Frontal Forcing of an Extreme Snow Event - Marrella - University of

20 Pages·1998·1.19 MB·English

FEBRUARY1998 MARTIN 329
The Structure and Evolution of a Continental Winter Cyclone.
Part II: Frontal Forcing of an Extreme Snow Event
JONATHAN E. MARTIN
DepartmentofAtmosphericandOceanicSciences,UniversityofWisconsin—Madison,Madison,Wisconsin
(Manuscriptreceived5September1996,infinalform9June1997)
ABSTRACT
Theproductionofanarrow,heavy,occasionallyconvectivesnowbandthatfellwithinamodestsurfacecyclone
on19January1995isexaminedusinggriddedmodeloutputfromasuccessfulnumericalsimulationperformed
usingtheUniversityofWisconsin—NonhydrostaticModelingSystem.Itisfoundthatthesnowbandwasproduced
by a thermally direct vertical circulation forced by significant lower-tropospheric warm frontogenesis in the
presence of across-front effective static stability differences as measured in terms of the equivalent potential
vorticity (PV). The sometimes convective nature of the snowband resulted from the development of freely
e
convectivemotionsforcedbyfrontalliftingoftheenvironmentalstratification.
Modeltrajectoriesdemonstratethatastreamofwarm,moistairascendedthroughthetrowalportionofthe
warm-occluded structure that developed during the cyclone life cycle. The lifting of air in the trowal was, in
thiscase,forcedbylower-troposphericfrontogenesisoccurringinthewarm-frontalportionofthewarmocclusion.
This trowal airstream accounts for the production of the so-called wrap-around precipitation often associated
withoccludedcyclonesand,inthiscase,accountedforthenorthernthirdoftheheavysnowband.
1. Introduction andBlakely1988),andlargesynoptic-scaleliftingmod-
ified by local terrain effects (Dunn 1988). Conceptual
WintercyclonesinthecentralUnitedStatesareoften
understandingoftheprecipitationdistributioninwinter
accompanied by a variety of hazardous meteorological
cyclones has also been distilled into a variety of fore-
elements including strong winds, subfreezing temper-
casting ‘‘rules of thumb.’’ These include thelikelihood
atures, freezing rain, torrential rains, and heavy snow.
that heavy snow will fall 200–250 km to the leftofthe
The explosively deepening cyclones that occasionally
track of the associated surface cyclone (NWSFO, Sul-
visit this region are nearly always attended by some
livan, Wisconsin). This region of a mature cyclone is
combination of these conditions (see, e.g., Schneider
often the occluded quadrant of the storm. It is not un-
1990; Marwitzand Toth 1993;MassandSchultz1993;
common for experienced forecasters to refer to the
Hakim et al. 1995). The much more common modest
heavy precipitation that often falls in this quadrant of
cyclones (i.e., modest as measured in terms of their
the storm as ‘‘wrap around’’ precipitation. This term
minimum sea level pressure) can also be accompanied
indicates that a portion of the cloud and precipitation
by dangerous weather. In fact, it is not uncommon for
shieldhaswrappeditselfaroundtothewestofthevortex
such ‘‘garden variety’’ winter cyclones to be attended
center. It also, incorrectly, connotes that precipitation
by near-blizzard conditions and resulting heavy snow-
falls (Moore and Blakely 1988; Shields et al. 1991; can be horizontally advected to different locations in a
Hakim and Uccellini 1992; Funk et al. 1995; Shea and cyclone.
Przybylinski 1993). In the 1950s, scientists at the Canadian Meteoro-
Previous studies of snowbands in cyclonic storms logical Service developed a conceptual model for the
have identified a number of processes responsible for structure of some North American cyclones that they
the production of precipitation. Among these are jet termedthe‘‘threefrontmodel’’(Godson1951;Penner
streak interactions (Hakim and Uccellini1992),fronto- 1955;Galloway1958,1960).Theyfoundthatthiscon-
genesis in the presence of small moist symmetric sta- ceptual model was operationally very useful. A com-
bility (Sanders and Bosart1985; Gyakum1987;Moore ponent of this conceptual model (devised by using
frontalcontourcharts)wasthetroughofwarmairaloft
(trowal), which represented the axis of highest poten-
tial temperature (cid:117)ahead of the upper cold front in a
Corresponding author address: Dr. Jonathan E. Martin, Dept. of warm-occludedcyclone.Thetrowalwasshowntohave
AtmosphericandOceanicSciences,UniversityofWisconsin—Mad-
a better correspondence to the sensible ‘‘weather’’as-
ison,1225WestDaytonSt.,Madison,WI53706.
E-mail:[email protected] sociated with a mature cyclone than was the surface
(cid:113)1998AmericanMeteorologicalSociety
330 MONTHLY WEATHER REVIEW VOLUME126
FIG.2.Mapofallcloud-to-groundlightningstrikesassociatedwith
thesnowbandinthe24-hperiod0000UTC19Januaryto0000UTC
20January1995.Crossesindicatepositionsofthelightningstrikes.
2. Observations of the snowband
From 1800 UTC 18 January until0600 UTC20Jan-
uary1995amodestsurfacecycloneaffectedthecentral
UnitedStates.Foradetaileddescriptionofthesynoptic
evolutionofthiscycloneanditsattendantfrontalstruc-
FIG.1.Observed24-hsnowfalltotalsfrom0000UTC19January
to0000UTC20January1995.Totalsare given incentimetersand ture, the reader is referred to Part I (Martin 1998). De-
arecontouredevery10cmexceptfortheoutermostcontour,which spite the modest nature of the surface disturbance (its
representstraceamountsto5cm. sealevelpressureneverfellbelow997hPa),thisstorm
was attended by a band of heavy snow that stretched
from Tulsa, Oklahoma,tonorthofGreenBay,Wiscon-
sin (Fig. 1). Despite the synoptic-scale length of this
warm-occludedfront(seePenner’sFig.1).Asamature
band (nearly 1100 km), it was only a few hundred ki-
cyclone continues to develop, the trowal can be
lometers wide. In fact, the half-width of the snowband
wrapped cyclonically around the cyclone center and
(defined as the lateral distance from the axis of maxi-
can end up in the northwest quadrant of the decaying
mum snowfall to a point receiving half the maximum
storm. The relationship between the trowal and heavy
amount)averagedonly63kmalongitsentirelengthand
snowbands, although a reasonable and intriguing ex-
was locally much narrower in central Missouri. Many
tension of prior work, has never been demonstratedin
locations affected by this precipitation featurereceived
the literature.
total snow accumulations from this single event that
InPartI(Martin1998),thefrontalstructureandevo-
exceeded seasonal climatology.
lutionofamodestcyclonethataffectedthecentralUnit-
Anadditionalcompellingcharacteristicofthissnow-
ed States on 19 January 1995 was described. In the
bandwasitsconsiderableconvectivenature,partialev-
present paper the extreme snowfall event that accom-
idenceforwhichisthedensityofcloud-to-groundlight-
panied this storm is examined. A finescale numerical
ning strikes associatedwithit(Fig.2).Althoughsnow-
model simulation of this cyclone is employed to de-
fall accompanied by lightning is not an exceedingly
scribe the meso–synoptic-scale dynamic and thermo-
anomalous occurrence in the central United States, the
dynamic circumstances that conspired to produce this precisefrequencyofoccurrenceandthemeanflashden-
recordevent.Insection2observationsofthesnowband sityofsucheventsisunknown.Inadditiontothecloud-
are presented. In section 3 we describe the numerical to-ground discharges shown in Fig. 2, in-cloud and
model used to simulate this case and verify the precip- cloud-to-cloudlightningwasobservedintermittentlyin
itation forecasts made by the model against the obser- southern Wisconsin between 1800 and 2200 UTC 19
vations.Analysisofthemechanismsresponsibleforthe January in association with this event (personal obser-
productionofthesnowfallispresentedinsections4and vation).
5. This analysis will include extensive model-basedair Central Missouri was hardest hit by the convective
parcel trajectories, which will illustrate the airflow snowfallassociatedwiththiscyclone.Atmanylocations
through the precipitation generation regions of the cy- theoccasionalthundersnowfellatratesexceeding2(cid:48)(5
clone. We will discuss the results of this analysis in cm) per hour. As an example of the extremity of this
section 6 and offer conclusions in section 7. event in central and southern Missouri a time series of
FEBRUARY1998 MARTIN 331
TABLE1.HourlytimeseriesofsurfaceobservationsatSpringfield,
Missouri(SGF),from2100UTC18Januaryto2100UTC19January
1995.TemperatureTanddewpointT aregivenindegreesCelsius,
d
windspeed(SPD)isgiveninmeterspersecond,winddirection(DIR)
is given in degrees with 360(cid:56) as north, sea level pressure (SLP) is
given in hectopascals, Rand Ssignify rain and snow, respectively,
withtheminusandplusindicatinglightandheavyintensities.OVC
indicatesovercastskiesandXindicatesobscuredsky.
Time T T DIR SPD SLP COV WX
d
2100 1(cid:56) 0(cid:56) 020 7.5 1012.6 OVC R(cid:50)
2200 1(cid:56) 0(cid:56) 020 7.0 1012.3 OVC S(cid:50)
2300 0(cid:56) (cid:50)1(cid:56) 020 9.0 1012.1 OVC S
0000 0(cid:56) (cid:50)1(cid:56) 010 8.5 1011.8 OVC S(cid:49)
0100 0(cid:56) (cid:50)1(cid:56) 010 8.0 1011.8 X S(cid:49)
0200 (cid:50)1(cid:56) (cid:50)1(cid:56) 020 8.5 1011.1 X S(cid:49)
0300 (cid:50)1(cid:56) (cid:50)2(cid:56) 360 7.5 1011.1 X S(cid:49)
0400 (cid:50)2(cid:56) (cid:50)2(cid:56) 360 8.0 1010.1 X S(cid:49)
0500 (cid:50)2(cid:56) (cid:50)2(cid:56) 350 9.0 1009.7 X S(cid:49)
0600 (cid:50)2(cid:56) (cid:50)3(cid:56) 360 8.0 1009.1 X S(cid:49)
0700 (cid:50)2(cid:56) (cid:50)2(cid:56) 340 8.0 1007.9 X S(cid:49)
0800 (cid:50)2(cid:56) (cid:50)2(cid:56) 340 8.5 1007.3 X S(cid:49)
0900 (cid:50)2(cid:56) (cid:50)2(cid:56) 340 9.5 1006.6 X S(cid:49)
1000 (cid:50)2(cid:56) (cid:50)2(cid:56) 340 9.5 1006.4 X S(cid:49)
1100 (cid:50)2(cid:56) (cid:50)2(cid:56) 340 9.5 1006.5 X S(cid:49)
1200 (cid:50)2(cid:56) (cid:50)2(cid:56) 330 10.0 1007.0 X S(cid:49)
1300 (cid:50)1(cid:56) (cid:50)2(cid:56) 330 10.0 1008.1 X S FIG.3.Geographiclocationsofthethreegridsusedinthenumeri-
1400 (cid:50)1(cid:56) (cid:50)2(cid:56) 320 10.5 1009.0 X S calsimulationdescribedinthetext.
1500 (cid:50)1(cid:56) (cid:50)1(cid:56) 320 10.0 1009.3 X S
1600 (cid:50)1(cid:56) (cid:50)1(cid:56) 320 10.0 1010.6 OVC S
1700 (cid:50)1(cid:56) (cid:50)1(cid:56) 320 9.5 1011.3 OVC S(cid:50) ried by the model include u, (cid:121), w, and (cid:112) (Exner
11890000 01(cid:56)(cid:56) (cid:50)(cid:50)11(cid:56)(cid:56) 332400 1100..00 11001111..54 OOVVCC SS(cid:50)(cid:50) function);ice–liquidpotentialtemperature(cid:117)il;andtotal
2000 2(cid:56) (cid:50)1(cid:56) 320 10.0 1011.7 OVC S(cid:50) water mixing ratio, as well as the mixing ratios for a
2100 2(cid:56) (cid:50)2(cid:56) 330 8.0 1012.4 OVC variety of precipitation particles.
Advectionofthescalarvariablesisaccomplishedusing
asixth-orderCrowleyscheme(Trembachetal.1987),and
hourly observations from Springfield, Missouri (SGF), the dynamic variables are advected using a second-order
isshowninTable1.From0000to1200UTC19January enstrophy-conserving leapfrog scheme (Sadourny 1975).
SGFreported13consecutivehoursofheavysnowwith Model physics include a radiation parameterization that
nearly zero visibility and wind gusts topping 20 m s(cid:50)1. predictslong-andshortwaveradiativetransferinacloudy
Intheend,nearly16(cid:48)(40cm)ofsnowfellatSpringfield. atmosphere(ChenandCotton1983),andapredictivesoil
Columbia, Missouri, received a record 19.7(cid:48) (50 cm), modelwithsurfaceenergybudget(TrembachandKessler
which forced the closing of the University of Missouri 1985). Liquid and ice processes are represented in the
for the first time in 17 years. modelbyanexplicitmicrophysicspackagethatdescribes
We now proceed to an analysis of the circumstances the evolution of cloud water, rainwater, pristine crystals,
that conspired to produce this extreme snowfall event. snow crystals, aggregate crystals, and graupel (Cotton et
This analysis will employ gridded output from a suc- al. 1986; Flatau et al. 1989). A version of the Emanuel
cessful numerical simulation of this cycloneperformed (1991) convective parameterization was employed, mod-
using the University of Wisconsin—Nonhydrostatic ifiedsuchthattheconvectionequilibrateswiththecyclone
ModelingSystem(UW-NMS),whichisdescribedinthe and frontal-scale vertical motion forcing (Tripoli 1996,
following section. personal communication). This modification reduces the
sensitivity of the parameterization to the amount of con-
vective available potential energy (CAPE) by tying the
3. Numerical model simulation
releaseofCAPEtothesynoptic-andfrontal-scalevertical
To further elucidate the circumstances leading tothe motions.
developmentofthesnowband,outputfromanumerical Three grids were used in the simulation. Grid 1 (outer
forecast of this cyclone made using the University of grid), grid 2 (middle grid), and grid 3 (inner grid) had
Wisconsin—Nonhydrostatic Modeling System (UW- horizontal resolutions of 160 km, 80 km, and 40 km,
NMS) is used. UW-NMS is described by Tripoli respectively.Thedatafromgrid3wasusedinthisstudy.
(1992a,b). The model employs a two-way interactive, The geographic locations of these grids are shown in
moveable nesting scheme, which allows for the simul- Fig 3.
taneous simulation of large synoptic-scale forcing as Themodelemployedgeometricheightasthevertical
well as frontal-scale forcing. Prognostic variables car- coordinate withdiscretelyblockedouttopographysim-
332 MONTHLY WEATHER REVIEW VOLUME126
themodeldata.Asaresult,thegriddedoutputfromthis
simulation will be used to describe the mesosynoptic
environment in which the record snowfall that charac-
terized this cyclone developed.
4. Forcing of the snowband
As noted in the previous section, the snowband that
accompanied this modest cyclone was noteworthy for
a number of reasons: 1) the snowfall totals were ex-
ceptional for the region, 2) the band was characterized
by intermittent convection throughout most of its life
history, and 3) the band had synoptic-scale length
((cid:59)1100 km) but mesobeta-scale width (especially in
terms of its half-width). In this section and the next, a
diagnosis of the dynamic and thermodynamic circum-
stancesthatledtothedevelopmentandaforementioned
FIG.4.UW-NMSmodelpredictedsnowfalltotalsfrom0000UTC characteristics of the snowband will be presented.This
19 January to 0000 UTC 20 January 1995. Labeled and contoured diagnosis will center on the role of frontogenesis in
asinFig.1.
generatingtherequisiteverticalmotionforprecipitation
generation. In the following subsection, frontogenesis
ilar to that used in NCEP’s Eta Model. Forty vertical and its relation to the forcing of vertical circulationsis
levels were used with the vertical grid spacing of 200 reviewed.
minthelowestfivegridlevelswithagradualgeometric
stretching(byafactorof1.07)abovesuchthatthenext
a. Frontogenesis and vertical circulations
18 levels had an average spacing of 404 m and the top
17 levels had a spacing of 700 m. The model top was Frontogenesis is the time rate of change of the mag-
located at 19.2 km. nitude of the potential temperature gradient(Petterssen
The model was initialized by interpolating directly 1936).Byvirtueofthewavelikestructureofmidlatitude
from the 90.5-km NCEP Eta initialization, which has cyclones, the horizontal flow associated with them al-
50-hPa vertical resolution. Horizontal wind compo- most always contains significant deformation. Elegant
nents,geopotentialheight,temperature,andrelativehu- theoreticalworkbySawyer(1956)andEliassen(1962)
midity were interpolated horizontally along constant has shown that differential thermal advection in a hor-
pressure surfaces to the locations of the model grid izontal deformation field can lead to a thermally direct
points. Data were then vertically interpolated to the vertical circulation by increasing the magnitude of the
model grid levels. The lateral boundarieswereupdated horizontal temperature gradient and inducing accelera-
every 6 h from the Eta gridded forecasts using a Ray- tions in the direction of the thermal wind vector. Such
leigh-type absorbing layer. The simulation was initial- differential thermal advection can increase |(cid:61)(cid:117)| (i.e., it
ized at 0000 UTC 19 January 1995 and was run for48 canbefrontogenetic).Therefore,whereverthehorizon-
h. Only the first 36 h of this two-day run are used in talflowactstoconcentratethehorizontalpotentialtem-
this study. Air parcel trajectories were calculatedusing peraturegradient, a thermallydirectverticalcirculation
u, (cid:121), and w from the model output, using a forward will likely develop. A 2D version of the Miller (1948)
differencing scheme with a time step of 24.5 min. frontogenesis function
Asystematicverificationofthemodelsimulationwill
d
not be presented here. The reader is referred to Martin F (cid:53) |(cid:61)(cid:117)|
2D dt
(1998) for comparisons between thesimulationandthe
actualobservations.Asfurtherevidenceoftheaccuracy 1 [ (cid:93)(cid:117)(cid:49)(cid:93)u(cid:93)(cid:117) (cid:93)(cid:121)(cid:93)(cid:117)(cid:50) (cid:93)(cid:117)(cid:49)(cid:93)u(cid:93)(cid:117) (cid:93)(cid:121)(cid:93)(cid:117)(cid:50)]
ofthissimulation,wepresentinFig.4thesnowfalltotal (cid:53) (cid:50) (cid:49) (cid:50) (cid:49)
map from the UW-NMS simulation. A gratifying sim- |(cid:61)(cid:117)| (cid:93)x (cid:93)x (cid:93)x (cid:93)x (cid:93)y (cid:93)y (cid:93)y (cid:93)x (cid:93)y (cid:93)y
ilarity exists between Figs. 1 and 4 testifying to the
(1)
fidelity of the simulation. Although exact agreement
does not exist between the observations and the simu- represents the contribution to frontogenesis from the
lation(mostnotably,themodelsnowfalltotals,although horizontal deformation. Here, the total derivative is
considerable, arenot asextremeastheobservedtotals) d (cid:93) (cid:93) (cid:93)
the horizontal dimensions of the band were quite ac- (cid:53) (cid:49) u (cid:49) (cid:121) . (2)
dt (cid:93)t (cid:93)x (cid:93)y
curatelysimulated.Suchagreementontheprecipitation
distributionsuggeststhatthemesoscaledynamicalforc- Calculations of frontogenesis in this paper are made
ing of the snowband was also accurately replicated in using (1) employing the total horizontal winds and po-
FEBRUARY1998 MARTIN 333
FIG.5.Two-dimensionalfrontogenesisfunctionat1.96kmat(a)0600UTC19January,(b)1200UTC19January,and(c)1800UTC19
January1995.Frontogenesis[K(100km)(cid:50)1day(cid:50)1]islabeledandcontouredevery6K(100km)(cid:50)1day(cid:50)1beginningat6K(100km)(cid:50)1day(cid:50)1.
CrosssectionsalongAA(cid:57)in(a)areshowninFig.8;alongBB(cid:57)in(b)areshowninFig.11;andalongCC(cid:57)in(c)areshowninFig.12.
tential temperatures from the 40-km simulation of this with the snowfall. Doppler radarreflectivityfromKan-
case. Although (1) ignores the modifying effect ofver- sas City, Missouri (MCI), at 0647 UTC 19 January is
tical velocity on gradient strength, it more clearly re- shown in Fig. 6. There were two rather distinct bands
veals the region and circumstances for the forcing of of precipitation in central Missouri at this time. Band
vertical circulations (especially in the middle tropo- 2 was the most intense band with reflectivities in the
sphere) than does the traditional Miller (1948) fronto- 35–40-dBZrange,uncommonforsnowfall.Band1was
genesis function. It should be noted that when the total further from the radar and beam attenuation may have
winds are replaced by the geostrophic winds in (1)and disguiseditsvigor.Itwascertainlyconvectiveinnature
(2), the quasigeostrophicfrontogenesisfunction[which
as it was associated with cloud-to-ground lightning at
involvestheQvector(Hoskinsetal.1978)]isreturned.
0647UTC.Bothbandswereapproximately35kmwide
Shown in Fig. 5 is a series of 6-h plots of fronto-
andtherewasaspacingofroughly35kmbetweenthem.
genesis at 1.96 km from the UW-NMS model simula-
Also portrayed in Fig. 6 are the mean isentropes in
tion. Although representing only one horizontal level,
the2–5-kmlayerasforecastedbytheUW-NMSmodel.
Fig. 5 demonstrates that the axis of maximum lower-
Thus,thesnowbandsdepictedbytheradarwereoriented
tropospheric frontogenesis was consistently coincident
nearly parallel to the 2–5-km thermal wind vector at
with the position of the snowband. In subsequent sec-
thistime.Suchacircumstancesuggeststhatamesoscale
tions of the paper, more detailed analysis ateachofthe
frontal instability, such as conditional symmetricinsta-
indicated times is presented in order to more fully ex-
bility (CSI), may have been acting to organize thepre-
plain the production of this snowband. This analysis
cipitation bands.
begins at 0600 UTC 19 January 1995.
CSIisa2D,semigeostrophicmesoscaleinstabilityin
which both gravitational and inertial buoyancy deter-
b. 0600 UTC 19 January 1995
mine the displacement of an air parcel (Bennetts and
Moderate to heavy snow was falling over much of Hoskins 1979; Emanuel 1983). Assuming no variation
centralandsouthwestMissouriat0600UTC19January. inthealong-shear(front)direction,theverticalandhor-
Between 0600 and 0700 UTC, a total of 10 cloud-to- izontalaccelerationsofasaturatedparcelofairaregiv-
ground lightning strikes were recorded in association en, respectively, by
334 MONTHLY WEATHER REVIEW VOLUME126
FIG. 6. WSR-88D radar reflectivity at 0647 UTC 19 January 1995 from Kansas City, Missouri (MCI). Reflectivity
(dBZ)islabeledandcontouredevery5dBZbeginningat15dBZ.Thindashedlinesare2–5-kmmeanlayerisentropes
at0600UTCcalculatedfromthe UW-NMSsimulation.Heavydashedlinesindicatedistinctbandsdesignatedband1
andband2.
dw
(cid:125) ((cid:117) (cid:50) (cid:117) )
dt eP eE
du
(cid:125) (M (cid:50) M ). (3)
dt P E
Here, M is the geostrophic pseudoabsolute momentum
defined as
M (cid:53) V (cid:49) fx, (4)
g
where f is the Coriolis parameter and x is positive in
the across-shear direction toward warm air. The sub-
scriptsPandErefertoparcelandenvironmentalvalues,
respectively.
Figure 7 shows a schematic cross sectionthroughan
environmentsusceptibletoCSI.Insuchanenvironment,
a saturated parcel displaced along the slanted path SS(cid:57)
wouldcontinuetoaccelerateinthatdirection.Asshown
in Fig. 7, the necessary condition for CSI is that the
isoplethsof(cid:117) aremoresteeplyslopedthantheisopleths
e
of M. Bennetts and Hoskins (1979), Emanuel (1983),
FIG.7.Schematicverticalcrosssectiondepictinganenvironment Shieldsetal.(1991),andMartinetal.(1992)notedthat
susceptibletoreleaseofCSI.Solidlinesareisoplethsof(cid:117).Dashed
e for 2D flow this condition is identicalto theequivalent
linesareisoplethsof M. Asaturatedparceldisplacedalongslanted
path SS(cid:57) will experience the indicated component and resultant ac- potential vorticity (PVe) being negative. Moore and
celerationsasdescribedinthetext. Lambert (1993) and McCann (1995) extendedthisidea
FEBRUARY1998 MARTIN 335
by performing analyses of PV and noting its utility in to considerable lifting, the relative humidities were, at
e
delineatingregionssusceptibletoCSI.Finally,itshould best,about80%(Fig.8d).SincereleaseofCSIrequires
be noted that an environment that is stable to purely the air to be saturated, this elevated region of CSI was
vertical and purely horizontal displacement (as in the likely not released at this time. Importantly, however,
schematic in Fig. 7) may still be unstable to slantwise the PV along the same cross section (Fig. 8b) dem-
e
displacement. onstratesthat,inthenearly2Denvironmentofthewarm
Thorpe and Emanuel (1985) performed a modeling front, negative values of PV occur in a similar region
e
study that physically explained why an environment as the shaded areas of instability portrayed in Fig. 8a.
characterizedbyCSImaybeimportantinthegeneration This circumstance will now be related to the observed
ofprecipitationbands.Theyshowedthatthecirculation narrowness of the band. The cross section of (cid:117) in Fig.
e
produced by a given frontogenetic forcing is sensitive 8b demonstrates the presence of a deep moist neutral/
to across-front differences in the slantwise stability—a potentiallyunstablelayeronthewarmsideofthefront.
fact implicit in the determinant oftheSawyer–Eliassen The 2D frontogenesisalong AA(cid:57)isshowninFig.8c.A
circulation equation. When the slantwise stability is deep layer of substantial frontogenesis was associated
loweronthewarm,saturatedsideofthefront(i.e.,when with the warm front at this time. An interestingfeature
the PV is lower on the warm side), an intense, hori- inFig.8cisthetworegionsofnegativePV thatstraddle
e e
zontally restricted updraft occurs in that airmasswhile thefrontogenesisaxis.ForPV tobeareliablemeasure
e
a gentle, widespread downdraft occurs in the colder, oftheeffectivestaticstabilityofairparcels,theairmust
drierairbehindthefront.ThisisbecausethePV ,being be saturated. This condition was met only on thewarm
e
ameasureoftheeffectivestaticstability,playsananal- side of the front. As a consequence, there was consid-
ogous role in the Sawyer–Eliassen equation to that erably less resistance to vertical displacement on that
playedbystaticstabilityinthequasigeostrophicomega side of the front. Thus, theory predicts the presence of
equation; where the PV is small, the response to fron- an intense, more horizontally restricted updraft on the
e
togenetic forcing is large, and vice versa. In order for warmsideandaconsiderablylessnoticeabledowndraft
massbalancetobemaintainedacrossthefront,thevig- on the cold side. The vertical motion from the model
orousupdraftmustbehorizontallyrestricted,leadingto provides exactly that distribution (Fig. 8d) although it
itsmanifestationasathin,bandedprecipitationfeature. is not detailed enough to resolve the multiple banded
Many studies have suggested that the production of structure observed in Fig. 6. Thus, it is likely that the
bandedprecipitationincyclonicstormsresultsfromthis narrowacross-frontdimensionofthesnowbandresulted
combination of frontogenetic forcing and an environ- from across-front differences in the effectivestaticsta-
mentsusceptibletoreleaseofCSIasdeterminedbythe bility (as measured in terms of PV ) in the presenceof
e
slope of M and (cid:117) lines (see, e.g., Sanders and Bosart a thermally direct frontogenetic circulation. Although
e
1985; Sanders 1986; Moore and Blakely 1988; Martin the absence of saturated regions of CSI in the frontal
etal.1992).Itisimportanttonotethatonlytheacross- environment of the simulation seems to suggest that
front difference in the effective static stability is nec- release of CSI was not afactor inthe productionofthe
essarytoproducenarrowbandsofprecipitationinfron- banded structure shown in Fig. 6, the model resolution
tal regions. used in this study precludes making a definitive state-
In recent numerical simulations of unforced, hydro- ment regarding the role of release of CSI.
static,nonlinearCSI,PerssonandWarner(1993)found Thecircumstancesthatforcedthewidthofthesnow-
that explicit resolution of CSI circulationscouldnotbe band do notnecessarilyexplainitsconvectivenatureat
achievedathorizontalgridspacingscoarserthan30km. 0600 UTC. A combination of model trajectories and
The current study, therefore, cannot offer evidence for real sounding data are now used to explain the devel-
theoccurrenceornonoccurrenceofCSIcirculations,or opment of the convective snow.
of the structures produced by such circulations, in this
case.Instead,inthefollowinganalysesthemodeloutput
c. Production of convective snow at
will be used to determine whether or not the necessary
approximately 0600 UTC
condition for CSI, referred to earlier, was found in the
frontal environment of this cyclone. Figure 8 presents Figure9isaplanviewoftrajectories,locatedwithin
cross sections along line AA(cid:57) in Fig. 5. Shown in Fig. the maximum updraft region along line AA(cid:57) at 0600
8aisthemodel-derivedMand(cid:117) analysesat0600UTC UTC, traced backward to their origins at0000 UTC19
e
19 January. The shaded regions were subjectively de- January.Fortuitously,theseparcelsoriginatedwithin25
termined to be susceptible to CSI by the conventional km of the real sounding site at Little Rock, Arkansas
methodofevaluatingslopesofMand(cid:117) isopleths.Con- (LZK). In this section, these three parcels will be in-
e
siderable portions of the entire shaded area are simply vestigated in an attempt to diagnose the development
regions of potential instability (i.e., (cid:93)(cid:117)/(cid:93)z (cid:44) 0),aspe- of the convective snowfall in central Missouri at about
e
cial case of CSI. The rather limited region of pure CSI 0600 UTC. Parcel1,originallylocatedat869hPa,was
(the darker area in Fig. 8a) was located in the 5–8-km lifted to 621 hPa between 0000 and 0600 UTC. Parcel
layer.Althoughbetween5and6kmthisairwassubject 2 rose from 801 to 599 hPa and parcel 3 from 729 to
336 MONTHLY WEATHER REVIEW VOLUME126
iseFnItGro.p8e.s(oa)f(cid:117)Cro(Kss),sleacbtieolend,aalnodngcoAnAto(cid:57)uirnedFiegv.e5rya,3ofKM.Daansdhe(cid:117)delfirnoemsatrheeiUsoWpl-eNthMsSofsMimu(mlatsi(cid:50)o1n)alatb0e6l0ed0aUnTdCco1n9toJuarneudareyve1r9y9150.Smolsi(cid:50)d1.liSnhesadaerde
e
areawassubjectivelydeterminedtobesusceptibletoCSI.DarkestshadedarearepresentsareasusceptibletopureCSI,lightershadedarea
representsareaofpotentialinstability.(b)Crosssection,along AA(cid:57)inFig.5a,ofPV and(cid:117) at0600UTC19January1995.Isentropesof
(cid:117) labeled and contoured as in Fig. 8a. Shaded area has negative PV;light shading eshows vealuesfrom 0 to (cid:50)0.25 PVU(1 PVU (cid:53) 10(cid:50)6
me2s(cid:50)1Kkg(cid:50)1),darkershadingshowsvaluesof(cid:50)0.25to(cid:50)0.50PVUe.(c)Crosssection,alongAA(cid:57)inFig.5a,of2Dfrontogenesis[K(100
km)(cid:50)1 day(cid:50)1] at 0600 UTC 19 January 1995 labeled and contoured every 6 K (100 km)(cid:50)1 day(cid:50)1. Thicker solid lines outline regions of
negativePV depictedinFig.8b.(d)Crosssection,alongAA(cid:57)inFig.5a,ofverticalmotion(cms(cid:50)1)labeledandcontouredevery5cms(cid:50)1.
e
Solid lines are upward vertical motions and dashed lines are downward vertical motions. Shading shows model relative humidity (RH),
labeledinpercentandshadedevery10%fromgreaterthan60%togreaterthan90%withlegendintopleft.
632 hPa during that same time interval. These parcel ity as the model underestimatesthepotentialinstability
displacements are taken directly from the trajectory inthatlayer.Theimplicationsofthesesubtledifferences
analyses. betweenmodelandobservedsoundingswereexamined
The model-derived sounding at the origin of the tra- byapplyingthemodelnetverticaldisplacementstoboth
jectories is shown in Fig. 10a. The relative humidity the model and observed soundings for the chosen par-
gradually diminishes from near 100% at the surface to cels. For instance, upon lifting parcel 1 assuming the
considerably lower values in the middle troposphere. model sounding, a state of deep neutrality is achieved.
The actual LZK sounding at 0000 UTC 19 January is Lifting parcel 2 against the model sounding creates a
showninFig.10b.SinceLZKwaslocatedinthewarm meager amount of free convection in a shallow layer.
sector, where horizontal variations of temperature and Parcel3remainsnegativelybuoyantwhenliftedagainst
dewpoint are minimal, its stratification was represen- the model sounding.
tative of the stratification along the entire trajectory Whenliftingtheseparcelsagainsttheobservedstrat-
paths shown in Fig. 9. With the exception of a weaker ification, the seemingly slight difference between the
low-levelinversion,themodelinitializedsoundingfair- model and observed soundings at 800-hPa renderspar-
ly well represents the lower-tropospheric vertical tem- cel 2 freely convective upon being lifted to 730 hPa
perature profile present in the observations. However, (easilyachievedbythatparcel).Thus,whenairmotions
inthe900–800-hPalayer,slightlysmallervaluesofrel- described by the model are made to lift the observed
ative humidity in the model sounding have important environmental stratification, free convection is easily
consequences for the diagnosis oftheconvectiveactiv- achievedatapproximately0600UTCinthecorrectob-
FEBRUARY1998 MARTIN 337
served location. Discretized models cannot avoid ver-
tical smearing of information. The verticalsmearingof
thermodynamic variables, particularly moisture vari-
ables, whichcanexhibitenormousgradientsinthever-
tical,canbeofgreatconsequenceinanystudythataims
to diagnose moist convective processes. The inability
of the model sounding to produce the observed con-
vection, itself a result of the model’s inability to rep-
licate (or even initialize) existing vertical gradients in
moisture, may well have been a contributing factor in
its underestimate of snowfall totals, which werepartic-
ularly erroneous in central Missouri.
d. 1200 UTC 19 January 1995
By 1200 UTC moderate to heavy snow continued in
central Missouri while light to moderate snow had in-
vadedeasternIowaandnorthwesternIllinois.Themain
snowbandwasstillorientedparalleltothe2–5-kmmean
isentropes (not shown) and so a similar set of circum-
stances, conducive to the forcing of a narrow updraft,
existed at 1200 UTC 19 January. For reference,the2D
FIG.9.The6-habsolutetrajectoriesofparcelslocatedinthemax- frontogenesis function at 1.96 kmat that timeisshown
imum updraft region of Fig. 8d at 0600 UTC traced backward to in Fig. 5b. A noticeable decreasein theintensityofthe
0000 UTC 19 January 1995. Location of Little Rock, Arkansas
frontogenesishadoccurredbythistimeasseparatemax-
(LZK),isindicated.
imadeveloped.Aseriesofverticalcrosssectionsalong
line BB(cid:57) in Fig. 5b are shown in Fig. 11.
To determine whether or not the environment was
susceptible to release of CSI, a cross section ofmodel-
FIG.10.(a) PseudoadiabaticdiagramdepictingtheUW-NMSmodelinitializedsoundingat0000UTC19January1995attheoriginof
trajectoriesshowninFig.9.TemperatureanddewpointsoundingsareindicatedbyTandT ,respectively.(b)AsinFig.10aexceptobserved
d
LittleRock,Arkansas(LZK),soundingat0000UTC19January1995.
338 MONTHLY WEATHER REVIEW VOLUME126
uaryFig.
nn
Jasi
9A
1
Cd)
T(
U5.
9
09
01
2
1y
atuar
5bJan
Fig.19
C
inUT
(cid:57)BB00
ne12
liat
ong5b
alg.
ptFi
xcein
e(cid:57)B
bB
8
g.ne
Fili
g
nn
io
Asal
b)ept
(c
5.ex
1998c
yg.
arFi
anuin
Js
9A
C1(c)
UTU.
V
00P
127595.
at(cid:50)0.19
b y
inFig.5(cid:50)0.50to19Januar
(cid:57)BmC
eBfroUT
linues00
alongVvaleat12
ptP5b
excesentsFig.
8aprein
g.re(cid:57)B
FigB
Asinshadingline
11.(a)Darkesteptalon
F.IG95.exc
9d
18