Nocturnal temperature inversions in a small, enclosed ...

Meteorol Atmos Phys 103, 195–210 (2009)DOI 10.1007/s00703-008-0341-4Printed in The Netherlands

Department of Geography, Michigan State University, East Lansing, MI, USA

Nocturnal temperature inversions in a small, enclosed basinand their relationship to ambient atmospheric conditions

W. Yao, S. Zhong

With 13 Figures

Received 2 October 2007; Accepted 13 October 2008Published online 20 January 2009 # Springer-Verlag 2009

Summary

Observations from a recent meteorological experiment inthe Meteor Crater, a small, near circular, enclosed basin,in northern Arizona of the United States, are used toinvestigate the status of the atmosphere in the basin asindicated by temperature inversion and its relationshipto ambient atmospheric conditions. Strong synoptic windsaloft do not necessarily imply that no inversion coulddevelop in the basin. Instead, the near-surface wind over theplain surrounding the basin is found to be a more importantfactor in determining whether the basin atmosphere wouldbe coupled to, or decoupled from, its ambient environment.A necessary condition for the decoupling and the develop-ment of a strong inversion overnight is that the meannighttime temperature over the plain needs to be less than5 m s�1. The low-level stability of the ambient environmentalso plays an important role. It was found that as long as thebulk Richardson number over the plain is less than 0.6,turbulence would exit over the plain and influence the basinatmosphere at least partially through the night, preventinga strong temperature inversion from developing in thebasin. The strength of the inversion in the basin appears tobe increase exponentially with the longwave radiation lossfrom the upper basin sidewalls. Over the basin floor, therelation between inversion strength and longwave radiationloss is poor mainly because the cooling of air over the basinfloor is a combination of radiative cooling and the drainageof cold air from the sidewalls.

1. Introduction

The formation and destruction of nocturnal tem-perature inversions in a mountain basin has beena subject of many previous studies. This is be-cause the high static stability associated withtemperature inversion, together with the basinsidewalls, limits atmospheric dispersions andtends to trap pollutants in the basin, which maylead to hazardous air pollution conditions (Reddyet al. 1995). When the atmosphere in the basin ismoist, strong inversions may lead to the forma-tion of fog and drizzle, affecting visibility andtransportation (Smith et al. 1997).

Majority of the previous studies on basintemperature inversions used either an analyticalapproach or numerical modeling. By solvingsemi-analytically an energy balance equation,Petkov�ssek (1978; 1985; 1992) estimated the en-ergy required for thermal dissipation and thewind speed for dynamic dissipation for tempera-ture inversions in an idealized basin. His analysessuggested that for typical Slovenia basins a windspeed of 7 to 9 m s�1 right above the basin isrequired to initiate dynamic dissipation of inver-sion from above. As the warmer air aloft pene-trates downward during turbulent dissipation, theinversion at the top of the remaining cold air pooltends to become stronger. As a result, winds aloft

Correspondence: Sharon Zhong, Department of Geography,

Michigan State University, 116 Geography Building, East Lansing,

M1 48824, USA (E-mail: [email protected])

need to increase continuously with time so thatshear-generated turbulence would remain greaterthan the buoyancy consumption to maintain tur-bulence erosion of the inversion from top. Theseresults were later confirmed by numerical model-ing study of inversion breakups by Vrhovec andHrabar (1996) and Rakovec et al. (2002). Alsobased on a semi-analytical approach, Zhong et al.(2003) estimated time scales involved in the dis-sipation of wintertime temperature inversions ina basin by turbulence erosion. Their calculationindicated that micro-scale turbulent erosion is arather slow process and that the rate of erosiondecreases rapidly with time as static stabilityincreases near the top of the inversion. Conse-quently, while weak and shallow inversion of afew tens of meters may be removed by turbulenceerosion from above in a matter of hours whenwinds are sufficiently strong, deep inversions withmoderate to strong temperature gradient are un-likely to be destroyed by turbulence erosion un-less combined with other larger-scale process.Recently, Z€aangl (2005) examined formation ofextreme temperature inversions in elevated sink-holes using idealized numerical modeling. Thenumerical simulations identified two crucial fac-tors for extreme cooling and strong temperatureinversions to occur in a sinkhole, namely, smallheat conductivity of the ground and more impor-tantly a drying mechanism to remove moisture inthe sinkhole during cooling process.

Compared to the number of analytical and nu-merical studies in the literature, observationalstudies of basin temperature inversions are rela-tive sparse. This is due mainly to the difficulty inmaking meteorological measurements in moun-tainous terrain in general and the usually harshenvironment with sometimes extremely coldtemperature on winter nights in a basin in partic-ular. Whiteman et al. (2001) studied two cold airpool episodes in the large Columbia Basin ofeastern Washington using limited observationsand indicated regional downslope winds, coldair advection aloft, and convective boundary lay-er growth as mechanisms for the destruction ofthese cold pools in a large basin. Clements et al.(2003) reported observations using a tetheredballoon and several surface stations during atwo-day exploratory study in Peter Sinks, a smallbasin of 1-km in diameter in northwestern Utah.The data showed that despite the calm conditionat night with near zero downward turbulent heat

flux, the basin atmosphere continued to coolthroughout the night. The cooling was attributedto radiative flux divergence. The inversion wasremoved through subsidence as well as convec-tive boundary layer growth the next morning(Whiteman et al. 2004). Observational data werealso collected in the Austria’s Gruenloch basin, a1-km diameter limestone sinkhole of 120 m deep(Steinacker et al. 2002). Although similar in sizecompared to the Peter Sinks basin, the moistenvironment in the Gruenloch basin producedmuch smaller cooling at night than in the dryPeter Sinks basin, and because most of the in-coming solar energy was converted to latent heat,the breakup of the inversion was caused primari-ly by subsidence instead of convective boundarylayer growth (Whiteman et al. 2004).

These exploratory experiments helped to ad-vance our knowledge about nocturnal coolingand temperature inversion inside a small basin.They were, however, limited by the number andtypes of measurements available for analyses andby the short duration of a few days. In addition,the interpretations of the previous observations inbasins were all complicated by the existence ofgaps or paths on the basin sidewalls, the asym-metry of the sidewall slopes and heights, and theheterogeneity of the vegetation cover.

During October 2006, an intensive meteoro-logical field campaign called the Meteor CraterExperiment, or METCRAX, was conducted inthe Meteor Crater on the Colorado Plateau innorthern Arizona to study the formation and de-struction of temperature inversions and the char-acteristics of seiches in a small, enclosed basin(Whiteman et al. 2007, 2008). A large numberof continuous observations were available fromin-situ and remote sensing instruments inside andoutside the crater. The symmetrical shape of thecrater and the uniform slopes and heights of thesidewalls with no gaps significantly simplifiedthe data interpretation. We present, in this paper,results from analyses of the data inside and out-side the crater with focus on conditions leadingto the formation and destruction of a temperatureinversion in the crater.

2. Sites and measurements

The Barringer or Meteor Crater, located on theColorado Plateau approximately 40 km east of

196 W. Yao, S. Zhong

Flagstaff, Arizona, is a small, enclosed, near-cir-cular-shaped basin formed by the impact of ameteorite some 50,000 years ago. The crater isapproximately 165 m deep and 1200 m in diame-ter at the rim level. The rim of the crater, whichhas no major paths, rises approximately 50 mabove the surrounding plain on the Colorado Pla-teau. The crater floors and sidewalls are primari-ly rocks that are sparsely covered with grassesand small bushes. The topography of the MeteorCrater is shown in Fig. 1 along with measure-ment sites used in this study.

From the 1st through the 31st of October,2006, continuous measurements of mean meteo-rological variables, turbulence fluxes, and radia-tive and soil fluxes were made inside the craterwith an array of micrometeorological flux towers

or ISFF (Integrated Surface Flux Facility) sup-plied, installed, and operated by the NationalCenter for Atmospheric Research (NCAR)’sEarth Observing Laboratory (EOL). The towerswere located on an east-west cross section (Fig. 1)with a 9 m tower near the center of the crater(FLR) and on the lower east and west sidewalls(EL and WL), and a 6 m tower on the upper slopeof the east and west sidewalls (EU and WU). Inaddition to the five flux towers inside the crater, a10 m tripod was used to collect data at the high-est point (1744 m above ground level or AGL) onthe crater rim (RIM). The specific site informa-tion is given in Table 1. On each tower, measure-ments of temperature, humidity, and 3D windcomponents were made at four levels usinghygro-thermometers and 3D sonic anemometers.

SW

ISS

Longitude

Latit

ude

35

35.01

35.02

35.03

35.04

35.05

35.06

35.07

35.08

1700

1680

1660

1570

16001600

1700 1700

1700

1700

1700

1700

WU WLFLR

EL EU

RIM

Fig. 1. Topographic map of the MeteorCrater (a) and the locations of theinstrumented towers inside the crater (b).Contour interval is 20 m in (a) and 10 min (b)

Table 1. Information for all the measurement sites used in the analyses

Site Site ID Longitude(��W)

Latitude(��N)

Altitude(m MSL)

Slopeangle (��)

Instrument

West slope upper WU 111.0270 35.0274 1609 22 flux towerWest slope lower WL 111.0255 35.0272 1572 4 flux tower

tethersondeCrater floor FLR 111.0225 35.0280 1563 – flux tower

tethersonderadiation tower

East slope lower EL 111.0198 35.0272 1572 7 flux towertethersonde

East slope upper EU 111.0184 35.0272 1600 24 flux towerSouthwest SW 111.0388 35.0103 1697 – SODAR

ISS ISS 111.0342 35.0722 1670 – radar wind profilerradiosondeautomatic weather station

Crater Rim RIM 111.0292 35.0295 1744 – Hydro-thermometer

Nocturnal temperature inversions in a small, enclosed basin 197

Measurements of moisture flux and radiationwere made at one level using a Krypton hygrom-eter, and net radiometer and=or pyranometer-pyr-geometer.

Vertical profiles of mean meteorological vari-ables through the depth of the crater atmospherewere obtained during Intensive ObservationalPeriods (IOPs) that would start several hours be-fore sunset and continue until several hours aftersunrise the next morning. The vertical soundingswere made using tethered-balloon sounding sys-tems that were operated simultaneously at threelocations (FLR, between EL and EU, and bet-ween WL and WU) at approximately 30 min timeinterval through each IOP. The IOPs were con-ducted on the basis of forecasts of quiescencesynoptic weather conditions to investigate for-mations of cold air pools, gravity waves, andseiches. A total of seven IOPs were conductedduring the month-long experiment.

Ambient meteorological conditions were mon-itored by a mini-sodar and a Radio AcousticSounding System (RASS) at a location about2.5 km southwest of the crater (SW in Fig. 1).The mini-sodar and RASS provide wind andtemperature profiles from about 10 m to 200–300 m above the surface at 10 m height interval.An ISFF was co-located at the mini-sodar=RASS

site to provide surface observations of mean andturbulence variables. This site was chosen basedon a climatological analysis of surface meteoro-logical data from the nearest national weatherservice station at Winslow, AZ which indicatedthat the prevailing surface wind direction forthe month of October is southwest. Upper-airmeteorological conditions were also observedby NCAR EOL’s Integrated Sounding System(ISS) consisting of a 915-MHz radar wind profil-er with RASS and a 10 m enhanced weather tow-er. The ISS site was located approximately 5 kmnorth–northwest of the crater (ISS in Fig. 1) toavoid radio interference with the RASS at the SW

mini-sodar site. During the seven IOPs, VaisalaRS92 GPS radiosondes were launched from theISS sites every three hours, starting at 1500 MSTand ending at 0900-1100 MST the next morning,to document the large-scale background atmo-spheric conditions.

Detailed descriptions of the experimental de-sign, measurement sites, instrumentation, andsampling strategies were given in Whiteman

et al. (2008). Although tethered-balloon sound-ings provided detailed vertical structure of tem-perature inversion, they were operated onlyduring the seven IOPs, all of which were undersimilar synoptic conditions with weak synoptic-scale winds. Since the focus of this study was onthe interaction of temperature inversions insidethe crater with the background atmospheric con-ditions, the analyses were primarily focused onthe continuous observations from the tower arrayinside the crater and the remote sensing instru-ments outside.

3. Results

3.1 Characteristics of the nocturnaltemperature inversions inside the crater

The general characteristics of the temperature in-version were examined by comparing tempera-ture time series from the five towers located atdifferent heights above the crater floor. The timeseries of 5-min average temperature at the 0.5 mlevel on all five flux towers are shown in Fig. 2for each day of the month. A temperature inver-sion developed on majority of the nights, withhigher temperature found at the two uppertowers, followed by the two lower towers, andthe lowest temperature found at the floor tower.The strengths of the inversion, however, variedsubstantially from one night to another. On somenights the temperature differed by as much as4–5 �C in the lowest 10 m between the floorand the two lower towers, and another 3 �C inthe next 30–35 m between the lower and the up-per towers, while on other nights, the tempera-ture differences between the towers were muchsmaller. Based on the strength and duration of thetemperature inversion, the nights may be dividedinto three regimes. The first regime was charac-terized by a strong temperature inversion thatusually began shortly after sunset and continueduntil the next morning. These nights will be re-ferred to as fully decoupled nights because thecrater atmosphere during these nights was com-pletely decoupled from the ambient atmosphereand winds inside the crater were usually veryweak or near calm. The second regime, whichwill be referred to as fully coupled regime, is op-posite to the first in that little or no temperatureinversion developed at night and the observed


temperatures on different towers collapsed to-gether and were similar to the temperatures ob-served on the rim and at the ISS site and SW siteoutside the crater over the plain. Between thesetwo regimes lies partially decoupled regimewhen the basin atmosphere became decoupled

from ambient environment during part of thenight, allowing a temperature inversion to buildup inside the crater for a period of time.

A more objective categorization of differentstatus of the crater atmosphere, or a more quan-titative measure of the temperature inversions

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31-10

-5

0

5

10

15

20

25

30

Date

Tem

pera

ture

(°C

)

EU

EL

FLR

WL

WUFig. 2. Time series of 5-minmean temperatures observedat 0.5 m level from all fivetowers inside the crater forOctober 2006

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30-500

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

6500

7000

Inve

rsio

n st

reng

th (

°C m

in)

Date

Fig. 3. Nocturnal temperatureinversion strengths as deter-mined by Eq. (1) for eachnight of October 2006


inside the crater, can be achieved using the inte-grated inversion strength, as described by

I ¼ðt2t1

ð�TTWU; EU � TFLRÞdt; ð1Þ

where t1 and t2 correspond to sunset and sunrisetimes at the location of the Meter Crater, TFLR isthe 0.5 m level temperature on the floor tower,and �TTWU, EU is the average 0.5 m-level tempera-ture at the two upper sidewall towers. The timestep for the integration is 5 min. This integralrepresents the area bounded by the FLR curveand the mean of the EU and WU curves in

Fig. 2. A zero value of I would imply either theatmosphere inside the crater was isothermalthroughout the night (zero area between the curvesall night long), or the temperature would in-crease with height during some periods at night(positive area) and at other times would decreasewith height (negative area) by similar amount. It isclear that the magnitude of I is proportional to thestrength of the temperature inversion. The largerthe value is, the stronger, on average, the nocturnalinversion was in the lower part of the crater be-tween the floor and the two upper sidewall towers.

Figure 3 shows the I value for each night ofthe month obtained from Eq. (1) using the

a b

c d

Fig. 4. 700 hPa composite (a) geopotential heights and (b) wind fields at 06 UTC (23 MST) for all 4 fully coupled cases;and (c) geopotential heights and (d) wind fields for all 15 fully decoupled cases. Contours in (a) and (c) are geopotentialheights with 25 m interval, and in (b) and (d) are wind speed at 2 m s�1 interval. The data are from NCEP=NCAR

reanalysis data


5-min mean tower data. A careful examinationof the I values and the temperature inversionson each night suggests the following thresholdvalues to be used to best represent the threeregimes discussed above: I� 1500 �C min forfully decoupled regime; I�100 �C min for fullycoupled regime; and 100 �C min <I<1500 �Cmin for partially decoupled nights. Based onthese criteria, four nights (2–3, 4–5, 5–6,16–17) during the month of October fell underthe fully coupled category and the rest weresplit between fully decoupled (15 nights) andpartially decoupled (11 nights) regimes. Thestrongest inversion with I¼ 6600 �C min devel-oped on the night of 22–23 Oct., which wasalso the best IOP night (IOP5) during the entirefield study.

3.2 Impact of local, regional, and synoptic flows

To understand synoptic conditions associatedwith the presence or absence of the basin tem-perature inversion, composite synoptic weathermaps were made for the three regimes usingNCEP (National Center for Environmental Pre-dictions) and NCAR Global Reanalysis data set(http:==www.cdc.noaa.gov=Composites=Hour=).Figure 4a, b shows the composite maps of 700-hPa geopotential heights and wind vectors forthe fully coupled nights. The main synoptic fea-ture for these nights was a deep trough locatedoff the coast of southern California, bringingstrong (12–14 m s�1) mid-level (700 hPa) south-westerly synoptic winds to the southwest US

including Arizona. These strong synoptic winds

a b c(2880) (6960) (10800)

NORTH

WEST

6%

12%

18%

24%

30%

6%

12%

18%

24%

30%

5%

10%

15%

20%

25%

SOUTH SOUTH SOUTH

8%

16%

24%

32%

40%

SOUTH

5%

10%

15%

20%

25%

SOUTH

NORTHNORTHNORTH

6%

12%

18%

24%

30%

SOUTHd e f(485) (2160) (1392)

≥ 8.0

≥ 7.0

7.0 –8.0

6.0 –7.0

5.0 –6.0

4.0 –5.0

3.0 –4.0

2.0 –3.0

1.0 –2.0

0.0 –1.0

C alm s : 0.00%

6.0 –7.0

5.0 –6.0

4.0 –5.0

3.0 –4.0

2.0 –3.0

1.0 –2.0

0.0 –1.0

-1.0 –0.0

(degree)INVERSION

(m s-1)WIND SPEED

NORTH NORTH

EAST WEST EAST EASTWEST

EAST EASTWESTEAST WESTWEST

Fig. 5. Nighttime (1800 MST–0600 MST) composite wind roses using the 1-min mean winds at ISS for (a) fully couplednights (4 nights) (b) partially decoupled nights (11 nights) and (c) fully decoupled nights (15 nights), and temperatureinversion computed as a function of wind direction at ISS for (d) fully coupled nights, (e) partially decoupled nights, and(f) fully decoupled nights. The temperature inversion is calculated using Eq. (1) but without time integration. The numberbelow each plot indicates the number of data points used to produce that plot. The smaller data samples for (d–f)compared to (a–c) is a result that the tower data are 5-min means while the ISS data are 1-min means


were able to penetrate down to the surface ofthe plain surrounding the Meteor Crater on theColorado Plateau [�1500 m above the mean sealevel (MSL)], producing disturbed conditions,weakening ambient stability, and allowing thecrater to be coupled to the ambient environment.In contrast, the composite synoptic map forthe fully decoupled nights (Fig. 4c, d) showsweak gradient of geopotential heights over theSouthwest just ahead of an approaching ridgefrom the west, with synoptic winds generally be-low 4 m s�1 over the area of the Meteor Crater innorthern Arizona.

The characteristics of the near-surface windspeed and direction over the surrounding plainfor the three regimes are illustrated by the com-posite wind roses using the 10 m wind data at theISS site for all the nights that belonged to thesame regime (Fig. 5a–c). For all three regimes,the prevailing wind direction over the plain wasfrom southwest, reflecting a combined influenceof synoptic-scale flows and regional and localcirculations. For the fully coupled nights, thesouthwesterly surface winds were a result ofdownward penetration of strong synoptic-scalewinds aloft as shown in Fig. 4b. For fullydecoupled and partially decoupled nights thatoccurred when composite synoptic winds were

weak, a moderately strong (4–6 m s�1) south-westerly low-level flow was usually observedby the weather tower and rawinsonde soundingsat the ISS site and the mini-sodar at the SW siteoutside the crater. This low-level flow typicallypeaked below 50 m AGL and extended to 100–200 m AGL. Detailed data analyses in com-bination with numerical modeling linked thislow-level flow to a regional-scale drainage flowconverging at night from higher terrain in theregion towards the Little Colorado River Valley(Savage et al. 2008). As will be shown later, thisregional-scale thermally driven flow had a largeimpact on the development of temperature inver-sion inside the crater. The small differences inthe ambient wind direction for the three regimessuggest that wind direction is not an importantfactor in determining whether the atmosphere in-side the crater would be decoupled from the en-vironment or not. This is not surprising given thatthe Meteor Crater has a near circular shape withno gaps=paths on its sidewalls and is surroundedby a near uniform topography. But for basinswith varying sidewall heights or major outlets,and=or imbedded in more complex terrain, theambient wind direction may be as importantas wind speed. Also plotted in Fig. 5 is thestrength of temperature inversion as a function

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300

5

10

15

20

25

Date

Win

d sp

eed

(m s

-1)

Surface station

Sodar (50 m)

Wind profiler (1000 m)

Fig. 6. Nighttime meanwind speed at differentheights over the plainsurrounding the crater (graycolumns at the bottomdenote fully coupled nights,dark columns represent fullydecoupled nights; and nocolumn for partiallydecoupled nights)


of wind direction separately by the three regimes(Fig. 5d–f). The strengths of the inversion do notappear to be sensitive to wind direction changewithin the same regime.

The role of local and regional-scale flows inthe formation of temperature inversion inside thecrater is further examined by using wind speedobservations from the surface station, the mini-sodar, and the radar wind profiler outside the cra-ter. Figure 6 shows the three regimes of the crateratmosphere and wind speed at three differentheights: near the surface of the surrounding plain(ISS 10 m tower), right above the crater rim(50 m level form the mini-sodar), and above thenocturnal boundary layer over the plain (1000 m-level from the radar wind profiler). The mini-sodar was located southwest of the crater withan elevation of 1697 m MSL and its 50 m levelwinds (1747 m MSL) would represent winds justabove the crater rim (�1730 m MSL). The 1000 mlevel radar profiler winds, which were obtained byinterpolating from the 924 and 1022 m profilerrange gates, is a good representation of the large-scale winds over the region at night because thelevel is well above the nocturnal boundary layerover the plain which was typically between 200–300 m deep as determined by the rawinsondesoundings launched from ISS site.

The near surface winds over the plain and thesodar winds above the crater rim appear to track

each other very well, but their variations were notalways in phase with the variations of large-scalewinds. As expected, all fully coupled nights oc-curred when the large-scale winds were strongwith the nighttime (1800 MST to 0600 MST thenext morning) averaged 1000 m wind greaterthan 12 m s�1 and the near-surface winds overthe plain and those just above the crater rimgreater than 5 m s�1. But nighttime mean windspeed alone was not always a reliable predictor.For example, the night of 6–7 Oct. was a partial-ly decoupled night, but the mean nighttime1000 m wind was near 20 m s�1 and the meanwind speed above the crater was close to5 m s�1. A careful examination of this night indi-cates that strong winds occurred initially at night,followed by a significant drop at all levels aftermidnight, which allowed an inversion to formand the night to gain partially decoupled statuswith a moderate I value. The 15 fully decouplednights appear to have the mean nighttime surfacewinds above the plain and those just above thecrater between 1–3 m s�1 with the exception oftwo nights when the mean winds were near4 m s�1. The large-scale winds, however, werenot always weak on the fully decoupled nights.For example, the 1000 m mean wind speed wasclose to 10 m s�1 on the fully decoupled nightsof 7–8 and 20–21 Oct., and was even higher(14 m s�1) on 29–30, also a fully decoupled

Calms 0–1 1–2 2–3 3–4 4–5 5–6 6–7 7–8 12–13 13–14 14–15 15–16 ≥160

5

10

15

20

25

30

35

40

Wind class (m s-1)

Per

cent

age

(%)

Fully coupledFully decoupledPartially decoupled

8–9 9–10 10–11 11–12

Fig. 7. Percentage of 1-minmean surface wind at ISS

for fully coupled, fullydecoupled, and partiallydecoupled nights


night. This apparent disconnection between thedevelopment of the nocturnal inversion and thestrengths of the large-scale wind may be partiallyattributed to the development of a stable bound-ary layer over the surrounding plain at night,which may prevent the large-scale wind frompenetrating down to the surface of the plain.Other factors, such as differential temperatureadvection, gravity waves in stable boundary lay-er, and the presence of local pressure gradientthat may oppose large-scale pressure gradient,may also contribute to the disconnection betweenthe near surface and upper-level winds. Theresults here suggest that knowledge of local orregional-scale flow in the nocturnal boundarylayer over the surrounding plain is as importantas upper-level synoptic wind for the forecastingof whether or not a strong temperature inversionwould develop in a basin.

Because of the importance of the near-surfacewind speed over the plain, the frequency distri-bution of the 1-min mean nighttime wind speedsat ISS for each of the three categories is shown in

Fig. 7. The distribution for the fully decoupledand partially coupled nights exhibits a near-nor-mal distribution while the distribution for the ful-ly coupled regime is skewed towards higher windvalues. The frequency for the fully decoupledregimes dropped sharply from 23.9% for windspeed between 4–5 m s�1 to only 6.6% for speedbetween 5–6 m s�1 with nearly 90% of all 1-minmean wind data from decoupled nights being lessthan 5 m s�1. The 10% higher winds for fullydecoupled nights must have happened for onlyshort periods of time, which may have temporal-ly weakened or even destroyed inversion in thebasin, but had little effect on the nighttime inte-grated inversion strength as determined usingEq. (1) that is employed for the classification.It is interesting to note that the frequency forthe three regimes is similar for wind speed be-tween 4 and 5 m s�1, suggesting that the status ofthe basin atmosphere, or the strength of the basininversion, is determined by not only wind speed,but also other factors such as stability discussedbelow.

0.25 0.5

Time (MST)

Hei

ght (

m)

15 18 21 0 3 6 9

1700

1750

1800

1850

1900

1950

2000

Time (MST)

Hei

ght (

m)

15 18 21 0 3 6 9

1700

1750

1800

1850

1900

1950

2000

Time (MST)

Hei

ght (

m)

15 18 21 0 3 6 91600

1650

1700

1750

Time (MST)

Hei

ght (

m)

0 3 6 91600

1650

1700

1750

a b

c dRi

Fig. 8. Bulk Richardsonnumber computed usingrawinsonde soundings atISS outside the craterfor (a) 22–23 Oct. orIOP5 and (b) 30–31 Oct.or IOP7, and usingtethersonde soundingsfrom the crater floor for(c) 22–23 Oct. and(d) 30–31 Oct.


3.3 The role of ambient stability

In addition to wind speed, ambient stability isexpected to have a large impact on the develop-ment of basin inversion. Vergeiner (1996) pre-sented a conceptual model for the erosion of acold air lake due to foehn penetration in a valleybased on the Froude number and the Richardsonnumber. Using the radiosonde sounding datafrom the ISS site and tethersonde profiles insidethe crater during IOPs, we calculated the bulkRichardson number outside and inside the craterand the results are shown in Fig. 8a and c for afully decoupled night (22–23 Oct., or IOP5) andin Fig. 8b and d for a partially decoupled night(30–31 Oct.). Neither the radiosondes nor thetethersondes were available for fully couplednights since the soundings were only operatedduring IOPs under quiescent synoptic conditions.Oct. 30–31, which was the last IOP and hadstronger synoptic winds compared to the otherIOP nights, was selected here to show differencebetween the quiescent synoptic condition (22–23Oct.) and the more disturbed conditions. The dis-tribution of the Richardson number inside thecrater determined using the tethersonde sound-ings from the floor site was substantially differentbetween the two nights. On 22–23 Oct., theRichardson number was greater than the criticalvalue of 0.25 through the entire depth almost allnight long, indicating a very stable atmosphereeverywhere inside the crater. On the night of 30–31, the tethersonde operations, which began at2140 MST, were interrupted from midnight until0500 MST due to strong winds, and was resumedafterwards until 0900 MST. The limited datashow that the Richardson number on this nightwas sub-critical inside the crater except within avery shallow layer (�10 m deep) near the sur-face. Over the plain, the Richardson numberscomputed using the three-hourly rawinsonde pro-files show that the largest difference in stabilitybetween the two nights occurred in the lowest15–30 m above the plain. On 22–23 Oct., theRichardson number was almost always super-critical from sunset to sunrise in the lowest1000 m above the plain. On the night of 30–31Oct., the Richardson number also exceeded 0.25the whole night from about 30 m above groundup to about 1000 m. However, in the 30 m deeplayer immediately above the surface of the plain,

the Richardson number was sub-critical. This re-sult is consistent with the wind speed analysespresented above and further suggests that the at-mospheric conditions immediately above the cra-ter in the nocturnal boundary layer over the plaininteract with the crater atmosphere to determinethe status of the crater atmosphere and thestrength of the nocturnal inversion.

Based on the results above, one may estimatethe critical wind speed over the plain required forthe coupling to occur using the bulk Richardsonnumber:

Rb ¼g

�

�� d

ð�uÞ2; ð2Þ

where d is the depth of the capping inversion atthe top of the crater and �� and �u are thedifference in potential temperature and windspeed across the depth d. For simplicity, assumecalm condition within the crater so that �usimply represents wind speed above the crater.Since, as indicated above, inversion removalwas enabled when bulk Richardson numberbecame sub-critical, the critical wind speedrequired can then be derived from Eq. (2) forRb<Rc, i.e.,

U >

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1

Rc

g

�� d

r¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiR�1c

g

�

��

dd2

r¼ R�1=2

c N d;

ð3Þwhere N ¼

ffiffiffiffiffiffiffig��d

qis the Brunt-V€aais€aal€aa frequency.

Equation (3) shows that the critical wind speedfor initiating inversion breakup depends on thestability and the thickness of the inversion, noton the size of the basin. The values of �� andd can be estimated using either the data fromthe tethersonde soundings during the seven IOP

nights, or when it was not a IOP night, from thetower on the crater rim and the two upper side-wall towers in the crater. The wind speed estimat-ed using Rc ¼ 0:25, i.e,

U ¼ R�1=2c

��Rc ¼ 0:25

Nd ¼ 2Nd ð4Þ

is compared to the actual observed wind speedduring fully coupled and fully decoupled nights(Fig. 9). The one-to-one line on which the ob-served wind speed equals to the estimate windspeed divides the plot into two areas: the areaunderneath the line is where the observed windswere lower than the estimated critical wind speed


and the area above the line means the opposite.All data points from decoupled nights fall in thearea underneath the one-to-one line, while major-ity of the data points from the coupled nights areabove the line. Not all the data points from thecoupled nights are above the one-to-one line.Instead, the thin line, which is represented byUobs¼ 0.66 Uest, separates the two groups of datawell. This implies that for the coupling to occur,the wind speeds do not have to be as strong assuggested by Eq. (4), but rather it only needs tobe above the thin line, or

Uobs>0:66Uest ¼ 0:66R�1=2c

��Rc ¼ 0:25

Nd

¼ R�1=2c

��Rc ¼ 0:57

Nd: ð5Þ

Notice that 0:66R�1=2c for Rc ¼ 0:25 is equiva-

lent to R�1=2c for Rc ¼ 0:57. This suggests that a

nocturnal boundary layer with Richardson num-ber less than 0.6 could enable a coupling of thebasin atmosphere with the ambient environment,whereas, a Richardson number greater than 0.6could allow the basin to be decoupled from itsambient environment and to develop a strongnocturnal inversion in the basin. This result ismore consistent with the suggestion in the litera-ture (e.g., Caron and Richards 1978; Louis 1979;Holstlag and de Bruin 1988; Galperin et al. 2007)that the cutoff value for turbulence in the atmo-

sphere may be larger than the value implied bythe critical Richardson number of 0.2–0.25 thatwas supported largely by laboratory studies.Using data from the Cabauw flux tower in theNetherland, Hostlag and de Bruin (1988) foundthat turbulence was present in the sampling dataup to a Richardson number of 0.7. In a recentarticle, Galperin et al. (2007) studied turbulenceusing a new spectral theory of turbulence thataccounts for strong anisotropy and waves andconcluded that turbulence survives for Ri � 1.

3.4 Turbulent properties duringcoupled=decoupled nights

As shown in Fig. 8, the stability of the crater at-mosphere was substantially different between thecoupled and decoupled nights. Turbulence prop-erties are expected to be also quite different.Figures 10 and 11 show average nighttime turbu-lent kinetic energy (TKE), friction velocity (u�),and sensible heat flux (w0t0) at the 0.5 m level onthe floor tower (FLR) for each night of the month.For the fully decoupled nights, Richardson num-ber exceeded the critical Richardson numbermost of the time (Fig. 8c) and the atmospherewas very stable. Turbulence was intermittent dur-ing these nights and the average TKE valueswere generally less than 0.5 m2 s�2 and friction

Estimated critical wind speed (m s-1)

Obs

erve

d w

ind

spee

d (m

s-1

)16.00

14.00

12.00

Coupled

Decoupled

U obs =

U est

U obs = 0.66 U est10.00

8.00

6.00

4.00

2.00

0.00

16.0014.0012.0010.008.006.004.002.000.00

Fig. 9. Estimated critical wind speedrequired for inversion breakup using Eq. (3)and the actual observed wind speed for eachof the fully coupled or fully decouplednights


velocity less than 0.1 m s�1. For fully couplednights, the nighttime average TKE value wasnear 1 m2 s�2 and u� ranged between 0.17 and0.28 m s�1, nearly doubled their counterpartsduring the decoupled nights. Similarly, the mag-nitude of the kinematic downward sensible heatflux w0t0 was quite low (usually less than0.01 m s�1 �C, or �10 W m�2) during decouplednights and was much higher on coupled nights.The very low sensible heat flux was also ob-served in the Peter Sinks basin when near-calmcondition was established in the basin at night(Clements et al. 2003). Despite the near zerodownward sensible heat flux during the decou-pled nights, the cooling of the basin atmosphere

continued through the night due possibly to long-wave radiation loss.

3.5 Radiative properties duringcoupled=decoupled nights

The relation between surface longwave radiationloss and the status of the crater atmosphere isexamined using the nighttime average of thenet longwave radiation loss computed from the5-min mean data and the inversion strength com-puted using Eq. (1). The results, shown in Fig. 12,indicate large night-to-night variations in thelongwave radiation loss with values rangingfrom 20 to 80 W m�2. Since the same trend and

1 2 3 4 5 6 7 8 9 1011121314151617181920212223242526272829300.04

0.035

0.03

0.025

0.02

0.015

0.01

0.005

0

w′t′

(m

s-1

°C

)

DateFig. 11. Similar to Fig. 10, but for sensibleheat flux

1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930

0

1

2

3

4

5

6

7

8

9

10TKE

0

0.05

0.10

0.15

0.20

0.25

0.30

0.35

u*

Date

TK

E (

m-2

s-2

)

u * (m

s-1

)

Fig. 10. The nighttime mean turbulentkinetic energy (TKE) and friction velocity(u�) at the 0.5 m level of the floor tower.(gray columns at the bottom denote fullycoupled nights, dark columns represent fullydecoupled nights; and no column forpartially decoupled nights)


amount of variations were observed at all sitesinside the crater, on the crater rim, as well asthe site outside the crater over the plain, thesenight-to-night variations in the net longwave ra-diation must have been related to variations incloudiness and humidity of the ambient environ-ment associated with changes in synoptic condi-tions. The amount of radiation loss on each nightdid not appear to be related to whether or not thebasin atmosphere was decoupled from the ambi-ent environment. For example, similar smallamount of longwave radiation loss was observedon 5–6 Oct., a fully decoupled night, and on 23–24 Oct., a fully coupled night. The best coupled(16–17 Oct.) and best decoupled (22–23 Oct.)

nights also experienced similar amount of radia-tion loss.

Although the net longwave radiation loss atnight does not appear to be a good indicator forwhether the basin atmosphere would be decou-pled from the ambient environment, the strengthof the temperature inversion appears to be posi-tively correlated to the loss of longwave ra-diation. Figure 13 shows scatter plots of thenighttime integrated inversion strength as a func-tion of nighttime integrated longwave radiationloss measured at each of the five towers insidethe crater. There is a clear correlation about theinversion strength and the radiation loss mea-sured at the two upper sidewall towers, with the

0 20 40 60 80 0 20 40 60 80 0 20 40 60 0 20 40 60 0 20 40 60

EU EL FL WL WU

-20.0

80.0

180.0

280.0

380.0

480.0

580.0

680.0

Rnet (W m-2)

Inve

rsio

n st

reng

h (°

C m

in)

Fig. 13. The nighttimeaveraged longwave radiationloss and the inversionstrengths as determined byEq. (1). The solid line is anexponential curve fitting tothe data

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

0

10

-10

20

30

40

50

60

70

80

90

Date

Rne

t (W

m-2

)

EUELFLRWLWURIMSW

Fig. 12. Nighttime averagednet longwave radiation at alltowers both inside andoutside the crater. The graycolumns at the bottomindicate the inversionstrength for each nightdetermined from Eq. (1)


inversion increasing exponentially with the in-crease of radiation loss. The relation is not asclear at the two lower tower sites, and it is poorat the floor site. This may be because the long-wave radiations loss does not account for all thecooling occur at these lower sites; cold air drain-age from sidewalls also contribute to the coolingof the lower basin.

4. Conclusions

The observations made during a recent meteoro-logical experiment in the Meteor Crater in north-ern Arizona show that nocturnal atmosphereinside the basin can be fully decoupled, partiallydecoupled, or fully coupled to the ambient envi-ronment. Strong synoptic wind speeds aloft, al-though indicative of strong synoptic forcing,does not necessarily have strong effect on thebasin atmosphere and prevent an inversion fromdevelop in the basin. Instead, the near-surfacewind over the plain surrounding the basin, whichis more susceptible to the influence of local orregional scale processes, is found to be a moreimportant factor in determining whether the ba-sin atmosphere would be coupled to, or decou-pled from, its ambient environment. A necessarycondition for the decoupling and the develop-ment of a strong inversion overnight is that themean nighttime temperature over the plain needsto be less than 5 m s�1.

In addition to wind speed, ambient stability, asindicated by Richardson number, also appears tobe an important factor. The data suggest that anocturnal boundary layer with Richardson num-ber less than 0.6 could enable a coupling of thebasin atmosphere with the ambient environment,whereas, a Richardson number greater than 0.6could allow the basin to be decoupled from itsambient environment with the development of astrong nocturnal inversion in the basin.

Turbulence intensity in the basin, as indicatedby turbulent kinetic energy and friction velocity,is found to be very weak on decoupled nightswith nighttime average TKE below 0.5 m2 s�2,u� smaller than 0.1 m s�1, and the downward sen-sible heat flux less than 10 W m�2. These valueswere more than doubled during coupled nights.

The inversion strength is found to increase ex-ponentially with the increase of longwave radia-tion loss at the two towers on the upper sidealls.

However the relation between the inversion andthe radiation loss on the basin floor is poor be-cause the cooling over the basin floor is not onlyaffected by the local radiative cooling, but alsoby the cold air drainage from the basin sidewalls.

Acknowledgements

We would like to thank Tom Horst, Bill Brown, GordonMacLean, Steve Oncley and a number of other NACRpersonnel for providing equipment, field support, anddata processing during the experiment, David and JohannaWhiteman, Sebastian Hoch, and Maura Hahnedberger, forcollecting and making tethersonde data available. We alsothank David Whiteman and Sebastian Hoch for valuablediscussions. Special thanks to Barringer Crater Corporation(Drew Barringer, President) and Meteor Crater Enterprises,Inc. (Brad Andes, President) for allow us to use the cratersite for this research. This research is supported by the U.SNational Science Foundation Physical and Dynamic Meteo-rology Division (S. Nelson, Program Manager) throughGrants ATM 0646206 and ATM 0444807.

References

Carson DJ, Richards PJ (1978) Modeling surface turbulentflows in stable conditions. Bound Layer Meteorol 14: 67–81

Clements CB, Whiteman CD, Horel JD (2003) Cold-air-poolstructure and evolution in a mountain basin: Peter Sinks.Utah J Appl Meteorol 42: 752–68

Galperin B, Sukoriansky S, Anderson PS (2007) On thecritical Richardson number in stably stratified turbulence.Atmos Sci Let 8: 65–69

Holtslag AAM, de Bruin HAR (1988) Applied modeling ofthe nighttime surface energy balance over land. J ApplMeteorol 27: 689–704

Louis JF (1979) Parametric model of vertical eddy fluxes inthe atmosphere. Bound Layer Meteorol 17: 187–202

Petkov�ssek Z (1978) Relief meteorologyically relevant char-acteristics of basins. Z Meteorol 28: 333–40

Petkov�ssek Z (1985) Die Beendigung von Luftverunreini-gungsperioden in Talbecken. Z Meteorol 35: 370–2

Petkov�ssek Z (1992) Turbulent dissipation of cold air lake in abasin. Meteorol Atmos Phys 47: 237–45

Rakovec J, Mer�sse J, Jernej S, Paradi�zz B (2002) Turbulentdissipation of the cold-air pool in a basin: comparison ofobserved and simulated development. Meteorol AtmosPhys 79: 196–213

Reddy PJ, Barbarick DE, Osterburg RD (1995) Develop-ment of a statistical model for forecasting episodes ofvisibility degradation in the Denver metropolitan area.J Appl Meteorol 34: 616–25

Savage CL, Zhong S, Yao W, Brown WJO, Horst TW,Whiteman CD (2008) An observational and numericalstudy of a regional-scale downslope flow in NorthernArizona. J Geophy Res 113: D14114; DOI: 10.1029=2007JD009623. pp. 17

Smith RB, Paegle J, Clark T, Cotton W, Durran D, Forbes G,Marwitz J, Mass C, McGinley J, Pan HL, Ralph M (1997)


Local and remote effects of mountains on weather: re-search needs and opportunities. Bull Am Meteorol Soc 78:877–92

Steinacker R, Dorninger M, Eisenbach S, Holzer AM,Pospichal B, Whiteman CD, Mursch-Radlgruber E(2002) A sinkhole field experiment in the Eastern Alps.Preprints, 10th Conf. on Mount. Meteor., 17–21 June2002, Park City, UT, pp. 91–92

Vergeiner I (1996) A conceptual dynamic model of foehnpenetration in a valley. Proc. ICAM 96 Meeting (Bled,September 3–9, 1996), pp. 127–34

Vrhovec T, Hrabar A (1996) Numerical simulations ofdissipation of dry temperature inversions in basins. Geo-fizika (Zagreb) 13: 81–96

Whiteman CD, Zhong S, Shaw WJ, Hubbe JM, Bian X,Mittelstadt J (2001) Cold pools in the Columbia Basin.Wea Forecast 16: 432–47

Whiteman CD, Pospichal B, Eisenbach S, Weihs P,Clements CB, Steinacker R, Mursch-Radlgruber E,

Dorninger M (2004) Inversion breakup in smallRocky Mountain and Alpine basins. J Appl Meteorol43: 1069–82

Whiteman CD, Sebastian WH, Maura H, Zhong S (2007)METCRAX 2006 – first results from the meteor craterexperiment. Preprints, 29th Conf. Alpine Meteor, 4–8June 2007, Chambery, France, 1: 93–96

Whiteman CD, Muschinski A, Zhong S, Fritts D, Hoch SW,Hahnenberger M, Yao W, Hohreiter V, Behn M, Cheon Y,Clements CB, Horst TW, Brown WOJ. METCRAX2006 – Meteorological Experiment in Arizona’s MeteorCrater. Bull Am Meteorol Soc 89: 1665–80. DOI:10.1175=2008BAMS2574.1

Z€aangl G (2005) Formation of extreme cold-air pools inelevated sinkholes: an idealized numerical process study.Mon Wea Rev 133: 925–41

Zhong S, Bian X, Whiteman CD (2003) Time scale forcold-air pool breakup by turbulent erosion. Meteorol Z 12:229–33

210 W. Yao, S. Zhong: Nocturnal temperature inversions in a small, enclosed basin

Nocturnal temperature inversions in a small, enclosed ...

Documents