UBC Faculty Research and Publications

Turbulence Structure and Exchange Processes in an Alpine Valley: The Riviera Project. Rotach, Mathias W.; Calanca, Pierluigi; Graziani, Giovanni; Gurtz, Joachim; Steyn, Douw G.; Vogt, Roland; Andretta, Marco; Christen, Andreas; Cieslik, Stanislaw; Connolly, Richard; De Wekker, Stephan F. J.; Galmarini, Stefano; Kadygrov, Evgeny N.; Kadygrov, Vladislav; Miller, Evgeny; Neininger, Bruno; Rucker, Magdalena; Van Gorsel, Eva; Weber, Heidi; Weiss, Alexandra; Zappa, Massimiliano 2004-09-30

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


Steyn_2004_BAMS-85-9-1367.pdf [ 887.46kB ]
JSON: 1.0041881.json
JSON-LD: 1.0041881+ld.json
RDF/XML (Pretty): 1.0041881.xml
RDF/JSON: 1.0041881+rdf.json
Turtle: 1.0041881+rdf-turtle.txt
N-Triples: 1.0041881+rdf-ntriples.txt
Original Record: 1.0041881 +original-record.json
Full Text

Full Text

1367SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |Local circulation systems and average temperatureand stratification conditions in valleys have beeninvestigated in some detail in the past, and arefairly well understood at present (e.g., Whiteman1990, 2000). In contrast, little has been done with re-spect to the associated turbulence structure and tur-bulent exchange processes in mountainous terrain(Rotach et al. 2000). Even from relatively recent largefield programs, such as the Atmospheric Studies overComplex Terrain (ASCOT) program (Clements et al.1989), only a few “representative” turbulence obser-vations are available.Often, the emphasis of atmospheric studies inmountainous terrain is focused on pollutant transport[e.g., Anfossi et al. (1998) for the Trans-Alpine At-mospheric Transport experiment (TRANSALP);Cionco et al. (1999) for the Meteorology and Diffu-sion over Non-Uniform Areas (MADONA) study] orAFFILIATIONS: ROTACH, CALANCA, GURTZ, ANDRETTA, WEBER,WEISS, AND ZAPPA—Institute for Atmospheric and Climate Science,Swiss Federal Institute of Technology, Zurich, Switzerland; VOGT,CHRISTEN, AND VAN GORSEL—MCR Laboratory, University of Basel,Basel, Switzerland; STEYN, DE WEKKER, AND RUCKER—AtmosphericScience Program, The University of British Columbia, Vancouver,British Columbia, Canada; GRAZIANI, CIESLIK, CONNOLLY, ANDGALMARINI—Joint Research Center, Ispra, Italy; E. N. KADYGROV,V. KADYGROV, AND MILLER—Central Aerological Observatory,Moscow, Russia; NEININGER—MetAir, Illnau, Switzerland*Current affiliation: Swiss Federal Office for Meteorology andTURBULENCE STRUCTUREAND EXCHANGE PROCESSESIN AN ALPINE VALLEYThe Riviera ProjectBY MATHIAS W. ROTACH,* PIERLUIGI CALANCA, GIOVANNI GRAZIANI, JOACHIM GURTZ, D. G. STEYN,ROLAND VOGT, MARCO ANDRETTA, ANDREAS CHRISTEN, STANISLAW CIESLIK, RICHARD CONNOLLY,STEPHAN F. J. DE WEKKER,+ STEFANO GALMARINI, EVGENY N. KADYGROV, VLADISLAV KADYGROV,EVGENY MILLER, BRUNO NEININGER, MAGDALENA RUCKER, EVA VAN GORSEL,HEIDI WEBER, ALEXANDRA WEISS,# AND MASSIMILIANO ZAPPAThe new detailed data set resulting from this field campaign, set in highly complex terrain,is helping improve turbulence schemes in meteorological and hydrological numerical models.Climatology, MeteoSwiss, Zurich, Switzerland+Current affiliation: Pacific Northwest National Laboratory, Richland,Washington#Present affiliation: MPI, Hamburg, GermanyCORRESPONDING AUTHOR: Mathias W. Rotach, Swiss FederalOffice for Meteorology and Climatology, MeteoSwiss, Kraebuehlstr.44, P.O. Box 514, CH-8044 ZurichE-mail: mathias.rotach@meteoswiss.chDOI:10.1175/BAMS-85-9-1367In final form 12 January 2004©2004 American Meteorological Society1368 SEPTEMBER 2004|atmospheric chemistry. For example, in the VerticalOzone Transport in the Alps (VOTALP) campaignsurface turbulence information was available at onlytwo selected sites within the Mesolcina valley (Furgeret al. 2000). Because pollutant distribution over anytype of terrain is largely determined by turbulent ex-change processes and their interaction with the meanflow, it is clear that there is a need to understand tur-bulence in complex terrain. What we do know, con-cerning turbulence structure in an alpine valley, is thatcommonly used approaches in boundary layer meteo-rology (such as the validity of scaling regimes;Holtslag and Nieuwstadt 1986) cannot be expected tohold, due to the violation of many underlying assump-tions. For flows over gentle hills linearized models andscaling considerations based thereupon are available,which identify different scaling regimes (e.g., Jacksonand Hunt 1975; Belcher and Hunt 1998), but thesecannot be expected to apply over steep mountains.Most theoretical work concerning valleys is devotedto the thermodynamics of local circulation due to dif-ferential heating (e.g., Vergeiner 1987). In these stud-ies the contribution of turbulent exchange is usuallytreated in a very simplistic manner.Numerical models of all scales—numericalweather prediction, climate, mesoscale, and largeeddy simulation models—use parameterizations,which are based on scaling approaches for flat andhorizontally homogeneous terrain in their surface ex-change and boundary layer schemes (Randall 2001;Beljaars and Viterbo 1998; Emeis and Rotach 1997).Large-scale models having a horizontal resolution of100 km or more will not resolve many of the oro-graphic features of complex topography. In theirhighly smoothed topography the first model level attypically 30 m above “ground” would need a param-eterization to account for bulk turbulent exchangefrom a series of valleys and ridges. Rather, they haveat present a surface layer description based on theMonin–Obukhov similarity theory (e.g., Louis 1979).This leads to sometimes unrealistically high values ofthe effective roughness length (e.g., Georgelin et al.1994). Similarly, nonhydrostatic models with highhorizontal resolution use the same surface exchangeschemes over the valley slope, valley floor, and nearbyridges irrespective of their having been developed forflat, horizontally homogeneous terrain.Hydrological models for the prediction of runoffin entire catchments typically use even more simpli-fied surface layer schemes to obtain surface fluxes ofsensible and latent heat from energy balance consid-erations (e.g., Gurtz et al. 1999). For such simulations,where mountainous terrain is not the exception butthe rule, it is clear that these approaches may some-times be problematic (e.g., Plüss and Mazzoni 1994),but, little data are presently available to improve theseschemes.The use of these surface exchange and boundarylayer schemes in numerical models can easily be criti-cized based on first principles, but they are employedonly due to the lack of better knowledge (Rotach 1995).As a step toward filling this gap, the Riviera projectwas planned and executed in conjunction with theMesoscale Alpine Programme (MAP), an internationalresearch effort to improve our knowledge of meteoro-logical and hydrological processes over the Alps (Binderand Schär 1996; Bougeault et al. 2001). It was decidedto investigate one valley in detail, rather than a num-ber of topographical features with different sizes, ori-entation, and other characteristics (Emeis and Rotach1997). Therefore, a wide range of instrumentation wasdeployed in a single valley in the southern Alps.SITE SELECTION. To link this research with hy-drological components of the project, a test valley wassought in the Lago Maggiore target area of MAP(Bougeault et al. 2001). The following valley selectioncriteria were applied:a) valley structure: as unobstructed, straight, andsymmetric as possible with equal height and slopeson both sides, and few side valleys;b) valley size: a small valley, with respect to optimalspatial resolution, given the limited instrumenta-tion available [However, it was felt that a valleywith a width comparable to the resolution ofmodern operational models (e.g., the highest reso-lution Swiss operational model: 7 km) was pref-erable]; andc) infrastructure: availability of suitable, easily acces-sible experimental sites on the slope(s).Based on these criteria the Riviera valley in south-ern Switzerland was chosen. It constitutes a part ofthe Ticino valley (Fig. 1) between Bellinzona andBiasca that discharges from the Saint Gotthard mas-sif in the central Alps to Lago Maggiore. It is approxi-mately straight and U-shaped, with a floor rangingfrom about 250 m above sea level (ASL) to ridge topsat about 2500 m ASL on both sides. The valley flooris 1.5 km wide, and the slopes are roughly 30∞ and 35∞on the eastern and western sides, respectively. TheRiviera valley contains some minor tributary valleyson both sides.The valley floor consists of agricultural land and anumber of villages and isolated farmhouses. A high-1369SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |way, a railroad, and the Ticino River run along thevalley axis. The slopes are mainly forest covered.Above roughly 1200 m ASL, meadows are inter-spersed with rocky rubble. One special feature of theRiviera valley is the fact that its outflow leads to an-other valley rather than the “classical” plain. Thisclearly influences the flow (Weigel and Rotach 2003)and will have to be analyzed in detail in the future.OBSERVATIONAL STRATEGY. Some of the“continuous observations,” as described in the follow-ing similarly titled section, started as early as 10 July1999 and were operating until 13 October 1999. Themajority of instrumentation at the continuous siteswas active from the beginning of August to the be-ginning of October 1999. Due to power outages atsome of the remote sites, not all instruments werecontinuously recording in this period. Two intensiveobservation periods (IOPs) were scheduled to coin-cide with the availability of the research aircraft(Research flights section). These periods lasted from15 to 31 August and from20 September to 8 October1999 and are referred to asR-IOP1 and R-IOP2, re-spectively. A total of 14flights of 2–4 h durationeach were performed oneight flight days, with all ofthe “additional instrumen-tation” (Table 2, see sectionson Scintillometry throughSodar for details) then be-ing available. Figure 2 de-picts the topography of theRiviera valley and indicatesthe main cross section ofobservations near Claro(close to sites A1 and A2),a small village at the foot ofthe eastern slope. Shownare the continuous sites andthe distribution of addi-tional instrumentation dur-ing the IOPs. Full detailconcerning the observa-tions (instrument levels,sampling frequencies, in-strument types, and cali-brations, etc.) can be foundon the project’s homepage (online at www.iac.ethz.ch/en/research/map_riviera/index.html), under “metadatareport.”Continuous observations. Eleven towers were estab-lished along a southeast–northwest cross section inthe Riviera valley through the village of Claro (Fig. 2)FIG. 1. Topography of southern Switzerland with theRiviera valley between the towns of Biasca andBellinzona. Cross section of surface stations (see text)runs through the valley approximately at the height ofthe little village of Claro.FIG. 2. Topography of the Riviera valley overlaid with land use types and the obser-vational sites with indications of the instrumentation. The color codes for theland use are green: deciduous forest, blue-green: coniferous forest, yellow:alpine vegetation, gray: rock, white: snow, light green: meadows, light gray:agricultural land, red: settlement, light red: industrial area, blue: water, black:roads. The village of Claro can be recognized below the symbol for site A2.1370 SEPTEMBER 2004|in order to obtain months-long observations of near-surface turbulence and meteorological/hydrologicalcharacteristics. This setup was chosen because thecross-valley variability of flow characteristics was an-ticipated to be more prominent than that along thevalley. Towers were located on the valley floor andthe eastern slope (Fig. 2). At the principal sites, hightowers with turbulence probes at several levels wereerected (Table 1) and typically accommodated obser-vations of the full radiation balance, precipitation,and standard meteorological parameters (pressure,profiles of temperature, humidity, and wind speed).Smaller towers with only one sonic anemometer levelwere operated mainly in regions of problematiclogistics. Altogether, 20 sonic anemometers and sixfast-response hygrometers continuously recordedturbulence statistics in the lowest 30 m of this valley’scross section for about 2 months. A field intercom-parison of different sonic anemometers will brieflybe introduced in the section titled Instrumentintercomparison.Detailed hydrological observations were per-formed at two of the principal sites (A1 and B, seeTable 1). These observations included profiles of soilmoisture and temperature (down to a 1.6-m depth),as well as soil heat flux, leaf wetness, and canopy pre-A1 28 3 levels 6 levels Full Yes Detailed MTP-5, Valley(2) balance RS, floor,sodar mixedagricultureA2 5 — 4 levels — — Scint., As A1T¢ TBB 30 3 levels 1 Net Yes Detailed — Slope,(1) reference forestC 6 1 level — Net Yes — — Slope,vineyardD 6 1 level ————MTP-5, Foot of(1) scint. slopeE1 12 2 levels — Full Yes Reduced Sodar Slope,(1) balance meadowE2 23 6 levels 8 levels Full — — 18 levels of T¢ Slope,(1) balance forestF1 6 m 1 level 3 levels Net — Some — Slope,sparsevegetationF2 11 2 levels 3 levels Full Yes Some — Slope,balance shrubF3 2 — 1 level Net — — — Slope,grassG 5 1 level 2 levels — — — 1 level of T¢ Slope,forestTABLE 1. Sites and observations in the Riviera valley: “No. hygr.” stands for the number of fast-responsehygrometers, Standard meteo refers to standard meteorological observations. Precip denotes precipita-tion, Hydro denotes hydrological observations (see text for details), and Additional obs. refers toadditional observations at the respective site as described in sections Scintillometry to Sodar MTP-5:microwave temperature profiler; RS: radio sounding; TB: tethered balloon, Scint.: scintillometricmeasurements; T¢: fast response temperature measurements.Site Height Turbulence Standard Radiation Precip. Hydro. Additional Surfaceof tower (m) (No. hygr.) meteo  obs. character1371SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |cipitation. At A1, profiles into the ground were ob-tained at various subsites with different surface cover.Together with simultaneously recorded meteorologi-cal parameters, these observations yielded an excel-lent opportunity to evaluate hydrological and coupledmeteorological/hydrological models (Zappa et al.2000; Carlaw et al. 2000). Additionally, soil moistureobservations were obtained episodically at site E1 andin three other valleys of the region, in order to bettercover the variety of surface types.The following sections describe the additional ob-servations that were performed during the IOPs.Scintillometry. Due to the expected spatial variabilityof turbulent fluxes, it was desirable to obtain, in ad-dition to the detailed array of point measurementsfrom sonic anemometers, spatially averaged informa-tion on turbulence characteristics. Two small-aper-ture, displaced-beam scintillometers (Scintec, SLS 20)were, therefore, operated at the valley floor (site A2)and at site D on a gentle slope. These scintillometersyield path-weighted sensible heat flux, momentumflux, and the dissipation rate of turbulent kinetic en-ergy over a distance between 50 and 200 m.There is some uncertainty surrounding the mea-surement principle of scintillometers in complex ter-rain (see Weiss et al. 2001). Large efforts were, there-fore, undertaken to investigate the performance of theinstruments prior to taking actual observations (Weisset al. 1999 and our Instrument intercomparisonsection).Figure 3 shows a typical example of the daily cycleof the sensible heat flux as measured simultaneouslyby a scintillometer at the valley floor (site A2), and ascintillometer and a sonic anemometer at the foot ofthe slope (site D). Generally, the difference betweenpath-averaged and point observations at site D iswithin the respective instrument uncertainties (sec-tion on Instrument intercomparison). Daytime val-ues of sensible heat flux on the slope (site D) are con-sistently larger than those observed on the valley floor(A2). Also, the valley floor site shows an earlier tran-sition to positive heat fluxes in the morning and tonegative fluxes in the afternoon. These differences re-flect the variability in net radiation (Matzinger et al.2003) due to differences in exposure and slope angle.At other sites with steeper slopes and more favorableexposure, these differences become even more pro-21 Aug 1999 1 ± Convective 0721–1031, T-B-A-B-T, Full, no. H2O;1434–1700 T-B-A-B-T full, no. H2O22 Aug 1999 2 ± Convective 0736–1010, T-A-B-T, No wind andH2O1136–1554 T-B-A-B-A-B-T full, no H2O25 Aug 1999 3 Convective 0649–0942, T-A-B-A-B-T, Full,1112–1541 T-B-A-B-A-B-T full21 Sep 1999 4 Mixed 0716–0905, T-B-A-B, Full,1117–1510 T-B-A-B-A-B-A-T no wind22 Sep 1999 5 Mechanical 0658–0925, T-A-B-A-T-B, Full,1115–1302 T-B-A-B-A-B low resolution wind28 Sep 6 Mechanical, 1202–1614 T-B-A-B-A-B-T Full, H2O lowtransition from qualityrainy period29 Sep 1999 7 Convective 0722–0945, T-B-A-B-T, Full,morning, 1156–1553 T-B-A-B-A-B-T fullmixed afternoon1 Oct 1999 8 Convective 1033–1409 T-B-A-B-A-B-T FullTABLE 2. Summary information on flight days. Flight patterns are described in section titled Researchflights.Date Flight day Type of day Flight times Flight Data coverage,no. (UTC) patterns failures1372 SEPTEMBER 2004|nounced (De Wekker et al. 2004, see also our sectionon Surface energy balance on a slope).Radio soundings. Radiosondes were launched from thevalley floor (site A1) at 3-hourly intervals during flightdays, starting at 0600 and ending at 2400 UTC. In ad-dition, on the day before a flight day, two soundingsat 1200 and 1800 UTC were launched to obtain somedetail of the developing situation. For all the R-IOP2soundings, release times were 1 h earlier (so that thesonde would reach the tropopause around 0000, 0600,1200, and 1800 UTC). A Vaisala MW11 receiving sys-tem was employed using RS80 sondes, which areequipped with standard pan-tilt unit (PTU) sensorsand an 8-channel digital GPS receiver. The accuracyof the raw data is 0.5 hPa (pressure), 0.2 K (tempera-ture), 3% (relative humidity), and 0.5 m s-1 (windspeed). The sounding system has a vertical range ofover 20 km. However, inabout 50% of the soundingsthe radio signal was lost be-fore the sonde reached thetropopause.Passive microwave tempera-ture profiler. A meteorologi-cal temperature profiler(Kipp & Zonen, MTP-5)was operated at the valleyf loor (site A1) during R-IOP1,and closer to the easternslope of the Riviera valley(site D) during R-IOP2. TheMTP-5 is a passive micro-wave sensor, which allowsdetermination of air tem-perature profiles from theground up to 600 m, at avertical resolution of 50 mand a temporal resolutionof 5 min. The instrumentsenses a cone of about500 m in length, yielding a“profile” of the air tempera-ture with some spatial av-eraging (Kadygrov and Pick1998). This instrumentsensed the daily evolution ofstatic stability in the lowerpart of the valley atmosphereat high temporal resolution.Because the MTP-5 is a rela-tively recently developedFIG. 4. Temperature information derived from MTP-5 data for the week 17–24 Aug 1999, at site A1, MAP-Riviera. (top) Time–height cross section: colorcoding from dark blue (12∞–13.5∞C) to orange (25.5∞–27∞C) in intervals of 1.5∞C.(bottom) Temperature inversion parameters. The temperature time seriesare shown for various heights (dark red = 0 m to dark blue = 600 m) in 50-mintervals. Also, the inversion height (blue line, right scale) and its maximumstrength (purple columns, left scale) are shown.FIG. 3. Time series of turbulent sensible heat flux (H)as measured by scintillometry on the valley floor (siteA2, dotted line) and on lightly sloping terrain (site D,solid line) on 1 Oct 1999. Dashed line: Sonic anemom-eter observation at site D.1373SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |instrument, its collocation with theradio sounding system at site A1 dur-ing R-IOP1 was exploited for a sys-tematic intercomparison.As an example of data from theMTP-5, a time–height cross sectionof temperature in the lowest 600 mat site A1 in the Riviera valley ispresented (Fig. 4). The daily cycle isvery pronounced with the expectedwell-mixed layer within the ob-served lowest 600 m (see also Fig. 9)and a stable near-surface layer dur-ing the night. The maximum tempera-ture difference across the inversion(9.8 K) was observed at 0555 LST(Local Standard Time, correspondsto UTC + 2) 6 October 1999 at aheight of 350 m. Continuous mea-surements at 5-min temporal resolution yield infor-mation on the very quick transition between near-surface nocturnal and daytime temperature regimes.Within about 20 min of sunrise or sunset (Weber2000, unpublished Msc Thesis) the stable (unstable)layer close to the surface is eroded and the lowest50 m of the valley atmosphere changes stability.Figure 5 shows an example of the sunrise transitionon 24 September. The newly developed morning un-stable layer near the surface is topped by an approxi-mately neutral region, while in the evening the tran-sition directly leads to an elevated inversion.Tethered balloon sounding. Profiles of temperature, hu-midity, wind speed, and direction were taken at siteA2 on the valley floor by operating an AtmosphericInstrumentation Research, Inc. (AIR), tethersonde(type TS-3A-SPH). Due to restrictions imposed by lo-cal air traffic control authorities, these ascents couldonly be performed on flight days during R-IOP2. Atypical ascent–descent cycle took about 30 min andreached a maximum height of 800 m above the valleyfloor. In combination with other “profiling systems,”the tethersonde data allow for a detailed investigationof the valley atmosphere’s spatial structure.Sodar. Two monostatic flat-array sodars (Scintec,FAS64) were situated at sites A1 and E1. Due to noisepollution these instruments could not be operatedcontinuously for day-and-night periods during flightdays. In addition, the sodar at site A1 exhibited sometechnical problems, which inhibited operation mostof the time. The instrument at site E1 was operatedwith a height-dependent vertical resolution (10 mnear the ground, and coarser with increasing height)and a temporal resolution of 20–30 min.Research flights. Research flights were performed withan instrumented light aircraft (Neininger et al. 2001)operated by MetAir for a total of eight flight days. Theaircraft is typically operated at a cruising speed of200 km h-1 and has an endurance of 4–5 h. The aircraftmeasured standard meteorological variables with a highenough sampling rate to allow for the derivation of tur-bulence statistics (for details on the instrumentation seethe MAP-Riviera Web site). Precise locations and heightwere continuously recorded for later mapping. In ad-dition to meteorological parameters, aerosol concen-trations and some trace gas concentrations were mea-sured as parameters that can be used to identify theorigin of air masses and transport processes. Duringsome flights a downward-looking IR camera was em-ployed to map the surface (radiation) temperature field.A flight day consisted of one of the following: amorning flight, an afternoon flight, or both. Generally,a flight started and ended with a profile flown up toabout 4000 m or to the ceiling height (pattern T). Inbetween, a succession of two different flight patternswas flown:i) pattern A: valley traverses at different heights(Fig. 6a) yielding a quasi-stationary valley crosssection of mean flow and turbulence characteris-tics; andii) pattern B: along-valley flight legs close to theslopes and in the center of the valley at differentheights yielding a three-dimensional picture of thevalley atmosphere (Fig. 6b).FIG. 5. Temperature profiles from the MTP-5 on 24 Sep 1999 duringthe morning transition period.1374 SEPTEMBER 2004|A typical flight then consisted of a succession offlight patterns such as T-A-B-A-B -T, with care takento ensure that flight legs were flown in similar loca-tions on the various days. Depending on the weatherconditions and other considerations, longer andshorter sequences were flown during the eight flightdays.Flight days were selected such that days withdistinctly different boundary layer characteristicswere captured, exhibitingthe following:a) “fully convective” condi-tions (weak synoptic forc-ing, clear sky, develop-ment of a valley windsystem),b) “fully mechanical” condi-tions (strong synopticforcing, overcast), andc) “mixed” conditions or, inboundary layer terminol-ogy, conditions of forcedconvection.A summary of the eightflight days is given (Table 2),where the number of differ-ent day types can be found.The majority (usually morethan 95%) of continuous observations (see similarlytitled section) were active during a flight day. Further-more, all the “additional observations systems” (Table3) were operating at least during the periods of ac-tual flights. In total, 14 flights were made, with ap-proximately 45 h of data.Tracer releases. The simulation of pollutant dispersionin complex terrain is notoriously difficult due to com-plicated flow and turbulence fields in such environ-ments. The detailed meteorological and turbulencedata, as obtained in the present project, yielded anexcellent background for testing and possibly improv-ing currently available dispersion models for complexterrain. Two release experiments (on 29 Septemberand 6 October) were carried out during the field phaseof the project. Both of these days were dominated byclear skies with a valley wind developing, thereby sim-plifying planning of sampler locations.The tracer was released from a point located nearthe village of Claro (point “R” in Fig. 7) at 1400 LSTfor both cases. The release height was 5 m AGL. Theactual release lasted for 25 min and was made at a con-stant rate of 2 g s-1. The tracer, belonging to the fam-ily of cyclic perfluorocarbons (PFCs), was per-fluorodymethylcyclohexane (PP3; see Girardi et al.1998). These substances are environmentally benign,very stable, nontoxic, insensitive to rain washout, andare detectable by chemical analysis at extremely lowconcentrations (10-16 v/v, i.e., 1.5 pg m-3).Ground-level samplers were deployed with positionsdictated by accessibility, with 6 along the valley axis overScintillometer, A2 and D H, M, e (path- All flight days +SLS20 weighted averages) additional daysduring R-IOPsTemperature profiler, A1 or D Profile of T R-IOP1 and R-IOP2MTP-5 (continuously)Radio sounding, A1 Profiles of T, RH, All flight daysMW11 WS, and WDTethered balloon A2 Profiles of T, RH, All flight daysWS, and WD during R-IOP2Sodar (A1) and E1 Profile of WS Some periods duringselected flight daysTABLE 3. Details of the additional observations, H and M denote theturbulent fluxes of sensible heat and momentum, respectively, “e” =dissipation rate of turbulent kinetic energy, T = temperature, RH =relative humidity, WS = wind speed, and WD = wind direction.Instrument Site(s) Variables DurationFIG. 6. Flight patterns of the research flight during theMAP Riviera field campaign (a) Cross-valley flight legs(pattern A) and (b) along—valley flight legs (pattern B).1375SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |a distance of roughly 10 km(Fig. 7) and 10 along the east-ern slope. Figure 8 showsthat the sampler positionsallowed for the investiga-tion of plume spread alongthe valley slope. Samplersalong the valley floor wereequipped with eight bags persampler, sampling air over30-min intervals. Samplersdistributed on the slope wereequipped with tubes. Eachsampler contains 12 tubesprogramed to sequentiallysample for 20 min (4-h sam-pling period).The three panels of Fig. 8show the time evolutionof the tracer mixing ratio(10–15l/l = fl/l) as measuredduring the two tracer releases at selected sampler sites.The two upper panels give the tracer mixing ratiomeasured by samplers on the eastern slope of the val-ley. The lower panel gives the time evolution of thetracer mixing ratio measured along the valley duringthe first release. No tracer was detected at the valleyfloor during the second release. The three panels re-veal that the sampling network nicely detected theplume’s passage. During the first release (Fig. 8, top)the tracer cloud is confined below 1000 m ASL andarrives at different times at the lower samplers.During the second release, on the other hand, thetracer peak is simultaneously detected by all samplerswithin the first three 20-min sampling periods (Fig. 8,middle). The mixing ratio then decreases at a constantrate at all altitudes. Along the valley axis (Fig. 8, bot-tom), the arrival of the peak mixing ratio is delayedfor larger travel distances. The first three samplers ex-hibit a clear peak, while those at some kilometers dis-tance (cf. positions in Fig. 7) show an approximatelyconstant concentration after the arrival of the plume.This suggests a relatively well-mixed lower portion ofthe valley atmosphere as was found on other days witha well-developed valley wind system (see Fig. 9).INSTRUMENT INTERCOMPARISON. TheMAP Riviera project was mainly devoted to the in-vestigation of spatial variability of turbulence statis-tics in highly complex terrain and its relation to meanflow. This was achieved by using a large number ofinstruments of a similar type, such as sonic anemom-eters, and by combining different observational tech-FIG. 7. Tracer release point (R) and the sampler locations for the release of 6Oct 1999. The sampling locations along the valley axis are indicated with an S,those on the slope with simple numbers.niques at different sites. It was, therefore, of major im-portance to assess the relative performance of variousinstruments prior to or during the field campaign. Itis not within the scope of this overview to present allof the corresponding efforts in detail, but a brief sum-mary of some of the most important aspects is pro-vided as follows.Sonic anemometer intercomparison. During a week inearly July 1999, 19 sonic anemometers of five differ-ent types were deployed on a small airfield with ho-mogeneous surface conditions. Details of the setupand procedures can be found in Christen et al. (2000,2001). The sonic anemometers were found to com-pare reasonably well in the derived turbulence statis-tics (Table 4). Those statistics that include the tem-perature are generally less reliable than those withonly wind components. Furthermore, the relativeerror of the wind statistics (and, more so, those in-cluding temperature) increases with smaller absolutewind speed. Overall, it was concluded that the rela-tive accuracy of the instruments was good enough todetect significant differences in turbulence statistics inthe actual field study.Scintillometers. Two displaced-beam scintillometers(Scintec, SLS 20) were compared to sonic anemom-eters, and with each other, in the same prefieldcampaign described above (Weiss 2002). Excellentcorrespondence was found between the two scintil-lometers with correlation coefficients of rM = 0.94(momentum flux M) and of rH = 0.99 (sensible heat1376 SEPTEMBER 2004|flux H), and rms differences ofabout 0.02 N m-2 (momentum flux)and less than 10 W m-2 (heat flux).This is considerably less than thescatter among the various sonic an-emometers. Also, the comparisonbetween sonic anemometers on theone hand and scintillometers on theother was satisfying: rmsH = 12 W m-2and rmsM = 0.04 N m-2, respectively.A more detailed analysis of thedataset revealed that the correspon-dence of turbulent fluxes was verygood for clearly stable and unstableconditions, but was less so for themomentum flux under near-neutralconditions (Weiss 2002). An addi-tional experiment that addressed thequestion of measurement height incomplex topography in connectionwith scintillometers is summarizedin the Sidebar.MTP-5 versus radio soundings. Datafrom the passive microwave tem-perature profiler MTP-5 were com-pared to temperature profiles fromsimultaneously launched radio-sondes at the same site (A1), yield-ing a total of 52 profiles for compari-son. Rms differences between datafrom the two systems were smallerthan 1.0 K everywhere (Kadygrov etal. 2001), regardless of the shape ofthe profile. When selecting onlyclosely linear profiles, the rms differ-ence was 0.39 K, while those profileswith a distinct inversion exhibited anrms difference of 0.47 K. The largestdifferences between the two systemswere observed for “mixed profiles.”In most of the compared profiles thetemperature gradients, that is, thestatic stability, were closely compa-rable. The favorable correspondencebetween the data from the two sys-tems allows a combination of theirrespective advantages for investiga-tion of temperature structure in theRiviera valley. While the MTP-5 hasa limited height range (600 m) but ahigh temporal resolution (5 min),the radiosondes can reach higher al-FIG. 8. Tracer concentrations at the various sampling points (as indi-cated in the inlets, cf. Fig. 7). The sampling time interval of the sam-plers along the valley (bottom panel) was 30 min., that on the slopeamounted to 20 min.1377SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |titudes but were launched only every 3 h during theflight days.RESEARCH OBJECTIVES AND PRELIMI-NARY RESULTS. Here, we summarize specificresearch objectives of the Riviera project. Rather thansimply listing research objectives we will illustratethem using data from the field observations.Turbulence structure in an alpine valley. Due to thesparse (essentially nonexistent) observational evi-dence concerning the turbulence structure in an alpinevalley, a first goal of the MAP-Riviera project mustconsist of establishing a phenomenological picture ofturbulence in such an environment. This provides abasis for exploring specific exchange mechanisms andalso serves as a basic picture, on which results fromnumerical models can be examined. The approachconsists in first defining characteristic flow regimessimilar to those in Table 2 (Andretta et al. 2001). Foreach of these flow regimes background boundarylayer structure in the valley is established with refer-ence to earlier work (e.g., Whiteman 2000).Figure 9 depicts the profiles of wind direction andpotential temperature on a clear summer day withweak synoptic forcing. Wind direction exhibits val-ley wind characteristics with down-valley wind (about340∞) during the night and up-valley wind (160∞) dur-ing the day. Valley wind depth reaches about 1600 mASL during night with a transition to geostrophicScintillometers use information on the spread of a laser beam over adistance of some 100 m to derive spatially averaged turbulence charac-teristics. In this algorithm the height of observation plays an importantrole, which is, obviously, difficult to determine in highly complex terrain.To address this problem, an experiment was performed on the occasionof the instrument intercomparison (see similarly titled section). Thereby,a scintillometer was set up with an inclined path, with the transmitter ata height of 1.35 m and the receiver at a height of 2.15 m, with theresulting mean height of the laser beam at 1.75 m. This latter heightcorresponded to that of a set of sonic anemometers (1.8 m). Figure SB1shows the resulting scatterplots of turbulent fluxes from this experiment.The root-mean-square differences for sensible heat and momentumfluxes amounted to 16 W m-2 and 0.02 N m-2, respectively, and were,thus, comparable to those for experiments with the leveled scintillom-eters (section titled Instrument intercomparison) under similar atmo-spheric conditions. These results indicate that over uneven terrain anaverage height of the laser beam determined over the path of thescintillometer may be a useful choice for the “measuring height.”MEASUREMENT HEIGHT FOR A SCINTILLOMETER BEAMFIG. SB1. Comparison of (top) momentum flux and (bottom) sensible heat flux froman inclined scintillometer (see text) to an average of five collocated sonic anemom-eter measurements.Maximum error of 2 4 8 15 11 15good instruments (%)Outliers (%) 14 14 18 27 27 89TABLE 4. Typical instrument-to-instrument uncertainty (rms differences) for turbulence statistics asderived from the field intercomparison on an airfield in southern Switzerland in Jul 1999. “Outliers”refers to three instruments that were found to have problems in one or more of the measured vari-ables. Statistics are based on about 40 half-hourly averages under optimal conditions (fetch, thresholdrequirements).u¯ (m s-1) su,v (m s-1) sw (m s-1) sq (K) u* (m s-1)æw¢q ¢æ (K m s-1)1378 SEPTEMBER 2004|wind direction (300∞) higher up. This latter is reachedabove the mean crest height (roughly 2000 m ASL).During the day valley wind direction extends up toabout 1600 m ASL. In the transition layer toward thegeostrophic wind direction some indications of returnflow can be observed.Potential temperature profiles show a distinctthree- or four-layer structure. During nighttime, astrongly stable layer a few hundred meters deep istopped by a less stable layer up to 2000 m ASL. Asharp transition then occurs toward the modestlystable lower free troposphere. For some ascents dur-ing the night (not shown) this transition occurs overa layer of a few hundred meters depth. The daytimevalley atmosphere is well mixed in the lowest 700 m.Higher up there is a deep stable layer (dQ/dz ª0.006 K m-1) up to about 1600 m ASL where a sharptransition occurs. There is some indication of a fourthlayer between 1600 and 2000 m ASL, that is, the meancrest height.The observed daytime potential temperature pro-files do not support the inversion breakup hypothesisof Whiteman (1990), which leads to a well-mixedlayer in the entire valley atmosphere for larger valleysin the afternoon. However, it does bear some similar-ity to a multilayer structure observed in a neighbor-ing valley during VOTALP (Furger et al. 2000) or tothe observations of Kuwagata and Kimura (1995) whodescribe a two-layer structure related to a cross-valleycirculation.From these observations it appears that dynamicand thermodynamic structures of the valley flow donot necessarily correspond. During nighttime thelayer between approximately 1600 m ASL and crestheight is characterized by a transition in wind direc-tion, but no clear signal can be seen in the stratifica-tion. During daytime, on the other hand, the valleywind layer (up to about 1600 m ASL) is well mixedup to a few hundred meters and stably stratified in itsupper part.Calanca et al. (2000) presented a preliminaryanalysis of turbulence structure in the afternoon val-ley atmosphere of 25 August 1999. As an example, weshow the kinematic turbulent momentum flux as ob-tained from airborne observations on a cross sectionthrough the Riviera valley collocated with the surfacetowers (Fig. 10). In a relatively shallow band a fewhundred meters deep, the turbulent flux is observedto be substantially different from zero. Comparisonto the noon sounding of the same day (Fig. 9) indi-cates that this turbulent layer roughly corresponds tothe well-mixed lowest portion of the valley atmo-sphere. Nonzero turbulent momentum flux is con-fined to the layer of the valley wind (Fig. 9). The larg-est momentum fluxes are found near the easternslope, and near the center of the valley atmosphere,that is, somewhat west of the observed “core” of thevalley wind on that day (Calanca et al. 2000).Over all, the turbulence structure is found to ex-hibit large spatial variability that is largely driven bythe differences in the surface energy balance (sectiontitled Surface energy balance on a slope) at the valleyfloor and the two slopes. Other flight days (not shown)exhibit a quite different structure in the valleyatmosphere’s turbulence structure, and its relation tothe mean flow field remains to be evaluated in detail.It is one of the primary objectives of the Rivieraproject to investigate to what extent high-resolutionnumerical models are capable of reproducing not onlyobserved mean flow fields, but also the turbulencestructure of an alpine valley. First simulations usingthe Regional Atmospheric Modeling System (RAMS;De Wekker et al. 2002) show a quite favorable corre-spondence in the mean variables (potential tempera-ture, wind speed components) for one case study. Theturbulent surface fluxes, on the other hand, exhibitFIG. 9. Profiles of (top) potential temperature and (bot-tom) wind direction at site A1 in the Riviera valley on25 Aug 1999. Release times are indicated in the inlets.1379SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |substantial differences betweenthe model and observation (DeWekker et al. 2002). The investi-gation of turbulent exchange pro-cesses, as outlined in the next sec-tion, will allow us to evaluate thecorresponding parameterizationsin numerical models in the future.Turbulent exchange processes ofheat, moisture, and momentum. Toinvestigate, and possibly improveon, turbulent exchange param-eterizations in numerical modelsthe underlying processes mustfirst be known and understood.Of primary importance in this re-spect is the surface exchange ofheat, momentum, and moisturefor the various characteristic sur-faces (valley floor, sunlit andshaded slopes). One major chal-lenge in this respect is the simul-taneous occurrence of slope ef-fects, surface inhomogeneity, andthe presence of a roughnesssublayer due to the variable na-ture of the slope surface (Rotach1995). However, this is the normalsituation in a real valley and thepresent dataset provides an excel-lent basis for addressing thesequestions. Mechanisms of turbu-lent exchange within the canopyand the roughness sublayer on theslope have been investigated indetail by Van Gorsel et al. (2003,2001, 2000).Andretta et al. (2001, 2000)have investigated the interactionbetween the valley wind and slopewind systems on clear summerdays with weak synoptic forcing.While turbulent moment trans-port at the valley floor (site A1) isdominated by the along-windcomponent of Reynolds stress(æu¢w¢æ), as is the case over flat, hori-zontally homogeneous terrain, a substantial contribu-tion of lateral shear stress (æv¢w¢æ) is observed at theslope site B and at other slope sites (van Gorsel et al.2003). This is due to directional shear introducedthrough the transition from the near-surface up-slopewind to a valley wind regime (up valley) at the higherlevels of observation. This raises the question ofwhether the conventional definition of a friction ve-locity u* = (æu¢w¢æ2 + æv¢w¢æ2)1/4 can be retained for such acomplicated flow. Figure 11 shows some preliminaryFIG. 10. Kinematic turbulent momentum fluxes on a cross sectionthrough the Riviera valley. The data are 2-h averages from the after-noon flight of 25 Aug 1999 (1200–1400 UTC, flight day 3, Table 2).The light blue diamonds indicate the position of data points fromwhich the fields are interpolated (if below the terrain: irregularity ofthe terrain mask). The surface towers are indicated by only one squareeven if three or more levels are available. Units are m2 s-2.FIG. 11. Scaled vertical velocity fluctuations at site B (see Fig. 2). Obser-vations from the upper two levels at 24 and 30 m ASL, respectively. Thescaling velocity is derived from the local observations of longitudinal andlateral Reynolds stress components (see text).1380 SEPTEMBER 2004|evidence that indeed the sum of (vertical) frictionaland directional shear stresses determine a character-istic local velocity scale. Andretta et al. (2001) showthat using the longitudinal stress component alone toderive a “friction velocity” (what makes sense onlyover homogeneous surfaces, when æv¢w¢æ vanishes) leadsto markedly inferior results (not shown). In Fig. 11,a distinction is made between “morning” and “after-noon” periods because Andretta et al. (2000) foundthese periods to exhibit distinct vertical profiles ofReynolds stress components. However, the locallyscaled vertical velocity seems to be insensitive to thisdifference.Conventional surface exchange schemes employedin numerical models rely on the Monin–Obukhovsimilarity theory to determine momentum fluxesfrom the mean wind profile and, thus, do not incor-porate the contribution of directional shear. They can,therefore, be expected to underestimate the frictionvelocity and, hence, mechanical turbulence. A moredetailed analysis is needed in future investigations ofthis problem.Surface energy balance on a slope. Clearly, the radia-tion balance in a valley and particularly on a slope isdifferent from that on a flat surface (Whiteman et al.1989a). This is due to both shortwave and longwavecontributions, which are influenced by surroundingtopography (Matzinger et al. 2003). While the formerexhibits temporal variation according to obstructionof direct solar radiation, incoming longwave radiationis not only determined by temperature and densityprofiles in the atmosphere, but is also modified byemission from nearby surfaces. In consequence, theenergy balance on a slope is different from that on aflat surface and a position-specific energy partition-ing can be expected at various valley locations.Figure 12 compares components of the energybalance at the valley floor site (A1) to that on theslope for a selection of 15 “valley wind days,” that is,clear days with weak synoptic forcing (see Andrettaet al. 2001 for a definition). At site A1 the sum ofsensible, latent, and ground heat fluxes makes upabout half the available radiative energy, and this isobserved on each of the individual days as well as inthe average over all valley wind days. Similar failureof closure of the energy balance has been observedat other, even less complicated, sites (e.g., Vogt et al.1996) provided that all of the components were di-rectly measured. Clearly, if the energy balance is de-termined using, for example, the Bowen ratiomethod, it will be closed by definition even if themeasurements are taken over highly complex terrain(e.g., Whiteman et al. 1989b).The consistent gap between available radiativeenergy and turbulent and ground heat fluxes cannotbe attributed to measurement errors alone in thepresent case. Although the various contributions toFIG. 12. Mean daily cycles of the components of the near-surface energy balance at (left) site A1 and (right) B.Data are averages over 15 valley wind days (see text). The turbulent fluxes are measured at the respective low-est level at each of the sites (3.5 m at A1 and 15 at B). Net radiation is observed at 2 m at A1 and at 15 m at B.1381SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |the energy balance are observed within a few metersand are relatively close to the surface, it is obviousfrom Fig. 12 that the idealized energy balance ap-proach (i.e., Rn = H + LE + G) for a flat and horizon-tally homogeneous surface cannot be applied in thetype of environment investigated here. The full en-ergy balance for a near-surface layer, including diver-gence of turbulent fluxes and advective processes, hasto be assessed in detail in order to devise appropriateapproaches for modeling purposes. Rotach et al.(2003) have demonstrated that vertical advection intoan idealized box between the height of observationand the surface is likely to make up the energy gap,as is evident from Fig. 12. However, small magnitudesof the vertical velocity make it extremely difficult todetermine the vertical advection term to reliableaccuracy.The situation at the slope site B is similar, albeitless extreme (Fig. 12). Due to the inclination of theslope net radiation reaches larger absolute values, andthe maximum is obtained later in the day than at thevalley floor (Matzinger et al. 2003). The turbulentfluxes, H in particular, at the canopy top are larger inmagnitude than at site A1, but their sum is still muchtoo small to close the simple energy balance equation.Again, while Fig. 12 shows an average over 15 valleywind days, the individual days exhibit very similardaily cycles of the contributions to the energy balance.The present dataset yields the possibility to inves-tigate in detail a proper description of the surface en-ergy balance in complex terrain. In particular, it willbe interesting to assess to what extent local circula-tions driven by variations in the surface energy bal-ance itself are important in explaining the observedenergy fluxes.Boundary layer height in a valley. The exchange of airbetween the valley boundary layer and atmospherealoft, and the exchange between neighboring valleysare key processes in the investigation of air pollutantdispersion in valley atmospheres in which a majorproportion of the population in mountainous terrainresides. Characteristics of this exchange are largelydetermined by the behavior of the boundary layerheight and thermally driven circulation in a valley.The discussion in connection with Fig. 9 hasshown that the inversion height may not, in general,be regarded as the boundary layer height. The depthof the well-mixed portion in the valley atmosphere(if significant at all) corresponds neither to that ofthe valley wind layer nor to the depth of a layer ofnonzero turbulence (Fig. 10). De Wekker (2002) andDe Wekker et al. (2004) have, therefore, investigatedin a case study the valley atmosphere using the vari-ous observations and a mesoscale model at high spa-tial resolution. The numerical model used was theRAMS (Pielke et al. 1992) with 4 two-way interac-tive nested grids down to a horizontal grid spacingof 0.333 km. Model-simulated boundary layer heightwas determined using a Richardson number criterion.For this case study the model was found to repro-duce broad flow features in the Riviera valley.However, the simulated spatial variability wasconsiderably larger than that observed, especially forthe dynamic fields and also for the near-surfaceturbulence structure, that is the surface heat flux(De Wekker et al. 2004).Model simulations indicated the boundary layerheight to be relatively uniform along the valley axisand across the valley. For a clear sunny summer daywith strong radiative heating and significant surfaceheat fluxes a modest “convective boundary layerheight” of about 1300 m was reached in the afternoon.This corresponds to the height of observednonnegligible turbulence on that day (Fig. 10). Note,however, that it does not correspond at all to theheight of a “well mixed” regime for potential tempera-ture (Fig. 9) or other scalars.Presently, efforts in numerical simulation of theobservations and numerical experimentation are in-tensified using the Advanced Regional PredictionSystem (ARPS) modeling system. Results from theseefforts will be reported in forthcoming publications.Hydrological processes. Hydrometeorological observa-tions conducted at MAP-Riviera sites include soilmoisture, soil temperature, soil heat flux, precipita-tion, and leaf interception, with all of these as timeseries having a high temporal resolution (Zappa et al.2000). These data, which are available concurrentlywith the meteorological observations at towers in theRiviera valley, comprise a complete dataset for test-ing and validating soil–vegetation–atmosphere trans-fer schemes (SVATSs) of various degrees of complex-ity. SVATs are key features in both atmospheric andhydrological modeling systems, although importantdifferences exist in the representation of the complex-ity of energetic and hydrological processes. SVATSsconstitute the model interface for describing inter-actions between hydrological processes at the soilsurface and subsurface, and atmospheric processeswithin the planetary boundary layer. They deal withspatial and temporal variations of evaporation, tran-spiration, interception, soil moisture, soil tempera-ture, and the soil energy balance. The respectiveinteractions generate continuous feedback on meteo-1382 SEPTEMBER 2004|rological processes at various scales, on the site’senergy and water balance, and on runoff generationprocesses. This highlights the importance of researchleading to a consistent coupling of atmospheric andhydrological processes and models in complexterrain.Figure 13 depicts a time series of some of the mea-sured and modeled hydrological components at siteA1 for the beginning of September 1999. The year1999 was rather wet when compared to the climato-logical average resulting in relatively large soilmoisture. The soil did not completely dry out, evenduring the 2-week period of fine weather (5–18 Sep-tember). The decrease in soil moisture and subse-quent recovery after the precipitation events are gen-erally well captured by the Precipitation RunoffEvapotranspiration Hydrotope model (PREVAH)(Gurtz et al. 1999), though the depletion of soil mois-ture is somewhat underestimated by the model dur-ing the dry period. Both methods to determine thelatent heat flux that use the energy balance equation(i.e., the Bowen ratio approach and the PREVAHmodel, which uses the Penman–Monteith approach)overestimate, as compared to the eddy correlationmeasurements. This is most likely related to thenonclosure of the near-surface energy balance at siteA1 (section titled Surface energy balance on a slope)during periods of fine weather. Figure 12 reveals thatthe sum of observed turbulent sensible and latent heatfluxes is substantially smaller than the observed avail-able energy Rn – G. Thus, if a model assumes energybalance closure in its simplest form, it will overesti-mate the contribution of turbulent fluxes. If it shouldturn out that the missing energy is due to local ad-vection and is a consequence of the valley wind sys-tem itself, this would have to give rise to a modifica-tion of these methods in complex terrain. Thisexample nicely shows the mutual benefits in differ-ent research areas that can be gained through the col-laborative nature of the MAP-Riviera project.SUMMARY AND OUTLOOK. In the presentpaper an overview is provided of the scientific back-ground and experimental arrangements of theMAP-Riviera project. This project aims at investigat-ing the near-surface and boundary layer turbulent ex-change processes in the presence of highly complextopography because it is the rule rather than the ex-ception in the Alps and other major mountainranges. During the experimental phase (July–October 1999), a high-quality dataset was collectedin the Riviera valley in southern Switzerland. It com-FIG. 13. (top) Time series of hydrological parameters for 1–23 Sep 1999 at site A1. (bottom) Observed soil mois-ture integrated over a 0.65-m-deep layer (dark line) and simulated with the PREVAH hydrological model (labeled“Penman–Monteith”); dark columns indicate precipitation events (right scale). (top) Latent heat flux observedusing eddy correlation (diamonds), calculated from temperature and humidity profiles using the Bowen ratioapproach (dark line) and determined with the PREVAH model (gray lines), which uses the Penman–Monteithequation.1383SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |prises detailed near-surface turbulence measure-ments, profiles of mean meteorological variablesthroughout the valley atmosphere, and airborne ob-servations of the mean and turbulence structure inthe bulk of the valley. Also, two tracer release experi-ments were conducted. Detailed hydrological mea-surements at various sites within the valley completethe experimental efforts.The Riviera dataset is presently being exploited tocharacterize and understand turbulent exchange pro-cesses near the surface, and between the valley atmo-sphere and free troposphere. Emphasis is given to theinteraction of the local dynamic and thermodynamicfields (e.g., a thermally driven valley wind system) andthe associated turbulence structure. The overall goalof the project lies in the evaluation and assessment ofturbulence exchange parameterizations in meteoro-logical and hydrological numerical models. Not onlydo we have a dataset at hand, which contains all thenecessary variables to critically assess such param-eterizations, but we also hope to use the results of theRiviera project to improve our understanding of theunderlying processes. This, in turn will potentiallylead to modified turbulence parameterizations fornumerical models in complex terrain. First simula-tions with the mesoscale atmospheric model RAMSwith simpler diagnostic models (not shown), and thehydrological runoff model PREVAH, are promisingin the sense that they reveal some overall skill in re-producing the observed mean flow features. This isespecially noteworthy because these models were usedbeyond their range of applicability. We mention hereonly the efforts that were necessary in order to obtaina stable integration for RAMS at the given high spa-tial resolution in the very steep orography of theRiviera valley. A careful analysis of both the observa-tions and the model results clearly bears the poten-tial to identify possible weaknesses in the numericsand physical parameterizations for mesoscale meteoro-logical modeling in highly complex terrain. We an-ticipate that other models of various degrees of complex-ity will be involved in this process in the near future.ACKNOWLEDGMENTS. Many individuals and insti-tutions have contributed to the success of the MAP-Rivieraproject. Funds were available from the Swiss National Sci-ence Foundation (Grants 21-54060.98, 21-55874.98, and20-63820.01), grants from the Natural Sciences and Engi-neering Research Council of Canada to D. G. Steyn and theEuropean Joint Research Center (JRC) in Ispra (I). Supportbefore and during the observational campaign of the localauthorities of Claro, Switzerland, and in particularMr. Pellegrini, is greatly appreciated. The passive micro-wave temperature profiler (MTP-5) was made available forthe project by Kipp & Zonen B.V. (Delft, the Netherlands)through the kind intervention of markasub AG (Basel,Switzerland) and Meteodat GmbH (Zürich, Switzerland).We are indebted to the Forschungszentrum Karlsruhe(Dr. N. Kalthoff) and to the Paul Scherrer Institute (Dr. M.Furger) for lending us a number of sonic anemometers.Karl Schroff and Hansjürg Frei from ETHZ, as well asValentino Badà and N. Cao from JRC were key people inthe preparation, construction, and maintenance of all ofthe countless surface towers, the instrumentation, and thenecessary facilities. Many thanks go to a large number ofstudents and other individuals who helped in the field tooperate and maintain the instrumentation, download thedata, and operate the radio sounding system: A. Bassi,S. Bethke, S. Carlaw, M. Dippon, C. Feigenwinter, A. Felber,C. Heinemann, S. Hoch, F. Imbery, P. Keller, N. Matzinger,E. Müller, M. Müller, P. Müller (from the workshop ofMCR-Lab), O. Rahs, M. Raymond, S. Regazzi, J. Sedlacek,S. Seneviratne, and S. Zimmermann. Finally, we acknowl-edge the very helpful and constructive comments of twoanonymous referees, which made the paper more readableand understandable.REFERENCESAndretta, M., S. Zimmermann, M. W. Rotach,P. Calanca, A. Christen, and R. Vogt, 2000: Investi-gation of the near-surface boundary-layer in anAlpine valley. MAP Newsletter, Vol. 13, 68–69.——, A. Weiss, N. Kljun, and M. W. Rotach, 2001: Near-surface turbulent momentum flux in an Alpinevalley: Observational results. MAP Newsletter, Vol. 15,122–125.Anfossi, D., D. Desiato, G. Tinarelli, G. Brusasca,E. Ferrero, and D. Sacchetti, 1998: TRANSALP 1989experimental campaign—II: Simulation of tracer ex-periments  with Lagrangian particle model.  Atmos.Environ., 32, 1157–1166.Belcher, S. E., and J. C. R. Hunt, 1998: Turbulent flowover hills and waves. Ann. Rev. Fluid Mech., 30, 507–538.Beljaars, A. C. M., and P. Viterbo, 1998: Role of theboundary layer in a numerical weather predictionmodel. Clear and Cloudy Boundary Layers, Proceed-ings of the Colloquium “Clear and Cloudy BoundaryLayers,” Amsterdam 26-29 August 1997, A. A. M.Holtslag and P. G. Duynkerke, Eds., Edita KNAW,287–304.Binder, P., and C. Schär, Eds., 1996: The Mesoscale AlpineProgramme: Design proposal. MAP, 77 pp. [Avail-able from MAP Programme Office, c/o Swiss Meteo-rological Institute, CH-8044 Zürich, Switzerland.]1384 SEPTEMBER 2004|Bougeault, P., and Coauthors, 2001: The MAP special ob-serving period. Bull. Amer. Meteor. Soc., 82, 433–462.Calanca, P., M. W. Rotach, A. Andretta, R. Vogt,E. Van Gorsel, and A. Christen, 2000: The turbulencestructure in an Alpine valley. Preprints, Ninth Conf.on Mountain Meteorology, Aspen, CO, Amer.Meteor. Soc., 235–236.Carlaw, S. M., N. Kouwen, and E. D. Soulis, 2000: Thesignificance of soil moisture in flood forecasting.Abstracts, Annual Scientific Meeting, Banff, AB,Canada, Canadian Geophysical Union.Christen, A., E. Van Gorsel, M. Andretta, P. Calanca,M. W. Rotach, and R. Vogt, 2000: Field intercom-parison of ultrasonic anemometers during the MAP-Riviera project. Preprints, Ninth Conf. on MountainMeteorology, Aspen, CO, Amer. Meteor. Soc., 130–131.——, ——, R. Vogt, M. Andretta, and M. W. Rotach, 2001:Ultra sonic anemometer instrumentation at steepslopes: Wind tunnel study-field intercomparison-measurements. MAP Newsletter, No. 15. [Availablelonline at www.map.ethz.ch/NL15/christen.pdf.]Cionco, R. M., and Coauthors, 1999: An overview ofMADONA: A multinational field study of high-resolution meteorology and diffusion over complexterrain. Bull. Amer. Meteor. Soc., 80, 5–20.Clements, W. E., J. A. Archuleta, and P. H. Gudiksen,1989: Experimental design for the 1984 ASCOT fieldstudy. J. Appl. Meteor., 28, 405–413.De Wekker, S. F. J., 2002: Structure and morphology ofthe convective boundary layer in mountainous ter-rain. Ph.D. dissertation, The University of BritishColumbia, 191 pp.——, D. G. Steyn, M. W. Rotach, J. D. Fast, and S. Zhong,2002: Observations and numerical modeling of thedaytime boundary layer structure in the Riviera Valley,Switzerland. Preprints, 10th Conf. on Mountain Me-teorology, Park City, UT, Amer. Meteor. Soc., 35–38.——, ——, J. D. Fast, M. W. Rotach, and S. Zhong, 2004:The performance of RAMS in representing the con-vective boundary layer structure in a very steep val-ley. J. Environ. Fluid Mech., in press.Emeis, S., and M. W. Rotach, 1997: Working group onplanetary boundary-layer (WG-PBL). MAP Newslet-ter, Vol. 6, 13–19.Furger, M., and Coauthors, 2000: The VOTALPMesolcina Valley Campaign 1996—Concept, back-ground and some highlights. Atmos. Environ., 34,1395–1412.Georgelin, M., E. Richard, M. Petitdidier, and A.Druilhet, 1994: Impact of subgrid-scale orographyparameterization on the simulation of orographicflows. Mon. Wea. Rev., 122, 1509–1522.Girardi, F., G. Graziani, D. van Veltzen, S. Galmarini,S. Mosca, R. Bianconi, R. Bellasio, and W. Klug, Eds.,1998: The ETEX project. Office for Official Publica-tions of the European Communities EUR Rep. 181-43, EN, 108 pp.Gurtz, J., A. Baltensweiler, and H. Lang, 1999: Spatiallydistributed hydrotope-based modelling of evapo-transpiration and runoff in mountainous basins.Hydrol. Processes, 13, 2751–2768.Holtslag, A. A. M., and F. T. M. Nieuwstadt, 1986:Scaling the atmospheric boundary layer. Bound.-Layer Meteor., 36, 201–209.Jackson, P. S., and J. C. R. Hunt, 1975: Turbulentwindflow over a low hill. Quart. J. Roy. Meteor. Soc.,101, 929–955.Kadygrov, E. N., and D. R. Pick, 1998: The potential fortemperature retrieval from an angular-scanningsingle channel microwave radiometer and somecomparisons with in-situ observations. Meteor. Appl.,5, 393–404.——, V. Kadygrov, E. Miller, H. Weber, and M. W.Rotach, 2001: The thermal structure of the atmo-spheric boundary layer in an Alpine valley: Resultsof continuous remote sensing measurements andcomparison with radiosonde data. IRS 2000: CurrentProblems in Atmospheric Radiation, W. L. Smith andYu M. Timofeyer, Eds., A. Deepak Publishing, 1097–1101.Kuwagata, T., and F. Kimura, 1995: Daytime boundarylayer evolution in a deep valley. Part I: Observationsin the Ina Valley. J. Appl. Meteor., 34, 1082–1091.Louis, J. F., 1979: A parametric model of vertical fluxes inthe atmosphere. Bound.-Layer Meteor., 17, 187–202.Matzinger, N., M. Andretta, E. Van Gorsel, R. Vogt,A. Ohmura, and M. W. Rotach, 2003: Surface radia-tion budget in an alpine valley. Quart. J. Roy. Meteor.Soc., 129, 877–895.Neininger, B., W. Fuchs, M. Bäumle, A. Volz-Thomas,A. H. S. Prevot, and J. Dommen, 2001: A small air-craft for more than just ozone: Metair’s Dimona af-ter ten years of evolving developments. Preprints,11th Symp. on Meteorological Observations and In-strumentation, Albuquerque, NM, Amer. Meteor.Soc., 123–128.Pielke, R. A., and Coauthors, 1992: A comprehensivemeteorological modeling system—RAMS. Meteor.Atmos. Phys., 49, 69–91.Plüss, C., and R. Mazzoni, 1994: The role of turbulentheat fluxes in the energy balance of high Alpine snowcover. Nord. Hydrol., 25, 25–38.Randall, D. A., 2001: Representing the turbulent bound-ary layer in global atmospheric models. Geophys. Res.Abstr., 3, p. 4163.1385SEPTEMBER 2004AMERICAN METEOROLOGICAL SOCIETY |Rotach, M. W., 1995: On the boundary layer over moun-tainous terrain—A frogs perspective. MAP Newslet-ter, Vol. 3, 31–32.——, P. Calanca, R. Vogt, D. G. Steyn, G. Graziani, andJ. Gurtz, 2000: The turbulence structure and ex-change processes in an Alpine valley: The MAP-Riviera project. Preprints, Ninth Conf. on MountainMeteorology, Aspen, CO, Amer. Meteor. Soc., 231–234.——, ——, A. H. Weigel, and M. Andretta, 2003: On theclosure of the surface energy balance in highly com-plex terrain. Extended Abstracts Vols. A and B, Int.Conf. on Alpine Meteorology and MAP Meeting 203,Veröffentlichungen der MeteoSchweiz, 66, 247–250.Van Gorsel E., A. Christen, E. Parlow, and R. Vogt, 2000:Micrometeorological measurements at an alpineslope. Preprints, Ninth Conf. on Mountain Meteorol-ogy, Aspen, CO, Amer. Meteor. Soc., 237–238.——, ——, ——, and ——, 2001: Momentum transportin and above a forest at a steep slope. MAP Newslet-ter, No. 15. [Available online at www.map.ethz.ch/NL15/gorsel.pdf.]——, ——, C. Feigenwinter, E. Parlow, and R. Vogt, 2003:Daytime turbulence statistics above a steep forestedslope. Bound.-Layer Meteor., 109, 311–329.Vergeiner, I., 1987: An elementary valley wind model.Meteor. Atmos. Phys., 36, 255–263.Vogt, R., C. Bernhofer, L. W. Gay, L. Jaeger, andE. Parlow, 1996: The available energy over a ScotsPine plantation: What’s up for partitioning? Theor.Appl. Climatol., 53, 23–31.Weigel, A. H., and M. W. Rotach, 2003: On the turbu-lence structure in a daytime alpine valley. ExtendedAbstracts Vols. A and B, Int. Conf. on Alpine Meteo-rology and MAP Meeting 203, Veröffentlichungender MeteoSchweiz, 66, 162–165.Weiss, A., 2002: Determination of thermal stratificationand turbulence of the atmospheric surface layer overvarious types of terrain by optical scintillometry.Ph.D. thesis, Swiss Federal Institute of Technology,ETH Dissertation 14514, 153 pp.——, M. W. Rotach, and M. Hennes, 1999: Comparisonof turbulence parameters derived from optical scin-tillation and eddy-correlation technique over flatterrain. Preprints, 13th Symp. on Boundary Layerand Turbulence, Dallas, TX, Amer. Meteor. Soc.,145–146.——, M. Hennes, and M. W. Rotach, 2001: Derivationof refractive index and temperature gradients fromoptical scintillometry for the correction of atmo-spheric induced problems in highly precise geodeticmeasurements. Surv. Geophys., 22, 589–596.Whiteman, C. D., 1990: Observations of thermally de-veloped wind systems in mountainous terrain. Atmo-spheric Processes over Complex Terrain, Meteor.Monogr., No. 45, Amer. Meteor. Soc., 5–42.——, 2000: Mountain Meteorology. Oxford UniversityPress, 355 pp.——, K. J. Allwine, L. J. Fritschen, M. M. Orgill, and J.R. Simpson, 1989a: Deep valley radiation and surfaceenergy budget microclimates. Part I: Radiation. J.Appl. Meteor., 28, 414–426.——, ——, ——, ——, and ——, 1989b: Deep valleyradiation and surface energy budget microclimates.Part II: Energy budget. J. Appl. Meteor., 28, 427–437.Zappa, M., N. Matzinger, and J. Gurtz, 2000:Hydrological and meteorological measurements atClaro (CH)—Lago Maggiore Target Area in theMAP-SOP 1999 RIVIERA experiment including firstevaluation. Hydrological Aspects in the Mesoscale Al-pine Programme-SOP Experiment, Tech. Rep. of theDepartment of Civil Engineering 10(2), Universityof Brescia, 33 pp.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics

Country Views Downloads
Austria 10 0
France 7 0
Canada 6 0
Japan 6 0
United States 5 22
Russia 4 0
Switzerland 2 1
China 2 0
City Views Downloads
Innsbruck 10 0
Montreal 6 0
Kyoto 6 0
Saint Petersburg 4 0
Grenoble 3 0
Roubaix 3 0
Dallas 3 0
Shenzhen 2 0
Zurich 2 1
Unknown 1 6
Charlottesville 1 1
Ashburn 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items