comparison of vertical soundings and sidewall air...

NOVEMBER 2004 1635W H I T E M A N E T A L .

q 2004 American Meteorological Society

Comparison of Vertical Soundings and Sidewall Air Temperature Measurements in aSmall Alpine Basin

C. DAVID WHITEMAN

Pacific Northwest National Laboratory, Richland, Washington

STEFAN EISENBACH, BERNHARD POSPICHAL, AND REINHOLD STEINACKER

Institute for Meteorology and Geophysics, University of Vienna, Vienna, Austria

(Manuscript received 4 November 2003, in final form 21 April 2004)

ABSTRACT

Tethered balloon soundings from two sites on the floor of a 1-km-diameter limestone sinkhole in the easternAlps are compared with pseudovertical temperature ‘‘soundings’’ from three lines of temperature dataloggerson the basin’s northwest, southwest, and southeast sidewalls. Under stable nighttime conditions with low back-ground winds, the pseudovertical profiles from all three lines were good proxies for free air temperature soundingsover the basin center, with a mean nighttime cold temperature bias of about 0.48C and a standard deviation of0.48C. Cold biases were highest in the upper basin where relatively warm air subsides to replace air that spillsout of the basin through the lowest-altitude saddle. On a windy night, standard deviations increased to 18–28C.After sunrise, the varying exposures of the dataloggers to sunlight made the pseudovertical profiles less usefulas proxies for free air soundings. The good correspondence between sidewall and free air temperatures duringhigh-static-stability conditions suggests that sidewall soundings can be used to monitor temperatures, temperaturegradients, and temperature inversion evolution in the sinkhole. Sidewall soundings can produce more frequentprofiles at lower cost than can tethersondes or rawinsondes, and extension of these findings to other enclosedor semienclosed topographies may enhance future basic meteorological research or support applications studiesin agriculture, forestry, air pollution, and land use planning.

1. Introduction

In the past, temperature data have been collectedalong mountainsides 1) to determine whether moun-tainside temperatures were useful proxies for free at-mosphere temperatures in the air surrounding or upwindof mountain areas and 2) to compare local microclimateson mountainsides having different orientations.

The first kind of study provided one of the originaljustifications for establishing mountain observatories inthe Alps, because they provided air temperature mea-surements that were used to determine the then-un-known vertical temperature structure of the earth’s loweratmosphere. Early experience (Barry 1992) showed thatmountainside measurements provided useful proxies fornearby free air soundings if the surface stations weresited well so as to remove microclimatic influences.Mountaintop temperatures were generally cooler thanconcurrently measured temperatures in the nearby freeatmosphere at the same elevation, with the temperaturedifference depending on cloud amount. On clear, un-

Corresponding author address: C. David Whiteman, Pacific North-west National Laboratory, P.O. Box 999, Richland, WA 99352.E-mail: [email protected]

disturbed days, mountaintop temperatures tended to bewarmer during daytime and cooler during nighttime thanthe nearby free atmosphere. For isolated mountains instrong winds, the lifting of the air up the mountainsideproduces relatively cold temperatures at the mountainsummit relative to the surrounding air at the same heightbecause of the dry isentropic or adiabatic temperaturedecrease produced as the air is lifted up the slopes.

The second kind of study has focused on the micro-climates of slopes. Near-surface air temperatures can beinhomogeneous in complex mountainous terrain whenwide spatial variations occur in the energy budgets ofthe underlying surfaces. For example, inhomogeneouspatterns occur during daytime when solar radiation fallswith different intensity on slopes of different aspect andinclination angles. McCutchan and Fox (1986) analyzedweather station data from the peak and from northeast,southeast, and northwest exposures at the base and mid-slopes of an isolated conical mountain in New Mexico.They showed that the average 0000 mountain daylighttime range of temperature differences among the sta-tions at the same altitudes during a July–August periodwas 1.58–28C, with the largest range at the mountainbase. The average daytime range of temperature differ-

1636 VOLUME 43J O U R N A L O F A P P L I E D M E T E O R O L O G Y

ences among the stations over the same period was 18–38C, with the largest differences occurring at both themidslope and mountain-base elevations when back-ground winds were light (#5 m s21).

The study presented here has a somewhat differentfocus from the two kinds of studies mentioned above.Our objective is to determine under what conditions airtemperature measurements from sidewalls can provideuseful proxies to atmospheric soundings in the free airover the center of enclosed basins. This objective is metby comparing pseudovertical temperature profiles ob-tained from three lines of temperature dataloggers onthe sidewalls of a small enclosed basin with concurrenttethered balloon soundings in the free air over the basincenter. Such studies support the broader goal of inves-tigating the evolution of atmospheric temperature struc-ture within various types of mountainous terrain. Dem-onstration of the utility of temperature datalogger linesfor studies of temperature structure evolution in en-closed and semienclosed topographies could advancemeteorological research and lead to improvements inunderstanding of the effects of the changing atmosphericstructure on biological communities and on applicationsin agriculture, forestry, air pollution dispersion, and landuse planning. Surface-based instruments greatly reducecosts as compared with rawinsondes or other in situ orremote sensing equipment and would produce a higherfrequency of ‘‘soundings’’ that would support detailedboundary layer and micrometeorological studies withincomplex topography areas.

2. Method

a. Experimental design

The experiment was designed to determine underwhat conditions pseudovertical temperature profiles ob-tained from surface stations on the sidewalls would pro-vide useful proxies for vertical soundings in the free airover the basin center. Free air soundings were madewith tethersondes while pseudovertical profiles weremade with three lines of surface-based temperature da-taloggers on the sidewalls. Concurrent observationswere made during two experimental periods that beganin the later afternoon and continued until late morningthe next day. The first period had clear skies and weakupper-level winds, and the second period had an eveningepisode of strong upper-level winds and cirrus clouds.During this second period, two tethersondes were flownat different locations on the basin floor to ascertainwhether the free atmosphere of the basin was horizon-tally homogeneous. During both experimental periods,the temperature datalogger lines were used to ascertainif basin sidewall exposure affected the comparisons.Weather stations were used to measure other atmo-spheric variables (in addition to temperature) both insideand above the basin.

b. Topography, instruments, and instrument locations

Experiments were conducted on 2–3 and 3–4 June2002 in the Gruenloch basin, a small enclosed basin onan elevated sloping plateau in the eastern Alps. A con-tour map showing the measurement sites in the basin isprovided in Fig. 1. A summary of the meteorologicalexperiments (Steinacker et al. 2002) is presently beingextended into a detailed journal article, so that a shortsummary will suffice here.

The Gruenloch (also called the Gstettneralm) is a 1-km-diameter limestone sinkhole or doline formed inkarst topography on the Hetzkogel plateau 5 km southof Lunz, Austria. The climate is humid, with the Gruen-loch receiving about 2100 mm of precipitation per year(Sauberer 1947). The Gruenloch has recorded some ofthe coldest minimum temperatures in central Europe(Sauberer and Dirmhirn 1954, 1956). Over the periodfrom 1928 through 1942, minimum thermometers leftin the Gruenloch were read 73 times. Twenty-sevenreadings were below 2408C, with the lowest reading at252.68C (Geiger 1965). Because of these temperatureextremes, distinctive vegetative patterns are found inthe doline. The floor of the basin at 1270 m MSL issubject to the most extreme temperatures and is coveredonly with grasses that can survive the winter under asnow cover; a 5-m-diameter pond is present at the lowestpoint of the basin. Dwarf firs are found on the lower-and midelevation sidewalls, with stately firs and tallwoods at the upper elevations (Geiger 1965). A thinlayer (;20 cm) of grass-covered soil covers limestonebedrock over much of the doline, so that it is possibleonly with effort to find suitable locations on the side-walls to install the wooden posts on which the temper-ature dataloggers were exposed; a deep layer of soil wasfound only at the lowest elevations of the basin floor.No snow was present in the basin during the June ex-periments.

Three lines of temperature dataloggers (HOBO ProTemp/External Temp dataloggers, Onset Computer Cor-poration, Bourne, Massachusetts) were placed on thenorthwest, southeast, and southwest sidewalls of the ba-sin. Whiteman et al. (2000) recently reported on labo-ratory and operational tests of these dataloggers, theiroperating characteristics, and their suitability for me-teorological investigations. The loggers were placed onthe northwest and southeast lines at varying altitudeintervals—5 m at the lowest elevations, changing to 10and 20 m at the higher elevations. On the southwestline, which extended to higher altitudes than the otherlines, HOBOS were separated by 14–44-m intervals.The datalogger thermistors were exposed at 1.3 m AGLin six-plate radiation shields (R. M. Young Company,Traverse City, Michigan) mounted on wooden posts. Atselected sites, a second HOBO was placed on the postat a height of 2.0 m AGL. The HOBOS stored instan-taneous temperature samples every 5 min using a sensorwith a 2-min time constant. The temperature accuracy


FIG. 1. Topographic map of the Gruenloch and surroundings, showing the locations of thetemperature dataloggers (black dots), automatic weather stations (W1–W3, black triangles), andtethersondes (TS1 and TS2, 3). The dataloggers were placed on lines running to the northwest(NW), southeast (SE), and southwest (SW) of the doline center. HOBOs on these lines are abovethe lowest logger on the doline floor (GL00) and are numbered NW01–NW16, SE01–SE16, andSW01–SW08.

of the HOBO is 60.48C over the range from 2108 to508C, with a resolution of better than 0.18C over therange from 08 to 408C. Whiteman et al.’s (2000) cali-brations show that the HOBOS have an increasinglynegative bias with rising temperatures. Because tem-peratures increase with height in the basin, there wouldbe a slight tendency for the HOBO line to underestimatevertical temperature gradients in the basin.

The northwest line, on the steepest sidewall of thebasin, was closest to the main tethered balloon soundingsite, TS1. The first 10 HOBOS on this line ran up thesidewall of the enclosed lower basin to the Lechnersaddle. This saddle is the lowest saddle in the basin andis the only opening in the basin below an altitude of120 m above the basin floor (ABF), the typical heightof temperature inversions in the Gruenloch. The re-maining six HOBOS on this line ran up the ridgelineof the basin from the saddle. The southeast line wasplaced on a longer, lower-angle slope. The highest-al-titude HOBO on this line, SE16 (see Fig. 1), was placedin the Seekopfalm saddle, the saddle separating theGruenloch from a smaller, higher-altitude doline (theSeekopfalm). The southwest line extended to higher al-titudes than the other lines. The first four of the HOBOSon this line were placed on a steep southeast-facingslope. The fifth HOBO was placed in the Ybbstaler sad-dle, with the remaining three HOBOS running up the

ridgeline from the saddle toward the weather station onthe Kleiner Huhnerkogel.

Two TS-3A tethered balloon sounding systems man-ufactured by Atmospheric Instrumentation, Inc. (nowpart of Vaisala, Inc.), of Boulder, Colorado, were usedto make a series of continuous up and down soundingsof temperature, humidity, pressure, wind direction, andwind speed from the floor of the basin to altitudes ofapproximately 200 m AGL at the locations shown inFig. 1. Single up and down soundings were completedover periods of about 20 min. Only the TS1 tetheredballoon system was flown on the first night; the TS1and TS2 tethered balloon systems were both flown onthe second night. Six soundings were made at TS1 onthe first night, and 11 were made on the second night.The accuracy of the tethersonde temperature sensor is60.58C.

Three automatic weather stations (MAWS 201, Vais-ala, Inc., Helsinki, Finland) were operated on the floorof the Gruenloch, in the Lechner saddle northwest ofthe basin center, and on the top of the Kleiner Huhner-kogel (1601 m MSL), a peak above the southwest side-wall. The weather stations recorded average tempera-ture, relative humidity, wind direction, and wind speedat 5-min intervals. The temperature accuracy of theweather stations is 60.38C.


FIG. 2. Meteorograms for 2–3 Jun 2002 from weather stations W1and W3, located at the floor of the Gruenloch (black line) and on thenearby summit of the Kleiner Huhnerkogel (gray line). Shown aretime series of temperature, relative humidity, wind direction, and windspeed.

c. Analysis limitations

There are several limitations of the analysis. Teth-ersonde temperature data come from a fast-responsethermistor that makes instantaneous samples at approx-imately 10-s intervals while performing continuous upand down soundings through the basin atmosphere,whereas the HOBOS take instantaneous samples at fixedlocations on the sidewalls at 5-min intervals. The com-parisons are made between tethersonde temperaturesthat are interpolated to the fixed altitudes of the HOBOSand HOBO temperatures that are closest in time to theinterpolated tethersonde temperatures. Thus, theHOBO–tethersonde temperature comparisons are al-ways within 2.5 min of each other. Only tethersondeascents (i.e., up soundings) during the period of night-time cooling were used for these comparisons.

The HOBO sites on the sidewalls were chosen toavoid nonrepresentative microclimates, within the prac-tical limitations imposed by the necessity to separateadjacent HOBOS by selected vertical altitude intervals.We were generally able to place the HOBOS on openslopes with widely spaced trees rather than in confineddrainage channels or in thick forests. Sites NW11 andNW12 were exceptions. NW11 was located on a ridge-line under a closed forest canopy, and NW12 was lo-cated close to an outcropping of limestone.

d. Synoptic conditions during the two nights

Synoptic weather conditions were somewhat differentfor the two experimental nights. For the 2–3 June ex-perimental period, the Lunz area was under a dry highpressure ridge that extended southward into the Alpsfrom southern Sweden. Scattered stratocumulus and cu-mulus dissipated early in the evening, and the remainderof the night was undisturbed, with clear skies and weaksynoptic winds. The night of 2–3 June was, thus, anideal night for studying an undisturbed inversion for-mation. In contrast, the 3–4 June experimental periodwas somewhat disturbed. Lunz was in an area of weakpressure gradients, but with a weak warm front andmoister air approaching from the west. Cirrus increasedin the early evening to cover from 4/8 to 7/8 of the skybetween 1800 and 2200 central European standard time(CEST). The onset of the cirrus clouds was accompaniedby a strengthening of the synoptic-scale wind, whichkept the inversion from building as quickly as on theprevious night. During this time, however, cooling con-tinued at the floor of the Gruenloch, wisps of fogformed, and the small pond began to steam. The cirrusband moved through, the winds in the Gruenloch weak-ened, and the cooling in the doline strengthened whenthe cirrus dissipated around 2200 CEST, leaving clearskies and light winds that persisted for the remainderof the night. At sunrise, the humidity in the cold poolwas near saturation, patchy wisps of fog were presenthere and there on the sidewalls, and the entire floor of

the doline was covered with rime. For reference, astro-nomical sunset occurred at 1946 CEST and astronomicalsunrise occurred at 0410 CEST.

3. Analysis of data

a. 2–3 June 2002

Figure 2 presents time series of meteorological var-iables observed at weather stations on the floor of theGruenloch and at the summit of the Kleiner Huhner-kogel from midafternoon on 2 June through midmorningon 3 June. Pressure decreased slowly throughout thisexperimental period (not shown). Temperatures near128C on the floor of the Gruenloch fell to 278C bysunrise, with the most rapid fall in the early evening.A much weaker diurnal temperature oscillation was seenon the nearby summit of the Kleiner Huhnerkogel. Day-time relative humidities of 50%–60% increased duringthe night to over 90% at the Gruenloch floor and 80%–90% on the nearby summit of the Kleiner Huhnerkogel.Winds on the floor of the Gruenloch were light andvariable in direction during the night, while northeast-erly winds of 2–4 m s21 prevailed on the Kleiner Huh-nerkogel.


Figure 3a presents data from a series of six tetheredballoon soundings on the night of 2–3 June. Soundingswere unavailable between 2200 and 0300 CEST. Thebasin cooled continuously and the vertical temperaturegradient in the basin strengthened during the night. Apersistent stepwise increase in temperature occurred inall soundings near 54 m ABF, the level of the Lechnersaddle. The winds within the lower, closed basin becamecalm. Humidity in this nearly stagnant pool was main-tained at 80%–90% and the mixing ratio decreasedthroughout the night (not shown) as moisture was re-moved at the surface through deposition and conden-sation. An outflow of basin air occurred just above theheight of the Lechner saddle, where southeast windswere observed with wind speeds below 2 m s21 up toan altitude of 80 m ABF.

A time–height plot of hour-average temperatures on2–3 June 2002 for the northwest line of HOBOS isshown in Fig. 4. The horizontal temperature differencesbetween the HOBOS on the sidewalls and the tetheredballoon over the center of the basin varied throughoutthe night on the three lines and are shown as a functionof height for the northwest and southeast lines in Fig.5. The absolute temperature differences on the northwestline (Fig. 5a) were less than about 28C throughout thenight. These differences are somewhat larger than theuncertainty due to the accuracies of the two instrumentsystems (section 2b). There was a tendency for HOBOSwithin the enclosed lower altitudes of the basin to agreebetter with tethered balloon temperatures, indicating thattemperatures were more horizontally homogeneouswithin the enclosed cold-air pool than for sites abovethe Lechner saddle outflow. The lowest HOBO on thebasin floor was typically about 18C colder than the teth-ersonde during the entire night. HOBOS just above thefloor at 1275–1294 m MSL had a tendency to be slightlywarmer (by several tenths of a degree) than the air overthe basin center, while the HOBOS above the Lechnersaddle were 18–28C colder than the free air. A similaranalysis on the southeast line (Fig. 5b) supports thefinding of relatively cold HOBO temperatures on thebasin floor. In contrast to the northwest line, however,relatively warm HOBO temperatures generally occurredon the southeast line above the saddle except at and justbelow the Seekopfalm saddle, where cold air overtoppedthe Seekopfalm doline and ran down into the Gruenloch.The site on the saddle (SE16) had temperatures that wereas much as 68C colder than at the same height over thebasin center. Following sunrise (0410 CEST), HOBO–tethersonde temperature differences became much morevariable. By 0700 CEST, individuals HOBOS deviatedby 138 to 248C from the free air values, depending onthe exposure of the HOBOS and their surroundings toinsolation (not shown).

The mean nighttime differences between the HOBOtemperatures and the concurrent tethered balloon tem-peratures at the same heights and times, and the con-fidence intervals for the mean for the six evening and

nighttime (1957–0402 CEST) soundings of 2–3 Juneare summarized for the northwest, southwest, and south-east lines of HOBOS in Fig. 6a. On all three HOBOlines, the nighttime-averaged HOBO temperatures werelower than the concurrent free air temperatures over thebasin center. In some cases, however, the temperaturedifferences were less than the likely measurement errors(section 2b). These cold temperature biases were largestat the highest altitudes and smallest within the topo-graphically enclosed lower basin where stability wasstronger. On the northwest line, mean nighttime HOBOtemperatures were within 1.28C of the free air values atall altitudes. The nighttime temperature differences be-low 1380 m averaged 20.408C with a standard deviationof 0.428C. Below the Lechner saddle, the nighttime tem-perature differences averaged 20.428C with a standarddeviation of 0.268C. In a similar way, on the southeastline below 1350 m MSL the mean difference was20.318C with a standard deviation of 0.418C. On thesouthwest line below 1400 m MSL, corresponding val-ues were 20.458 and 0.378C. The tethersonde reachedabove 1400 m on only one of the six nighttime sound-ings, so that the values above 1400 m have little sta-tistical significance. Another characteristic of the tem-perature differences is the relatively cool HOBO tem-perature at the basin floor. This difference is attributedpartly to the differing measurement heights—the HOBOat 1.3 m AGL and the balloon sonde, because of rigginglimitations, at about 2 m AGL. In individual temperatureprofile comparisons, the temperature differences at theuppermost HOBO sites were often larger than temper-ature differences at lower altitudes (this is also seen inthe averages shown in Fig. 6a). These larger temperaturedifferences are produced by the intrusion of upper windsinto the higher altitudes of the doline. In addition to thewarm- or cold-air advection associated with these in-trusions, their deflection up and down the slopes pro-duces temperature changes caused by isentropic expan-sion and compression, respectively.

b. 3–4 June 2002

Figure 7 shows the time series of meteorological var-iables observed at weather stations on the floor of theGruenloch and at the summit of the Kleiner Huhner-kogel from midafternoon on 3 June through midmorningon 4 June. Atmospheric pressure was steady (notshown). Temperatures on the floor of the Gruenloch fellfrom 168C in the afternoon to 258C by sunrise. Tem-peratures at the summit fell from a daytime maximumof about 148C to a sunrise minimum of about 78C. Thetemperature difference between the summit and the ba-sin floor increased throughout the night as the basininversion strengthened. Relative humidity rose from50% during afternoon to nearly 100% during the nighton the basin floor and rose from 60% in the afternoonto 75% in the night on the Kleiner Huhnerkogel. At thesummit, winds veered from east-southeasterly to west-


FIG. 3. Tethersonde TS1 up soundings of temperature, wind direction, and wind speed during the cooling periods of (a) 2–3Jun and (b) 3–4 Jun. The starting times of the up soundings (CEST) are indicated in the legends.


FIG. 4. Time–height plot of hour-averaged temperatures (8C) in theGruenloch basin on 2–3 Jun 2002 as determined from the northwestline of HOBOs.

FIG. 5. HOBO–tethersonde temperature differences on the (a) northwest and (b) southeast lines for 2–3 Jun 2002 and on the (c) northwestand (d) southeast lines for 3–4 Jun 2002. The dashed horizontal line is the height of the basin floor. The thin gray lines indicate the timesand heights for which data comparisons were available for drawing the contours. Note the 6-h period of missing data during the middle ofthe night in (a) and (b).

erly during the night, with wind speeds generally in therange from 1 to 4 m s21, but with a period of strongerwinds reaching 10 m s21 between 1900 and 2100 CEST.Despite the stronger winds on the Kleiner Huhnerkogel,

the basin floor was shielded by the surrounding terrainfrom the stronger winds aloft and winds remained below1 m s21 during most of the experimental period at thebasin floor weather station, with variable wind direc-tions.

Figure 3b shows the sequence of 11 tethersondesoundings on 3–4 June. The generally stronger, but var-iable, wind speeds in the early evening of this night(relative to 2–3 June, Fig. 3a) affected the developmentof the temperature profiles. In the first sounding, windswere below 1 m s21 to a height of 60 m, with strongerwinds aloft. These stronger winds intruded into the low-est altitudes of the sinkhole in the second and thirdsoundings, mixing the atmosphere and keeping the tem-perature profiles near isothermal. Additional cooling oc-curred in the lowest 40 m of the fourth sounding (1934CEST) as winds dropped below 1 m s21 in the lowest80 m. A temperature jump that appeared at the 100-mlevel in this sounding, like the jump at the 80-m levelon the 2024 sounding of the previous evening, appearedat a height corresponding to the base of the stronger


FIG. 6. Mean temperature differences between HOBOs and TS1tethersonde soundings (solid lines) and their 90% confidence intervals(dashed lines) as a function of height for (a) 2–3 Jun and (b) 3–4Jun 2002 for each of the HOBO lines (NW, SW, and SE). The meansand confidence intervals are computed over the nighttime soundingsfrom early evening until sunrise for all heights at which three or moreHOBO–tethersonde comparisons were available. The horizontaldashed line indicates the elevation of the basin floor. The horizontaldotted lines show, from left to right, the heights of the Lechner,Ybbstaler and Seekopfalm saddles, respectively.

FIG. 7. Same as Fig. 2, but for 3–4 Jun 2002.

winds aloft. The fifth, sixth, and seventh soundings(2030–2131 CEST) were made as strong winds againintruded into the lower altitudes of the doline, with theeffect of confining the strongest cooling to a shallowlayer at the doline floor. Winds weakened in the lateevening and remained light for the rest of the night,allowing the inversion to grow more normally, as forthe previous night. The inversion deepened throughoutthe remainder of the night, with cooling being distrib-uted mostly evenly with height. The mixing ratio withinthe doline decreased throughout the night (not shown),as noted for the previous night.

The temperature differences between the sidewallpseudoprofiles obtained from the northwest and south-east lines and the tethered balloon ascents are shown inFigs. 5c and 5d, respectively. The strong horizontal

winds that descended into the basin in the early eveningcaused a warming in the basin center relative to thesidewalls, presumably caused by a circulation in whichair descended over the basin center and reascended atthe slopes. The sinking motions in the stable air overthe basin center produced warming, while the compen-satory ascending air over the slopes produced relativecooling. The relative cooling was somewhat strongerover the northwest line because the prevailing windswere generally from the southeast. Downward turbulentfluxes at the sidewalls, promoted by the shear generationof turbulence, probably also played a role in local cool-ing adjacent to the slope. Both before and after theintroduction of the strong winds, the temperature dif-ferences approximated those observed on the previous,undisturbed night, with horizontal temperature differ-ences of 618C. These differences, as before, should becompared with the measurement errors stated in section2b.

The 0114 CEST sounding sampled an event for whichthe HOBO lines were relatively warm as compared withthe basin center. The amount and vertical distributionof the warming varied between the northwest and south-east lines. The relative warming locally reached as highas 1.58C on the northwest line and 2.68C on the southeastline. We could not determine the length of the event,because the relative warming was seen in only onesounding. The event occurred at a time when a 40-m-


deep stagnant layer was present at the bottom of thesinkhole. Above this layer, southeast winds varied inspeed with height. The higher stability in the basin atthis time and the lower wind speeds suggest that thisevent, in contrast to the earlier event in which substantialvertical motions were present in the basin, involved hor-izontal eddies (i.e., eddies with a vertical axis).

Figure 6b summarizes the mean temperature differ-ences between the three lines of HOBOs and the teth-ersondes for the night of 3–4 June. Also shown are theconfidence intervals for the mean, as averaged over the10 nighttime soundings. Like the previous night, HOBOtemperatures tended to be colder than free air temper-atures at the same altitudes, although a band of relativelywarm temperatures persisted all night on the lower side-walls inside the enclosed lower basin. As compared withthe previous night, temperature differences were un-steady on this windier night. The relatively high stan-dard deviations about the mean on this night, whichproduced wider confidence intervals in comparison with2–3 June were caused by alternating periods during thenight when HOBO–tethersonde temperature differenceschanged sign within the basin. On the northwest line,standard deviations of temperature differences betweenthe HOBOs and tethersonde soundings were about 0.58Cat the lowest three HOBOs and around 18C at the higherHOBOs. On the southeast line, the standard deviationincreased from 0.58C at the basin floor to 1.98C justabove the Lechner saddle, with a decrease back to 18Cabove that. The southeast line was in sunlight duringthe first sounding of the afternoon (1753 CEST), pro-ducing elevated temperatures at some of the HOBOsrelative to the tethersondes. A period of relatively coldHOBO temperatures in the range from 21 to 238C wasfirst seen on the southeast line above 1296 m MSL inthe 1856 CEST sounding. The relatively cold HOBOtemperatures were later noted on the northwest line afterthe 1934 CEST sounding. They persisted through the2131 CEST sounding on both lines. The 0114 CESTsounding, by contrast, found HOBO temperatures atnearly all elevations on both sidewalls to be warmer (byas much as 2.68C) than the tethersonde sounding. Again,like the 2–3 June case, very cold sidewall temperatureswere found below the Seekopfalm on the southeast line,indicating a flow of cold air out of the Seekopfalm basin.Last, on the southwest line, the HOBOs that extendedto higher altitudes experienced relatively cold sidewalltemperatures, especially just below the Ybbstal saddlewhere one of the HOBOs was poorly exposed in a minortributary.

Another interesting feature of the HOBO–tethersondetemperature differences, most apparent on the northwestline, was the warmer (i.e., relative to the tethersondeascents) HOBO temperatures that persisted all night atelevations 5–20 m above the basin floor. A similarHOBO temperature excess on the southeast line wasfirst noted 5 m above the basin floor in the early evening.

The axis of maximum temperature excess grew duringthe night to reach 1305 m MSL by sunrise.

A second tethersonde (TS2; Fig. 1) was operated onthe night of 3–4 June on the same schedule as the firsttethersonde. This tethersonde was operated southeast ofthe basin center at a base altitude of 16 m ABF, withsomewhat smaller depth soundings than at TS1. Ouranalysis (not shown) of the temperature differences be-tween the HOBO lines and the TS2 tethersondes is inbroad agreement with the findings at TS1. On average,the HOBOs were 08–18C colder than the tethersonde onthe northwest and southeast lines, with a standard de-viation of 18–28C. The shorter soundings and larger al-titudinal differences between HOBOs on the southwestline allowed comparisons to be made for only threeHOBOs. The mean standard deviations were, again, 18–28C, but the mean temperature differences were closeto zero on this line, except at the intermediate site, wherethe HOBO was nearly 18C warmer than the tethersonde.

c. Comparison of vertical profiles from three HOBOlines

If lines of surface-based temperature dataloggers ona sidewall are used as proxies for free air temperaturesoundings in a basin, will the proxies differ dependingon which sidewall is chosen for the line? This questionis investigated in Fig. 8, where hour-averaged pseudo-profiles from the three lines of HOBOs are comparedat 3-h intervals for 2–3 June 2002. During daytime(0800, 1100, and 1400 CEST), noticeable temperatureexcursions occur not only between lines, but also be-tween HOBOs on the same line. In the lower basin,temperatures tend to be generally higher on the north-west line in the morning and early afternoon becauseof its southeast- and south-facing aspect, which is fa-vorable for receipt of solar radiation. The temperaturedifferences between HOBOs on a given line, on theother hand, are affected by the nonuniform exposure ofthe HOBOs to shadows that propagate across the terrainfrom the surrounding topography and, especially, fromnearby trees. At night, the temperature profiles from thelines of dataloggers are in much better agreement thanduring the daytime. Nonetheless, there are some sys-tematic differences between the sidewall lines. Thesoutheast line is warmer than the northwest line througha 50-m-deep layer above the Lechner saddle. This is,no doubt, caused by the outflow of cooled air throughthe saddle, a phenomenon discussed further by Pospi-chal et al. (2003). The line of HOBOs on the saddleand the adjacent ridgeline are exposed to this cool flowof air, and the HOBOs on the southeast line may alsoexperience some warming associated with descendingair on the southeast sidewall that supports the flowthrough the saddle.

Poorly exposed HOBOs can be clearly identified fromsequences of pseudovertical temperature profiles asshown in Fig. 8. The poor exposure of the uppermost


FIG. 8. Comparison of pseudovertical soundings from the three HOBO lines on 2–3 Jun 2002. The black line is the southeast line, thedashed line is the northwest line, and the solid squares are on the southwest line.

HOBO on the southeast line (SE16) is readily apparentin the figure. This HOBO’s temperature (plotted at thetop of each of the solid curves) is affected by cold-airoutflow from the Seekopfalm at night, as mentioned insection 3a.

The quantitative comparisons between tethersondeand HOBO profiles, and between different lines of HO-BOs, show that errors in the pseudoprofiles are smallrelative to the inversion strength. Thus, the pseudov-ertical profiles are suitable for detecting the formationof inversions in the doline, as well as for making usefulmeasurements of inversion strengths, depths, and ver-tical temperature gradients. Although it is clear thatsome accuracy is lost relative to vertical profiles overthe basin center, gains are made in terms of the higher

frequency and lower cost of the surface-based temper-ature profiles.

4. Discussion

In this section, we provide an explanation for theobserved temperature differences between sidewalls andthe free air and provide guidance on extending theGruenloch results to other topographic situations.

a. Temperature differences

In undisturbed conditions, nighttime temperature dif-ferences between the slopes and the basin center can becaused by two main processes. The first is the well-


known development of shallow temperature deficitsabove a slope caused by the net loss of longwave ra-diation from the slope and the (partially) compensatingdownward turbulent flux of sensible heat from the over-lying atmosphere, which cools the layer of atmosphereabove the slope. This nighttime temperature deficit overa slope relative to air at the same elevation adjacent tothe slope drives the well-known downslope flows thatdevelop over sloping surfaces (Whiteman 2000). Theseflows are usually strongest in the early evening beforethe basin inversion builds up over the slopes, and it istypical for the downslope flows to weaken or die inbasins and for the air to become stagnant once the in-version builds up within the basin (see, e.g., Clementset al. 2003). The second process, important in drainedvalleys or basins, is the sinking of potentially warmerair into a valley to compensate for the cooled air thatdrains out of the valley in down-valley flows. Enclosedbasins generally attain much lower nighttime tempera-tures than do drained valleys or basins, because thissource of warm air is eliminated in topography that isnot drained by down-valley flows (Whiteman 2000).These two processes maintain the temperature differ-ence between the slope and the basin center in theGruenloch. In the lower basin, the temperature differ-ence is maintained solely by the first process. In thebasin above the Lechner saddle, both processes are atwork, producing larger temperature differences that aremaintained all night, especially on the low-angle, high-elevation slopes.

In a similar way, the development of a warm layerover the slope when it is illuminated after sunrise drivesup-slope flows (Whiteman 2000). The net radiant inputonto the slopes during daytime is much larger than thenet nighttime loss, and thus variations in soil type, al-bedo, vegetation, and other factors that affect the receiptand disposition of net radiation into sensible heat fluxthrough the surface radiation budget become importantin producing thermals and spatial variations in temper-ature over the slopes.

b. Extending the Gruenloch results

The suitability of pseudovertical profiles from sur-face-based dataloggers as proxies for free air profilesover valleys and basins depends on many factors, someof which have been investigated in this paper. Key con-siderations include the following:

• Pseudovertical profiles will usually have a mean tem-perature deficit (i.e., a cold bias) at nighttime relativeto the air at the same level away from the slope be-cause a shallow temperature deficit forms over theslope.

• For useful profiles, dataloggers must be installed inappropriate exposures outside of localized microcli-mates. Several general principles for the exposure ofthese instruments can be given. If possible, instru-

ments should be exposed on open hillsides with ho-mogeneous soil or vegetation where the complicatinginfluences of inhomogeneous forest canopies are ab-sent. Tributary valleys, gullies, or other smaller chan-nels for shallow drainage flows should be avoided;small ridges projecting from an open slope will pro-vide a somewhat more representative exposure to thefree basin atmosphere. Other more subtle exposureproblems may be present, as represented by Gruenlochsite SE16 where cold-air flows into the doline froman adjacent doline. Poorly chosen sites can often bedetected by inspecting the time evolution of pseu-dovertical temperature profiles obtained from largenumbers of dataloggers on an altitudinal cross section.Experimental and instrumental errors of various types(e.g., datalogger elevation errors, temperature offsets,and radiation shield problems) can be determined byinspecting pseudovertical profiles on days on whichthe basin atmosphere is expected to be well mixed(cloudy, windy periods). During such periods, thepseudovertical temperature gradients should be ap-proximately isentropic (Whiteman et al. 2000), withtemperature decreasing smoothly with height at therate of 9.88C km21.

• The Gruenloch analyses show that under nighttimeconditions of strong stability with weak winds aloftand with properly exposed dataloggers, the suitabilityof proxy soundings does not depend sensitively on thelocations of the sidewall data lines. In strong inver-sions, buoyancy tends to maintain horizontal iso-therms, and strong inversions tend to reduce the in-fluence of winds aloft. The pseudovertical profiles are,thus, expected to be most useful in strong inversionconditions. Perfectly horizontal isotherms within thebasin will cause the pseudovertical temperature pro-files from different sidewalls to become coincident.They can also become coincident if the isotherms arehorizontal in the bulk of the basin cross section at atime when equal temperature deficits form over all ofthe slopes.

• Pseudovertical profiles can be expected to be muchmore representative of the free atmosphere duringnighttime than during daytime, because, when the at-mosphere becomes stable, buoyancy or gravitationaleffects tend to produce horizontal isotherms. Pseu-dovertical profiles may also provide useful proxies forfree air soundings during the daytime in winter, es-pecially when a uniform snow cover maintains night-time stability and reduces the temperature contrastsusually associated with radiation receipt on different-aspect slopes and different types of ground cover.

• Meteorological disturbances tend to perturb the iso-therms or isentropes. For example, strong winds abovean inversion may cause waves to form at the inversiontop that may perturb the sidewall temperatures forsome distance below the inversion top. This, however,is usually strongly damped with distance downwardinto the basin (Whiteman et al. 2001). Also, strong


flows carried up the outside slopes of a basin or valleycan produce cold temperatures at the basin or valleyridgeline because the air carried up the slope will cooladiabatically at the rate of 9.88C km21. This contin-uous source of cold air at the ridgeline may be colderthan the air in the basin or valley below and could betransported in gravity flows down the inner sidewall,as Bossert and Cotton (1994a,b) have previously doc-umented and simulated for Colorado topography.

• Because differences between pseudovertical and freeair temperature profiles are enhanced by the intrusionof strong winds from aloft, the openness of a basinor valley to wind intrusions will have an importanteffect. Pseudovertical profiles can, therefore, be ex-pected to be most suitable as proxies for free airsoundings when a basin or valley is enclosed by sur-rounding sheltering topography. Larger-diameter ba-sins are likely to provide less shelter from strong back-ground winds, and the upper sidewalls are expectedto be more exposed to strong wind influences than arethe lower sidewalls.

5. Conclusions

Tethered balloon soundings in a limestone sinkholein the eastern Alps on a clear, undisturbed night werecompared with pseudovertical temperature soundingsfrom three lines of temperature dataloggers exposed at1.3-m height on the basin’s northwest, southwest, andsoutheast sidewalls. Under stable nighttime conditions,the pseudovertical profiles from all three lines weregood proxies for free air temperature soundings overthe basin center. The HOBO–tethersonde temperaturedifferences were frequently within the measurement er-ror expected from the reported accuracies of the HOBOand tethersonde temperature sensors. The pseudoverticalprofiles had a mean cold temperature bias of about 0.48Cas compared with free air soundings within the basin,with a standard deviation of about 0.48C. Because thesidewall temperature errors are small relative to inver-sion strengths within the sinkhole, the sidewall sound-ings are expected to be useful in estimating free airtemperatures and temperature gradients within the sink-hole. The best correspondence between the pseudo-vertical and free air profiles was in the confined cold-air pool in the lowest 54 m of the sinkhole where at-mospheric stability was highest. Above the confinedcold-air pool, there was a tendency for the sidewalltemperatures to be a little cooler, presumably becausewarmer air sinks into the center of the basin to replaceair that is cooled on the slopes and flows out of thebasin at altitudes above the lowest pass. The mean tem-perature bias and the nighttime standard deviation ofthe temperature bias varied little from line to line, sug-gesting that the orientation and inclination angles at thesidewall lines are relatively unimportant in these high-stability, low–wind speed conditions.

On a second night with higher synoptic winds, a sim-ilar cold temperature bias was again observed on thesidewalls, but the nighttime standard deviation of thebiases was much higher (18–28C) because of the time-varying exposure of the sidewalls to wind intrusionsfrom aloft.

In the experiments described, the locations of the da-taloggers on the sidewalls were carefully chosen toavoid nonrepresentative or localized microclimates.Nonetheless, one site (SE16) had an extreme local mi-croclimate that was over 68C colder than the free airover the basin center at the same altitude, caused byinflow into the basin from an adjacent higher-altitudebasin. After sunrise, the differing microclimates of theindividual surface sites, especially their exposure to sun-light or shade, make the pseudovertical profiles less use-ful as proxies for free air soundings.

The good correspondence between the sidewall andfree air temperatures in the basin suggests that the side-wall soundings will prove to be useful in monitoringtemperatures and vertical temperature gradients over thebasin during clear, nighttime, stable conditions withweak winds aloft. Because sidewall soundings can bemade more frequently and extended over many nightsat lower cost than free air soundings with tethersondesand rawinsondes, they provide valuable advantages forsome types of meteorological analyses. For example,HOBO profiles in the Gruenloch basin (Eisenbach et al.2003) have been used to classify different types of coldpool disturbances, including foehn intrusions, mixingevents caused by wind shear, turbulent erosion episodes,frontal passages, and cloud formation/dissipation. Anextended analysis of these classifications for an 8-monthperiod using hourly averaged pseudovertical profilesfrom the Gruenloch is now being prepared for publi-cation.

Extensions of the results in this paper to larger basins,to open basins that are better exposed to synoptic windinfluences, and to stable wintertime soundings in snow-covered terrain are left to future investigations. The al-ternative approach of using ground-based remote sound-ing systems, such as radio acoustic sounding systemsor microwave radiometers, to make proxy temperaturesoundings of valley and basin atmospheres has, so far,not been investigated.

Acknowledgments. We thank the other organizers andparticipants of the 2001–02 Gruenloch experiments, in-cluding Drs. M. Dorninger and M. Hantel at the Uni-versity of Vienna, Drs. E. Mursch-Radlgruber and P.Weihs at the Agricultural University of Vienna, andMag. K. Baumann and Mr. F. Traher at Austria’s CentralInstitute for Meteorology and Geodynamics in Vienna.Mister A. Holzer provided synoptic weather informationfor the two experimental days. Students from the Uni-versity of Vienna are, especially, thanked for their manycontributions to the field program, which included setupand take-down of the instruments, operation of the teth-


ersondes, and downloading of the HOBO data. Person-nel at the Biological Station in Lunz, Austria, providedadvice and encouragement. Mister P. Kupelwieser isthanked for providing access to the experimental area.

One of the authors (CDW) acknowledges researchsupport from the U.S. Department of Energy (DOE)under the auspices of the Atmospheric Sciences Pro-gram of the Office of Biological and EnvironmentalResearch. His contributions were made at Pacific North-west National Laboratory (PNNL), which is operatedfor the DOE by Battelle Memorial Institute. Two of theauthors (SE and BP) thank DOE, PNNL, the Universityof Vienna, and the government of Lower Austria forfellowships served at PNNL in 2002 and 2003.

REFERENCES

Barry, R. G. 1992: Mountain Weather and Climate. 2d ed. Routledge,402 pp.

Bossert, J. E., and W. R. Cotton, 1994a: Regional-scale flows inmountainous terrain. Part I: A numerical and observational com-parison. Mon. Wea. Rev., 122, 1449–1471.

——, and ——, 1994b: Regional-scale flows in mountainous terrain.Part II: Simplified numerical experiments. Mon. Wea. Rev., 122,1472–1489.

Clements, C. B., C. D. Whiteman, and J. D. Horel, 2003: Cold airpool structure and evolution in a mountain basin: Peter Sinks,Utah. J. Appl. Meteor., 42, 752–768.

Eisenbach, S., B. Pospichal, C. D. Whiteman, R. Steinacker, and M.Dorninger, 2003: Classification of cold air pool events in theGstettneralm, a sinkhole in the eastern Alps. Vol. A, ExtendedAbstracts, Int. Conf. on Alpine Meteorology and the MAP-Meet-

ing 2003, Brig, Switzerland, MeteoSwiss, MeteoSwiss Publi-cation 66, 157–160.

Geiger, R., 1965: The Climate near the Ground. Harvard UniversityPress, 482 pp.

McCutchan, M. H., and D. G. Fox, 1986: Effects of elevation andaspect on wind, temperature and humidity. J. Climate Appl. Me-teor., 25, 1996–2013.

Pospichal, B., S. Eisenbach, C. D. Whiteman, R. Steinacker, and M.Dorninger, 2003: Observations of the cold air outflow from abasin cold pool through a low pass. Vol. A, Extended Abstracts,Int. Conf. on Alpine Meteorology and the MAP-Meeting 2003,Brig, Switzerland, MeteoSwiss, MeteoSwiss Publication 66,153–156.

Sauberer, F., 1947: Kleinklimatische Niederschlagsuntersuchungen imLunzer Gebiet (Microclimatic precipitation investigations in theLunz area). Die Umwelt, 11, 410–415.

——, and I. Dirmhirn, 1954: Uber die Entstehung der extremen Tem-peraturminima in der Doline Gstettner-Alm (On the occurrenceof extreme temperature minimums in the Gstettner-Alm Doline).Arch. Meteor. Geophys. Bioclimatol., Ser. B, 5, 307–326.

——, and ——, 1956: Weitere Untersuchungen uber die kaltluftan-sammlungen in der Doline Gstettner-Alm bei Lunz im Niede-rosterreich (Further investigations of the cold air buildup in theGstettner-Alm doline near Lunz in lower Austria). Wetter Leben,8, 187–196.

Steinacker, R., M. Dorninger, S. Eisenbach, A. M. Holzer, B. Pos-pichal, C. D. Whiteman, and E. Mursch-Radlgruber, 2002: Asinkhole field experiment in the eastern Alps. Preprints, 10thConf. on Mountain Meteorology, Park City, UT, Amer. Meteor.Soc., 91–92.

Whiteman, C. D., 2000: Mountain Meteorology: Fundamentals andApplications. Oxford University Press, 355 pp.

——, J. M. Hubbe, and W. J. Shaw, 2000: Evaluation of an inex-pensive temperature data logger for meteorological applications.J. Atmos. Oceanic Technol., 17, 77–81.

——, S. Zhong, W. J. Shaw, J. M. Hubbe, X. Bian, and J. Mittelstadt,2001: Cold pools in the Columbia basin. Wea. Forecasting, 16,432–447.

comparison of vertical soundings and sidewall air...

Documents