JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 102, NO. D24, PAGES 29,155-29,166, DECEMBER
26, 1997
Lake-induced atmospheric circulations during BOREAS
Jielun Sun,• Donald H. Lenschow,2 L. Mahrt, 3 Tim L. Crawford, 4
Kenneth J. Davis,2 Steve P. Oncley,2 J. I. MacPherson,s
Qing Wang,6 Ron J. Dobosy,4 and R. L. Desjardins7
Abstract. Lake-inducedatmosphericcirculationsover three lakes rangingfrom 3 to 10
km width are analyzedusingdata from three aircraft during the 1994 Boreal EcosystemAtmosphereStudy(BOREAS). A well-defineddivergentlake breeze circulationis
observedover all three lakes during the day. Under light wind conditions,the lake breeze
is not very sensitiveto the water temperature, and the strengthof the divergenceover the
lake decreaseswith increasinglake size. The boundary-layerdevelopmentover the
surroundingland can be very important for generatinga horizontal pressuredifference
which drives the lake breeze.
Diurnal
and seasonal variations
of lake breezes
are
investigatedon the basisof repeated passesfrom the different aircraft at different
altitudesfrom late springto early fall of 1994. The lake breeze divergenceincreaseswith
time during the day and reachesa maximum around 1300 LST. The latent heat flux over
10-km-wideCandle Lake increasessteadilyfrom springto fall as the lake temperature
increases.The latent heat flux over the land reachesa maximum during the summer due
to evapotranspiration.
The lake effect on area-averagedfluxessometimesleadsto a
negative heat transfer coefficientfor an averagingscaleof severaltimes the lake width.
Pielke,1985;Arritt, 1987].Arritt [1987]concludedthat the exact
lake surfacetemperature doesnot criticallyinfluencethe charAtmosphericboundarylayer flow can be stronglyinfluenced acteristicsof the lake breeze as long as the lake air temperaby the spatially varying energy exchangebetween heteroge- ture is coolerthan the inland air temperatureand that the lake
neous ground surfacesand overlying air. Lakes can lead to breeze is most sensitiveto depth of the internal boundarylayer
particularlystrongvariationsin surfaceheat flux.
over the lake.
The summer daytime air over a lake typically is cooler and
Rabin et al. [1990] and citationstherein noted that lakes on
denser than the air overlying relatively warm adjacent land
scalesof 1-10 km could induce cloud-free areas in a region of
surfaces.As a result, a horizontal pressuregradient is gener- otherwise convective cloudiness associated with surface heatated by the air densitydifference which forces onshoreflow,
ing. They also noted that the minimum lake size required to
commonly called the lake breeze. Lake breeze studies have
affect the cloud pattern is proportional to the wind speed;on
been reported for Lake Erie [Biggsand Graves, 1962], Lake
windier days,only the large lakes affect the cloud pattern. For
Michigan [Lyons,1972;Keen and Lyons, 1978], and Lake Onsmall lakes the large-scaleflow may essentiallyeliminate the
tario [Estoqueet al., 1976]. A commonlyapplied lake breeze
horizontal gradient of air temperature due to rapid removal of
index,cr-= [/2/(c/,AT),wasintroduced
byBiggs
andGraves air immediately above the lake. Furthermore, lake breezes
[1962]. Here I/is a characteristicwind speedsuchasthe hourly-averagedspeedbetween 1000 and 1600 LT, or the geostro- associatedwith small lakes are more likely affected by backphicwindfrom1200UT surface
mapsat aninlandlocation,ce groundtransientmesoscalecirculations.Arritt et al. [1996]have
is the specificheat of dry air, and A T is the differencebetween also found cloud suppressionassociatedwith the lake breeze.
As part of the 1994 Boreal Ecosystem-AtmosphereStudy
the maximum surface air temperature at the inland location
(BOREAS),
four aircraft flew over a region in Saskatchewan
and the mean water temperature.
The lake breeze and its related thermal internal boundary and Manitoba, Canada, containingnumerouslakes [Sellerset
layer have alsobeen studiedwith numericalmodels [Segaland al., 1995]. The extensivedata collected by the aircraft offer
more detailed temporal and spatial observationsof the flow
abovelakes than previouslyavailable.We examine the generIprogramin Atmospheric
andOceanicSciences,
Universityof Colation of the lake breeze, the strengthof suchcirculationsas a
orado, Boulder.
2NationalCenterfor Atmospheric
Research,
Boulder,Colorado. function of the large-scaleflow and lake-land temperature
3Collegeof OceanicandAtmospheric
Sciences,
OregonStateUni- contrast, and contrasts in vertical fluxes over the lake and the
versity, Corvallis.
adjacent area.
4Atmospheric
Turbulenceand DiffusionDivision,NOAA ARL,
The atmosphericdata analyzedin this paper are from the
Oak Ridge, Tennessee.
aircraft flying repeatedlyalong the sameflight track (section
5NationalResearchCouncil,Ottawa.
r'Meteorology
Department,NavalPostgraduate
School,Monterey, 2). The generalcharacteristics
of meteorologicalvariablesover
California.
the lake and the surroundingland are describedin section3. A
7LandResources
ResearchCentre,AgricultureCanada,Ottawa.
new scalingargumentfor predictingthe strengthof the lakeCopyright 1997 by the American GeophysicalUnion.
induced atmosphericcirculation is developed,and important
factors on the generation of the lake breeze are discussedin
Paper number 97JD01114.
0148-0227/97/97JD-01114509.00
section 4. Diurnal variation of the lake-induced atmospheric
1.
Introduction
29,155
29,156
SUN ET AL.: LAKE-INDUCED
ATMOSPHERIC
106000
'
.,•?
::
?::
•,•
•.:•:;/•i?y
105030
'
CIRCULATIONS
105000
'
,•'• •Lake
.,:::/•
Bi;ern
• •• 'c•
i•White
.Ai
Gull
•ake.
-,v
54ø00
'-
•:::•.:.:
/
Lake
" :• •
•••
• ' •1•15
2•
Kilomete•s•
Figure 1. Schematicof CandleLake flight track.
circulationis investigatedin section5. Seasonalvariationsof
the lake breeze are examined in section 6. The influence of the
eddycorrelationmethodswherethe unweightedmeanis computed over a windowsizeequal to the lake width.
lake on area-averagedfluxesis documentedin section7, and
results are summarized
in section 8.
2.2.
Data Comparisons Between Aircraft
The Electra flew a BarnesEngineeringModel PRT5 radi-
2.
Data Description
2.1.
General
Information
ometerduringthe first intensivefield campaign(IFC-1) and
both the PRT5 and a HelmannradiometerduringIFC-2 and
IFC-3. Surfaceradiationtemperaturesfrom the two radiometersshowverygoodagreement.DuringIFC-1 the PRT5 sometimesfailed due to instrumentoverheating.
bert, Saskatchewan,
Canada.Candle Lake, surroundedby
Duringthe threefield campaigns
in 1994the Twin Otter and
mixed canopies, is about 10 km across in the northeast- the Electraflew over CandleLake on the sameday on four
southwest direction and about 20 km in the northwestoccasionsand the LongEZ and the Electra on nine occasions.
southeastdirectionand is the largestlake in the SSA. In order In general,the Twin Otter andLongEZ CandleLake radiation
This studyfocuseson Candle Lake in the BOREAS southern studyarea (SSA), located 70 km northeastof PrinceAl-
to study the lake breeze for different sizesof lakes the data
temperatures are about IøC lower than the Electra. On the
overWhiteGull Lake(5 km in diameter)andHalkerrLake(3
km in diameter) will be comparedwith those over Candle
Lake. All three lakesare on the sameflighttrackusedby all
aircraft(Figure 1). White Gull Lake is in a regionof black
basisof the four intercomparison
flightsbetweenthe LongEZ
andthe Twin Otter the air temperature
andthe watervapor
mixingratioare,on the average,about1.75øCand1.2g kg-1
higheron the Twin Otter. The latentheatflux compares
well
betweenthe two,whilethe sensible
heatfluxappearsto be 38
spruceand mixedvegetationabout 12 km NE of CandleLake,
andHalkett Lake is surrounded
by aspenabout50 km SW of
W m-2 higherfromtheLongEZthantheTwinOtter[Kelly
et
Candle
al., 1996;Dobosy
etal., thisissue].No corrections
to anyof the
Lake.
Measurementsfrom the NOAA LongEZ, CanadianNRC aircraftdata are made in this study.
Twin Otter [MacPherson,1996], and NCAR Electra aircraft
wereused.Both the Twin Otter andthe LongEZflewat about
30 m abovethe surface,andthe Electraflewat altitudesrang- 3. General Characteristicsof Meteorological
ing from 100 m to 3 km abovethe surface.Duringthe three Variables Around Candle Lake
intensivefield campaigns,
therewere 13, 120,and 66 low-level
Air temperature
averaged
overall the CandleLakepasses
is
legsoverCandleLakefor the Twin Otter, LongEZ,andElec- about 0.7øC warmer over the land than over the lake at the
tra, respectively,
and 37 Electraflight legshigherthan 100 m 30 m levelandabout0.3øCwarmerat the 100m level(Figure
and within the boundarylayer. The data coverbetween0900 2 andTable1). Thevariationof the air temperature
overboth
LST and 1700 LST from late May to mid-September.
Wind the land and the lake followsthe variationof the ground
components,air temperature,humidity,and CO2 concentra- temperature(Figure 2). Becauseof the stablystratifiedflow
tion are availableat 25 s- • on the Electra,16 s-• on the Twin overthe cool,smoothlake andthe unstablystratifiedflowover
Otter,and40s- • ontheLongEZ.In addition,
theO3concen- thewarm,roughlandsurface,thefluctuations
of windcompotration was recorded on the Electra and the Twin Otter.
nents,air temperature,humidity,and CO2 and 03 concentraIn order to documentthe land-lakesurfacetemperature tion are smalleroverthe lake thanoverthe land (Table3).
contrasts,track segmentsover land aboutthe width of the lake
on both sides of the lake were selected. Fluxes over the lake
Sensibleheat, latent heat, momentum,carbon dioxide, and
ozonefluxesaveragedover all the CandleLake passesare
and over the adjacentland segments
were calculatedusing muchsmalleroverCandleLakethanovertheland(Table2).
SUN ET AL.: LAKE-INDUCED
(a>
Twin
ATMOSPHERIC
5
5
02468
l0
12
14
25
25
150
2
4
6
10
8
10
12
415
.........
00
2
6
10
14
560
540
2
4
6
4ot-/ \
30!
20(_ .'
0
2
I•
4
6
IFCI
(b)
35
8
l0
8
•
10
12
14
1FC2
--• IFC31
30 m over the land; the heat source is above 30 m over the lake
and below 30 m over the land; and the moisture source is from
LongEZ
•
,
.
--•
ß
•
35
25
25
15
15
21}
5
20
0
21}
40
25
-_
:•.•
10
60
80
100
the surface(Figure 3). The sinksand sourcesare the advection
of warm, dry, and low CO 2 and O3 air from land over the lake
2O
15
10
between
5
0
-5
25
20
15
15
0
2t}
40
60
80
'
100
0
•{}
h{}
60
80
5{}{}
580
........................
'•.................
•[7"'
'•[."•"
120
12{}
0.5
0
-t}.5
over the lake.
q580'
• -lg
4.
52{}1•
....20-'40 60 80• i00• h2•
21}
•= -40•=
• IFCI4<
,FC2
-I-,FC3q
Lake-Induced Atmospheric Circulation
4.1.
Scale Analysis
The horizontal pressuredifference (AP) generatedby the
thermal
(c)
-
0
Electra
10
20
30
30 m and 100 m.
The specifichumidity is about the sameover the land as over
the lake exceptsometimesin the morning(section5) when it
is higher over the lake. Carbon dioxide and ozone concentrations are, respectively,about 2 ppmv and 0.3 ppbv higher over
the lake than over the land at the 30 m level due to the uptake
of the carbon dioxide by the vegetationand ozone deposition
to leaf and groundsurfaces.The strongdownwardentrainment
of ozone-rich air from the overlying free atmosphere makes
the ozoneconcentrationat the 100 m level higher over the land
than over the lake even though downward vertical motion
associatedwith the lake breeze may bring down ozone-richair
12t}
20
contrast
4t}
5t}
6t}
7t}
-I0
-10
2t}
-3t}
v-:•--••'-'• •- • ' ] ' 5{},•c: 4
' Vt-2
10[•C
• ....
L""•:•_•_]"2
VI•_,2•
_•'-"•Y}10
..............
10
3o •
20
• ,FCI •<
30
the lake and the land can be esti-
-
40
,FC2 >I•
50
60
WC3
Candle Lake Pass Number During BOREAS
71
Z2
pg dz-
ApgH -'-' -pogHAT/To
(1)
21
: 5 9
0
between
matedby combiningthe hydrostaticequationand the horizontal equationof motion(neglectingthe Coriolisterm) alongthe
direction of flight at aircraft flight level z•:
AP = A
,[--z•
29,157
The standarddeviationsof the pass-averagedvariablesin Table 1 and fluxesbetween the passesin Table 2 include diurnal
and day-to-dayvariations and flux random sampling errors
from eachpass(sections5 and 6). The influenceof cold front
passages
on surfaceradiation temperatureand air temperature
is evident in Figure 2b.
The sensibleheat flux averaged over all the Candle Lake
passesis weak downwardover the lake and strongupward over
the land, which is also found by McDermottand Kelly [1995].
The moisture flux is upward over both the lake and the land.
The CO2 and 03 fluxesare downwardover both the land and
the lake at the 100 m level but upward over the lake at the 30 m
level (Figure 3). The negativeskewnessof the vertical velocity
at 30 m over the lake (Table 3) implies that the strongest
vertical motions are downdrafts.This suggeststhat the main
turbulencesourceover the lake is above 30 m. Combining the
skewnessof the vertical wind componentin Table 3 and the
vertical flux convergencein Table 2, we infer that the O3 and
CO2 sinksare between30 m and 100 m over the lake and below
Otter
15
CIRCULATIONS
Figure 2. (opposite) Mean surface radiation temperature
(T,), air temperature(T•), specifichumidity(q•), and COg
and O3 concentrationsoverboth the land (solidlines,subscript
g) andCandleLake (dashedlines,subscript
l) for (a) the Twin
Otter, (b) the LongEZ exceptfor O3, and (c) the Electra
exceptfor CO2 for all the Candle Lake passes.The difference
betweenthe variablesover the land and over the lake (dotted
lines,subscriptd) is alsoplotted. The abscissais the Candle
Lake pass number for all passesduring the 1994 BOREAS
campaigns.The surface radiation temperatures from pass
numbers 4, 5, and 10 from the Electra are low due to instru-
ment overheating.The thin horizontalline is the zero difference line.
29,158
SUN ET AL.: LAKE-INDUCED
Table
1.
Mean
Variables
and Standard
ATMOSPHERIC
Deviations
CIRCULATIONS
of Mean
Variable
Between
Passes
(in Parentheses)for All CandleLake Passes
Twin Otter (at 30 m)
Electra (at 100 m)
LongEZ (at 30 m)
Variables
Land
Lake
Land
Lake
Land
Lake
Ta
20.92
20.18
19.54
18.87
19.00
18.73
(C)
(3.67)
(3.61)
(3.50)
(3.49)
(4.12)
(4.06)
qa
6.50
6.49
6.79
6.82
7.09
7.11
(g kg-•)
(1.66)
(1.71)
(2.13)
(2.11)
(2.14)
(2.12)
COg
(ppmv)
O3
(ppbv)
365.9
(10.24)
33.96
(7.44)
367.7
(11.28)
34.21
(7.96)
357.9
(8.68)
34.17
(7.43)
358.8
(2.5)
33.87
(7.69)
359.3
(5.14)
UAU/L = -(1/po)AP/L.
(2)
4.2.
362.0
(7.70)
Correlations Among Lake Breeze, Air Temperature
and Ambient
Here A representsthe horizontalvariation acrossa lake with
width L along the flight track, z2 is the top of the lakeinfluencedinternal boundarylayer at which level the air temperaturesover the lake and surroundingland are equal,H =
z2 - z• is the depth of the lake-inducedinternal boundary
layer above the aircraft flight level, and A9 and AT are the
verticallyintegrated horizontal densityand temperaturevariations, respectively.The subscript0 representsa reference
state.The variablesP, T, and U are pressure,air temperature,
and wind speedalong the directionof flight, respectively,and
# is gravity. The state equation
Wind
The divergenceover Candle Lake is estimatedfrom the least
squareslinear slopeof the along-flight-trackwind component
as a function
of distance
across the lake. The
ambient
wind
speedU alongthe flight track is averagedover the land on both
sidesof the lake. The scatterin Figure 4 probablyreflectsthe
importance of the missingvariable H, the internal boundary
layer depth [see also Arritt, 1987]. Arritt et al. [1996] have
observedlake breeze circulations250-300 m deep around
1445EDT withwindspeeds
of 5 m s-1 overLakeOkeechobee,
about 40 km in diameter.
The cool air over Candle
Lake can
sometimesbe observedeven at the 700 m Electra flight level.
The variation in H can be causedby variationsin large-scale
flow,lake-landsurfacetemperaturedifference,and the diurnal
is usedin the derivationof (1) togetherwith the assumption and day-to-dayvariation of the stratificationof the internal
that the densityvariation is dominated by the temperature
boundarylayer. The scatterin Figure 4 may alsobe due to the
variation
fact that the divergencealongthe flight track is estimatedwith
A9/9o = -AT/To.
(4) the wind componentdirectlyoverthe lake althoughthe aircraft
sometimesflew abovethe lake-modifiedair for part of the lake
Combining(1) and (2),
transverse,particularlyunder strongwind conditions.Under
this situationthe divergenceis displacedand occursonly over
gHAT
that part of the lake where the lake air is not completely
AU/L- LToU'
(5) replaced
by the advectedland air. Becauseof lack of temperHere A U/L is the contribution to the divergenceof the flow ature profilesover the lake the replacementof the vertically
over the lake alongthe flight track.Equation(5) indicatesthat integrated temperature difference AT with the temperature
for a fixed lake size, the divergent flow increaseswith the difference at the aircraft level is another source of the scatter
verticallyintegratedhorizontalair temperaturedifferenceand in Figure 4.
The traditional lake breeze index, tr in Table 4, is estimated
the depth of the air modified by the cool lake and decreases
for each Candle Lake passusingthe total ambientwind velocwith ambientwind speed.
Table 2. AverageFluxesand StandardDeviationsof Fluxes(in Parentheses)Over Land
and Over
Candle
Fluxes
Lake
for all Candle
LongEZ (at 30 m)
Electra (at 100 m)
Land
Land
Land
174.3
(W m-2)
(41.4)
Latent
202.4
heat flux
(55.4)
Friction velocityu.
(m s-•)
Carbon
dioxide
(m s-• ppmv)
Ozone
flux
(ppbvrns-j)
flux
Passes
Twin Otter (at 30 m)
Sensible heat flux
(W m-2)
Lake
Lake
- 15.8
(24.2)
26.3
(43.5)
Lake
Lake
110.72
-6.2
105.3
- 1.83
(43.3)
(28.5)
(59.4)
(33.6)
140.8
(65.2)
42.3
(62.1)
159.2
(76.1)
49.86
(58.92)
0.39
0.08
0.38
0.15
0.28
0.12
(0.26)
(0.1)
(0.22)
(0.15)
(0.16)
(0.11)
-0.11
(0.04)
-0.17
(0.05)
0.01
(0.02)
0.003
(0.023)
-0.04
(0.06)
0.003
(0.1)
-0.03
(0.06)
-0.18
(0.09)
-0.01
(0.09)
-0.007
(0.11)
SUN ET AL.: LAKE-INDUCED
ATMOSPHERIC
CIRCULATIONS
29,159
Table 3. StandardDeviation o-and SkewnessS of VariablesFrom Each PassAveraged
Over All Flight Legs Over Land and Over Candle Lake
Twin Otter
Statistics
Land
Lake
o-(w) (m s-l)
S(w)(m3 s-3)
o-(u)(m s-•)
o-(v)(m s-•)
0.82
0.3
1.76
1.6
o'(Ta) (C)
0.39
S(Ta)(C3)
o'(qa)(g Kg-1)
S(qa)(g3Kg-3)
0.27
0.25
0.20
o-(CO2)(ppmv)
S(CO2)(ppmv
3)
cr(O3)(ppbv)
Land
0.29
-0.31
1.31
1.0
Electra
Lake
Land
0.95
0.16
1.73
2.07
0.49
-0.06
1.37
1.52
0.89
0.41
1.39
1.29
0.51
0.16
1.14
1.02
0.26
0.39
0.33
0.28
0.25
0.075
0.23
0.45
0.41
0.27
0.13
0.31
0.27
0.45
0.49
0.26
0.23
0.4
0.24
0.34
0.80
0.62
0.97
1.38
0.44
0.58
-0.01
-0.01
0.01
-0.02
-0.01
-0.2
0.72
1.1
0.99
0.09
0.1
0.15
ity over the land and the differencebetween the air temperature over the land and the lake surfaceradiation temperature.
The linear correlationcoefficientsbetweenthe divergenceover
0.0005 -
0.00025
-
o
the lake and the lake breeze index, A T, U, and A T/U in Table
4, indicatethat the divergenceis best correlatedwith the ambient wind at 30 m altitude and with the air temperature
differenceat 100 m altitude. The strengthof the advectioncan
modify the air temperaturedifferenceand vertical mixingover
the lake very effectivelyat the lower level, while the air temperature difference at 100 m is more correlatedwith the vertical developmentof the thermal influenceof the lake. The net
observed effect is that the column
Lake
-0.08
0.97
S(O3)(ppbv
3)
LongEZ
[]
o
o
0-
-0.00025
-
-0.0005
-1
of the cool air over the lake
growsduringthe increaseof the vertical mixingin the morning.
The deepeningof the internal boundary layer over the lake
strengthensthe horizontal pressuregradient which leads to
strongdivergentflow over the lake even if the horizontal temperature difference at lower levelsdoes not increase.
To interpret the relationshipbetweenthe lake breeze circulation and the strengthof the large-scaleflow, one must recognizeseveralcompetinginfluences.The significantwind can
tilt the internal boundarylayer over the lake at the upper level
and reduce the air temperature difference at the lower level.
These influencesact to reduce the generation of lake breeze
flow and lead to a general negative correlation between the
lake breeze divergenceand the speedof the large-scaleflow.
However, the depth of the stable layer over the lake, H, may
o
o
o
Ta (Land) - Ta (Lake) (C)
0.0006 0.0004-
0.0002-
o
o
o
0-
o
o
o
o
-0.0002-
-0.0004
0
WindSpeed
( rns'1 )
0.0005
0.0004,
0.0003-
Q,H
•
Q,H
100m
• CO2,
4•03
• CO2,
303q/x•
• •
Q H
4x•
•
CO2,
03
0.0002-
o
[] I• o o
o
0.0001o
warm
cool
0
warm
-1
Q,H
•
Pland•
CO2,
03
4!,
Plake • Pland
LakeBreeze
<
q,co•,o,
H
LakeBreeze
•
Q,H
CO•,03
4[•
[Ta(Land)-Ta(Lake)]/[Wind
Speed]
(C sm'l)
II
II
Figure 3. Schematic diagram of the fluxes and airflows
around Candle Lake. The solid arrow and the open arrow
representthe airflow and the flux, respectively.
Figure 4. Along-flight-pathcomponentof the divergentflow
over Candle Lake as a function of the air temperature(Ta)
differencebetweenthe land and lake areas(top), wind speed
alongthe flight track (middle), and their ratio (bottom). Each
point representsbin averagesover an interval along x axis.
Squares,Twin Otter; diamonds,LongEZ; circles,Electra.
29,160
Table
SUN ET AL.: LAKE-INDUCED
4.
Linear
Correlation
Coefficients
ATMOSPHERIC
Between
temperaturestratificationand residualturbulenceat the upper
level over the lake.
Divergence and Other Parameters
Linear
Correlation
Coefficients
Parameters
Twin Otter
LongEZ
Electra
Lake breeze index (•r)
-0.39
-0.33
-0.09
0.18
0.25
AT
U
-0.67
A T/U
0.47
-0.36
0.29
-0.33
0.27
0.14
increasewith wind speed due to shear-inducedmixing. This
secondaryeffect of the strengthof the large-scaleflow acts to
increasethe strengthof the lake breeze circulationsince the
cool internal boundarylayer over the lake is deepened.
4.3.
CIRCULATIONS
Pressureadjustmentsresultingfrom subsidenceand consequent adiabaticwarming in the presenceof stablestratification
within or at the top of the internal boundarylayer can act as a
negative feedback to the lake-induced circulation. The adiabatic warmingdue to the subsidenceover the lake actsto limit
the generationof surfacehighpressureover the lake. Sincethis
effect preferentially inhibits the developmentof circulations
driven by small-scalesurfaceheterogeneity[Smithand Mahrt,
1981], it more effectivelyreducesthe influence of small lakes
and is probably important on the scale of Candle Lake. This
negativefeedback appearsto contribute to the preferred sea
breeze scalefound in the analysisof Rotunno [1983] and Dalu
et al. [1991].Therefore the pressureover the lake is the balance
between the lowering of pressure caused by the adiabatic
warming and the increasingpressurecausedby the cool lakemodified
Upper Level Circulation Induced by Lake Breeze
air associated with the weak downward
The downward air motion induced by the low-level divergenceis seen on the August 31, 1994, Electra flight where the
differencebetweenthe vertical wind componentover the land
and over the lake increaseswith height (Table 5). This downward motion is associatedwith the vertical change of sign of
the horizontal temperature difference AT. The air is cooler
the layer just above the lake.
lake.
Halkett
4.4.
heat flux into
Influence of Boundary Layer Development on Lake
Breeze
The developmentof the low-level divergent flow over the
lake dependson the depth of the layer in which the air temover the lake than over the land at lower levels and warmer at
perature over the lake is substantiallydifferent from the air
higher levels.These resultssuggestadiabaticwarming associ- temperatureover the land. This depth can be associatedwith
ated with downward motion over the lake.
the developmentof the internal boundarylayer over the lake
The strongverticalmixingand developmentof a well-mixed or the growth of the heated boundarylayer over the land. As
boundarylayer over land during the morning rapid transition an example,on May 25, 1994,when the ambientwind wasweak
period can also cool the air at higher levels[Lenschowet al., (about1 m s l), the divergence
is about2 and 2.4 times,
1979;Ehret et al., 1996].As a result,the air over the lake above respectively,strongerover White Gull Lake and Halkett Lake
the nocturnal inversionlayer can be warmer than the rapidly than over Candle Lake (Figure 5). The strengthof the diververticallymixed air over the land. However, all the higher-level gencecan be easilyseenfrom the stronggradient of the alongElectra flightsare closeto local noon; therefore the warmer air flight-track wind acrossthe lake shown in Figure 5. The air
over the lake at higher levels is likely to be due to adiabatic temperatureat 30 m over all three lakesis aboutthe sameeven
warming associatedwith the downward circulation over the thoughthe water temperatureof the shallowerWhite Gull and
The standard
deviation
of the vertical
motion
is smaller
Lakes
is about
5øC warmer
than
Candle
over
Lake.
The
similar air temperature between large and small lakes may be
the lake than over the land (Table 5), includingthe levels due to the weak vertical mixingin the stablelayer over the lake
where the air temperature is warmer over the lake than over and less influence from the land under the weak wind condithe land. The difference of the standard deviation of the air
tion. The aboveobservationsimply that the internal boundary
temperaturebetweenthe land and the lake, however,changes layer depthsover the three lakesare roughlythe samebecause
in the vertical.The standarddeviationof the air temperatureis of the cancelationof the lake size on both sidesof (5). The
smaller within the lake-modified
cool air at the lower level over
divergenceapparentlydecreaseswith increasinglake size unthe lake and becomes larger when the air temperature is der this weak wind condition. The results also indicate that the
warmer over the lake than over the land at the higher level. developmentof the boundarylayer over the land may play an
The higher standarddeviationof the air temperatureover the activerole in the buildup of the horizontal land-lake pressure
lake at 1600 m (Table 5) is apparentlyassociatedwith stable difference.The cold lake simplymaintainsthe overlyingcool
Table 5. StandardDeviation rr and SkewnessS of Air Temperature and Vertical Motion
Over Land and Over Lake and Air Temperature Difference Between Land and Lake
From Electra on August 31, 1994
T. (C)
Land
w (m s-')
Lake
Land
Lake
Level
(m)
100
530
1200
1600
AT,
•r(r,)
S(T,)
•r(T,)
S(T,)
Aw
•r(w)
S(w)
•r(w)
S(w)
0.46
0.17
-0.04
-0.35
0.27
0.14
0.18
0.13
0.49
0.39
1.84
2.11
0.14
0.12
0.24
0.35
0.4
0.29
1.35
1.31
0.12
0.18
0.13
0.29
1.06
1.48
0.94
0.98
0.43
0.42
1.01
0.45
0.43
0.55
0.43
0.84
0.032
-0.41
0.03
0.43
SUN ET AL.' LAKE-INDUCED
ATMOSPHERIC
air becauseof inefficient horizontal and vertical mixing with
weak large-scaleflow.
The
observed
lake
breezes
over White
Gull
297
296
for the lake breeze
but rather
29,161
(a)
295
and Halkett
294
293
Lakes also imply that the exactwater temperature of the lake
is not crucial
CIRCULATIONS
292
it is the contrast
291
ß
290
between the boundary layer developmentover the warm land
surroundingthe lake and the stablecool air over the lake. This
is consistentwith the results of Segal and Pielke [1985] and
Arritt [1987]. The size of the lake for which the divergentflow
can be observedappearsto be controlledby the ambientwind
and the stabilityof the air over the lake sinceboth factorscan
affect the warm-up of the lake-modifiedair.
16
Diurnal
Variation
of Lake-Induced
o
•l 0 l>
[]
[]
0
0 A V<I 0
[]
[]
[]
[]
A
16 ß
0.0006]
Atmospheric Circulation
•' 0.00041
(c)
The traditional lake breeze index (o-) is defined in terms of
the daily maximum air temperature and the hourly averaged
wind in the morning. Therefore it cannot be used to characterize the diurnal
variation
of the lake breeze.
BOREAS
1994
Electra
Candle
Lake
missions with
Electra (370m)
Electra (700m)
Electra(1000m)
Electra(1300m)
?
Electra(2000m)
• -0.0002
1 o
..
D ø []
• -0.0004
•
10
Ii
o
1•
contains 1 Twin Otter and 16 LongEZ at 30-m-level and 10
100-m-level
LongEZ (30m)
Electra (100m)
(b)
300]
295
1
290
-•
285
1
280 •
10
5.
[]
•
1•
1•
1•
Time (LST)
at least 3 re-
peated passes.On the basisof theserepeatedpassesthe diurnal variation of the lake-induced atmosphericcirculation can
be summarized.As a representativeexample,the diurnalvariation of the lake breeze on May 31, 1994, is shownin Figure 6.
Apparently, becauseof inertia the nocturnal land breeze
(d)
o
io
ß
LongEZ (30m)
l0
Electra (100m)
•
Twin Otter,
6
.
I
25 May 1994
'
I
0.6
I
0.4
•
48
(e)
•
46
•
44
o
•
40
A
Electra (370m)
Electra (700m)
38
Electra(1000m)
Electra(1300m)
Electra(2000m)
[]
-6
2
I
•
I
6
8
10
[]
[]
[]
Time (LST)
30
II
[]
12
'0.4
17.5
I
[]
25
I
•
16.5
20 •
16
15
15.5
Figure 6. Diurnal variationsof (a) air temperature,(b) surface radiation temperature, (c) the divergenceover Candle
Lake, (d) specifichumidityboth over the land (solid symbols)
and over the Candle Lake (open symbols),(e) 03 concentrations,and (f) the differenceof the CO2 concentrationoverthe
land and over the lake on May 31, 1994.
10
14.5
4
2
Halkett
Lake
6
8
Candle
Lake
5
10
12
White Gull
Lake
Distance (Km)
continuesin the morning even after the air over the land
becomes warmer
from
Figure 5. The Twin Otter pass along the flight track over
Halkett Lake, Candle Lake, and White Gull Lake on May 25,
1994.As in equation(2), here U is the wind componentalong
the flight track, positive toward northeast.NDVI is the normalized differenceof vegetationindex. The variablesover the
land, 5 km upstreamand 5 km downstreamfrom eachlake, are
also plotted for comparison.
than the air over the lake. The land breeze is
seen as the convergentflow at 30 m and the divergentflow at
1000 m over Candle Lake in Figure 6. As the air is heated by
the warm land surface,the flow over the lake graduallychanges
the land breeze
to the lake breeze.
At
1230 LST
the
divergentand convergentflow over the lake is seenat 30 m and
at 100 m, respectively.The observationof convergenceat the
higherlevel over the lake is consistentwith the existenceof the
downward adiabatic warming describedin section 4. The divergencegenerally increasesin the morning, reachesa maximum around 13 LST, then decreases in the afternoon. This
29,162
SUN ET AL.' LAKE-INDUCED
(a)
ATMOSPHERIC
the land decreaseswith time due to increasingdepth of the
vertical mixing. At the same time, the cool lake air is warmed
due to both advectionand downwardmixing of the warm air.
As a result,the land-lake air temperaturedifferencedecreases
at the lower level. However, the warm layer over the land
deepens,the vertically integrated land-lake air temperature
difference increasesas doesthe internal boundarylayer depth
H. Therefore the divergence at the lower level increases
steadily in the morning. This example further explains the
relatively poor correlation between the divergence and the
low-level air temperaturedifferencebetweenthe land and the
Twin Otter, 17 September 1994
• =2ms '1
6•
• , •
.41
I
0
,_
10
•
I
]6•
t
20
30
CIRCULATIONS
0'2
lake areas discussed in section 4.
As the horizontal divergencebecomesestablishedover the
lake, the advection of the warm land air over the lake surface
is opposedby the divergentlake breeze. The mixing between
the cool lake air and the warm land air is reduced,'and AT
starts to increase
at the 30 m level. This
increase
of the air
temperature difference is commonlyobservedin the early afternoon.
35,
Specific humidity and CO2 concentration are generally
higher over the lake than over the land in the morning (Figure
6). A Twin Otter passaround noon over Candle Lake under
the light wind conditionshowsa sharpdrop of specifichumidity aswell as CO2 and 03 concentrationsat the west end of the
lake (Figure 7). The sharpfront is the resultof easterlymoving
dry land air with low CO2 and high 03 and the westerlymoving
onshorelake breeze with high humidity and CO 2 and low 03
30 •
O• 540•
•0 "
10
•0
Distance (Km)
(b)
300,
Twin Otter, 17 September 1994
,
,
,
,
lake-modified
0
5
10
•5
20
400
•-2',"•';""¾":'•.,',•"%1
.200
•_ ,
0
• ,
0.4•'X•
/
00
5 =•
•
,
25
30
35 '
..................
'¾•
• , 7•--•''--I-----,0
:0 I
'
;0 I
N"'A •
I
5
•
I
25
30
ß , , , • • T-Z"-,---,•
-10 5 10 15 20 25
30
5 •'• -0.4
-0.6
0
0.4
5
10
15 20
35'
,
?"•"T",i , • ,
10 15 20 25 30
5'.4
'
5'.4
Distance (Km)
Figure 7. The Twin Otter flight over Candle Lake around
noon on September17, 1994; (a) mean variables(zonal and
meridional wind componentsu and v, air temperature T•,
surfaceradiation temperature T•, pressureP, specifichumidi• q•, CO2, and O3 concentrations),(b) fluxes.The reference
point 0 km along the flight track is arbitrarily chosenas the
west end of the land on the west side of Candle
Lake.
variation of the divergenceis consistentwith the observations
at Lake Okeechobee[Arritt et al., 1996].
The divergenceincrease in the morning occurseven when
the horizontal air temperature difference at 30 m decreases
with time. In the early morning the layer of warm air over the
land is too shallowto create a significanthorizontal pressure
gradient and there is no lake breeze to opposethe warm air
advectionfrom the land. Gradually, the rate of warming over
air.
The high CO2 and humidity and low O 3 concentrationover
the lake may be related to the remnants of the nocturnal air,
which has high CO2 concentrationfrom the land respiration,
high moisture due to evaporation, and low O3 due to ozone
depositionon leavesand the ground surface.This air is transported horizontallyand retainedin the stablenocturnalboundary layer over the lake. If the lake water is warmer than the
surroundingland at night, then the unstableair over the lake
may help distributethe high CO 2 and water vapor and low O 3
air into higher levels over the lake [Sun et al., 1997]. Nevertheless,only small fluxesare observedover Candle Lake during the aircraft passeswhen higher CO 2 and moisture and
lower O3 over the lake are observed(Figure 7). The high
humidity over the lake in the morning may also be related to
the weak upward moistureflux being confinedto a thin stratified layer over the lake surface.
Both humidity and CO 2 decrease,and 03 increasesover
both the land and the lake duringthe day.Thesevariationsare
causedby the plant uptake of CO2, possiblephotochemical
production of ozone, and entrainment and downward mixing
of dry and O3-rich air. The air is sometimesdrier over the lake
than over the land in the early afternoon becauseof the mean
downwardmotionat the higherlevelsover the lake (section4)
and partly by less evaporation over the cool lake compared
with the large midday evapotranspirationover the land. As a
part of the lake breeze circulation, the convergentflow associated with the subsidingwarmer, drier air with higher ozone
concentrationis seen at 1420 LST at 1300 m in Figure 6. The
CO2 concentration above 100 m is about the same over the
lake as over the land during the day (Figure 6). The sharp
increaseof the CO2 concentrationover the land at 1430 LST at
30 m in Figure 6 may be related to partial stomatalclosuredue
to developmentof clouds.
The large heat capacityof the lake water delaysboundary
layer developmentover the lake. Sincethe airplane flightsdo
SUN ET AL.' LAKE-INDUCED
ATMOSPHERIC
CIRCULATIONS
not occur before midmorning, the transition between noctur-
over Candle
nalanddaytime
boundary
layercannot
beobserved
overthe
ff
2o
ationsof CO2 and 03 concentrationsbetweenflight passesare
•
10
larger
overthelakethanthose
overtheland(Table
1,section
3). i
0
tember 17 case,shownin Figure 7. The concentrationof water
over Candle Lake
Lake
120
3o
land but can be observed over the lake. As a result, the vari-
Horizontalasymmetry
of specifichumidity,CO2 and 03
concentrations
dueto themeanwindcanbe seenin theSep-
29,163
100
• -10
=
ß•
-20
20
]0
']
-20
vapor,
CO2,andO3arehigher
downstream
fromthelakethan •, -30
upstreambecauseof the advectionof moist, CO2- and O3-rich
air by the westerly ambient wind. The asymmetryis not very
clear for many of the other casessince the concentrationdifferencesbetweenover the lake and over the land are typically
small during the day.
40
II
II
II
II
II
I'll
If
II
II
ii
I
i'1
II
I
Seasonal
II
II
III
II
I1 II11
II
I II
II
I
over the !and
over the !and
20O
20O•.09
Variations
The lake temperatureis about 10øCcolder than the typical
middayland temperatureat the end of May (Figure 2). The
lake temperatureincreasesslowlyfrom springto summerand
decreasesslowlyin late summer.The slowvariationof the lake
temperature is due to the large heat capacity of the water
(Figure2, section3). As a result,the moistureflux overCandle
Lake increasessteadilywith time duringthe three IFCs (Figure
8). Followingthe slowseasonalvariationof the lake temperature, the air-lake temperature difference decreaseswith time
from springthroughfall sincethe air temperatureoverthe lake
is more controlledby the advectionfrom the land (Figure 2,
section3). That is, the climatologicalstabilityof the boundary
layer over the lake changesfrom stablein the springto unstable in the fall, with occasionaldisruptionsby cold-front passages.The seasonaltrend for the weak sensibleheat flux is not
apparent due to the limited number of flight missionswith
more than three repeatedpassesrequiredfor Figure 8 in order
to reduce random samplingerrors in the fluxes.
Associatedwith the full developmentof leaves,canopystomata release more water vapor and take more CO2 and O3
during the summer,which leads to a larger upward moisture
flux and large downwardCO2 and 03 fluxes(Figure 8). However, the soil temperature dependenceof the respiration may
balancethe decreaseof the CO: concentration.As a result,the
seasonalvariation of the CO: flux is not well defined (not
shown).
Sincethe lake coolsslowlyat night, the lake temperature is
warmer than the land in the early morningfor many of the days
after the end of July. However, for most of the cases,the
upwardheat and moisturefluxesduringthe daytimeare larger
over the land than over the lake even when the lake temperature is warmer than the land surface radiation temperature
[Lenschowet al., 1996].Sun and Mahrt [1995] havefound that
the spatiallyaveragedsurfaceradiation temperature over the
land duringBOREAS canbe stronglyinfluencedby the shaded
cool ground surfaces,while the heat flux is dominatedby the
sunnycanopytop. As a result,the heat flux canbe upwardeven
though the averaged surface radiation temperature may be
coolerthan the air temperature.The fractionof shadedground
surfaceis particularlylarge in the morningwhen the Sun angle
is low. Therefore the sunnycanopytop over the land leads to
upward heat flux althoughthe averagedsurfaceradiation temperature is lower over the land than over the lake.
However, some specialcasesare observedwhere the fluxes
over the lake are larger than thoseover the land. For example,
on September3, 1994, the land surfacewas about 6øCcolder
IIill
IFC1 7/20-8/2
IFC2 8/31-9/17
IFC3
5/25-6/10
150
6.
-40
%d
.,-."'ø
50
1111
!1
II'11
Itl
I1
II
11111
II
I!
I
50 iiIi !111-1
ii ii11
!111
ii i•!1
iii
Figure 8. Seasonalvariationsof sensibleheat (top left) and
latent heat (top right) fluxesover Candle Lake and sensible
heat (bottom left) and latent heat (bottom right) fluxesover
land based on LongEZ observations.The incrementbetween
tickson the x axisrepresentsa 2-dayperiod.The datesof the
intensivefield campaignsare indicatedin the abscissa.
than the lake temperaturein the middle of the day with heavy
cloudcover(Figure 9). For thiscase,the specifichumidityand
air temperature are higher over the lake than over the land.
The sensibleand latent heat fluxesover the lake are, respectively,abouttwo and four timeslarger than thoseover the land
due to the cool land air over the warm water. The momentum,
CO2, and 03 fluxesare about the same over both surfaces.
7.
Fluxes As Function of Flux Averaging Scales
Applying the bulk formulae for fluxes over heterogeneous
surfacesis problematical[Mahrt and Sun, 1995a,b]. The bulk
formulae are developed for calculation of turbulent fluxes
basedon local meanvariables.Theserelationshipsbetweenthe
fluxesand the local mean variablesmay not be the sameas the
relationshipbetween the area-averagedfluxes and the areaaveragedmean variables.With surfaceheterogeneity,the relationship between the grid-averaged fluxes and the gridaveraged air-ground temperature and humidity differences
and wind speedin numericalmodelsdependson the grid size
[Mahrtand Sun, 1995a,b]. The requiredtransfercoefficientsin
the bulk formula may not follow surfacesimilaritytheory.
In order to examine the grid size dependenceof the bulk
relationshipin the presenceof lakesthe transfercoefficientis
examined in this section as a function of the averaginggrid
size, as simulated by choosingsectionsof aircraft record of
different lengths.The fluxesand the mean variablesrequired
in the bulk formula are averagedover lengthsof L, 2L, ---,
alongflight tracks,whereL is the lake width. For an averaging
width of L the area coversCandle Lake only. For the width of
2L the area coversthe lake and the adjacentland along the
flight track. As the averagingwindowwidth increases,the per-
29,164
SUN ET AL.: LAKE-INDUCED
(a)
ATMOSPHERIC
CIRCULATIONS
from unweightedmeanswith a fixed averagingwidth of 1.2 km.
The cutoff length of 1.2 km capturesmostof the turbulent flux
based on argumentssimilar to those presentedby Sun et al.
[1996].The wind is vector averagedas in numericalmodels
Electra, 3 September 1994, at 100 m
.•=2ms -1
V = (•2 + Vyy2)•/2
0
(6)
Here Vxand Vyare the wind components
in the zonaland
-
10
20
30
12
f ' I
11
40
..... '
10 .,,4,,•
.....•'"'C•q<
90
10
20
7.5
•.•
•M•
•q
7.25
6.5•
'•
•
0
10
16
"..•'a,•.
, '
30
40
',
',
20
30
meridional directions,and the overbar representsthe spatial
averaging.Transfer coefficientsare calculatedfrom the fluxes,
air-ground differences,and wind componentswhich are averaged over repeatedpassesduring a given flight.
As an example, the lake effect on the area-averagedflux
from the LongEZ on May 26, 1994, is shown in Figure 10.
When the grid size coversonly the lake, the grid area is homogeneous,the air over Candle Lake is stable with weak
downwardsensibleheat at the aircraft flight level of 30 m; the
air is warmer than the lake water and the airwater temperature
difference is small. As the averagingwindow increasesto include more warm land surfaces, the fluxes and the gridaveragedvariablesare dominatedby strongupward sensible
heat, latent heat, and momentum fluxesand large air-ground
differences.Therefore as the averagingscale increasesfrom
•02
10
08
',. •
•
0.5
'
40
• -1
50
'
26
S48•
24
546
•
544•
, 10
I , 20I
i ' 40i
100% lake to almost 100% of the land, where the lake effect
0•
•
30
becomesnegligible, the averaged sensibleheat flux and the
mean air-ground temperature difference change signs from
Distance (Km)
(b)
Electra, 3 September 1994, at 100 m
LongEZ, 26 May 1994
• I 75f •
i 400
•
0-2•
0
•
10
i .• • , i . f 2,:•
0.06
(a)
5 ,, 0.06
20 '
0.04
0 :.q0.04
;0
0'2
• • • , • ' . 2,
•
20
,
t
30
t
40
'
• -2
50
o.o
'sb o.o2
o
.lO
.0.021,
I , I , I , t , : ,
(b) 2
00
10
20
30
40
3
4
5
6
0.041
0.03, . , . , . , • , , , 112.5
10,
50
'2
7.5
0.02
0
I• w•
•
•
-0.
.0.2
0
i
40
.
,
.
10
20
30
40
'•
50
Distance (Km)
Figure 9. (a) Electrazonaland meridionalwind components
u and v, air (T•), and surfacetemperatures(T•), specific
humidi• (q•), NDVI (normalized difference of vegetation
index), CO2 and O3 concentrationsover Candle Lake at an
altitudeof 100 m on September3, 1994.CandleLake is the low
NDVI
area in the center. The low NDVI
2.5
;
, ,,
-15
7
;
0
1.75
.
,
[}c)
, I ' I ' I ' I ' I ' 6 0.15
1.5[
5•
............
'""-.
-5
0.1
0.05
075
'"•,'""'"
t , • , •''"
"•'
'' 4 0
ß 1
2
3
4
Grid
5
6
7'
Size
area at 40 km cor-
respondsto White Gull Lake. (b) Sensibleand latent heat, Figure 10. (a) Sensibleheat flux, the differencebetweenair
momentum (expressedas friction veloci•), carbon dioxide, and surfaceradiationtemperatures,and observedheat transfer
coefficientas a functionof averagingscale;(b) moistureflux,
and ozone fluxes across Candle Lake.
the difference between specific humidity and the saturated
specifichumidityusingsurfaceradiationtemperature,and observedmoisturetransfercoefficient;(c) momentumflux,wind
centage of the lake in the averagingarea decreases.In this speed, and observed drag coefficient. All the variables are
approach, averagingalong the flight track is assumedto be averagedfrom four repeated passes,and the abscissarepreequivalent to averagingacrossa grid area. To exclude me- sentsthe averagingscaleequal to L, 2L, 3L, -.., where L is
soscalefluxes, the turbulence is calculatedas perturbations the lake width.
SUN ET AL.: LAKE-INDUCED
ATMOSPHERIC
CIRCULATIONS
29,165
negative to positive at some intermediate averaging lengths. column air temperaturewhich leads to a horizontal pressure
This intermediate averaginglength correspondsto the grid difference that drives the lake breeze circulation.
area where both !akc and land are important and is several
The lake breeze stronglydependson the diurnal developtimes the lake width. However, the increase of the sensible ment of the difference between the boundary layers over the
heat flux with averagingscalecan be faster or slower than the land and over the lake. In the morning the residualland breeze
increaseof the air-temperaturedifferencewith averagingscale. is commonly observed even as the air first becomes warmer
As a result, the transfer coefficient must become negative at over the land than over the lake. The carbon dioxide and water
some grid sizesin order to use the bulk formula to obtain the vapor concentrationscan be higher over Candle Lake than
measured
sensible heat flux. The dramatic
variation
of the
over the land in the morning due to retention of nocturnal air
exchangecoefficientfor heat happensonly when the heat flux over the lake and the muchlargerverticalmixingover the land.
is downward
over the lake and the, air is warmer
than the lake.
The exchangccoefficientsfor moistureand momentumgenerally increasesteadilywith averagingscale.
In addition to the averagingproblem in the sensibleheat
flux, the grid-averagedwind speedbascdon (6) decreaseswith
averaging scale [Mahrt and Sun, 1995a, b]. If the grid just
coversthe lake, the divergencefrom the lake breeze leadsto a
very small grid-averagcd wind speed. As the grid averaging
scaleincreasesto includethe land surface,the vector-averaged
wind speedis closerto the ambient large-scalewind. Therefore
the area-averagedwind speedmay increaseas the percentage
of the land surfacc in the grid increases.Because of the restriction of the observationsto a singleflight track, the cancelation for the averagingscaleequal to the lake width in Figure
10 is limited only to the direction of the flight track. As the
averagingscalecontinuesto increase,the area-averagedwind
speeddecreasesdue to random fluctuationsof the wind direction. This problem is not related to lakes specificallybut more
to area averagingin general. Therefore the exchangecoefficients may increasewith averagingscaleunlessthe fluxesand
the air-ground differencesdepend significantlyon grid size.
The problcms related to the area averagingcan be complicated if a grid size covcrsseveral lakes, which could happen
with grid arcas in the BOREAS study area. The example in
Figure l{I showstypical difficulticswith scalingup in casesof
strong hctcrogcncityassociatedwith lakcs. In situationswhere
lakes dominate the surfaceheterogeneity,the mosaicapproach
in numerical modeling I Wctzelaml Chat•g, 1988;Avissarand
Pielke, 1989;Chtttssen,1991;Dttcottdrt]el al., 1993;Huang and
l,yoas, 1995] may bc suitable for solving the area-averaging
problems.
8.
Summary
Lake.
The
turbulence
is weak over Candle
Lake
than
over the surroundingland even at the 1000 m level.
The divergenceof the lake breeze circulation is negatively
correlated with the wind speed. Under weak wind conditions,
the lake divergencedecreascswith increasinglake size and is
not sensitiveto the lake water temperature. The air temperature over three different
sized lakes is about the same since the
timescale of the vertical mixing in the stratified air over the
lake is longer than the timescaleof the land-air advection.The
lake breeze
is sensitive
to the relative
horizontal
difference
around
1300 LST.
The vertical
land contributes
to the horizontal
difference
turbulent
fluxes are much smaller
than over the land. Weak
downward
sensible
over the lake
heat flux is ob-
servedat the 30- and 100-m flight levels.As the lake warmsup
from springto fall, the latent heat flux over the lake increases
steadily,while over the land, the latent heat flux peaks during
the summerseasondue to transpiration.
The area-averagedfluxesstronglydepend on the averaging
length when significantland-lake contrast occursin the averaging area. For a grid size several times the lake width, the
exchange coefficient for heat in the bulk formula must be
negative to predict the correct grid-averagedheat flux. The
negative exchange coefficient occurs when the sensible heat
flux and the air-groundtemperaturedifferencesare dominated
by the different surfacetypes.
Numerous lakes in BOREAS influence not only the flow
throughland and lake breezes,which alsoinducelocal changes
in cloud cover, but also the area-averaged fluxes. In the
BOREAS southern and northern study areas, lakes cover
about 10% and 20% of the study areas, respectively.If we
assumethe upward sensibleheat flux over the land and downward over Candle Lake on May 26, 1994 (Figure 10), to represent lake and the land areas in BOREAS, respectively,the
area-averagedsensibleheat flux over boreal forest is estimated
to be 12-24% lower than if the area were covered by forest.
However, a more accurate estimate of the area-averagedheat
flux over the BOREAS area requiresmore information on the
spatial variation of the surface fluxes over the boreal forest.
Besidesthe surfacetype, the sensibleheat flux can depend on
the depth and turbidity of the lake. For example,the sensible
heat fluxes over White
Gull
Lake
are found
to be
higher than those over Candle and Halkerr Lakes since this
lake is shallow
and turbid.
Acknowledgments. Jielun Sun and Larry Mahrt were supportedby
NASA grant NAG 5-2300. Donald H. Lenschow,Steve Oncley, Qing
Wang, and K. J. Daviswere sponsoredby NASA grant PO S-12857-F.
Qing Wang and K. J. Davis'sparticipationwas alsosupportedin part
by NCAR's Advanced Study Program. The NOAA LongEZ observations, operated by Tim Crawford and Ron Dobosy,were completed
under BOREAS programfunding.Fundingfor Twin Otter operations
in BOREAS was providedby the National ResearchCouncil of Canada, Agriculture Canada, and the Natural Sciencesand Engineering
ResearchCouncil of Canada.Discussions
with Ray Arritt were greatly
appreciated.The National Center for Atmospheric Research is supported by the National ScienceFoundation.
in
the boundary layer development and column air temperature
between the lake and the land, not the air temperature difference at lower levels.The rapid boundary layer deepeningover
the warm
pressuregradient increases.This pressuregradient drives the
divergence over the lake which reaches maximum values
and latent
The above study has shown that cool lakes, here ranging
from 3 to 1() km in width, can generate well-defined lake
breezes on days with weak large-scaleflow. As a part of the
lake brcezc circulation, downward motion of warm, dry air,
and high ozone concentration is observed at 1000 m over
Candle
As the air over the land continues to warm, the horizontal
of the
References
Arritt, R. W., The effect of water surfacetemperatureon lake breezes
and thermal internal boundarylayers,BoundaryLayer Meteorol.,40,
101-125, 1987.
29,166
SUN ET AL.: LAKE-INDUCED
ATMOSPHERIC
CIRCULATIONS
Arritt, R. W., M. Segal, M. Leuthold, C. J. Anderson, and R. W.
Mahrt, L., and J. Sun, Dependenceof exchangecoefficientson averTurner, An observationalstudyof the Lake Okeechobeelake breeze
aging scale or grid size, Q. J. R. Meteorol. Soc., 121, 1835-1852,
1995b.
and its effect on deep convection,paper presentedat the Conferenceon CoastalOceanicand AtmosphericPrediction,Am. Meteo- McDermott, M. L., and R. D. Kelly, Fluxes over a heterogeneous
rol. Soc.,Atlanta, Ga., 28 January-2 February 1996, 1996.
forest,paper presentedat the 11th Symposiumon BoundaryLayers
and Turbulence, Am. Meteorol. Soc., Charlotte, N. C., March 27-31,
Avissar,R., and R. A. Pielke, A parameterizationof heterogeneous
1995.
land surfacesfor atmosphericnumerical models and its impact on
regionalmeteorology,Mon. WeatherRev., 117, 2113-2136, 1989.
Rabin, M. S., S. Stadler,P. J. Wetzel, D. J. Stensrud,and M. Gregory,
Biggs,W. G., and M. E. Graves,A lake breezeindex,J.Appl. Meteorol.,
Observedeffectsof landscapevariabilityon convectiveclouds,Bull.
1,474-480,
1962.
Am. Meteorol. Soc., 71,272-280, 1990.
Claussen,M., Estimationof areally-averagedsurfacefluxes,Boundary
Layer Meteorol.,54, 387-410, 1991.
Rotunno, R., On the linear theory of land and sea breeze, J. ,4tmos.
Dalu, G. A., R. A. Pielke, R. Avissar, G. Kallos, M. Baldi, and
Segal, M., and R. A. Pielke, The effect of water temperature and
synopticwinds on the developmentof surfaceflows over narrow,
elongatedwater bodies,J. Geophys.Res.,90, 4907-4910, 1985.
Sellers, P., et al., The Boreal Ecosystem-Atmosphere Study
(BOREAS): An overviewand early resultsfrom the 1994 field year,
A. Guerrini, Linear impact of thermal inhomogeneitieson mesoscaleatmosphericflowwith zero synopticwind,Ann. Geophys.,9,
641-647, 1991.
Dobosy,R. J., T. L. Crawford,J. I. MacPherson,R. Desjardins,R. D.
Kelly, S. P. Oncley, and D. H. Lenschow,Intercomparisonamong
the four flux aircraftat BOREAS in 1994,J. Geophys.
Res.,thisissue.
Ducoudrd, N. I., K. Laval, and A. Perrier, SECHIBA,
a new set of
parameterizations of the hydrologic exchanges at the land/
atmosphereinterfacewithin the LMD atmosphericgeneralcirculation model, J. Clim., 6, 248-273, 1993.
Ehret, G., A. Giez, C. Kiemle, K. J. Davis, D. H. Lenschow, S. P.
Oncley, and R. D. Kelly, Airborne water vapor DIAL and in situ
observationsof a sea-landinterface, Contrib.Atmos.Phys.,69, 215228, 1996.
Estoque,M. A., J. Gross,and H. W. Lai, A lake breeze over southern
Lake Ontario, Mon Weather Rev., 104, 386-396, 1976.
Huang, X., and T. J. Lyons,The simulationof surfaceheat fluxesin a
land surface-atmospheremodel, J. Appl. Meteorol.,34, 1099-1111,
1995.
Keen C. S., and W. A. Lyons, Lake/land breeze circulationson the
westernshoreof Lake Michigan,J. Appl. Meteorol.,17, 1843-1855,
Soc., 40, 1999-2009, 1983.
Bull. Am. Meteorol. Soc., 76, 1549-1577, 1995.
Smith, B., and L. Mahrt, A studyof boundarylayer pressureadjustments, J. ,4tmos. Soc., 38, 334-346, 1981.
Sun, J., and L. Mahrt, Relationshipof surfaceheat flux to microscale
temperaturevariations:Application to BOREAS, BoundaryLayer
Meteorol., 76, 291-301, 1995.
Sun, J., L. Mahrt, S. K. Esbensen,
J. F.
Howell,
C. M.
Greb, R. Grossman,and M. A. LeMone, Scale dependenceof airsea fluxes over the western equatorial Pacific,J. ,4tmos.Soc., 53,
2997-3012, 1996.
Sun, J., R. Desjardins,and L. Mahrt, The influenceof lakes on atmospherictransportcarbon dioxide in BOREAS, paper presentedat
the SpecialSymposiumon BoundaryLayer and Turbulence,Am.
Meteorol. Soc.,Long Beach,Calif., Feb. 2-7, 1997.
Wetzel, P. J., and J. T. Chang,Evapotranspirationfrom non-uniform
surfaces:A first approachfor short-termnumericalweather prediction, Mon. WeatherRev., 116, 600-621, 1988.
1978.
Kelly, R. D., J. I. MacPherson, R. J. Dobosy, and T. L. Crawford,
BOREAS 1994 intercomparisonamong three flux aircraft, paper
presentedat the 22nd Conferenceon Agricultural and Forest MeteorologyWith Symposiumon Fire and Forest Meteorology, Am.
Meteorol. Soc., Atlanta, Ga., Jan. 28-Feb. 2, 1996.
Lenschow,D. H., B. B. Stankov,and L. Mahrt, The rapid morning
boundarylayer transition,J. Atmos. Soc.,36, 2108-2124, 1979.
Lenschow,D. H., Q. Wang, S. P. Oncley, K. J. Davis, and J. Mann,
Lake-induced modification of the boundary layer over the boreal
forest,paper presentedat the 22nd Conferenceon Agriculturaland
Forest MeteorologyWith Symposiumon Fire and ForestMeteorology, Am. Meteorol. Soc.,Atlanta, Ga., Jan. 28-Feb. 2, 1996.
Lyons, W. A., The climatologyand prediction of the Chicago lake
breeze,J. Appl. Meteorol.,11, 1259-1270, 1972.
MacPherson,J. I., NRC Twin Otter operationsin BOREAS 1994,
NRC Rep. LTR-FR-129, Natl. Res. Counc. of Can., Ottawa, Can.,
J. Sun,Programin Atmosphericand OceanicSciences,Universityof
Colorado, Boulder, CO 80309-0311. (e-mail: jls@monsoon.
colorado.edu)
K. J. Davis, D. H. Lenschow,and S. P. Oncley, National Center for
AtmosphericResearch,Boulder, CO 80307-3000.
L. Mahrt, College of Oceanic and AtmosphericSciences,Oregon
State University, Corvallis,OR 97331.
T. L. Crawford and R. J. Dobosy, Atmospheric Turbulence and
DiffusionDivision,NOAA ARL, Oak Ridge, TN 37830.
J. I. MacPherson, National Research Council, Ottawa, Ontario,
KiA036, Canada.
Q. Wang, MeteorologyDepartment,Code MR/Qg, Naval Postgraduate School,Monterey, CA 93943-5114.
R. L. Desjardins,Land ResourcesResearch Centre, Agriculture
Canada, Ottawa, ON, KiA 0C6, Canada.
1996.
Mahrt, L., and J. Sun,Multiple velocityscalesin the bulk aerodynamic
relationshipfor spatiallyaveragedfluxes,Mon. WeatherRev., 123, (ReceivedMarch 28, 1996; revisedFebruary 15, 1997;
3032-3041, 1995a.
acceptedMarch 3, 1997.)