of the sea-breeze circulation is evident inthe region of enhanced easterly flow overlying the max-imum in the westerly component. The most noticeablefeature in the meridional wind component is the sea-

breeze circulation near the southern gulf coast, whichlies on the western side of the cross section.

By 2200 LST, the air at low levels over the land hasbegun to stabilize, and over the peninsula there is amarked westward gradient of temperature below 1800m, the maximum occurring at low levels near the westcoast (Fig. 3b). The region of supergeostrophic4 (.8 ms21) easterly winds has extended across much of thepeninsula and is overlain by a region with a subgeos-trophic easterly component of flow (Fig. 4b). The west-erly sea-breeze component is now shallower but extendsacross much of the gulf in this cross section. The y crosssection shows a northerly ‘‘jet’’ over the western sideof the peninsula underlying a region of marked south-erlies. The development of these features can be attrib-uted both to the meridional variation of the orography

4 Relative to the imposed easterly background flow.


FIG. 4. Vertical isotach cross sections of the zonal (u) and meridional (y) wind components below 3.5 kmalong the same latitude circle as in Fig. 3 and at the same times: (a) 1800 LST on day 1, (b) 2200 LST onday 1, (c) 0200 LST on day 2, and (d) 0600 LST on day 2. The dashed lines indicate negative values, i.e.,an easterly or northerly component of flow.


FIG. 4. (Continued)

FIG. 5. Space–time cross section of (a) u isotachs and (b) y isotachs at a height of 100 m above the surface calculatedalong the same latitude circle as in Fig. 3 for the 72-h period commencing at 1400 LST on day 1. The dashed linesindicate negative values. The vertical lines mark the coastlines.

and to inertial effects, which manifest as a Coriolis turn-ing of the wind. As time proceeds, during the early hoursof the morning, the pronounced jump in the isentropesnear the western side of the peninsula translates west-ward (Figs. 3c and 3d), accompanied by the westwardextension of the easterly surge over the peninsula (Figs.4c and 4d).

The regions of low-level northerly and higher-levelsoutherly components associated with this surgestrengthen during the afternoon and evening, becomingsteady after midnight. Also they increase in extent asthey progress westward. Note that, even as late as 0600LST, the maximum easterly component continues to lieover the higher ranges to the east, whereas the maximumin the northerly component translates westward across

the gulf (see section 5). The near-surface transition froma southerly to a northerly component over the gulf cor-responds with the model analog of the morning gloryconvergence line. The region of strongest temperaturegradient in the lowest few hundred meters remains closeto the west coast of the peninsula and does not translatewith the surface convergence line.

Figure 5 shows a space–time cross section of the uand y isotachs at a height of 100 m above the surface,calculated along a latitude circle that passes throughMornington Island in the southern part of the gulf, forthe 72-h period commencing at 1400 LST on day 1.Note that the principal features of the isotach patternare similar on each day of the simulation. In the u fieldsthese features include the development around midday


FIG. 6. Comparison of the surface wind speed and direction, po-tential temperature, and water vapor mixing ratio on days 2 and 3 ofthe model calculation with the mean (thick line) of a five-week periodat Burketown in September–October 1991.

FIG. 7. (a) Day 2 to day 4 average vertical profile of potentialtemperature at Burketown at 0430 LST in the control calculation (thinline) compared with the mean soundings observed there on morningglory days during the 1991 experiment (thick line) and the mean ofthe first four morning glory days during this experiment (dashed line).(b) Average of the day 3 and day 4 vertical profiles of potentialtemperature at 0000 LST at Macaroni Station on the western side ofthe Cape York Peninsula.

of the sea breeze on the west coast of the peninsula andover the gulf, the acceleration of the onshore flow inlandof the east coast of the peninsula during the daytime,and the formation in the early evening of a line of sharpconvergence at the leading edge of this flow over thewestern side of the peninsula. Subsequently, this con-vergence line, essentially the sea-breeze front from theeast coast sea breeze, crosses the peninsula and interactswith the west coast sea breeze to form what we shallrefer to as the morning glory convergence line, al-though, in reality, the line may be accompanied by aline of convective clouds (the NACL) instead of themorning glory wave cloud. This convergence line con-tinues to advance westward with a speed of about 10m s21. At Mornington Island township (grid point 21)the transition from southwesterlies to southeasterlies oc-curs a little after 0700 LST on day 1 and about an hourearlier on days 2 and 3, all within the range of observedtimes. The morning glory convergence line is a pro-nounced feature also in the y fields (Fig. 5b), where themaximum north–south gradient occurs near the westcoast of the peninsula around 0200 LST on day 1 andaround midnight on days 2 and 3 (as is the case for themaximum gradient in the u component). Note that themaximum and minimum of the meridional component

are displaced westward with the convergence line,whereas the maximum easterly component remains tiedto the orography on the eastern side of the peninsulaand is largest during the daylight hours, a feature seenalso in the space–height cross sections in Fig. 4.

b. Comparisons with observations

The conventional data network in the gulf region israther sparse but detailed data have been obtained atspecific locations during various field experiments.These data include, inter alia, surface measurements oftemperature, humidity, pressure, wind speed, and winddirection at Burketown for a five-week period in Sep-tember and early October 1991; daily 0430 EST radio-sonde soundings at Burketown for the same period; anddaily radiosonde soundings at Macaroni Station for athree-week period in October 1981 (Clarke 1983). Whilewe cannot expect good agreement in all detail, if onlyfor the reason that the assumed uniform geostrophicwind is not observed over the region in any particularcase, it is of interest to determine the extent to whichthe model mimics the observed diurnal variation andthe observed soundings.

A comparison of the surface wind speed and direc-tion, potential temperature, and water vapor mixing ratioon days 2 and 3 of the model calculation with the five-week mean obtained at Burketown in 1991 is shown inFig. 6. The model data are based on hourly output. Onthe whole, the comparison is remarkably good, the di-urnal variation of all parameters being well captured by


FIG. 8. Time–height isotach cross sections of (a) the zonal (u) and(b) meridional (y) wind components, and (c) the corresponding is-entrope cross sections for the 72-h period commencing at 1400 LSTon day 1 at Burketown. Dashed lines in (a) and (b) indicate negativecontour values, i.e., an easterly or northerly component of flow asappropriate.

the model, notably the wind direction, although the lateafternoon wind speeds are too light. Also, the model-predicted potential temperatures are a little too cool dur-ing the afternoon. The mixing ratio is a few grams perkilogram more than that observed, a feature that is prob-ably a result of the sea surface temperature (SST) in thegulf being a little cooler than that in the model. In themodel, the SST is the same for all bodies of water andis based on November conditions, whereas the obser-vations were carried out in September and early October.

The precise timing of the model morning glory onsetis difficult to determine from Fig. 6, but it would appearto be between 0400 and 0600 LST. At Burketown, north-easterly morning glories are relatively uncommon be-fore 0500 LST; typically the surge arrives within thetime period 0500–1000 LST, the mean time being about0730 LST. Thus, the onset in the model is at the early

end of the observed range. This suggests that the 8 ms21 easterly flow is on the high side of the normal rangeof values over the region.

Figure 7 compares the day 2 to day 4 average verticalprofile of potential temperature at Burketown at 0430LST with the mean soundings observed there on morn-ing glory days during the 1991 experiment. In the mod-el, the potential temperatures below 3 km are 18–38Ccooler than those observed, the largest difference beingbelow 1 km. The mixed layer, itself, is better definedin the observations, but an average for the early periodof the experiment shows closer agreement with the mod-el (see Menhofer et al. 1997).

Figure 7 shows also the average of the day 3 and day4 vertical profiles of potential temperature at 0000 LSTat Macaroni Station on the western side of the peninsula.The actual depth of the mixed layer is at the lower end


FIG. 9. Legend as for Fig. 8 except for Macaroni Station.

of the range that is typically observed over the westernside of the peninsula during the late dry season; in Oc-tober this is around 3–4 km (Clarke 1983). The reasonis probably because of the relatively coarse vertical res-olution of the boundary layer above 1500 m. The profileon day 1 of the simulation has a mixed-layer depth ofonly 2.5 km.

Figure 8 shows time–height isotach cross sections ofu and y and the corresponding isentrope cross sectionsfor the 72-h period commencing at 1400 LST on day 1at Burketown where many observational data have beenobtained (see Smith 1988 for references). Notable fea-tures of the isotachs include the onset of a shallow west-erly component of flow (u . 0), which commences inthe late afternoon as the northerly (onshore) sea-breezeflow and its remnants rotate anticyclonically throughnorth. Prior to the morning glory onset between 0400

and 0600 LST, this low-level flow has become a south-westerly, as is normally observed (see, e.g., Clarke etal. 1981). The easterly component of the morning glorysurge is evident up to 3 km, being a northeasterly belowabout 2 km and a southeasterly above. Thus the upperpart of the mixed layer is influenced by conditions tothe south over land. This is of consequence for the de-velopment of a suitable waveguide for the morning glo-ry waves, which includes a deep near-neutrally stablelayer above the surface-based stable layer. The shallowsouthwesterly flow enhances the waveguide propertiesas well (Crook 1988).

Principal features of the isentropes include the for-mation of a deep well-mixed layer during the daytimeand the development during the night of a shallow ra-diation inversion; a stable layer in the lowest 500 m orso, which is most pronounced in the early morning


FIG. 10. Legend as for Fig. 5 except for the calculation without orography.

FIG. 11. Legend as for Fig. 5 except for the calculation with a northeasterly geostrophic flow.

hours; and a relatively deep layer of air above this withonly weak vertical gradients extending to between 2.5and 3.0 km. All these features are in accord with ob-servations. Note that the stable layer is mostly a resultof the inland penetration of sea-breeze air from the gulf,whereas the nearly neutral layer above it is the legacyof air that has been well mixed over land on the previousday. Note also that the morning glory surge is accom-panied by a decrease in potential temperature in thelowest few hundred meters, a consequence of the ad-vection and lifting of low-level air from the gulf (Fig.3d). It is apparent in Fig. 8c that the depth of the nearlyneutral layer steadily increases during the simulation ifone judges this depth by following the 306-K contour

on days 1 and 2 and then the 305-K contour on day 3.A possible reason for this is discussed below.

Figure 9 shows similar time–height isotach cross sec-tions for Macaroni Station on the western side of thepeninsula. Here, the onset of the morning glory surgeoccurs at between 2200 and 2300 LST, which is slightlyearlier than is typically observed. The 11 disturbancesdocumented at Macaroni during a 12-day period in Oc-tober 1981 occurred between 2304 and 0212 LST, themean time being 0051 LST (Clarke 1983, Table 4). Thesurge is manifest as a strong northeasterly flow belowabout 2 km with a southeasterly flow aloft and is pre-ceded by a shallow (,200 m deep) inland flow—thesea breeze from the gulf. The winds revert to near east-


FIG. 12. Vector plots of the surface wind in the calculation with a northeasterly flow at 0200 LST: (a) on day 2 and(b) on day 3.

erly at all levels as the convective mixing reaches itspeak. At Macaroni, the northeasterly surge is accom-panied by cold-air advection in a layer 1.0–1.5 km deep.

As at Burketown, the depth of the nearly neutral layerat Macaroni steadily increases during the simulation ifone judges this depth by following the 305-K contourin Fig. 9c. The reason for this appears to be that weused a constant sea surface temperature, which is char-acteristic of the gulf everywhere in the model. In reality,the gulf temperature is a degree or two warmer thanthat of the Coral Sea. This exposes a weakness of thepresent formulation, because the initial sounding weused is not in radiative–convective equilibrium in themodel and the imposed geostrophic flow warms up alittle during the integration. This is reflected also in thegrowth in depth of the mixed layer over the Coral Seaseen in Fig. 3; note that the mixed-layer depth in theinitial sounding is only 300 m. We have evidence thatthis will not have a large influence on the strength ofthe surge. One of us (JN) carried out a number of two-dimensional sensitivity studies in which the initialsounding was varied, but the collision of the sea breezesoccurred at more or less the same time and place, andthe resulting disturbances had similar strengths.

5. Orographic effects

The effects of orography on the mesoscale circula-tions over the peninsula can be judged by comparingthe control simulation with the equivalent one in whichthe orography is omitted. Figure 10 shows a space–timecross section of the near-surface u and y isotachs forthis calculation, which is again along the latitude circlethat passes through Mornington Island. These fieldsshould be compared with the corresponding ones in Fig.

5. A number of differences are evident. In the run with-out orography, the large daytime maximum in the east-erly flow on the eastern side of the peninsula is absent,showing that, in the control calculation, this feature isan orographic effect. In addition, the region of westerlywinds marking the sea breeze along the west coast ofthe peninsula remains offshore and does not extend overmore than half of the gulf in this section. The maximumwind speeds (2 m s21 on day 1) are also weaker thanthose in the control calculation that exceed 4 m s21, andthey occur in the early evening, around 1800 LST. Inthe absence of orography, the line of strong convergencestill occurs, but it develops entirely over the gulf andcorresponds with a transition from a southeasterly to anortheasterly wind.

The calculation without orography is similar in mostrespects to that in NS, but because the easterly flow is3 m s21 stronger than that in NS, the westward pro-gression of the convergence line is more rapid. In par-ticular, comparison of this calculation with that in NSshows little evidence that the sea-breeze circulations onthe western side of the gulf have a significant effect onthose on the eastern side and in particular on the for-mation of convergence lines over the gulf. Thus anypersisting effects of inertial flow induced over the gulfby these sea breezes do not appear to be of consequencefor the flow patterns over the peninsula and eastern gulfregion.

6. Northeasterly and southeasterly basic flows

Some of the essential differences between calcula-tions with a uniform northeasterly or southeasterly flowand the control calculation are brought out by a com-parison of the space–time cross section of the near-sur-


FIG. 13. Legend as for Fig. 5 except for the calculation with a southeasterly geostrophic flow.

face u and y isotachs along the latitude circle that passesthrough Mornington Island. Figure 11 shows these fieldsfor an 8 m s21 northeasterly flow. In this case, the near-surface flow patterns are broadly repeatable from dayto day, as in the control calculation. The westerly sea-breeze flow over the gulf penetrates a little farther inlandthan in the control calculation but does not extend sofar westward. A surface convergence line forms againover the western side of the peninsula in the late eveningand propagates westward across the gulf. As would beexpected, the southerly component of the winds pre-ceding this line is weaker than that in the control cal-culation. In addition, the strength of the convergenceweakens earlier. The surface winds at 0200 LST on days2 and 3 for this calculation are shown in Fig. 12. Notethat the convergence line over the gulf is extensive andaffects the area along the southern gulf coast.

The situation is somewhat different in the case of an8 m s21 southeasterly flow, for which the near-surfaceu and y isotachs along the latitude circle that passesthrough Mornington Island are shown in Fig. 13. Whilesome of the features in these fields are repeatable fromday to day, others are not. A strong westerly sea-breezeflow forms in the late morning and penetrates muchfarther inland than in the control calculation. Moreover,the strength of this flow increases from day to day. Ondays 1 and 2, the winds over the gulf have a mostlysoutherly component, but on the afternoon of day 3 andearly morning of day 4 this component reverses sign,an indication that the long inertial period in the gulfregion is important in this case. As in the other cases,the formation of a convergence line over the peninsulaoccurs in the evening and again this line translates west-ward during the night. However, the line does not extendas far southward as the southern gulf coast, as is evidentin the surface wind fields shown in Fig. 14. In particular,

the convergence line does not reach Burketown on eitherday shown, a feature that is in accord with the obser-vation that morning glory disturbances there are rarelyobserved when the winds are southeasterly. Note alsothe differences between the flow over the gulf at 2200LST on the two days. On day 2 the flow over the centraland western parts has a pronounced southerly compo-nent, whereas on day 3 the winds there are northerly inthe south and light westerly in the north.

Observations show that during the late dry season,morning glory generation is almost a nightly occurrenceover Cape York Peninsula, but only a fraction of thesedisturbances apparently survive until sunrise on the fol-lowing morning, judging by their arrival at Burketown.Furthermore, recent field observations, including dailypredisturbance radiosonde soundings at Burketown donot show any significant difference in waveguide struc-ture between days on which disturbances survive tocross the southern part of the gulf and those that do not.Our model calculations suggest that the broadscale winddirection is a key factor here. They show that for south-easterly flow, convergence lines form on the peninsulabut do not extend to the southern gulf coast. The broad-scale wind speed may be important also. In calculationsnot reported here, a reduction in the geostrophic windspeed delays the formation of the morning glory con-vergence line on the western side of the peninsula. Dis-turbances that arrive there beyond a certain time wouldnot reach Burketown until the morning glory waveguidehas been destroyed in the Burketown region by renewedconvective mixing.

7. Circulation over the northern gulf region foreasterly flow

The NACL was the subject of an observational studyin October 1986 (Holland et al. 1986; Drosdowsky et


FIG. 14. Vector plots of the surface wind in the calculation with a southeasterly flow at 2200 LST (a) on day 1 and(b) on day 2, and at 0600 LST (c) on day 2 and (d) on day 3. The differences between the wind fields at correspondingtimes are discussed in the text.

al. 1989) and as part of this experiment, serial radio-sonde soundings were carried out at Weipa. An extraor-dinary feature of these soundings was the occurrenceof an enhanced easterly component of flow with its max-imum amplitude around 900 mb (1 km) in the morninghours, typified by the 0300 and 0900 LST soundings(see Drosdowsky et al. 1989, Fig. 2). This low-level jetwas not accompanied by a prominent signal in the me-ridional component of the flow; hitherto, no explanationfor it has been advanced. Its occurrence is significantin that rawinsondes at that location may not be repre-sentative of the surrounding region—that is, the regionwithin a radius of half the distance to the next rawin-sonde station. Interestingly, a similar feature is presentin the control calculation. This is evident from the time–height isotach cross sections of the zonal and meridionalwind components for the 72-h period commencing at

1400 LST on day 1 at Weipa (shown in Fig. 15). Thereason for the feature is suggested by Fig. 16, whichshows a space–time cross section of the u and y isotachsat a height of 500 m above the surface calculated alonga latitude circle that passes through Weipa for the 72-hperiod commencing at 1400 LST on day 1. The accel-erated easterly flow occurs first over the eastern side ofthe peninsula and extends westward with time (Fig.16a). The evolution of the flow in this cross section issimilar to that described earlier (illustrated in Fig. 17),which shows longitude–height cross sections of the is-entropic surfaces below 3.5 km along the latitude ofWeipa at four hourly intervals between 1400 LST onday 1 and 0200 LST on day 2, and in Fig. 18, whichshows isotach cross sections of the zonal wind com-ponent for the period 1800 LST on day 1 to 1400 LSTon day 2. One essential difference between the behavior


FIG. 15. Time–height isotach cross sections of the zonal and meridional wind components for the 72-h periodcommencing at 1400 LST on day 1 at Weipa. Dashed lines in (a) and (b) indicate negative contour.

FIG. 16. Space–time cross section of (a) u isotachs and (b) y isotachs at a height of 500 m above the surface, calculatedalong the meridian that passes through Weipa in the north of the gulf for the same period as in Fig. 15. The dashedlines indicate negative values. The vertical lines mark the coastlines.

in these fields and that in the Mornington Island crosssection is that the peninsula is much narrower at thelatitude of Weipa. It is clear that the acceleration of theeasterlies is associated with the zonal temperature gra-dient that develops over the eastern side of the peninsuladuring the late morning and early afternoon. The max-imum gradient translates westward with time and is sit-uated near the west coast of the peninsula at 1800 LST,the remnants of the convectively well-mixed layer hav-ing been advected out over the gulf by this time (Fig.17b). The disturbance continues to propagate westwardovernight as a long wavelike disturbance. Examination

of the isotach fields in Fig. 18 shows an eastward tiltof the disturbance with height. For an easterly basicflow, this is indicative that the disturbance is radiatingenergy vertically.

The easterly surge attains its maximum developmentover the whole peninsula at 0600 LST and is confinedlargely to a layer below 1 km in height, which is con-sistent with the observations. However, a second max-imum in the easterlies develops over the gulf in asso-ciation with a feature that originally formed the returnbranch of the sea-breeze circulation near the west coastof the peninsula. By 1000 LST, this feature is already


FIG. 17. Vertical cross sections of the isentropic surfaces below 3.5 km along the meridian that passesthrough Weipa in the north of the gulf at selected times: (a) 1400 LST on day 1, (b) 1800 LST on day 1,(c) 2200 LST on day 1, and (d) 0200 LST on day 2.

over Gove on the west coast of the gulf and has theform of an easterly jet located at about 2 km. Again,the feature is significant because it calls into questionthe representativeness of the routine 2300 UTC rawin-sonde soundings at Gove for the surrounding region.

The remains of the easterly surge over the peninsulapersist in weakened form until at least 1000 LST (Fig.18f), and the surge is at its weakest at 1400 LST onday 2, albeit a little stronger than at the same time asday 1. Thereafter, a similar diurnal sequence is resumed.The low Coriolis acceleration in these latitudes explainsthe relatively weak signatures in the meridional windthat occur in the model (Fig. 16b) and those observedduring the 1986 experiment (not shown.)

8. The longevity of morning glory disturbances

The present calculations, together with the results ofa recent observational study, lead us to new interpre-tations concerning the dynamics of the morning glory.

Studies of solitary wave dynamics, believed to be ap-plicable to the component bore waves of the disturbance,suggest that the waves should decay rapidly by the up-ward radiation of energy since the waveguide on whichthey propagate in the gulf region is always sufficientlyleaky (see, e.g., Mitsudera and Grimshaw 1991; R.Grimshaw 1996, personal communication). Estimatesbased on linear wave theories suggest that sinusoidalwaves having the same wavelength as morning glorywaves (typically 10–20 km) would lose a significantfraction of their energy by upward radiation before theytravel one wavelength (see, e.g., Crook 1988). Theseresults appear to be at odds with observations that showthe waves to be relatively long lived. Furthermore, ra-diosonde soundings in the pre–morning glory environ-ment at Burketown show that northeasterly morning glo-ries may reach Burketown even on days when the nearlyneutrally stable layer above the surface-based stable lay-er is absent (Menhofer et al. 1997), as in the modelcalculation (see section 4b). Our calculations lead us to


FIG. 18. Vertical isotach cross sections of the zonal wind component below 3.5 km along the same latitudecircle as in Fig. 16 at selected times: (a) 1400 LST on day 1, (b) 1800 LST on day 1, (c) 2200 LST on day1, (d) 0200 LST on day 2, (e) 0600 LST on day 2, (f) 1000 LST on day 2, and (g) 1400 LST on day 2.The dashed lines indicate negative values.


FIG. 18. (Continued)

hypothesize that the convergence line and bore are oneand the same feature and are continuously forced by theenergetic mesoscale circulations that develop over thepeninsula. Indeed, the calculations show that these cir-culations are thermally forced, a manifestation of thesea breezes over the peninsula and gulf seaboard. Wehypothesize further that the energy losses associatedwith the leakiness of the waveguide in morning glorywave disturbances are at least partially offset by energygains associated with the evolving mesoscale patternsgenerated by sea-breeze circulations. These hypothesescannot be tested using the present model: first, the modelresolution is far too coarse to represent the individualwaves, and second, the model is hydrostatic, an as-sumption that would preclude the formation of solitarywaves in deep fluids (see, e.g., Christie 1992). Note wedo not suggest that the waveguide is of no importanceat all; in fact the numerical model calculations by Crook(1988, Fig. 2) show that wave generation is favored bya strong waveguide. However, they show also that inthe absence of wave trapping, a borelike disturbancestill develops, even though the amplitude of the borewaves is significantly reduced in this case.

9. Conclusions

We have presented a comprehensive numerical studyof the thermally forced mesoscale circulations that arecommon flow features of the lower atmosphere over theGulf of Carpentaria region of northern Australia formuch of the year. The results, which indicate the regularformation of a nocturnal convergence line at low levelsover the gulf, complement and extend those of our ear-lier study. Calculations for a range of uniform geo-strophic flows over the region, with directions typicalof the dry season conditions, show that the developmentover the gulf of westward-moving lines of low-levelconvergence is the rule. This explains why traveling

convective- and wave-cloud lines are commonly ob-served over the gulf. The convergence lines are asso-ciated with sea-breeze circulations that develop overCape York Peninsula and around the gulf. For easterlyand northeasterly geostrophic flows, the circulations thatdevelop are broadly repeatable from day to day, despitethe relatively long inertial period (nearly two days) inthe region. However, this is not the case for a south-easterly flow. It does not appear that the sea-breeze cir-culations on the western side of the gulf have a signif-icant effect on those on the eastern side and in particularon the formation of convergence lines over the gulf.

The presence of the orography on Cape York Pen-insula enhances the low-level easterly flow over the east-ern side of the peninsula but delays the formation ofthe morning glory convergence line on the western side.

The quasi-regular, diurnally varying patterns of low-level flow over the gulf region, and in particular at coast-al locations around the gulf, call into question the rep-resentativeness of rawinsonde soundings carried outthere. This is because the flow patterns have timescalescomparable with or less than the frequency of rawin-sonde ascents (i.e., 6 h) and they have a spatial scalethat is appreciably smaller than the separation betweenrawinsonde stations in the region. Moreover, these flowfeatures may not be captured by current numerical fore-cast models. Similar questions might be asked of sound-ings at other locations in the Tropics where thermallyinduced circulations are large and not well representedin forecast models.

Finally, our calculations lead us to reinterpret earlierideas concerning the longevity of the waves. They sug-gest that the bore-wave system should be regarded ascontinuously forced by the mesoscale circulations thatarise in the area so that energy loss by the individualwaves on account of the vertical radiation of wave en-ergy do not lead to rapid wave decay as might otherwisebe the case. In turn, the mesoscale circulations are as-sociated principally with thermally forced circulationsassociated with the sea breezes over the peninsula andgulf seaboard.

Acknowledgments. We are grateful to Dr. AndrewCrook at the National Center for Atmospheric Researchand two anonymous reviewers for their perceptive com-ments on the original version of the manuscript.

REFERENCES

Christie, D. R., 1992: The morning glory of the Gulf of Carpentaria:A paradigm for non-linear waves in the lower atmosphere. Aust.Meteor. Mag., 41, 21–60.

Clarke, R. H., 1983: Fair weather nocturnal inland wind surges andatmospheric bores: Part II. Internal atmospheric bores in northernAustralia. Aust. Meteor. Mag., 31, 147–160., 1984: Colliding sea-breezes and the creation of internal at-mospheric bore waves: Two-dimensional numerical studies.Aust. Meteor. Mag., 32, 207–226., R. K. Smith, and D. G. Reid, 1981: The morning glory of the


Gulf of Carpentaria. An atmospheric undular bore. Mon. Wea.Rev., 109, 1733–1757.

Crook, N. A., 1988: Trapping of low-level internal gravity waves. J.Atmos. Sci., 45, 1533–1541.

Drosdowsky, W., and G. J. Holland, 1987: North Australian cloudlines. Mon. Wea. Rev., 115, 2645–2659., , and R. K. Smith, 1989: Structure and evolution of NorthAustralian cloud lines observed during AMEX Phase I. Mon.Wea. Rev., 117, 1181–1192.

Holland, G. J., J. L. McBride, R. K. Smith, D. Jasper, and T. Keenan,1986: The BMRC Australian Monsoon Experiment: AMEX.Bull. Amer. Meteor. Soc., 67, 1466–1472.

Louis, J. F., 1979: A parametric model of vertical eddy fluxes in theatmosphere. Bound.-Layer Meteor., 17, 187–202.

McNider, R. T., and R. A. Pielke, 1981: Diurnal boundary-layer de-velopment over sloping terrain. J. Atmos. Sci., 38, 2198–2212.

Menhofer, A., R. K. Smith, M. J. Reeder, and D. R. Christie, 1997:‘‘Morning-glory’’ disturbances and the environment in whichthey propagate. J. Atmos. Sci., 54, 1712–1725.

Mitsudera, H., and R. Grimshaw, 1991: Effects of radiative dampingon resonantly generated internal gravity waves. Stud. Appl.Math., 84, 183–206.

Noonan, J. A., and R. K. Smith, 1986: Sea-breeze circulations overCape York Peninsula and the generation of Gulf of Carpentariacloud line disturbances. J. Atmos. Sci., 43, 1679–1693., and , 1987: The generation of North Australian cloud linesand the ‘morning glory.’ Aust. Meteor. Mag., 35, 31–45.

Physick, W. L., J. A. Noonan, J. L. McGregor, P. J. Hurley, D. J.Abbs, and P. C. Manins, 1994: LADM: A Lagrangian Atmo-spheric Dispersion Model. CSIRO Division of Atmospheric Re-search Tech. Paper 24, 137 pp. [Available from The Chief, CSI-RO Division of Atmospheric Research, Private Bag 1, Mor-dialloc, Victoria, 3195, Australia.]

Smith, R. K., 1988: Travelling waves and bores in the lower atmo-sphere: The ‘morning glory’ and related phenomena. Earth-Sci.Rev., 25, 267–290., and M. A. Page, 1985: Morning glory wind surges and theGulf of Carpentaria cloud line of 25–26 October 1984. Aust.Meteor. Mag., 33, 185–194., M. J. Coughlan, and J. Evans-Lopez, 1986: Southerly nocturnalwind surges and bores in northeastern Australia. Mon. Wea. Rev.,114, 1501–1518., M. J. Reeder, N. J. Tapper, and D. R. Christie, 1995: CentralAustralian cold fronts. Mon. Wea. Rev., 114, 19–38.

generation of low-level mesoscale convergence lines over …roger/... · 1998-03-11 · thermally...

Documents