• Ballard, S. P., , B. W. Golding, , and R. N. B. Smith, 1991: Mesoscale model experimental forecasts of the Haar of northeast Scotland. Mon. Wea. Rev., 119 , 21072123.

    • Search Google Scholar
    • Export Citation
  • Banta, R. M., , L. S. Darby, , P. Kaufmann, , D. H. Levinson, , and C-J. Zhu, 1999: Wind-flow patterns in the Grand Canyon as revealed by Doppler lidar. J. Appl. Meteor., 38 , 10691083.

    • Search Google Scholar
    • Export Citation
  • Binder, P., , and C. Schär, Eds.,. 1996: MAP design proposal. MeteoSwiss, 75 pp. [Available from the MAP Programme Office, MeteoSwiss, CH-8044, Zurich, Switzerland.].

    • Search Google Scholar
    • Export Citation
  • Bougeault, P., , P. Binder, , and J. Kuettner, Eds.,. 1998: MAP science plan. MeteoSwiss, 64 pp. [Available from MAP Programme Office, MeteoSwiss, CH-8044, Zurich, Switzerland.].

    • Search Google Scholar
    • Export Citation
  • Bougeault, P., and Coauthors, 2001: The MAP Special Observing Period. Bull. Amer. Meteor. Soc., 82 , 433462.

  • Browning, K. A., , and R. Wexler, 1968: The determination of kinematic properties of a wind field using Doppler radar. J. Appl. Meteor., 7 , 105113.

    • Search Google Scholar
    • Export Citation
  • Clark, T. L., , W. D. Hall, , and R. M. Banta, 1994: Two-and three-dimensional simulations of the 9 January 1989 severe Boulder windstorm: Comparison with observations. J. Atmos. Sci., 51 , 23172343.

    • Search Google Scholar
    • Export Citation
  • Cohn, S. A., , S. D. Mayor, , C. J. Grund, , T. M. Weckwerth, , and C. Senff, 1998: The Lidars in Flat Terrain (LIFT) experiment. Bull. Amer. Meteor. Soc., 79 , 13291343.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , and C. F. Mass, 1998: Windstorms along the western side of the Washington Cascade Mountains. Part I: A high-resolution observational and modeling study of the 12 February 1995 event. Mon. Wea. Rev., 126 , 2852.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , and C. F. Mass, 2000: High-resolution observations and numerical simulations of easterly gap flow through the Strait of Juan de Fuca on 9–10 December 1995. Mon. Wea. Rev., 128 , 23982422.

    • Search Google Scholar
    • Export Citation
  • Cressman, G. P., 1959: An operational objective analysis system. Mon. Wea. Rev., 87 , 367374.

  • Darby, L. S., , W. D. Neff, , and R. M. Banta, 1999: Multiscale analysis of a meso-β frontal passage in the complex terrain of the Colorado Front Range. Mon. Wea. Rev., 127 , 20622081.

    • Search Google Scholar
    • Export Citation
  • Darby, L. S., , R. M. Banta, , and R. A. Pielke, 2002a: Comparisons between mesoscale model terrain sensitivity studies and Doppler lidar measurements of the sea breeze at Monterey Bay. Mon. Wea. Rev., 130 , 28132838.

    • Search Google Scholar
    • Export Citation
  • Darby, L. S., , K. J. Allwine, , and R. M. Banta, 2002b: Relationship between tracer behavior in downtown Salt Lake City and basin-scale wind flow. Preprints, 10th Conf. on Mountain Meteorology, Park City, UT, Amer. Meteor. Soc., 12–15.

    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and N. A. Bond, 2001: Research aircraft observations and numerical simulations of a warm front approaching Vancouver Island. Mon. Wea. Rev., 129 , 978998.

    • Search Google Scholar
    • Export Citation
  • Durran, D. R., , T. Maric, , R. M. Banta, , L. S. Darby, , and R. M. Hardesty, 2003: A comparison of ground-based Doppler lidar and airborne in situ wind observations above complex terrain. Quart. J. Roy. Meteor. Soc., 129 , 693713.

    • Search Google Scholar
    • Export Citation
  • Flamant, C., and Coauthors, 2002: Gap flow in an Alpine valley during a shallow south föhn event: Observations, numerical simulations and hydraulic analogue. Quart. J. Roy. Meteor. Soc., 128 , 11731210.

    • Search Google Scholar
    • Export Citation
  • Garnier, B. J., , and A. Ohmura, 1968: A method of calculating the direct shortwave radiation income of slopes. J. Appl. Meteor., 7 , 796800.

    • Search Google Scholar
    • Export Citation
  • Gayno, G. A., 1994: Development of a higher-order, fog-producing boundary layer model suitable for use in numerical weather prediction. M.S. thesis, Dept. of Meteorology, The Pennsylvania State University, 104 pp.

    • Search Google Scholar
    • Export Citation
  • Goodin, W. R., , G. J. McRae, , and J. H. Seinfeld, 1979: A comparison of interpolation methods for sparse data: Application to wind and concentration fields. J. Appl. Meteor., 18 , 761771.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev., 121 , 764787.

  • Grell, G. A., , J. Dudhia, , and D. R. Stauffer, 1995: A description of the fifth-generation Penn State/NCAR mesoscale model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 122 pp.

    • Search Google Scholar
    • Export Citation
  • Jackson, P. L., , and D. G. Steyn, 1994: Gap winds in a fjord. Part I: Observations and numerical simulation. Mon. Wea. Rev., 122 , 26452665.

    • Search Google Scholar
    • Export Citation
  • Klemp, J. B., , and D. R. Durran, 1983: An upper boundary condition permitting internal gravity wave radiation in numerical mesoscale models. Mon. Wea. Rev., 111 , 430444.

    • Search Google Scholar
    • Export Citation
  • Kuhn, M., 1989: Föhnstudien. Wissenschaftliche Buchgesellschaft, 504 pp.

  • Mayr, G. J., , I. Vergeiner, , A. Gohm, , J. Vergeiner, , and R. Mayr, 2003a: Gap flow—An overview and preliminary results. Österr. Beitr. Meteor. Geophys., 29 , 1554.

    • Search Google Scholar
    • Export Citation
  • Mayr, G. J., and Coauthors, 2003b: Gap flow measurements during the Mesoscale Alpine Programme. Meteor. Atmos. Phys., in press.

  • Mlawer, E. J., , S. J. Taubman, , P. D. Brown, , M. J. Iacono, , and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102 , 1666316682.

    • Search Google Scholar
    • Export Citation
  • Neiman, P. J., , R. M. Hardesty, , M. A. Shapiro, , and R. E. Cupp, 1988: Doppler lidar observations of a downslope windstorm. Mon. Wea. Rev., 116 , 22652275.

    • Search Google Scholar
    • Export Citation
  • Orlanski, I., 1975: A rational subdivision of scales for atmospheric processes. Bull. Amer. Meteor. Soc., 56 , 527530.

  • Pan, F., , and R. B. Smith, 1999: Gap winds and wakes: SAR observations and numerical simulations. J. Atmos. Sci., 56 , 905923.

  • Piringer, M., 1994: Selected results of a sodar intercomparison experiment. Meteor. Z., 3 , 132137.

  • Piringer, M., , K. Baumann, , U. Pechinger, , and S. Vogt, 2001: Meteorological and ozone sounding experience during a strong foehn event—A MAP case study. Meteor. Z., 10 , 445455.

    • Search Google Scholar
    • Export Citation
  • Post, M. J., , and R. E. Cupp, 1990: Optimizing a pulsed Doppler lidar. Appl. Opt., 29 , 41454158.

  • Reisner, J., , R. M. Rasmussen, , and R. T. Bruintjes, 1998: Explicit forecasting of supercooled liquid water in winter storms using the MM5 mesoscale model. Quart. J. Roy. Meteor. Soc., 124 , 10711107.

    • Search Google Scholar
    • Export Citation
  • Rogers, D. P., and Coauthors, 1998: Highlights of Coastal Waves 1996. Bull. Amer. Meteor. Soc., 79 , 13071326.

  • Seibert, P., 1985: Fallstudien und statistische Untersuchungen zum Südföhn im Raum Tirol. Ph.D. thesis, University of Innsbruck, 368 pp.

    • Search Google Scholar
    • Export Citation
  • Seibert, P., 1990: South foehn studies since the ALPEX experiment. Meteor. Atmos. Phys., 43 , 91103.

  • Sprenger, M., , and C. Schär, 2001: Rotational aspects of stratified gap flows and shallow föhn. Quart. J. Roy. Meteor. Soc., 127 , 161187.

    • Search Google Scholar
    • Export Citation
  • Stanski, H. R., , L. J. Wilson, , and W. R. Burrows, 1989: Survey of common verification methods in meteorology. World Weather Watch Tech. Rep. 8, Tech. Doc. 358, World Meteorological Organization, Geneva, Switzerland, 114 pp.

    • Search Google Scholar
    • Export Citation
  • Steenburgh, W. J., , D. M. Schultz, , and B. A. Colle, 1998: The structure and evolution of gap outflow over the gulf of Tehuantepec, Mexico. Mon. Wea. Rev., 126 , 26732691.

    • Search Google Scholar
    • Export Citation
  • Wilks, D. S., 1995: Statistical Methods in the Atmospheric Sciences. Academic Press, 467 pp.

  • Zängl, G., 2002a: Stratified flow over a mountain with a gap: Linear theory and numerical simulations. Quart. J. Roy. Meteor. Soc., 128 , 927949.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., 2002b: An improved method for computing horizontal diffusion in a sigma-coordinate model and its application to simulations over mountainous topography. Mon. Wea. Rev., 130 , 14231432.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., 2002c: Idealized numerical simulations of shallow föhn. Quart. J. Roy. Meteor. Soc., 128 , 431450.

  • Zängl, G., 2003: Deep and shallow south foehn in the region of Innsbruck: Typical features and semi-idealized numerical simulations. Meteor. Atmos. Phys., 83 , 237261.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., , A. Gohm, , and G. Geier, 2003: South foehn in the Wipp Valley–Innsbruck region: Numerical simulations of the 24 October 1999 case (MAP-IOP 10). Meteor. Atmos. Phys., in press.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., , B. Chimani, , and C. Häberli, 2004: Numerical simulations of the foehn in the Rhine Valley on 24 October 1999 (MAP IOP 10). Mon. Wea. Rev., 132 , 368389.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Topographic map of the MAP Brenner Pass target area. Shown are model domains 5 and 6 of the MM5 simulations. Elevation contours are gray shaded, starting at 600 m with an increment of 200 m. The 1600 m MSL contour line is white. The solid rectangle in domain 5 indicates the location of domain 6, which is displayed in the upper-right corner. The dashed rectangle represents the subdomain shown in Figs. 2 and 8. Filled circles show the location of instrumented sites: Sterzing (STZ, 944 m), Brennerbad (BRB, 1310 m), Brenner (BRE, 1373 m), Sattelberg (SAB, 2107 m), Steinach (STE, 1116 m), Gedeir (GED, 1084 m), Ellbögen (ELB, 1080 m), Patsch (PAT, 913 m), and Innsbruck (IBK, 609 m). Italic letters indicate the location of the Wipp Valley (WV), Inn Valley (IV), Stubai Valley (SV), and Gschnitz Valley (GV)

  • View in gallery

    Topographic map of the Wipp Valley region. Solid lines with circles indicate four selected flight legs (P1P2, P3P4, P5P6, and P7P8) of the NCAR Electra aircraft at approximately 1500 UTC on 24 Oct 1999; the average cruising altitude was 4.9 km MSL. The dashed line with markers every 10 km indicates the location of the cross section displayed in Fig. 18. The dashed rectangle represents the subdomain displayed in Figs. 2123. Cross markers show the location of mountain summits: Patscherkofel (PAK, 2252 m), Nockspitze (NOC, 2403 m), Serles (SER, 2717 m), Bentlstein (BEN, 2436 m), Nösslachjoch (NOS, 2231 m), and Sattelberg (SAB, 2107 m)

  • View in gallery

    ECMWF analysis for 1200 UTC on (a), (b) 24 Oct 1999 and (c), (d) 25 Oct 1999. Height of the 500-hPa level is represented as contour lines with 50-gpm increments and wind barb for the Brenner Pass grid point at the 700-hPa level in (a) and (c); half barb, full barb, and triangle denote 2.5, 5, and 25 m s−1, respectively. Contour lines with 2-hPa increments in (b) and (d) represent sea level pressure

  • View in gallery

    Hand analysis of surface fronts based on ECMWF analysis data for (a) 24 and (b) 25 Oct 1999. Fronts are drawn as isochrones, with dotted, dashed, dashed–dotted, and solid lines at 0000, 0600, 1200, and 1800 UTC, respectively

  • View in gallery

    Cross-Alpine pressure gradient (Pa km−1) between 1800 UTC 23 Oct 1999 and 0600 UTC 26 Oct 1999. (a) Meso-α-scale gradient as a function of time derived from ECMWF analysis gridpoint data (solid line) and observations (dashed line) at locations near Verona and Munich. Meso-β-scale gradient derived from observations at Sterzing and Patsch as a (b) function of time and (c) function of wind speed observed at Sattelberg (plus signs), Ellbögen (circles), and Innsbruck (crosses). The linear regression curves in (c) have inclination 0.53 for SAB, 0.91 for ELB, and 2.36 for IBK. Meso-α-scale and meso-β-scale gradients were calculated from pressure data reduced to sea level and to 1000 m MSL, respectively. Positive values indicate a northward-directed pressure gradient force

  • View in gallery

    Time series of MAP weather station data from 1800 UTC 23 Oct 1999 to 0600 UTC 26 Oct 1999. Plotted are (a) sustained wind speed (m s−1), wind direction (°), and potential temperature (K) for the station Sattelberg (solid), Ellbögen (dotted), and Innsbruck (dashed). Light gray areas superposed with dark gray areas indicate periods with south foehn at Ellbögen and Innsbruck, respectively

  • View in gallery

    Vertical soundings from radiosonde ascents at Gedeir (solid line) and Sterzing (dashed line) at 1400 and 1500 UTC 24 Oct 1999, respectively: (a) potential temperature (K), (b) wind speed (m s−1), and (c) wind direction (°).

  • View in gallery

    MM5 fields of the 0000 UTC run at 1500 UTC 24 Oct 1999. Orography is gray shaded starting at 1200 m with 600-m increments. The 600-m contour line is light gray. Horizontal winds at (a) the surface, (c) z = 2000 m MSL, and (e) z = 3000 m MSL. Wind barbs defined as in Fig. 3. (b) Surface potential temperature, with contour increment 1 K. Vertical wind speed at (d) z = 2000 m MSL and (f) z = 3000 m MSL; contour increment is 2 m s−1, positive (negative) contour lines are solid (dashed). Boxed numbers represent horizontal wind speed (m s−1) in (a) and potential temperature (K) in (b), observed at selected MAP weather stations

  • View in gallery

    Meso-β-scale pressure gradient as (a) a function of time and (b), (c) a function of wind speed for the period 0600 UTC 24 Oct 1999 to 0000 UTC 25 Oct 1999. Comparison of observed pressure gradient (solid line) calculated from pressure data measured at PAT and STZ with values derived from MM5 gridpoint data of the 0000 UTC run (dashed line) and of the 1800 UTC run (dotted line) in (a). Simulated pressure gradient as a function of wind speed at SAB (plus signs), ELB (circles), and IBK (crosses) for the 0000 UTC run in (b) and for the 1800 UTC run in (c). The linear regression curves in (b) have inclination 0.69 for SAB, 1.13 for ELB, 1.44 for IBK, and in (c) 0.68 for SAB, 1.11 for ELB, 1.42 for IBK.

  • View in gallery

    Comparison of observations from MAP weather stations (solid line) with MM5 gridpoint data of the 0000 UTC run (dashed line) and 1800 UTC run (dotted line) for the period 0600 UTC 24 Oct 1999 to 0000 UTC 25 Oct 1999. Plotted is wind speed upstream of and at the pass for (a) STZ, (b) BRE, and (c) SAB and downstream for (a) STE, (b) ELB, and (c) IBK. Model data were taken from domain 5 in (a) and (d)–(f), and from domain 6 in (b) and (c).

  • View in gallery

    As in Fig. 10, but for potential temperature

  • View in gallery

    Vertical profile of the horizontal wind at the lidar site GED as function of time for the period 0900–1800 UTC 24 Oct 1999: (a) Doppler lidar observations and MM5 gridpoint data of the (b) 0000 and (c) 1800 UTC run. Horizontal wind speed is represented as contour lines with an increment of 2 m s−1 and as shaded contours for velocities exceeding 15, 20, and 25 m s−1. Wind barbs indicate horizontal wind direction and speed and are defined as in Fig. 3

  • View in gallery

    Vertical profile of the horizontal wind at the sodar site BRB as function of time for the period 0900–1800 UTC 24 Oct 1999: (a) Doppler sodar observations and MM5 gridpoint data of the (b) 0000 and (c) 1800 UTC run. Horizontal wind speed is represented as contour lines with increments of 1 m s−1. Shaded contours and wind barbs as in Fig. 12

  • View in gallery

    Vertical cross section along NCAR Electra flight leg P1P2 (cf. Fig. 2). (a) SABL backscatter intensity (relative backscatter) at 532 nm between 1523 and 1532 UTC 24 Oct 1999. MM5 (b) 0000 and (c) 1800 UTC run at 1500 UTC 24 Oct 1999: Potential temperature is represented as contour lines with 1-K increment and horizontal wind speed as shaded contours for velocities exceeding 15, 20, and 25 m s−1. Crosses in (a), (b), and (c) mark the AML top height derived from observed backscatter intensities. The white line in (a) represents either topography or the height of the cloud tops. Clouds are labeled with “CL.” The white triangles on the abscissas in (b) and (c) indicate the location of the pass

  • View in gallery

    As in Fig. 14, but for cross section P3P4. Observations collected between 1453 and 1457 UTC 24 Oct 1999

  • View in gallery

    As in Fig. 14, but for cross section P5P6. Observations collected between 1447 and 1449 UTC 24 Oct 1999

  • View in gallery

    As in Fig. 14, but for cross section P7P8. Observations collected between 1441 and 1444 UTC 24 Oct 1999

  • View in gallery

    Vertical cross section along the Wipp Valley, as indicated by the dashed line in Fig. 2. The cross section is oriented at an azimuth angle of 320° (178°) north (south) of the Doppler lidar site GED, which is located at x = 20 km. Orography is displayed as filled black areas. Radial wind velocities (a)–(d) observed with TEA CO2 lidar and (e)–(h) simulated by the MM5 0000 UTC run: contour lines with 5 m s−1 increments; negative values are dashed; shaded contours for velocities above (below) 15, 20, 25 m s−1(−15, −20, −25 m s−1). Arrows in (e)–(h) represent winds along the cross section. (i)–(l) Potential temperature (contour lines, 1-K increment) and horizontal wind speed (shaded for speeds exceeding 15, 20, 25 m s−1) simulated by the MM5 0000 UTC run. Figure rows from top to bottom represent times of approximately 0900, 1200, 1500, and 1700 UTC, respectively

  • View in gallery

    Vertical soundings at Sterzing at 0900 UTC 24 Oct 1999. Observations (solid), MM5 0000 UTC run (dashed), and 1800 UTC run (dotted). (a) Potential temperature (K), (b) wind speed (m s−1), and (c) wind direction (°).

  • View in gallery

    (a) Vertical transect across the Alps, as indicated by the solid line in (b). Data are taken from domain 4 of the MM5 0000 UTC run for 0900 UTC 24 Oct 1999. Potential temperature is represented as contour lines with 1-K increments, horizontal wind speed as gray shaded contours for winds above 10, 15, and 20 m s−1, and arrows represent winds along the cross section in (a). (b) Model orography of domain 4, with elevation contours starting at 200 m MSL with 200-m increments. Indicated are Sterzing (STZ), Innsbruck (IBK), and the Sarntaler Alps (SA). The vertical bar near STZ in (a) shows the location of the stable inversion layer

  • View in gallery

    Horizontal cross section at z = 1500 m MSL for the subdomain indicated by the dashed rectangle in Fig. 2. Orography is gray shaded, starting at 1200 m with 600-m increments. Radial wind velocities (a)–(d) observed with TEA CO2 lidar and (e)–(h) simulated by the MM5 0000 UTC run: contour lines with 5 m s−1 increments; negative values are dashed; thick contour lines for velocities above (below) 25 m s−1 (−25 m s−1). (i)–(l) Horizontal wind speed (contour lines; 5 m s−1 increments; thick lines for values above 25 m s−1) and horizontal wind vectors simulated by the MM5 0000 UTC run. The Doppler lidar site GED is marked as a filled circle. Figure columns from left to right represent times of approximately 0900, 1200, 1500, and 1700 UTC, respectively

  • View in gallery

    As in Fig. 21, but for z = 2000 m MSL

  • View in gallery

    As in Fig. 21, but for z = 2500 m MSL

  • View in gallery

    Error measures of radial wind velocities (m s−1) as a function of time for 24 Oct 1999. (a), (b) Derived from vertical cross sections along the Wipp Valley, as shown in Fig. 18. (c), (d) Derived from horizontal cross sections at z = 1500, 2000, and 2500 m MSL, as shown in Figs. 2123. Rmse of radial velocities (a) and (c), as well as ME of the absolute value of radial velocities (b) and (d). Compared are lidar observations to the 0000 UTC run (circles) and to the 1800 UTC run (squares), as well as the 0000 UTC run to the 1800 UTC run (stars).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 103 103 4
PDF Downloads 52 52 3

South Foehn in the Wipp Valley on 24 October 1999 (MAP IOP 10): Verification of High-Resolution Numerical Simulations with Observations

View More View Less
  • 1 Department of Meteorology and Geophysics, University of Innsbruck, Innsbruck, Austria
  • | 2 Meteorological Institute, University of Munich, Munich, Germany
  • | 3 Department of Meteorology and Geophysics, University of Innsbruck, Innsbruck, Austria
© Get Permissions
Full access

Abstract

A case study of a south foehn windstorm observed across the Brenner Pass in the Wipp Valley near the Austrian–Italian border is presented based on a detailed comparison and verification of high-resolution numerical simulations with observations. The event of 24 through 25 October 1999 was part of the Intensive Observing Period 10 of the Mesoscale Alpine Programme (MAP). The simulations were performed with the fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5). The observations were collected with a ground-based scanning Doppler lidar, an airborne aerosol backscatter lidar, a Doppler sodar, several weather stations, and two radiosounding systems. The study provides a synoptic-scale and mesoscale overview of the event and focuses on a comparison of simulated and observed fields for a 9-h period on 24 October 1999. The quantitative agreement between the numerical results and the observations is discussed in terms of root-mean-square error (rmse) and mean error (ME). Rmse values are high during the early stage of the event (∼7 m s−1), have a transient peak for about 1 h at 1400 UTC, and are minimal at the fully developed foehn stage near 1500 UTC (∼5 m s−1). The discrepancies at the beginning are likely to be related to deficiencies in the model profile on the upstream side of the pass, exhibiting a too low inversion and a too shallow southerly flow. The transient error peak at 1400 UTC is related to a mismatch in the timing of the enhancement of the upper-level winds. Moreover, evidence is found for an overestimation of the mass flux through the lower Brenner gap, which is the narrowest and deepest part of the incision in the main Alpine crest, and a subsequent underestimation of the flow descent into the Wipp Valley on the leeward side of the Brenner Pass. Considering mass continuity, the latter effect is probably a result of the former. Nevertheless, the model captures most of the striking foehn features: Simulated isentropes and aerosol backscatter measurements consistently indicate regions of flow descent, across-valley asymmetries, and hydraulic jump–like features. The across-valley asymmetry of the foehn strength near the Wipp Valley exit is particularly well reproduced by the model. The primary reason for the stronger winds on the eastern sidewall is the asymmetry in the position of the mountain ridges protruding into the valley together with the westward bending of the valley axis.

Corresponding author address: Alexander Gohm, Department of Meteorology and Geophysics, University of Innsbruck, Innrain 52, A-6020 Innsbruck, Austria. Email: alexander.gohm@uibk.ac.at

Abstract

A case study of a south foehn windstorm observed across the Brenner Pass in the Wipp Valley near the Austrian–Italian border is presented based on a detailed comparison and verification of high-resolution numerical simulations with observations. The event of 24 through 25 October 1999 was part of the Intensive Observing Period 10 of the Mesoscale Alpine Programme (MAP). The simulations were performed with the fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5). The observations were collected with a ground-based scanning Doppler lidar, an airborne aerosol backscatter lidar, a Doppler sodar, several weather stations, and two radiosounding systems. The study provides a synoptic-scale and mesoscale overview of the event and focuses on a comparison of simulated and observed fields for a 9-h period on 24 October 1999. The quantitative agreement between the numerical results and the observations is discussed in terms of root-mean-square error (rmse) and mean error (ME). Rmse values are high during the early stage of the event (∼7 m s−1), have a transient peak for about 1 h at 1400 UTC, and are minimal at the fully developed foehn stage near 1500 UTC (∼5 m s−1). The discrepancies at the beginning are likely to be related to deficiencies in the model profile on the upstream side of the pass, exhibiting a too low inversion and a too shallow southerly flow. The transient error peak at 1400 UTC is related to a mismatch in the timing of the enhancement of the upper-level winds. Moreover, evidence is found for an overestimation of the mass flux through the lower Brenner gap, which is the narrowest and deepest part of the incision in the main Alpine crest, and a subsequent underestimation of the flow descent into the Wipp Valley on the leeward side of the Brenner Pass. Considering mass continuity, the latter effect is probably a result of the former. Nevertheless, the model captures most of the striking foehn features: Simulated isentropes and aerosol backscatter measurements consistently indicate regions of flow descent, across-valley asymmetries, and hydraulic jump–like features. The across-valley asymmetry of the foehn strength near the Wipp Valley exit is particularly well reproduced by the model. The primary reason for the stronger winds on the eastern sidewall is the asymmetry in the position of the mountain ridges protruding into the valley together with the westward bending of the valley axis.

Corresponding author address: Alexander Gohm, Department of Meteorology and Geophysics, University of Innsbruck, Innrain 52, A-6020 Innsbruck, Austria. Email: alexander.gohm@uibk.ac.at

1. Introduction

The spatial resolution of the present-day numerical weather prediction models has outpaced routine meteorological networks. A thorough verification of the numerical results therefore requires higher-resolution observations that can in general only be collected in dedicated field campaigns. In contrast to in situ measurements, remote sensing instruments such as radar, lidar, sodar, and optical sensors are able to map atmospheric parameters continuously over a wide domain. These instruments are operated as ground-based and airborne, as well as satellite-based. The present paper uses observations from such instruments in order to assess to what extent the temporal evolution and spatial structure of small-scale orographic flows can be simulated by a state-of-the-art model run in a very high resolution mode.

Among the great number of mesoscale weather phenomena, the investigation of flow over complex terrain has been a prominent field of research. Combined numerical and observational studies have been conducted for downslope windstorms in the lee of elongated mountain ranges (e.g., Clark et al. 1994), for gap flows through topographic contractions (e.g., Jackson and Steyn 1994; Steenburgh et al. 1998; Pan and Smith 1999), for windstorms through gaps and in valleys embedded in mountain ridges (e.g., Colle and Mass 1998; Flamant et al. 2002; Zängl et al. 2003, 2004), for sea level gap flows (e.g., Colle and Mass 2000; Doyle and Bond 2001), and for sea-breeze flows affected by coastal and inland mountains (e.g., Darby et al. 2002a). However, only a few investigations have included a detailed quantitative verification of the numerical results with high-resolution observations (e.g., Colle and Mass 2000). In view of the fact that mesoscale weather forecast models will reach valley-resolving resolutions in the next years, such verification studies are highly important, particularly for weather situations triggering downslope windstorms and complex gravity wave structures.

The 70-day Special Observing Period (SOP) of the Mesoscale Alpine Programme (MAP) in fall 1999 (Bougeault et al. 2001) opened a unique opportunity to study flow over and around mountains in a previously unrivaled scale. One of the primary scientific objectives of the MAP field experiment was to provide datasets for the validation and improvement of high-resolution numerical weather prediction models (Binder and Schär 1996). The Brenner Pass region in the Austrian–Italian Alps was chosen as one of the MAP target areas for the investigation of south foehn (Figs. 1 and 2). The Brenner is a deep incision in the main Alpine crest [∼3 km above mean sea level (MSL)], with a broader (∼15 km) “upper” gap down to 2.1 km MSL and an embedded narrow (∼2 km) “lower” gap down to 1.4 km MSL. The dynamics of foehn winds in the Wipp Valley through and over the Brenner Pass are hypothesized to be similar to that of hydraulic-like gap flows affected by topographic contractions if the cross-mountain flow is “shallow” (Bougeault et al. 1998; Flamant et al. 2002). However, foehn winds rather show characteristics of “deep” downslope windstorms with vertically propagating internal gravity waves if the impinging southerly flow extends significantly above the main Alpine crest (Seibert 1990; Zängl 2003). During the MAP SOP, the Wipp Valley between Sterzing (Italy) and Innsbruck (Austria) was densely instrumented. One of the key instruments was the ground-based scanning Doppler lidar operated by the National Oceanic and Atmospheric Administration's Environmental Technology Laboratory (NOAA/ETL), which explored the three-dimensional structure of the foehn flow. First analyses of its unique dataset are presented in Flamant et al. (2002), Durran et al. (2003), and Mayr et al. (2003a,b). A Doppler sodar provided by the Zentralanstalt für Meteorologie und Geodynamik (ZAMG) continuously measured vertical profiles of the horizontal winds a few kilometers south of the Brenner Pass. The National Center for Atmospheric Research's (NCAR) Electra aircraft carried a down-looking aerosol backscatter lidar to map the aerosol structure within the foehn flow downstream and the height of the “foehn wall” cloud tops upstream of the pass.

Among the strongest MAP foehn events was the Intensive Observing Period (IOP) 10 case, 24 through 25 October 1999. Wind speeds exceeding 30 m s−1 were observed by the Doppler lidar. This work presents a detailed case study of this event based on a comparison and verification of high-resolution numerical simulations with observations. The dataset consists of measurements from the Doppler lidar, the Doppler sodar, and the airborne backscatter lidar, as well as from automatic weather stations and radiosoundings. The main focus lies on the period 0900–1800 UTC 24 October 1999, corresponding to the operation hours of the Doppler lidar, but the analysis of surface data is extended over 18 h to complete the description of the foehn evolution. The paper is organized as follows: Section 2 introduces the numerical model and the observations. The third section gives a general overview of the IOP 10 case, describing the synoptic-scale environment and the mesoscale structure of the foehn event. Section 4 describes the comparison and verification methodology. The results of the comparison are discussed in section 5. A summary and conclusions are given in section 6.

2. Model and measurement description

a. The fifth-generation mesoscale model

The numerical simulations in this case study were conducted with the fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5) version 3.3 (Grell et al. 1995). MM5 is a nonhydrostatic model based on a terrain-following sigma coordinate system. Six interactively nested model domains were used with horizontal mesh sizes of 64.8, 21.6, 7.2, 2.4, 0.8, and 0.267 km, respectively. The coarsest domain (domain 1) was centered at 47.5°N, 11.5°E and covered an area of about 4000 km × 4000 km, which is most of the European continent. Figure 1 shows the topographies of the two innermost model domains (5 and 6). Most of the model data shown in this study were derived from domain 5, which covers the Alpine chain and part of the northern plains in the vicinity of the Inn and Wipp Valleys. Domain 6 covers a small region centered at the Brenner Pass, so as to allow for a proper resolution of the flow through the Brenner gap. The model orography of domains 1–2 (3–6) was interpolated from terrain data with 5′ (30″) resolution. Elevations in domains 5 and 6 along the Inn Valley, Wipp Valley, and their tributaries were manually corrected based on readings from a terrain map with a scale of 1:100 000. Information on land use was obtained from United States Geological Survey (USGS) data with the same horizontal resolution as for orography. Additional snow cover data were not needed, since snow cover was restricted to glaciated areas in October 1999. In the vertical, 39 unevenly spaced full-sigma levels were used, corresponding to 38 half-sigma levels. The lowermost half-sigma level, which will be referred to as surface level, is about 14 m above ground. The vertical distance between the model levels is about 50 m close to the ground and increases up to 800 m near the upper boundary, which is located at 100 hPa.

A cloud microphysical scheme was used that solves prognostic equations for six categories of hydrometeors (the so-called Reisner2 scheme; Reisner et al. 1998). Whereas convection was resolved explicitly in the finer domains 4–6, a cumulus parameterization scheme was applied in the coarser domains 1–3 (Grell 1993). Further, a turbulent kinetic energy (TKE) based parameterization of the planetary boundary layer was used (the Gayno–Seaman scheme; Gayno 1994; Ballard et al. 1991). The radiation scheme accounts for the interactions with moisture and clouds (Grell et al. 1995; Mlawer et al. 1997). It was modified by the second author to include the effects of sloping orography on the flux of direct solar radiation (Garnier and Ohmura 1968). At the upper boundary, an improved version of Klemp and Durran's (1983) radiative boundary condition was applied (Zängl 2002a). The implementation of the horizontal diffusion was modified in order to reduce the numerical errors in narrow valleys (Zängl 2002b). Numerical diffusion was computed truly horizontally for temperature, water vapor, and cloud water mixing ratios, and along sigma coordinates (original MM5 scheme) for the remaining variables.

The initial and boundary conditions were taken from the operational European Centre for Medium-Range Weather Forecasts (ECMWF) analysis on standard pressure levels. They are available at six-hourly half-degree intervals in latitude and longitude. The three lowermost standard pressure levels of the analysis (850, 925, and 1000 hPa) lie below the ECMWF model topography in the Alpine region and therefore contain extrapolated data that are unreliable. To fix these problems, the temperature and humidity fields of the initial conditions were modified on these levels in the Alpine region with radiosonde data. Two model simulations were conducted with different initialization times: 1800 UTC 23 October 1999 and 0000 UTC 24 October 1999, hereafter referred to as the “1800 UTC run” and “0000 UTC run,” respectively. One purpose of using two instead of one single simulation is to elucidate the time advantage—if it exists at all—of a second model run that is initialized 6 h closer to the foehn event than the first run. Further, two slightly different simulations can shed light on the consistency and predictability of foehn patterns.

b. Instruments

From the 35 automatic weather stations (see Mayr et al. 2003b) deployed in the Brenner Pass area during the MAP SOP, we used only a subset of seven stations (STZ, BRE, SAB, STE, ELB, PAT, and IBK; see Fig. 1). The comparison herein is based on their 10-min-average dataset. Profiles from radiosonde ascents at Sterzing (STZ) and Gedeir (GED) (see Fig. 1), up- and downstream of the pass, are incorporated in this study as well.

The NOAA/ETL Transverse Excitation Atmospheric Pressure (TEA) CO2 scanning Doppler lidar was located in the Wipp Valley at GED about halfway between the Brenner Pass and Innsbruck (see Fig. 1). The system is described in Post and Cupp (1990). A more MAP-related description is given by Durran et al. (2003) who investigated the accuracy of the lidar by comparison with aircraft observations for three MAP gap flow events. TEA CO2 has been used in several studies of flow over complex terrain, such as in the Grand Canyon (Banta et al. 1999), on the Colorado Front Range (Neiman et al. 1988; Darby et al. 1999), in the Salt Lake Basin (e.g., Darby et al. 2002b), and in the Wipp Valley (Flamant et al. 2002; Durran et al. 2003). The lidar emits pulses of infrared light at 10.59 μm. The signal is backscattered from aerosols that move with the flow. The Doppler-shifted frequency of the backscattered signal reveals the radial wind velocity component along the direction of the lidar beam. In our definition, winds toward (away from) the lidar have positive (negative) radial velocities. The signal of each beam is split into 300-m range gates. Various types of scans were conducted at constant elevation angle [plan position indicator (PPI)] as well as constant azimuth angle [range–height indicator (RHI)], typically within two sectors centered at azimuth angles upvalley (178°) and downvalley (320°) of the lidar site.

The ZAMG Doppler sodar, a “Phased Array 2” (PA2) model, was located at Brennerbad (BRB), about 4 km south of Brenner Pass (see Fig. 1). Piringer (1994) gives a description of this system. PA2 data were also used in a MAP case study of a strong foehn event in the Rhine Valley (Piringer et al. 2001). The PA2 sodar emits acoustic pulses in the audio band and detects the intensity and Doppler frequency shift of the backscattered echoes to derive among other parameters vertical profiles of the horizontal winds. High ambient background noise generated by heavy traffic over the Brenner Pass as well as by acoustic reflections from the nearby steep sidewalls of the narrow valley limited the performance of the sodar and kept the average maximum range below 500 m AGL. For the MAP SOP the instrument was set to measure winds based on 25-m range gates and a 30-min averaging interval.

The NCAR Scanning Aerosol Backscatter Lidar (SABL) was operated during the MAP SOP on the NCAR Electra aircraft in a nadir-pointing mode at a typical cruising altitude of 5 km MSL. A brief description of the system can be found, for example, in Cohn et al. (1998) and Rogers et al. (1998). SABL emits pulses of light at 532 (green) and 1064 nm (infrared), and detects the signal backscattered from aerosols, air molecules, and hydrometeors. We used data from the green channel only. During MAP the instrument revealed the aerosol structure in the lower troposphere along and across the Wipp Valley with typical resolutions of 7.5 m vertically and ∼120 m horizontally. Aerosol mixed layer (AML) top heights can be estimated from these data. During foehn events the AML top heights frequently coincided with the height of an elevated temperature inversion (e.g., Flamant et al. 2002). The AML top height is located where the second vertical derivative of the backscatter intensity is zero (i.e., at a negative peak of the intensity gradient). In the presence of thick clouds, this location is at the height of the cloud tops.

3. Overview of the event

a. Synoptic-scale environment

South foehn in the Alps is often observed in connection with the approach of a pronounced midtropospheric pressure trough from the Atlantic Ocean. The life cycle of a “classical” type of south foehn contains two stages. During the early stage, foehn is observed as gap flow only near deep indentations in the Alpine chain such as the Brenner or Gotthard Pass. Cross-mountain winds are confined to below the main Alpine crest with westerly winds aloft. This first stage is classified as “shallow” foehn (e.g., Seibert 1985, 1990; Sprenger and Schär 2001; Zängl 2002c). As the pressure trough advances eastward, winds near crest level and above turn southerly and the transition into the “deep” foehn stage occurs. The flow is now perpendicular to the Alps at least throughout the whole lower troposphere. Nevertheless, besides this classical pattern of foehn evolution, the two stages can occur on their own.

In the MAP IOP 10 case (24–25 October 1999), the shallow foehn period preceding the actual event was very short-lived and restricted to the early morning of 24 October. In fact, a well-established large-scale trough had caused deep foehn already on the previous day (23 October). However, foehn winds were interrupted in the afternoon of 23 October because of the passage of a short-wave trough that was embedded in the main trough. Associated with this system was an occluded cold front, which passed the central Alps at about 1800 UTC 23 October. Winds near the summits temporarily turned to west and to northwest but shifted back to southwest in the early morning of 24 October. The ECMWF analysis for 1200 UTC 24 October 1999 in Fig. 3a shows the large-scale pressure trough at 500 hPa located west of the Alps and the short-wave trough, causing the previous foehn interruption, northeast of the Alpine range. In the Brenner Pass target area, the analysis indicates southwesterly winds with 15 m s−1 at 700 hPa (i.e., near crest level). In the Alpine area, the surface pressure distribution in Fig. 3b shows the so-called foehn nose, where isobars are aligned nearly parallel with the Alpine chain, forming a northward-directed pressure gradient. These synoptic features indicate a deep foehn in the Brenner Pass area. The large-scale trough extends farther southward 24 h later, but its axis is nearly at the same place as on the previous day (Fig. 3c). At the surface, the pressure nose is weaker than on the day before (Fig. 3d).

Figures 4a and 4b illustrate the progress of the frontal systems between 24 and 25 October 1999. Surface fronts, drawn as isochrones at a 6-h interval, were manually derived from ECMWF analysis data. During 24 October a partly occluded frontal system approached the Alps from the Atlantic Ocean and was distorted as it impinged on the arc-shaped mountain ridge. By then, the occlusion responsible for the foehn interruption between 23 and 24 October was already east of the Alps (Fig. 4a). The warm front belonging to the new system approached the Alps from the south during the afternoon of 24 October, advecting moisture from the Mediterranean Sea toward the southern Alpine slopes. At the same time the occluded part north of the Alps moved as far as western Germany but soon became inactive. During 25 October (Fig. 4b) the still-active parts of this frontal system, that is, the warm front and the cold front, became nearly stationary since both were almost aligned parallel with the southwesterly upper-level flow. Moisture advection from the south caused rain and drizzle even north of the Alpine ridge. Foehn events with precipitation on the leeward side of the main crest are known as “Dimmerföhn” (e.g., Kuhn 1989).

Unlike the “classical” type of foehn, the IOP 10 event was not terminated by the passage of a pronounced cold front followed by a midtropospheric trough. In fact, the main trough formed a cutoff low southwest of the Iberian Peninsula and did not pass the Alps. In the evening of 25 October the warm front reached the main Alpine crest from the south (see Fig. 4b). At the same time cool air from the northern Alpine foreland filled the Inn and Wipp Valleys [see section 3b(2)]. This marked the end of the foehn event in the valleys.

b. Mesoscale structure

1) Cross-alpine pressure gradient

A characteristic of south foehn or any other similar mountain-induced windstorm is a low-level pressure gradient perpendicular to the mountain ridge in the direction of the main flow. On the mesoscale, this pressure gradient is mainly caused by the forced ascent of the impinging flow, which cools the lower troposphere upstream of the obstacle, and by the forced descent of the flow to the lee, which warms the lower troposphere on the downwind side. This part of the pressure gradient is caused by gravity wave dynamics. At low levels, the gradient can be hydrostatically enhanced because of orographic blocking of cool air upstream of the Alps. In addition, the cross-Alpine pressure gradient may be reinforced on the synoptic scale because of a large-scale depression northwest of the Alps. In the case considered here, a significant part of the pressure gradient turned out to be related to the generation of large-amplitude gravity waves north of the Alpine crest (see sections 5c and 5d). Another important contribution arose from blocked cold air over the Po Valley, particularly on 24 October (see Figs. 3b and 3d). After describing the temporal evolution of the pressure field, we will demonstrate that the former part of the pressure gradient is closely related to the strength of the foehn winds.

Figures 5a and 5b show the meso-α-scale and meso-β-scale (Orlanksi 1975) pressure gradient along the Brenner Pass cross section as function of time for the IOP 10, respectively. The former was calculated from measurements of the SYNOP stations Munich (MUN) and Verona (VER), as well as with ECMWF analysis data using differences in mean sea level pressure observed and analyzed at these two sites. This pressure gradient includes contributions from both gravity waves and low-level blocking. The meso-β-scale gradient was derived from differences in surface pressure reduced to 1000 m MSL measured at the MAP weather stations Patsch (PAT) and Sterzing (STZ) and mainly reflects gravity wave effects. The horizontal distance between MUN, north of the Alps, and VER, south of the Alps, is approximately 300 km, whereas the distance between PAT, north of Brenner, and STZ, south of Brenner, is 36 km (see Fig. 1). Positive values indicate a northward-directed pressure gradient force. The beginning of the foehn event is marked by a continuous increase of the pressure gradient on the meso-α and meso-β scale. The general trend of the meso-α-scale gradient as seen by SYNOP observations is well documented by the ECWMF analysis. Some discrepancies exist, especially during the night and the morning of 25 October. The maximum of nearly 4 Pa km−1 occurred on 24 October between 1200 and 1500 UTC. This corresponds to a mean sea level pressure difference between MUN and VER of approximately 12 hPa. The meso-β-scale gradient, however, with its maximum of 17 Pa km−1 at 1520 UTC 24 October was approximately 4 times larger than the meso-α-scale gradient. This implies that the isobars are packed most tightly near the main Alpine crest and that their spacing increases toward the northern and southern Alpine forelands. The approach of the warm front toward the Po Valley south of the Alps and of the occluded front toward the northern Alpine forelands (see Fig. 4) changed the far up- and downstream air masses. This decreased the pressure gradient on the meso-α scale between the first and the second day of the event but did not affect the gradient on the meso-β scale. The end of the south foehn event was marked with a disappearing pressure gradient on both scales at approximately 2100 UTC 25 October after cold air had filled the downstream valleys.

Figure 5c shows the meso-β-scale pressure gradient from Fig. 5b as function of wind speed measured at three MAP stations and shown in Fig. 6a. The mountain station Sattelberg (SAB), located near Brenner (BRE) at an elevation about halfway between the pass and the main crest, describes the general flow through the Brenner gap. The valley station Ellbögen (ELB), located 21 km north of BRE, illustrates the foehn condition at the surface in the northern half of the Wipp Valley. The station Innsbruck (IBK), located in the Inn Valley, describes the situation at the exit of the Wipp Valley. A distinct correlation exists between the pressure gradient and the strength of the foehn winds. The correlation coefficients are 0.95 for ELB, 0.79 for SAB, and 0.72 for IBK. For the square of the wind speed, which is a measure for the kinetic energy, the correlation coefficients are 0.93 for ELB, 0.81 for SAB, and 0.58 for IBK. This result suggests that the pressure difference between PAT and STZ is a useful indicator for the strength of the foehn winds, especially in the northern part of the Wipp Valley near ELB. The different temporal evolution of the meso-α- and meso-β-scale gradient indicates weaker correlation of the foehn strength with the former gradient than with the latter. A quantitative estimate would have little significance because of the few data points in Fig. 5a.

2) Surface winds and potential temperature

Figure 6 documents the life cycle of the IOP 10 foehn event based on data from the MAP weather stations SAB, ELB, and IBK. As usual, the onset of south foehn in the Wipp Valley was gradual. Between 2130 UTC 23 October and 0630 UTC 24 October, the southerly flow at ELB continuously intensified but remained potentially cooler than at SAB. During this period, the air descending from the upper gap near SAB increasingly mixed with the air flowing through the lower gap over the pass and along the valley floor. The strengthening of surface winds and the increase of flow descent and mixing was related to buoyancy accelerations in connection with mountain lee waves. At the surface, the effect of vertical mixing to the lee of the pass was first observed at 0350 UTC 24 October when ELB became potentially warmer than Brenner (not shown). The subsequent period after 0630 UTC, which lasted until 1730 UTC 25 October, represents the fully developed foehn stage where potential temperatures at ELB were at least equal or greater than at SAB. The onset of foehn in the Inn Valley at IBK is generally abrupt. It occurred at 1100 UTC 24 October when winds suddenly turned from westerly to southerly directions and potential temperature rose to the values of SAB. IOP 10 is counted among of the strongest foehn events of the MAP SOP. The maximum of the sustained winds was 27 m s−1 at SAB, 19 m s−1 at ELB, and 9 m s−1 at IBK (see Fig. 6). Gusts were observed up to 37 m s−1 at SAB and 27 m s−1 at ELB. Because of nocturnal cooling of the air mass in the upper (western) Inn Valley, a foehn break was recorded at IBK between 0250 and 0850 UTC 25 October when the foehn flow temporarily detached from the surface, and a cooler, westerly flow was established. The nocturnal foehn break in the Inn Valley east of IBK was associated with the inflow of cooler air from the lower Inn Valley (Zängl et al. 2003). The end of the south foehn event at ELB and IBK in the afternoon of 25 October was marked by a rapid drop of potential temperature as the Alpine valleys filled with cooler air from the northern Alpine forelands. Nevertheless, southwesterly winds persisted on the Alpine summits and aloft until the following day (see SAB in Fig. 6b).

3) Vertical structure

Figure 7 depicts the vertical structure of the foehn flow in the afternoon of 24 October based on radiosonde data from ascents downstream (GED) and upstream (STZ) of the Brenner Pass. The foehn flow at the surface near GED was potentially warmer than the air mass in the basin of STZ. Descent of the foehn air downstream of Brenner caused warming of a layer, especially between 2.4 and 3.3 km MSL. The acceleration of the flow downstream of the pass formed a pronounced jet in the Wipp Valley (see Fig. 7b) with winds from SSE up to 2.3 km MSL, which partly exceeded 25 m s−1. The two profiles of potential temperature in Fig. 7a have a three-layer structure. In both cases a strongly stable layer separates two less stable regions. This strongly stable layer is located between 3.1 and 4.3 km MSL at STZ and between 2.4 and 2.9 km MSL at GED. Following, for example, Pan and Smith (1999), such a stratification justifies the calculation of the Froude number F = U/gH as defined for a reduced-gravity shallow-water system, where U and H are the mean wind speed and the depth of a layer extending from the surface to the center of the stable layer, and g′ is the reduced gravity calculated from the step of potential temperature across the stable layer. For STZ (GED) we get U = 9 (22) m s−1, H = 2.8 (1.5) km, and g′ = 0.46 (0.15) m s−2, which gives a Froude number of F = 0.2 (1.5). Therefore, in terms of shallow-water dynamics, the flow in the bottom layer is subcritical (F < 1) upstream of the pass but becomes supercritical (F > 1) downstream of the pass. Hydraulic theory would therefore explain the strong winds to the lee of the pass as flow transition from a sub- to a supercritical state. However, the interpretation of the present deep foehn case based on single-layer hydraulics is ambiguous since the cross-barrier flow on top of the stable layer is substantial and not decoupled from the flow in the lower layer. Therefore, the discussion of certain flow features in section 5, such as jumps in the AML top heights, seems to be more appropriate in the light of internal gravity waves forming in a continuously stratified flow.

4) Simulated flow field

The structure of the fully developed foehn as predicted by the 0000 UTC run for 1500 UTC 24 October is shown in Fig. 8. At the surface, the southerly flow accelerates from Brenner to Innsbruck and becomes strongest near the exit of the Wipp Valley on the eastern valley slopes (Fig. 8a). As this jet hits the mountain range north of Innsbruck, it splits up into a westerly and an easterly current. Figure 8b shows that the surface air in the Wipp Valley is nearly adiabatic. It is potentially warmer than the air south of Brenner (up to 4 K), but potentially cooler than the air in the Stubai Valley (up to 3 K) and at Innsbruck (∼1 K). Besides the northern Wipp Valley, additional regions with strong surface winds can be identified. They are located on the leeward side of three mountain ridges with the embedded summits of Patscherkofel (PAK), Nockspitze (NOC), and Serles (SER) (see Fig. 2). These locations also show up as warm areas in the surface potential temperature field (Fig. 8b) and as strong wind areas at an altitude of 2 km MSL (Fig. 8c). The flow acceleration and warming is associated with strong descent of more than 8 m s−1 (Fig. 8d). Farther downstream, this flow descent abruptly changes into an upward motion of comparable magnitude. Vertical motions generally decrease between 2 and 3 km MSL (Figs. 8d and 8f). Horizontal winds at 3 km MSL are strongest above the highest terrain east and west of the Wipp Valley (Fig. 8e).

The location of the strongest winds in our simulation of a deep foehn case differs from a simulation presented by Flamant et al. (2002) for the shallow foehn case on 30 October 1999 (MAP IOP 12). There, the strongest winds occurred near Nösslachjoch (NOS; 2231 m), which is a summit embedded in a mountain ridge protruding into the Wipp Valley from the western side (see Fig. 2). In our simulation, surface wind speeds are approximately 10 m s−1 in the region of NOS, whereas they are much higher (up to 20 m s−1) in the lower eastern part of the Wipp Valley.

4. Comparison methodology

The comparison of each pair of datasets presented in this study is done in a qualitative as well as quantitative fashion. These two datasets, say A and B, can be either a specific simulation and a set of measurements or two different types of simulations. The qualitative comparison is done by investigating the general agreement of temporal and spatial patterns for various parameters. The quantitative comparison and verification is based on the calculation of objective error measures, exclusively for wind data, such as the root-mean-square error (rmse) and the bias or mean error (ME) (e.g., Stanski et al. 1989; Wilks 1995). The rmse, defined as
i1520-0493-132-1-78-e1
indicates the typical magnitude of the error between the two datasets, A and B. In its general form, the rmse is calculated for a field of N spatially distributed grid points and for a period of M time steps. The ME, defined as
i1520-0493-132-1-78-e2
indicates the average direction of the deviation of the first from the second dataset. The ME is negative (positive) if dataset A under- (over-) estimates dataset B on average. In the present work we calculated rmse and ME values for single time steps as well as for the period 0900–1700 or 0900–1800 UTC 24 October 1999. For the calculation of the ME we replaced A and B by their absolute values |A| and |B| if the two datasets were fields of radial wind velocities, since they change sign between up- and downvalley directions. The temporal variability in a specific dataset is estimated by comparing a spatially distributed field with the field of the same dataset but at a later time step. This is similar to an autocorrelation based on a time lag of Δt hours or m time steps. Mathematically this variability can be expressed by Eqs. (1) and (2), whereby Bi,j is replaced by Ai,j+m. To avoid confusion with the definition of the error measures, we replace the terms rmse and ME with rmsv for the root-mean-square variability and with MV for the mean variability, respectively. In our study we chose a specific time lag of Δt = 1 h. The interpolation of the observations and model data onto a common Cartesian grid was mostly done by calculating distance-dependent weighted averages (e.g., Goodin et al. 1979) with a Cressman-type weighting function (Cressman 1959). The model data used for the comparison were taken from the model domain 5 unless it is explicitly noted otherwise.

5. Comparison with observations

In section 5a we will clarify whether the model is able to reproduce the meso-β-scale pressure gradient, a single but basic scalar measure of the foehn flow, as well as several parameters measured at selected weather stations. In section 5b, we will compare vertical profiles of the horizontal winds retrieved from Doppler lidar and sodar observations with model soundings. Section 5c gives a qualitative comparison of AML top heights calculated from lidar observations with the simulated distribution of potential temperature. In section 5d we will examine the qualitative and quantitative agreement of observed and modeled radial wind velocities.

a. Surface data

Figures 911 show a comparison of observations from selected weather stations to model data of the 00 and 18 UTC runs from surface grid points closest to the measuring sites. All model data were taken from domain 5 except for the sites BRE and SAB, where data from domain 6 were used. Because of the smoothed model orography, the elevations of the surface grid points and the heights of the weather stations differ between 10 and 150 m. Both simulations are able to reproduce the magnitude of the observed meso-β-scale pressure gradient along the Brenner Pass cross section (Fig. 9a) with values between 7 and 17 Pa km−1. The maximum in the simulations occurs about 3 h earlier than in the observations. In contrast to the observed gradient, the simulated one starts to weaken after 1800 UTC. In agreement with Fig. 5c, which was derived from observations, the correlation of the simulated pressure gradient is best with the simulated wind speed at ELB and less with the speed at SAB and IBK (Figs. 9b and 9c). The correlation coefficients of the 0000 (1800) UTC run are 0.86 (0.87) for ELB, 0.69 (0.77) for SAB, and 0.64 (0.09) for IBK. The model captures the acceleration of the surface flow along the Wipp Valley between STZ and ELB and the strong winds exceeding 20 m s−1 in the upper part of the Brenner gap at SAB (Fig. 10). It should be noted that the fine mesh of domain 6 (0.267 km) is necessary to get the right winds on individual mountain peaks such as SAB. Winds at SAB are 7 to 10 m s−1 (∼50%) weaker in domain 5 than in domain 6. Figure 10d shows a major discrepancy in the simulated and observed evolution of the flow in the middle of the Wipp Valley. The two simulations initially overestimate the strength of the winds at STE by up to 10 m s−1; however, they approach the observed values after 1200 UTC. This error is likely to be a direct consequence of a too low temperature inversion in the upstream model profile (see section 5d). The discrepancies of modeled and observed wind speeds at the station IBK might be attributed to the special location of this site near the splitting point of the foehn current where the high spatial wind variability (see Fig. 8a) makes a direct comparison to nearby model grid points difficult. Nevertheless, the onset of foehn at IBK marked by a rapid warming at 1100 UTC 24 October is well represented in both simulations (Fig. 11f). However, the nocturnal foehn break marked by a temperature drop occurs 4–5 h too early in the model (cf. Figs. 11f and 6c). This large discrepancy of the beginning of the foehn break is limited to the immediate vicinity of IBK and is much less pronounced east and west of IBK (see Zängl et al. 2003b). Both simulations capture the temporal evolution and magnitude of potential temperature at STZ, BRE, and SAB (Figs. 11a–11c) before 1700 UTC. Afterward, the model becomes too warm upstream and near the pass, which explains the drop of the pressure gradient below the observed value in Fig. 9a. Downstream of the pass the model is generally too cold by about 2 K (Figs. 11d and 11f). We believe that this is partly due to an overestimation of the mass flux through the lower Brenner gap, as indicated by sodar measurements (see section 5b). Consequently, the descent of potentially warm air passing the upper gap is underestimated, as indicated by backscatter lidar measurements (see section 5c). Further, the model might underestimate the amount of turbulent vertical mixing of stably stratified air in the narrow valley, as its boundary layer parameterization scheme was developed for flat terrain.

b. Vertical profiles of the horizontal wind

1) Downstream: Doppler lidar

At hourly intervals, two individual vertical profiles of the horizontal wind were derived by applying the velocity–azimuth display (VAD) technique (Browning and Wexler 1968) onto full 360° conical scans that were executed at elevation angles of 20° and 25°. The subsequent averaging of these two profiles to a single one filled gaps at regions where one of the two profiles had missing data and provided a somewhat smoother profile. For the comparison of the lidar profiles with model data, we retrieved model profiles from the grid point nearest to the lidar site. We interpolated the data of the lidar profiles first vertically to the model levels and then temporally to the full hours between 0900 and 1800 UTC 24 October 1999. Figures 12a–12c show the corresponding observed and simulated profiles. Gaps with missing lidar observations are apparent between an elevation of about 3.5 and 5.5 km MSL, where the backscatter signal is weak because of the low aerosol concentration. Because of backscattering from ice clouds between 5.5 and 8 km MSL, the received signal is again strong enough to derive wind data. The overall vertical wind structure measured by the lidar is captured fairly well in both simulations. Below the main Alpine crest (∼3 km MSL) the winds are generally S to SSE because of channelling of the flow by the Wipp Valley. Near the crest level and above, winds turn to SW and then W near the tropopause. With the evolution of the foehn between 0900 and 1800 UTC the observations as well as the two model runs show a general increase of the wind speeds, especially in the lower troposphere, and a change of wind direction from SW to S near crest level. Error measures for the whole 9-h comparison period are derived in Table 1 for profile levels where lidar data points are available (lidar range) and for all model levels (model range), respectively. When comparing the two simulations with observations, the overall rmse is in both cases approximately 5 m s−1 for wind speed and less than 15° for wind direction. Both model runs slightly underestimate the wind strength. The bias is small for the 1800 UTC run (−1.0 m s−1) but nearly twice as large for the 0000 UTC run (−1.7 m s−1). A positive bias in the wind direction between +5° and +7° indicates a stronger westerly wind component in the simulations compared to the observations. The two simulations compared with each other have an rmse only slightly smaller (∼4 m s−1) than when compared with observations. Negative ME values (∼−1 m s−1) in table columns 4 and 5 indicate that winds are slightly weaker in the 0000 UTC run than in the 1800 UTC run. The aforementioned bias in wind direction between model and observations has a time dependency that is not obvious from Table 1. For the layer below 4 km MSL, the ME is strongest at 0900 UTC, with +14° in both simulations, and decreases later on. The tendency of the model to have a directional bias at an early foehn stage is discussed in more detail in section 5d.

2) Pass: Doppler sodar

For the comparison of PA2 Doppler sodar observations with vertical soundings of the two simulations, the sodar data was interpolated onto model levels in the same manner as for the Doppler lidar data. The model profiles were retrieved from model domain 6. Figure 13 shows the temporal evolution of the corresponding wind profiles. Because of the limited vertical depth of the sodar measurements, the profiles depict the strongly channeled flow through the Brenner gap only below an altitude of about 2 km MSL. The wind direction is basically SW, which coincides with the orientation of the valley axis at the location BRB (see Fig. 1). With increasing wind speeds during the day, the vertical range of the sodar drops from about 2 to 1.7 km MSL, presumably a consequence of beam wandering in high-wind conditions. Using all profile levels where sodar data points are available (sodar range), the average of the wind speed is 10.4 m s−1 for the observations and 11.9 (12.0) m s−1 for the 0000 UTC (1800 UTC) simulation. This causes a total ME of approximately 1.5 m s−1 (cf. Table 2), which is an overestimation of the strength of the flow through the gap of approximately 15%. The cross-sectional area of the lower gap, that is, below an altitude of 2100 m MSL, is approximately 1.3 km2 in a 30-m-resolution topography. It is overestimated in domain 6 by 30% because of smoothing of the model orography. These two errors result in a significant overestimation of the mass flux through the lower gap by 50%. A closer inspection of the profiles reveals discrepancies in the exact temporal evolution. The model winds within the gap are everywhere too strong in the beginning and too strong in the upper part at the end. For both simulations the overall rmse for wind speed between observations and model is approximately 6 m s−1 and is therefore only slightly higher than the values derived from lidar measurements. The rmse for wind direction, however, is almost twice as large (∼20°). The errors between the two simulations are generally smaller than between simulated and observed profiles.

c. Backscatter intensities versus potential temperature

The analysis of airborne SABL observations presented herein is based on four selected flight legs (Fig. 2) that were flown at a nearly constant altitude of 4.9 km MSL between 1440 and 1532 UTC 24 October 1999. These measurements are compared with MM5 model fields for 1500 UTC. Since the 1-Hz dataset of relative backscatter intensities provided by the NCAR Atmospheric Technology Division (ATD) was partly substantially contaminated with noise, we calculated moving averages over five profiles and 15 range gates (=112.5 m) to smooth the backscatter fields and to increase the signal-to-noise ratio. From the averaged profiles we derived vertical intensity gradients and AML top height estimates. The AML was hardly one homogeneous layer with nearly constant backscatter intensity, but rather contained several individual layers. Therefore, for a certain profile, we calculated not only the absolute minimum of the intensity gradient but also the secondary minima. In most cases the absolute minimum was the best estimate for the AML top height. However, in several cases the absolute minimum marked the height of a shallow layer inside the overall AML and a secondary minimum marked the AML top.

1) Along-valley

In Fig. 14 the along-valley structure of observed backscatter intensities is compared with simulated fields of potential temperature and horizontal wind speed. This transect (P1P2) is aligned along the eastern part of the Wipp Valley at a SSE to NNW direction (see Fig. 2). The general structure of the backscatter intensity field in Fig. 14a has remarkable similarity to the lidar observations presented by Flamant et al. (2002, their Fig. 6) for the MAP IOP 12 event. In contrast to their case, a stratiform cloud layer exists in IOP 10 with its top located at about 3.5 km MSL upstream of Brenner. The strongest flow descent occurs north of 47.13°N, which is near the middle of the Wipp Valley, and is indicated by descending isentropes and decreasing AML top heights. This subsidence causes the cloud layer to dissolve and the flow to accelerate. The 0000 and 1800 UTC runs show very similar structures. A region of strong winds exceeding 25 m s−1 exists on the lee side of the western slope of the Patscherkofel (PAK) mountain (near 47.23°N) and a steeply amplified gravity wave has been established above the Inn Valley. The latter is a hydraulic jump–like feature whose position is affected by the mountain range north of Innsbruck, where the flow is forced to rise with updrafts exceeding 5 m s−1. The existence of the simulated jump above the Inn Valley is supported by the reappearance of clouds and the rise of the AML top heights (see Fig. 14a). Generally, the observed AML and cloud-top heights do not exactly follow the simulated isentropes. This might indicate an underestimation of the flow descent by the model. However, this discrepancy might also arise from the fact that the interpretation of our cloud and AML top heights as material streamlines is strictly limited. Effects like turbulent mixing at the top of the jet flow, such as to the lee of PAK, can break up and destroy the sharp boundary between the AML and the ambient atmosphere, which makes the identification of a distinct AML top height difficult or even unjustifiable.

2) Across-valley

Figure 15 shows the southernmost across-valley transect P3P4, which crosses the Gschnitz Valley and Wipp Valley north of the two mountain peaks Nösslachjoch (NOS) and Bentlstein (BEN) (see Fig. 2). The AML top heights as well as the modeled isentropes have a depression near the center of the Wipp Valley (∼11.45°E) caused by the descent of the foehn flow along the Wipp Valley (i.e., normal to the transect). It is likely that the difference in the strength of the observed and simulated depression is again an indication for the underestimation of the flow descent by the model. The transect P5P6 in Fig. 16 crosses the Stubai Valley, the lee of the Serles (SER) mountain, and the Wipp Valley. Contrary to the previous transect, the AML top and the isentropes are lower on the western than on the eastern side of the Wipp Valley. This asymmetry is caused because the flow on the western side descends as it passes the SER mountain. Winds are therefore much stronger to the lee and west of SER than in the Wipp Valley. Discrepancies between isentropes and AML top heights occur especially in the western part of the transect. There, no distinct AML structure is observed, which leads to large scattering in the AML height calculation. This might be caused by strong turbulent mixing above the descending jet, which could break up the boundary between AML and ambient atmosphere. Some evidence for the existence of turbulence in the model is the supercritically steepened gravity wave above the Stubai Valley in Figs. 16b and 16c. Figure 17 shows the transect P7P8. It crosses the peak of Nockspitze (NOC) and the exit of the Wipp Valley, and continues in the Inn Valley along the northern slope of PAK. A pronounced feature in the two simulations is the near-surface jet-flow with velocities exceeding 25 m s−1, which emanates from the Wipp Valley into the Inn Valley. The sloping of the isentropes across the Wipp Valley is supported by the lidar observations. As will be shown in the next section, this asymmetry is related to stronger winds on the eastern than on the western sidewalls.

d. Spatial distribution of radial wind velocities

The TEA CO2 Doppler lidar was operated continuously between 0800 and 1800 UTC 24 October 1999. In a first step, we generated interpolated fields of radial velocities at full hours between 0900 and 1700 UTC on a Cartesian grid using all available RHI and PPI scans within a time window of approximately 1 h centered at this full hour. Using several scans instead of a single one increases the data coverage and smoothes small-scale transient features. For 1800 UTC the data coverage in the southern Wipp Valley was too sparse to generate lidar cross sections. In a second step, radial wind velocities were derived from simulated wind fields. For the calculation of error measures, these radial velocities were then interpolated onto the same Cartesian grid as for the lidar data. In the following we will only show fields of the 0000 UTC run, although we calculated error measures for both simulations.

1) Vertical cross sections

The vertical cross section shown in Fig. 18 is essentially parallel to the Wipp Valley axis. Its direction is 178° to the south and 320° to the north of the Doppler lidar site (see Fig. 2). The temporal evolution of the flow is illustrated from the top to the bottom figure row. The flow in the valley is from the south (left) to the north (right). Radial wind velocity changes its sign at the lidar site, which is located on the abscissa at 20 km. In the fully developed foehn stage near 1500 UTC, the model qualitatively reproduces a layer of strong winds that descends near the lidar site and forms pronounced near-surface jet in the northern part of the valley. The simulated wind vectors support the idea that the descent of the jet actually corresponds to downward motions. Apart from the fully developed stage, a clear mismatch exists in the temporal evolution of the predicted and observed flow field. In the beginning, around 0900 UTC, the model underestimates the strength of the winds in the valley, especially at higher levels. The layer of strong winds is too thin and too confined to the ground. At later times the model increases the depth of the jet layer north of the lidar site to realistic values but finally makes the jet too deep and too strong. The source for the discrepancy in the early stage is the difference in the location of the strongly stable layer and the depth of the southerly flow in the upstream profile. At 0900 UTC, the stable layer in both model profiles is located approximately 1 km too low compared to the radiosounding of STZ (Fig. 19), and thus the southerly flow is too shallow. Between 0900 and 1500 UTC the stable layer in the model is gradually lifted to realistic heights, and, in accordance with observations, the stability of this layer decreases (cf. Figs. 18i–k, near left boundary). The shallow foehn period in the simulations lasted until around noon, whereas the observed one ended before 0900 UTC. To further elucidate the reasons for this deficiency, Fig. 20 shows the flow field in the next coarser domain (D4). Evidently, the inversion forms dynamically as the flow approaches Sterzing, and thus the main Alpine crest, because of the influence of the Sarntaler Alps south of Sterzing. Since the lack of an inversion near the southern edge of the Alps is in accordance with radiosoundings (not shown), it follows that the physical process forming the inversion over the Alps is inaccurately represented in the model in a way that places the inversion too low. One possible factor is that the horizontal resolution of D4 (2.4 km) is not high enough to realistically represent this process. It is also conceivable that small errors in the upstream model wind and divergence field, which are not detectable by comparing model profiles with radiosoundings, might contribute to the error in the formation process. It is noteworthy that the temporal evolutions of the flow structure in the 1800 and 0000 UTC runs are very similar, and therefore the discrepancies with observations are also similar. The major difference between the two simulations is the depth of the jet layer downstream of the lidar site, which is about 400–600 m deeper around 1500 UTC in the 0000 UTC run than in the 1800 UTC run.

2) Horizontal cross sections

The horizontal cross sections at the vertical levels z = 1500, 2000, and 2500 m MSL shown in Figs. 2123 illustrate the across-valley variability of the foehn winds. The subdomain of these transects is depicted as a dashed rectangle in Fig. 2. At the lowest level the flow is basically channeled along the Wipp Valley axis. Model and observations both show the acceleration of the flow along the valley at all three levels. A further consensus is the across-valley asymmetry in the foehn strength north of the lidar site with stronger winds on the eastern sidewall of the Wipp Valley. This feature was observed also in other MAP IOP events (e.g., Flamant et al. 2002; Durran et al. 2003; Mayr et al. 2003b). At z = 2000 m MSL simulated wind vectors show that, as the valley bends westward north of the lidar site, the flow continues straight. It impinges on the western slope of the PAK mountain (Figs. 22i–22l) and accelerates, similar to a hydraulic flow passing a ridge. On the western valley side this acceleration has already occurred farther south, that is, to the lee of the SER mountain. This asymmetric flow acceleration leads to isotachs that are aligned from southwest to northeast (see especially Figs. 22k and 22l) and is the reason for the sloping of the isentropes and the AML top heights in Figs. 16 and 17. Based on this result, we believe that the main reason for the across-valley asymmetry in the flow strength is the asymmetry of the topography, that is, the displacement of the ridges protruding into the valley from the eastern and the western side together with the bending of the valley axis. The influence of differences in the shape of the two gap sidewalls on the cross-gap velocity distribution was also shown by Colle and Mass (2000) for a gap flow event in the Strait of Juan de Fuca. In this case, the coastal foothills of the Olympic Mountains, which form one sidewall of the strait, were responsible for strong lee wave response and wake formation, which led to a velocity difference across the strait near its exit. In contrast to our view, Flamant et al. (2002) suggest that the primary mechanism for the flow “deflection” to the eastern sidewall of the Wipp Valley is the interaction between the mountain range extending from SER southwestward and the southwesterly synoptic flow. Because this asymmetric behavior is also found for an idealized deep foehn with no distinct westerly component in the background winds [see Figs. 9a and 8c of Zängl (2003)], we suppose that at least for deep foehn cases the westerly wind component is not responsible for this asymmetric behavior of the flow in the valley. As mentioned earlier, the model simulates shallow foehn at the beginning, which is supported by southwesterly wind vectors at z = 2500 m MSL between 0900 and 1200 UTC (Figs. 23i and 23j). In contrast, observations show southerly flow along the valley with no distinct westerly component, as indicated by the west–east alignment of the zero-radial-velocity contour line at the lidar site (Figs. 23a and 23b).

3) Error and variability measures

In the following, the agreement of simulations and observations is examined in terms of rmse and ME for fields of radial wind velocities. Figures 24a and 24b show the time dependency of the error based on hourly data retrieved from vertical transects, as shown in Fig. 18. Figures 24c and 24d display the error evolution calculated from horizontal transects at three levels, as shown in Figs. 2123. The above-mentioned discrepancy between observations and simulations near 0900 UTC causes rather high rmse values of 7 m s−1 and negative ME values between −2 and −4 m s−1. A high rmse and a positive ME at 1700 UTC indicate that the model overestimates the strength of the flow at the end of the comparison period. The agreement between simulations and observations is best near 1500 UTC where the rmse is minimal (∼5 m s−1) and the ME is nearly zero. Averaged over the period 0900–1700 UTC, the rmse for the comparison of the 0000 and 1800 UTC runs with observations is in both cases approximately 6 m s−1. The quality of the two simulations is therefore very similar, and no distinct effect due to the different initialization time becomes apparent. The mean rmse for the model–model comparison is about 3 m s−1. Based on the same dataset we calculated hourly variability measures as described in section 4. On average, the variability in the observed and modeled fields is very similar, with mean rmsv of 3.3 m s−1 for the lidar fields and 2.7 m s−1 for both model datasets. At 1300 UTC, the rmsv (MV) has a distinct maximum of 5.6 (+4.2) m s−1 for the vertical transects of observed radial velocities because of a general increase of the wind strength between 1300 and 1400 UTC, especially at upper levels. For the two simulations, however, the rmsv and the MV is not maximal at this time, since the model does not capture the sudden flow strengthening. Consequently, the simulated flow at 1400 UTC is too weak compared to the observed one. This causes a maximum of the rmse and a negative peak of the ME at 1400 UTC, especially for the vertical transects (Figs. 24a and 24b).

6. Summary and conclusions

A case study is presented of a south foehn windstorm that occurred in the Brenner Pass region on 24 through 25 October 1999 and was subject to the Intensive Observing Period 10 of the Mesoscale Alpine Programme. The primary aim is to compare and verify high-resolution numerical simulations with remote sensing and in situ observations in order to examine the capability of a state-of-the-art model to simulate flow over complex terrain and to study the dynamics of foehn winds in an Alpine valley.

IOP 10 is counted among the strongest windstorm cases of the MAP Special Observing Period. The Doppler lidar observed wind velocities exceeding 30 m s−1. This case is classified as Dimmerföhn because of the pronounced cloud cover and its associated drizzle to the lee of the main Alpine ridge. Furthermore, it is classified as a deep foehn that was preceded by a very short-lived shallow foehn phase. The maximum observed pressure difference between the northern and southern Alpine foreland was 12 hPa, which corresponds to a meso-α-scale pressure gradient of 4 Pa km−1. The observed meso-β-scale gradient was approximately 4 times larger. A high correlation (0.95) was found between the meso-β-scale pressure gradient and the foehn strength in the northern part of the Wipp Valley. The Froude number analysis indicated subcritical flow at Sterzing (upstream) and supercritical flow at Gedeir (downstream of the pass). The simulations with the MM5 identified several regions with strong surface winds: the Wipp Valley exit and the lee side of the mountains Patscherkofel, Nockspitze, and Serles. In contrast to the shallow foehn case discussed by Flamant et al. (2002), the present deep foehn event had weaker winds at Nösslachjoch, a summit on the western side of the upper Wipp Valley, than at the Wipp Valley exit.

The 6-h difference of the initialization time between the two simulations did not show up as an advantage for the most recent simulation (the 0000 UTC run). The spatial structure and the temporal evolution of the foehn flow was very similar in both model runs. This also indicates that a spinup phase of 6 h was sufficient to approach the flow pattern of the earlier initialized simulation.

The numerical model was capable of reproducing the magnitude of the meso-β-scale pressure gradient. The breaktrough of the foehn winds in the Inn Valley was modeled at the right time. Simulated surface winds at most of the selected weather stations had realistic magnitudes. Simulated isentropes and aerosol backscatter measurements showed qualitative agreements. Both revealed regions of along-valley flow descent, across-valley flow asymmetries, and hydraulic jump–like features. The across-valley asymmetry in the flow acceleration near the Wipp Valley exit, which led to stronger winds on the eastern than on the western sidewall, was well represented in the simulations. In contrast to an earlier explanation of this feature (Flamant et al. 2002), it is supposed that its reason is the across-valley asymmetry in the alignment of the protruding mountain ridges and the bending of the valley axis. It is believed that this flow asymmetry would occur even with near-crest-level winds having no westerly components. In the earlier explanation, however, this seems to be a necessary ingredient.

The main discrepancy between the simulations and the observations was attributed to the fact that the model predicted a shallow foehn phase that was not observed within the comparison period 0900–1800 UTC. This difference was due to an unrealistic upstream model profile around 0900 UTC where the cross-mountain winds were too shallow. The reason was not an error in the far-upstream condition south of the Alps, where the agreement with observations was good, but an inaccuracy in the physical process forming a stable inversion layer as the flow approached the main Alpine crest from the southern foothills. This process placed the inversion immediately south of the main Alpine crest too low compared with observations. The best agreement between model and observations in terms of rmse was found for the fully evolved stage near 1500 UTC. One hour prior, however, the rmse was large since the model did not represent the sudden foehn strengthening at upper levels between 1300 and 1400 UTC. At the end of the comparison period, the model tended to overestimate the strength and depth of the valley jet. Further, discrepancies were observed between measured AML top heights and simulated heights of isentropes. The reason might be twofold. It is likely that the model underestimated the descent of the foehn flow to the lee. On the other hand, the derived AML top heights might not always coincide with stream surfaces, especially in regions with strong turbulence. A negative bias of surface potential temperature north of the pass is suspected to be caused by an overestimation of the mass flux through the lower gap, by an underestimation of the foehn descent, and by an underestimation of the turbulent vertical mixing in the boundary layer of the narrow valley. Considering mass continuity, the underestimated descent is probably a result of the overestimated flow through the pass. This result shows the importance of a proper resolution of the model topography to capture the mass flux through narrow gaps and stresses the need for improvements of boundary layer parameterization schemes to include the effect of the surrounding topography.

Acknowledgments

The first author is deeply indebted to Robert Banta, Lisa Darby, and Michael Hardesty from NOAA/ETL for the scientific support to acquire the skills of the Doppler lidar data analysis. We would like to thank Craig Walther from NCAR/ATD for preparing and providing the aerosol backscatter lidar data. Kathrin Baumann from ZAMG is acknowledged for providing the Doppler sodar instrument. We are indebted to Stephen Mobbs and Samantha Arnold from the University of Leeds for deploying the bulk of the surface weather stations in the Wipp Valley. Dale Durran from the University of Washington is acknowledged for his contribution to additional radiosoundings in the Brenner Pass region and for directing the Electra flights. This research was supported by the Austrian Science Fund FWF under Grants P13489 and P15077.

REFERENCES

  • Ballard, S. P., , B. W. Golding, , and R. N. B. Smith, 1991: Mesoscale model experimental forecasts of the Haar of northeast Scotland. Mon. Wea. Rev., 119 , 21072123.

    • Search Google Scholar
    • Export Citation
  • Banta, R. M., , L. S. Darby, , P. Kaufmann, , D. H. Levinson, , and C-J. Zhu, 1999: Wind-flow patterns in the Grand Canyon as revealed by Doppler lidar. J. Appl. Meteor., 38 , 10691083.

    • Search Google Scholar
    • Export Citation
  • Binder, P., , and C. Schär, Eds.,. 1996: MAP design proposal. MeteoSwiss, 75 pp. [Available from the MAP Programme Office, MeteoSwiss, CH-8044, Zurich, Switzerland.].

    • Search Google Scholar
    • Export Citation
  • Bougeault, P., , P. Binder, , and J. Kuettner, Eds.,. 1998: MAP science plan. MeteoSwiss, 64 pp. [Available from MAP Programme Office, MeteoSwiss, CH-8044, Zurich, Switzerland.].

    • Search Google Scholar
    • Export Citation
  • Bougeault, P., and Coauthors, 2001: The MAP Special Observing Period. Bull. Amer. Meteor. Soc., 82 , 433462.

  • Browning, K. A., , and R. Wexler, 1968: The determination of kinematic properties of a wind field using Doppler radar. J. Appl. Meteor., 7 , 105113.

    • Search Google Scholar
    • Export Citation
  • Clark, T. L., , W. D. Hall, , and R. M. Banta, 1994: Two-and three-dimensional simulations of the 9 January 1989 severe Boulder windstorm: Comparison with observations. J. Atmos. Sci., 51 , 23172343.

    • Search Google Scholar
    • Export Citation
  • Cohn, S. A., , S. D. Mayor, , C. J. Grund, , T. M. Weckwerth, , and C. Senff, 1998: The Lidars in Flat Terrain (LIFT) experiment. Bull. Amer. Meteor. Soc., 79 , 13291343.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , and C. F. Mass, 1998: Windstorms along the western side of the Washington Cascade Mountains. Part I: A high-resolution observational and modeling study of the 12 February 1995 event. Mon. Wea. Rev., 126 , 2852.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , and C. F. Mass, 2000: High-resolution observations and numerical simulations of easterly gap flow through the Strait of Juan de Fuca on 9–10 December 1995. Mon. Wea. Rev., 128 , 23982422.

    • Search Google Scholar
    • Export Citation
  • Cressman, G. P., 1959: An operational objective analysis system. Mon. Wea. Rev., 87 , 367374.

  • Darby, L. S., , W. D. Neff, , and R. M. Banta, 1999: Multiscale analysis of a meso-β frontal passage in the complex terrain of the Colorado Front Range. Mon. Wea. Rev., 127 , 20622081.

    • Search Google Scholar
    • Export Citation
  • Darby, L. S., , R. M. Banta, , and R. A. Pielke, 2002a: Comparisons between mesoscale model terrain sensitivity studies and Doppler lidar measurements of the sea breeze at Monterey Bay. Mon. Wea. Rev., 130 , 28132838.

    • Search Google Scholar
    • Export Citation
  • Darby, L. S., , K. J. Allwine, , and R. M. Banta, 2002b: Relationship between tracer behavior in downtown Salt Lake City and basin-scale wind flow. Preprints, 10th Conf. on Mountain Meteorology, Park City, UT, Amer. Meteor. Soc., 12–15.

    • Search Google Scholar
    • Export Citation
  • Doyle, J. D., , and N. A. Bond, 2001: Research aircraft observations and numerical simulations of a warm front approaching Vancouver Island. Mon. Wea. Rev., 129 , 978998.

    • Search Google Scholar
    • Export Citation
  • Durran, D. R., , T. Maric, , R. M. Banta, , L. S. Darby, , and R. M. Hardesty, 2003: A comparison of ground-based Doppler lidar and airborne in situ wind observations above complex terrain. Quart. J. Roy. Meteor. Soc., 129 , 693713.

    • Search Google Scholar
    • Export Citation
  • Flamant, C., and Coauthors, 2002: Gap flow in an Alpine valley during a shallow south föhn event: Observations, numerical simulations and hydraulic analogue. Quart. J. Roy. Meteor. Soc., 128 , 11731210.

    • Search Google Scholar
    • Export Citation
  • Garnier, B. J., , and A. Ohmura, 1968: A method of calculating the direct shortwave radiation income of slopes. J. Appl. Meteor., 7 , 796800.

    • Search Google Scholar
    • Export Citation
  • Gayno, G. A., 1994: Development of a higher-order, fog-producing boundary layer model suitable for use in numerical weather prediction. M.S. thesis, Dept. of Meteorology, The Pennsylvania State University, 104 pp.

    • Search Google Scholar
    • Export Citation
  • Goodin, W. R., , G. J. McRae, , and J. H. Seinfeld, 1979: A comparison of interpolation methods for sparse data: Application to wind and concentration fields. J. Appl. Meteor., 18 , 761771.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations. Mon. Wea. Rev., 121 , 764787.

  • Grell, G. A., , J. Dudhia, , and D. R. Stauffer, 1995: A description of the fifth-generation Penn State/NCAR mesoscale model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 122 pp.

    • Search Google Scholar
    • Export Citation
  • Jackson, P. L., , and D. G. Steyn, 1994: Gap winds in a fjord. Part I: Observations and numerical simulation. Mon. Wea. Rev., 122 , 26452665.

    • Search Google Scholar
    • Export Citation
  • Klemp, J. B., , and D. R. Durran, 1983: An upper boundary condition permitting internal gravity wave radiation in numerical mesoscale models. Mon. Wea. Rev., 111 , 430444.

    • Search Google Scholar
    • Export Citation
  • Kuhn, M., 1989: Föhnstudien. Wissenschaftliche Buchgesellschaft, 504 pp.

  • Mayr, G. J., , I. Vergeiner, , A. Gohm, , J. Vergeiner, , and R. Mayr, 2003a: Gap flow—An overview and preliminary results. Österr. Beitr. Meteor. Geophys., 29 , 1554.

    • Search Google Scholar
    • Export Citation
  • Mayr, G. J., and Coauthors, 2003b: Gap flow measurements during the Mesoscale Alpine Programme. Meteor. Atmos. Phys., in press.

  • Mlawer, E. J., , S. J. Taubman, , P. D. Brown, , M. J. Iacono, , and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102 , 1666316682.

    • Search Google Scholar
    • Export Citation
  • Neiman, P. J., , R. M. Hardesty, , M. A. Shapiro, , and R. E. Cupp, 1988: Doppler lidar observations of a downslope windstorm. Mon. Wea. Rev., 116 , 22652275.

    • Search Google Scholar
    • Export Citation
  • Orlanski, I., 1975: A rational subdivision of scales for atmospheric processes. Bull. Amer. Meteor. Soc., 56 , 527530.

  • Pan, F., , and R. B. Smith, 1999: Gap winds and wakes: SAR observations and numerical simulations. J. Atmos. Sci., 56 , 905923.

  • Piringer, M., 1994: Selected results of a sodar intercomparison experiment. Meteor. Z., 3 , 132137.

  • Piringer, M., , K. Baumann, , U. Pechinger, , and S. Vogt, 2001: Meteorological and ozone sounding experience during a strong foehn event—A MAP case study. Meteor. Z., 10 , 445455.

    • Search Google Scholar
    • Export Citation
  • Post, M. J., , and R. E. Cupp, 1990: Optimizing a pulsed Doppler lidar. Appl. Opt., 29 , 41454158.

  • Reisner, J., , R. M. Rasmussen, , and R. T. Bruintjes, 1998: Explicit forecasting of supercooled liquid water in winter storms using the MM5 mesoscale model. Quart. J. Roy. Meteor. Soc., 124 , 10711107.

    • Search Google Scholar
    • Export Citation
  • Rogers, D. P., and Coauthors, 1998: Highlights of Coastal Waves 1996. Bull. Amer. Meteor. Soc., 79 , 13071326.

  • Seibert, P., 1985: Fallstudien und statistische Untersuchungen zum Südföhn im Raum Tirol. Ph.D. thesis, University of Innsbruck, 368 pp.

    • Search Google Scholar
    • Export Citation
  • Seibert, P., 1990: South foehn studies since the ALPEX experiment. Meteor. Atmos. Phys., 43 , 91103.

  • Sprenger, M., , and C. Schär, 2001: Rotational aspects of stratified gap flows and shallow föhn. Quart. J. Roy. Meteor. Soc., 127 , 161187.

    • Search Google Scholar
    • Export Citation
  • Stanski, H. R., , L. J. Wilson, , and W. R. Burrows, 1989: Survey of common verification methods in meteorology. World Weather Watch Tech. Rep. 8, Tech. Doc. 358, World Meteorological Organization, Geneva, Switzerland, 114 pp.

    • Search Google Scholar
    • Export Citation
  • Steenburgh, W. J., , D. M. Schultz, , and B. A. Colle, 1998: The structure and evolution of gap outflow over the gulf of Tehuantepec, Mexico. Mon. Wea. Rev., 126 , 26732691.

    • Search Google Scholar
    • Export Citation
  • Wilks, D. S., 1995: Statistical Methods in the Atmospheric Sciences. Academic Press, 467 pp.

  • Zängl, G., 2002a: Stratified flow over a mountain with a gap: Linear theory and numerical simulations. Quart. J. Roy. Meteor. Soc., 128 , 927949.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., 2002b: An improved method for computing horizontal diffusion in a sigma-coordinate model and its application to simulations over mountainous topography. Mon. Wea. Rev., 130 , 14231432.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., 2002c: Idealized numerical simulations of shallow föhn. Quart. J. Roy. Meteor. Soc., 128 , 431450.

  • Zängl, G., 2003: Deep and shallow south foehn in the region of Innsbruck: Typical features and semi-idealized numerical simulations. Meteor. Atmos. Phys., 83 , 237261.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., , A. Gohm, , and G. Geier, 2003: South foehn in the Wipp Valley–Innsbruck region: Numerical simulations of the 24 October 1999 case (MAP-IOP 10). Meteor. Atmos. Phys., in press.

    • Search Google Scholar
    • Export Citation
  • Zängl, G., , B. Chimani, , and C. Häberli, 2004: Numerical simulations of the foehn in the Rhine Valley on 24 October 1999 (MAP IOP 10). Mon. Wea. Rev., 132 , 368389.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Topographic map of the MAP Brenner Pass target area. Shown are model domains 5 and 6 of the MM5 simulations. Elevation contours are gray shaded, starting at 600 m with an increment of 200 m. The 1600 m MSL contour line is white. The solid rectangle in domain 5 indicates the location of domain 6, which is displayed in the upper-right corner. The dashed rectangle represents the subdomain shown in Figs. 2 and 8. Filled circles show the location of instrumented sites: Sterzing (STZ, 944 m), Brennerbad (BRB, 1310 m), Brenner (BRE, 1373 m), Sattelberg (SAB, 2107 m), Steinach (STE, 1116 m), Gedeir (GED, 1084 m), Ellbögen (ELB, 1080 m), Patsch (PAT, 913 m), and Innsbruck (IBK, 609 m). Italic letters indicate the location of the Wipp Valley (WV), Inn Valley (IV), Stubai Valley (SV), and Gschnitz Valley (GV)

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 2.
Fig. 2.

Topographic map of the Wipp Valley region. Solid lines with circles indicate four selected flight legs (P1P2, P3P4, P5P6, and P7P8) of the NCAR Electra aircraft at approximately 1500 UTC on 24 Oct 1999; the average cruising altitude was 4.9 km MSL. The dashed line with markers every 10 km indicates the location of the cross section displayed in Fig. 18. The dashed rectangle represents the subdomain displayed in Figs. 2123. Cross markers show the location of mountain summits: Patscherkofel (PAK, 2252 m), Nockspitze (NOC, 2403 m), Serles (SER, 2717 m), Bentlstein (BEN, 2436 m), Nösslachjoch (NOS, 2231 m), and Sattelberg (SAB, 2107 m)

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 3.
Fig. 3.

ECMWF analysis for 1200 UTC on (a), (b) 24 Oct 1999 and (c), (d) 25 Oct 1999. Height of the 500-hPa level is represented as contour lines with 50-gpm increments and wind barb for the Brenner Pass grid point at the 700-hPa level in (a) and (c); half barb, full barb, and triangle denote 2.5, 5, and 25 m s−1, respectively. Contour lines with 2-hPa increments in (b) and (d) represent sea level pressure

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 4.
Fig. 4.

Hand analysis of surface fronts based on ECMWF analysis data for (a) 24 and (b) 25 Oct 1999. Fronts are drawn as isochrones, with dotted, dashed, dashed–dotted, and solid lines at 0000, 0600, 1200, and 1800 UTC, respectively

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 5.
Fig. 5.

Cross-Alpine pressure gradient (Pa km−1) between 1800 UTC 23 Oct 1999 and 0600 UTC 26 Oct 1999. (a) Meso-α-scale gradient as a function of time derived from ECMWF analysis gridpoint data (solid line) and observations (dashed line) at locations near Verona and Munich. Meso-β-scale gradient derived from observations at Sterzing and Patsch as a (b) function of time and (c) function of wind speed observed at Sattelberg (plus signs), Ellbögen (circles), and Innsbruck (crosses). The linear regression curves in (c) have inclination 0.53 for SAB, 0.91 for ELB, and 2.36 for IBK. Meso-α-scale and meso-β-scale gradients were calculated from pressure data reduced to sea level and to 1000 m MSL, respectively. Positive values indicate a northward-directed pressure gradient force

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 6.
Fig. 6.

Time series of MAP weather station data from 1800 UTC 23 Oct 1999 to 0600 UTC 26 Oct 1999. Plotted are (a) sustained wind speed (m s−1), wind direction (°), and potential temperature (K) for the station Sattelberg (solid), Ellbögen (dotted), and Innsbruck (dashed). Light gray areas superposed with dark gray areas indicate periods with south foehn at Ellbögen and Innsbruck, respectively

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 7.
Fig. 7.

Vertical soundings from radiosonde ascents at Gedeir (solid line) and Sterzing (dashed line) at 1400 and 1500 UTC 24 Oct 1999, respectively: (a) potential temperature (K), (b) wind speed (m s−1), and (c) wind direction (°).

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 8.
Fig. 8.

MM5 fields of the 0000 UTC run at 1500 UTC 24 Oct 1999. Orography is gray shaded starting at 1200 m with 600-m increments. The 600-m contour line is light gray. Horizontal winds at (a) the surface, (c) z = 2000 m MSL, and (e) z = 3000 m MSL. Wind barbs defined as in Fig. 3. (b) Surface potential temperature, with contour increment 1 K. Vertical wind speed at (d) z = 2000 m MSL and (f) z = 3000 m MSL; contour increment is 2 m s−1, positive (negative) contour lines are solid (dashed). Boxed numbers represent horizontal wind speed (m s−1) in (a) and potential temperature (K) in (b), observed at selected MAP weather stations

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 9.
Fig. 9.

Meso-β-scale pressure gradient as (a) a function of time and (b), (c) a function of wind speed for the period 0600 UTC 24 Oct 1999 to 0000 UTC 25 Oct 1999. Comparison of observed pressure gradient (solid line) calculated from pressure data measured at PAT and STZ with values derived from MM5 gridpoint data of the 0000 UTC run (dashed line) and of the 1800 UTC run (dotted line) in (a). Simulated pressure gradient as a function of wind speed at SAB (plus signs), ELB (circles), and IBK (crosses) for the 0000 UTC run in (b) and for the 1800 UTC run in (c). The linear regression curves in (b) have inclination 0.69 for SAB, 1.13 for ELB, 1.44 for IBK, and in (c) 0.68 for SAB, 1.11 for ELB, 1.42 for IBK.

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 10.
Fig. 10.

Comparison of observations from MAP weather stations (solid line) with MM5 gridpoint data of the 0000 UTC run (dashed line) and 1800 UTC run (dotted line) for the period 0600 UTC 24 Oct 1999 to 0000 UTC 25 Oct 1999. Plotted is wind speed upstream of and at the pass for (a) STZ, (b) BRE, and (c) SAB and downstream for (a) STE, (b) ELB, and (c) IBK. Model data were taken from domain 5 in (a) and (d)–(f), and from domain 6 in (b) and (c).

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 11.
Fig. 11.

As in Fig. 10, but for potential temperature

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 12.
Fig. 12.

Vertical profile of the horizontal wind at the lidar site GED as function of time for the period 0900–1800 UTC 24 Oct 1999: (a) Doppler lidar observations and MM5 gridpoint data of the (b) 0000 and (c) 1800 UTC run. Horizontal wind speed is represented as contour lines with an increment of 2 m s−1 and as shaded contours for velocities exceeding 15, 20, and 25 m s−1. Wind barbs indicate horizontal wind direction and speed and are defined as in Fig. 3

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 13.
Fig. 13.

Vertical profile of the horizontal wind at the sodar site BRB as function of time for the period 0900–1800 UTC 24 Oct 1999: (a) Doppler sodar observations and MM5 gridpoint data of the (b) 0000 and (c) 1800 UTC run. Horizontal wind speed is represented as contour lines with increments of 1 m s−1. Shaded contours and wind barbs as in Fig. 12

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 14.
Fig. 14.

Vertical cross section along NCAR Electra flight leg P1P2 (cf. Fig. 2). (a) SABL backscatter intensity (relative backscatter) at 532 nm between 1523 and 1532 UTC 24 Oct 1999. MM5 (b) 0000 and (c) 1800 UTC run at 1500 UTC 24 Oct 1999: Potential temperature is represented as contour lines with 1-K increment and horizontal wind speed as shaded contours for velocities exceeding 15, 20, and 25 m s−1. Crosses in (a), (b), and (c) mark the AML top height derived from observed backscatter intensities. The white line in (a) represents either topography or the height of the cloud tops. Clouds are labeled with “CL.” The white triangles on the abscissas in (b) and (c) indicate the location of the pass

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 15.
Fig. 15.

As in Fig. 14, but for cross section P3P4. Observations collected between 1453 and 1457 UTC 24 Oct 1999

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 16.
Fig. 16.

As in Fig. 14, but for cross section P5P6. Observations collected between 1447 and 1449 UTC 24 Oct 1999

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 17.
Fig. 17.

As in Fig. 14, but for cross section P7P8. Observations collected between 1441 and 1444 UTC 24 Oct 1999

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 18.
Fig. 18.

Vertical cross section along the Wipp Valley, as indicated by the dashed line in Fig. 2. The cross section is oriented at an azimuth angle of 320° (178°) north (south) of the Doppler lidar site GED, which is located at x = 20 km. Orography is displayed as filled black areas. Radial wind velocities (a)–(d) observed with TEA CO2 lidar and (e)–(h) simulated by the MM5 0000 UTC run: contour lines with 5 m s−1 increments; negative values are dashed; shaded contours for velocities above (below) 15, 20, 25 m s−1(−15, −20, −25 m s−1). Arrows in (e)–(h) represent winds along the cross section. (i)–(l) Potential temperature (contour lines, 1-K increment) and horizontal wind speed (shaded for speeds exceeding 15, 20, 25 m s−1) simulated by the MM5 0000 UTC run. Figure rows from top to bottom represent times of approximately 0900, 1200, 1500, and 1700 UTC, respectively

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 19.
Fig. 19.

Vertical soundings at Sterzing at 0900 UTC 24 Oct 1999. Observations (solid), MM5 0000 UTC run (dashed), and 1800 UTC run (dotted). (a) Potential temperature (K), (b) wind speed (m s−1), and (c) wind direction (°).

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 20.
Fig. 20.

(a) Vertical transect across the Alps, as indicated by the solid line in (b). Data are taken from domain 4 of the MM5 0000 UTC run for 0900 UTC 24 Oct 1999. Potential temperature is represented as contour lines with 1-K increments, horizontal wind speed as gray shaded contours for winds above 10, 15, and 20 m s−1, and arrows represent winds along the cross section in (a). (b) Model orography of domain 4, with elevation contours starting at 200 m MSL with 200-m increments. Indicated are Sterzing (STZ), Innsbruck (IBK), and the Sarntaler Alps (SA). The vertical bar near STZ in (a) shows the location of the stable inversion layer

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 21.
Fig. 21.

Horizontal cross section at z = 1500 m MSL for the subdomain indicated by the dashed rectangle in Fig. 2. Orography is gray shaded, starting at 1200 m with 600-m increments. Radial wind velocities (a)–(d) observed with TEA CO2 lidar and (e)–(h) simulated by the MM5 0000 UTC run: contour lines with 5 m s−1 increments; negative values are dashed; thick contour lines for velocities above (below) 25 m s−1 (−25 m s−1). (i)–(l) Horizontal wind speed (contour lines; 5 m s−1 increments; thick lines for values above 25 m s−1) and horizontal wind vectors simulated by the MM5 0000 UTC run. The Doppler lidar site GED is marked as a filled circle. Figure columns from left to right represent times of approximately 0900, 1200, 1500, and 1700 UTC, respectively

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 22.
Fig. 22.

As in Fig. 21, but for z = 2000 m MSL

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 23.
Fig. 23.

As in Fig. 21, but for z = 2500 m MSL

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Fig. 24.
Fig. 24.

Error measures of radial wind velocities (m s−1) as a function of time for 24 Oct 1999. (a), (b) Derived from vertical cross sections along the Wipp Valley, as shown in Fig. 18. (c), (d) Derived from horizontal cross sections at z = 1500, 2000, and 2500 m MSL, as shown in Figs. 2123. Rmse of radial velocities (a) and (c), as well as ME of the absolute value of radial velocities (b) and (d). Compared are lidar observations to the 0000 UTC run (circles) and to the 1800 UTC run (squares), as well as the 0000 UTC run to the 1800 UTC run (stars).

Citation: Monthly Weather Review 132, 1; 10.1175/1520-0493(2004)132<0078:SFITWV>2.0.CO;2

Table 1.

Error measures derived from vertical profiles of the horizontal wind at the lidar site GED for the period 0900–1800 UTC 24 Oct 1999 (see Fig. 12). Listed are rmse and ME for wind speed (Wspd) and for wind direction (Wdir). Compared are lidar observations with the 0000 UTC run (column 2) and with the 1800 UTC run (column 3), as well as the 0000 UTC run with the 1800 UTC run for the lidar range (column 4) and for all model levels (column 5). The lidar range and the model range cover a total of 164 and 300 data points, respectively

Table 1.
Table 2.

Error measures derived from vertical profiles of the horizontal wind at the sodar site BRB for the period 0900–1800 UTC 24 Oct 1999 (see Fig. 13). Listed are rmse and ME for wind speed (Wspd) and for wind direction (Wdir). Compared are sodar observations with the 0000 UTC run (column 2) and with the 1800 UTC run (column 3), as well as the 0000 UTC run with the 1800 UTC run for the sodar range (column 4) and for all model levels (column 5). The sodar range and the model range cover a total of 82 and 300 data points, respectively

Table 2.
Save