• Albrecht, B. A., R. S. Penc, and W. H. Schubert, 1985: An observational study of cloud-topped mixed layers. J. Atmos. Sci.,42, 800–822.

  • Andreas, E. L., C. A. Paulson, R. M. Williams, R. W. Lindsay, and J. A. Businger, 1979: The turbulent heat flux from Arctic leads. Bound.-Layer. Meteor.,17, 57–91.

  • Baumgardner, D., J. W. Strapp, and J. E. Dye, 1985: Evaluation of the forward scattering spectrometer probe. Part II: Corrections for coincidence and dead-time losses. J. Atmos. Oceanic Technol.,2, 626–632.

  • Belair, S., P. Lacarrere, J. Noilhan, V. Masson, and J. Stein, 1998: High-resolution of surface and turbulent fluxes during HAPEX-MOBILHY. Mon. Wea. Rev.,126, 2234–2253.

  • Curry, J. A., and G. F. Herman, 1985: Relationship between large-scale heat and moisture budgets and the occurrence of Arctic stratus clouds. Mon. Wea. Rev.,113, 1441–1457.

  • ——, E. E. Ebert, and G. F. Herman, 1988: Mean and turbulence structure of the summertime Arctic cloudy boundary layer. Quart. J. Roy. Meteor. Soc.,114, 715–746.

  • ——, J. L. Schramm, and E. E. Ebert, 1993: Impact of clouds on the surface radiation balance of the Arctic Ocean. Meteor. Atmos. Phys.,51, 197–217.

  • ——, W. B. Rossow, D. Randall, and J. L. Schramm, 1996: Overview of Arctic Cloud and radiation characteristics. J. Climate,9, 1731–1764.

  • Ebert, E. E., and J. A. Curry, 1993a: An intermediate one dimensional thermodynamic sea ice model for investigating ice–atmosphere interactions. J. Geophys. Res.,98 (D4), 10 085–10 109.

  • ——, and ——, 1993b: A parameterization of ice cloud optical properties for climate models. J. Geophys. Res.,97 (D4), 3831–3836.

  • Fletcher, N. H., 1962: The Physics of Rainclouds. Cambridge University Press, 386 pp.

  • Fowler, D. L., D. A. Randall, and S. A. Rutledge, 1996: Liquid and ice cloud microphysics in the CSU general circulation model. Part I: Model description and simulated microphysical processes. J. Climate,9, 489–529.

  • Francis, P. N., P. Hignett, and A. Macke, 1998: The retrieval of cirrus cloud properties from aircraft multi-spectral reflectance measurements during EUCREX ’93. Quart. J. Roy. Meteor. Soc.,124, 1273–1291.

  • GEWEX News, 1997: Satellite cloud data for regional studies. GEWEX News,7, 1–3.

  • Ghan, S. J., L. R. Leing, and Q. Hu, 1997: Application of cloud microphysics to NCAR community climate models. J. Geophys. Res.,102 (D14), 16 507–16 527.

  • Gultepe, I., and G. V. Rao, 1993: Moisture and heat budgets of a cirrus cloud from aircraft measurements during FIRE. Quart. J. Roy. Meteor. Soc.,119, 957–974.

  • ——, and D. O’C. Starr, 1995: Dynamical structure and turbulence in cirrus clouds: Aircraft observations during FIRE. J. Atmos. Sci.,52, 4159–4182.

  • ——, and G. A. Isaac, 1999: The influence of air mass origin on the microphysical structure of Arctic boundary layer clouds. 33d CMOS Congress, Montreal, PQ, Canada, CMOS, 22.

  • ——, A. J. Heymsfield, and D. H. Lenschow, 1990: A comparison of vertical velocity in cirrus obtained from aircraft and lidar measurements. J. Atmos. Oceanic Technol.,7, 58–67.

  • ——, D. O’C. Starr, A. J. Heymsfield, T. Uttal, T. P. Ackerman, and D. L. Westphal, 1995: Dynamical characteristics of cirrus clouds from aircraft and radar observations in micro and meso-gamma scales. J. Atmos. Sci.,52, 4060–4078.

  • ——, G. A. Isaac, A. Korolev, S. Cober, and J. W. Strapp, 1998: Effects of ice crystal shape on the radiative characteristics of low level ice clouds. Preprints, 14th Conf. on Cloud Physics, Everett, WA, Amer. Meteor. Soc., 372–375.

  • Hense, A., and E. Heise, 1984: A sensitivity study of cloud parameterizations in general circulation models. Beitr. Phys. Atmos.,57, 240–258.

  • Heymsfield, A. J., 1975: Cirrus uncinus generating cells and the evolution of cirriform clouds. Part II: The structure and circulations of the cirrus uncinus generating head. J. Atmos. Sci.,32, 809–819.

  • ——, and J. L. Parrish, 1978: A computational technique for increasing the effective sampling volume of the PMS two-dimensional particle size spectrometer. J. Appl. Meteor.,17, 1566–1572.

  • Houze, R. A., 1981: Structure of atmospheric precipitating systems:A global survey. Radio Sci.,16, 671–689.

  • ——, S. A. Rutledge, T. J. Matejka, and P. V. Hobbs, 1981: The mesoscale and microscale and organization of clouds and precipitation in midlatitude cyclones. Part III: Air motions and precipitation growth in a warm-frontal rainband. J. Atmos. Sci.38, 639–649.

  • Hudak, D. R., and R. Nissen, 1996: Doppler radar applications in major winter snow storms. Atmos. Res.,41, 109–130.

  • ——, R. E. Stewart, G. W. K. Moore, and E. T. Hudson, 1995: Synoptic considerations of storms in the southern Beaufort Sea—expectations for BASE. Proc. Fourth Conf. on Polar Meteor. Oceanogr., Dallas, TX, Amer. Meteor. Soc., 234–237.

  • ——, R. List, and T. Krauss, 1996: Microphysical properties of Arctic cyclones. Preprints, 12th Int. Conf. on Clouds and Precipitation, Zurich, Switzerland, International Commission on Clouds and Precipitation, 158–161.

  • Isaac, G. A., and R. A. Stuart, 1996: Relationships between cloud type and amount, precipitation and surface temperature in the Mackenzie River valley–Beaufort Sea area. J. Climate,9, 1921–1941.

  • Kiehl, J. T., J. J. Hack, G. B. Bonan, B. A. Boville, B. P. Briegleb, D. L. Williamson, and P. J. Rasch, 1996: Description of the NCAR Community Climate Model (CCM3). NCAR Tech. Note NCAR/TN-420+STR, National Center for Atmospheric Research, 152 pp. [Available from NCAR, P.O. Box 3000, Boulder, CO 80307-3000.].

  • Korolev, A. V., J. W. Strapp, G. A. Isaac, and A. Nevzorov, 1998: The Nevzorov airborne hot wire LWC/TWC probe: Principles of operation and performance characteristics. J Atmos. Oceanic Technol.,15, 1496–1511.

  • ——, G. A. Isaac, and J. Hallett, 1999: Ice particle habits in Arctic clouds. Geophys. Res. Lett.,26, 1299–1302.

  • Liu, A. K., C. Y. Peng, B. Chapron, E. M. Christensen, and N. E. Huang, 1995: Direction and magnitude of wind stress over wave groups observed during SWADE. Global Atmos. Ocean. Sys.,3, 175–194.

  • MacPherson, J. I., 1993: Use of a wing-mounted airflow pod for airborne wind and flux measurements. Preprints, Eighth Symp. on Meteor. Obser. and Instrumentation, Anaheim, CA, Amer. Meteor. Soc., 169–174.

  • McInnes, K. L., and J. A. Curry, 1995: Modelling the mean and turbulence structure of the summertime Arctic cloud boundary layer. Bound.-Layer Meteor.,73, 125–143.

  • Meyers, M. P., P. J. DeMott, and W. R. Cotton, 1992: New primary ice-nucleation parameterizations in an explicit cloud model. J. Appl. Meteor.,31, 708–721.

  • Mezrin, M. Yu, 1997: Humidity measurements from aircraft. Atmos. Res.,44, 53–59.

  • Pinto, J. O., and J. A. Curry, 1995: Atmospheric convective plumes emanating from leads. Part II: Microphysical and radiative processes. J. Geophys. Res.,100 (C3), 4633–4642.

  • ——, and ——, 1997: Role of radiative transfer in the modeled meso-scale development of summertime Arctic stratus. J. Geophys. Res.,102 (C3), 13 861–13 872.

  • ——, ——, and K. L. McInnes, 1995: Atmospheric plumes emanating from leads. Part I: Thermodynamic structure. J. Geophys. Res.,100 (C3), 4621–4631.

  • Schnell, R. C., R. G. Barry, M. W. Miles, E. L. Andreas, L. F. Radke, C. A. Brock, M. P. McCormick, and J. L. Moore, 1989: Lidar detection of leads in Arctic Sea ice. Lett. Nature,339, 530–532.

  • Stewart, R. E., R. W. Crawford, K. K. Szeto, and D. R. Hudak, 1996:Horizontal aircraft passes across 0°C regions within winter storms. Atmos.-Ocean,34, 133–159.

  • Thomson, A. D., and R. List, 1996: Raindrop spectra and updraft determination by combining Doppler radar and distrometer. J. Atmos. Oceanic Technol.,13, 465–476.

  • Tsay., S.-C., K. Stamnes, and K. Jayaweera, 1989: Radiative energy budget in the cloudy and hazy Arctic. J. Atmos. Sci.,46, 1002–1018.

  • Wilson, L. D., J. A. Curry, and T. P. Ackerman, 1993: Satellite retrieval of lower-tropospheric ice crystal clouds in the polar regions. J. Climate,6, 1467–1472.

  • Zerr, R. J., 1997: Freezing rain: An observational and theoretical study. J. Appl. Meteor.,36, 1647–1661.

  • Zhang, T., S. A. Bowling, and K. Stamnes, 1997: Impact of the atmosphere on surface radiative fluxes and snowmelt in the Arctic and subArctic. J. Geophys. Res.,102 (D4), 4287–4302.

  • View in gallery

    (a) Large-scale meteorological conditions for 0000 UTC 9 Sep 1994. Arrows are for horizontal wind (m s−1) at the 500-mb level. Dashed and solid lines are for temperature (°C) and sea level pressure (mb), respectively. (b) Large-scale meteorological conditions for 0000 UTC 25 Sep. Arrows are for horizontal wind (m s−1)at the 500-mb pressure level. Dashed and solid lines are for temperature (°C) and sea level pressure (mb), respectively.

  • View in gallery

    (a) Rawinsonde observations for 1200 UTC 8 Sep. (b) Rawinsonde observations for 0000 UTC 25 Sep.

  • View in gallery

    (a) PPI scan for 1840 UTC and (b) RHI scan for 1946 UTC for the 8 Sep case. The color bar shows the reflectivity scale in dBZ units. In (b), the vertical axis is height and the horizontal axis is distance from the radar.

  • View in gallery

    (a) PPI scan for 0344 UTC and (b) RHI scan for 0500 UTC for the 25 Sep case. The color bar shows the reflectivity field in dBZ units. In (b), the vertical axis is height and the horizontal axis is distance from the radar.

  • View in gallery

    LANDSAT image of the DC0.83 for an area of 20 km × 20 km for the 8 Sep case: (a) for 3D image with vertical axis representing DC0.83, where x and y axes represent pixel number; (b) digital counts vs NS distance from LANDSAT observations after the mean is taken out. Three different scales are seen (see text for details); and (c) the wavelet analysis of the observations obtained for a segment about 35 km in length, with frequency in the y axis and distance in the x axis.

  • View in gallery

    The same as Fig. 5 except for the 24 Sep case.

  • View in gallery

    Vertical profiles of wa, horizontal wind (Uh), wind direction (D), equivalent potential temperature (θe) vapor mixing ratio (qυ), and aerosol number concentration (Na) for (a) 8 Sep, (b) 24 Sep, (c) 25 Sep, are shown in boxes 1–6, respectively. The arrow shows aircraft ascent/descent. Note that winds in (b) not reliable due to instrument wetting; therefore, arrows are not shown.

  • View in gallery

    Time series of aircraft observations of RHw, T, z, LWCN, LWCf, Ni, and TWCN for the 8 Sep case. Horizontal lines show the regions of RHw > 95%. The aircraft speed is approximately 85 m s−1.

  • View in gallery

    Spectral energy density vs wavenumber for the in-cloud wind observations (U, V, and wa) at about 2.8 km (1000–2000 s) for the 8 Sep case. Dashed lines are for −3 and −5/3 slopes.

  • View in gallery

    Same as Fig. 9 but for clear air wind observations at about 8 km (8000–9000 s) for the 24 Sep case.

  • View in gallery

    (a) Time series of wa vs time for the 8 Sep case. The thick solid line represents scales at about 20 km. The combination of thick and thin solid lines gives the original time series from aircraft. (b) Wavelet analysis of wa.

  • View in gallery

    For the 8 Sep flight at about 0.4 km, (a) the number of points vs binned SHF values, (b) probability and cumulative probability values of SHF, and (c) time series of SHF. See text for other symbols.

  • View in gallery

    The same as Fig. 12 except for LHF.

  • View in gallery

    The same as Fig. 8 except for the 24 Sep flight.

  • View in gallery

    (a) Time series of wa vs time for the 24 Sep flight. The thick solid line represents scales at about 20 km. A combination of thick and thin solid lines gives the original time series from the aircraft. (b) Wavelet analysis of wa.

  • View in gallery

    For the 24 Sep flight at about 8 km, (a) the number of points vs binned SHF values, (b) probability and cumulative probability values of SHF, and (c) time series of SHF. See text for other symbols.

  • View in gallery

    The same as Fig. 8 except for the 25 Sep flight.

  • View in gallery

    (a) Ice crystal concentration (Ni) vs vertical air velocity. See Table 5 and text for the assumptions made in the use of Eq. (7) to obtain ice crystal number concentration. Straight dark solid line is for the maximum Ni value observed during BASE. (b) Ice crystal number concentration vs temperature for the 2D-C data collected for the entire BASE project. Filled circles are averaged Ni for 5°C intervals.

  • View in gallery

    The number of points estimated in the 0.5-dBZ intervals for radar observations for the 8 and 25 Sep cases are shown in (a) for 1439–1552 and 1630–2245 UTC and in (b) for 0313–0430 and 1240–1341 UTC, respectively. Observations used in (a) are collected along six volume scans at 12 elevation angles from 3.0° to 67.2°. Observations in (b) are collected along seven volume scans at 13 elevation angles from 1.0° to 67.2°. Number of points (Np) vs cell size (L) and height (h) for (c) the 8 Sep case on 1840 UTC and for (d) the 25 Sep case on 1250 UTC. The Np is obtained using Ze > 10 dBZ.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 202 202 1
PDF Downloads 38 38 1

Dynamical and Microphysical Characteristics of Arctic Clouds during BASE

View More View Less
  • 1 Cloud Physics Research Division, Meteorological Service of Canada, Toronto, Ontario, Canada
© Get Permissions
Full access

Abstract

In this study, observations from aircraft, Doppler radar, and LANDSAT are used to better understand dynamical and microphysical characteristics of low-level Arctic clouds for climate change studies. Observations during the Beaufort and Arctic Storms Experiment were collected over the southern Beaufort Sea and the northern Mackenzie River Basin during 1 September–14 October 1994. Measurements from the cases of 8 September and 24–25 September are analyzed. In situ observations were made by instruments mounted on a Convair-580 research aircraft. Reflectivity and radial winds were obtained from an X-band Doppler radar located near Inuvik. The reflectivity field from LANDSAT observations concurrent with the aircraft and radar observations was also obtained. Dynamical activity, representing vertical air velocity (wa) and turbulent fluxes, is found to be larger in cloud regions. The sizes of coherent structures (e.g., cells) are from 0.1 to 15 km as determined by wavelet analysis and time series of aircraft data. This size is comparable with LANDSAT and Doppler radar–derived cell sizes. Reflectivity in embedded cells for the 8 September case was larger than that of single convective cells for the 24–25 September case. The effective radius for ice crystals (droplets) ranged from 37(7.5) μm to 70(9.5) μm for both cases. Using observations, parameterization of the ice crystal number concentration (Ni) is obtained from a heat budget equation. Results showed that Ni is a function of wa, radiative cooling, particle size, and supersaturation. The large-scale models may have large uncertainties related to microphysical and dynamical processes (e.g., particle size and vertical air velocity, respectively), which can directly or indirectly influence radiative processes. Overall, the results suggest that the microphysical and dynamical properties of Arctic clouds need to be further explored for climate change studies.

Corresponding author address: Dr. Ismail Gultepe, AES/ARMP Cloud Physics Research Division, 4905 Dufferin St., Toronto, ON M3H 5T4, Canada.

Email: ismail.gultepe@ec.gc.ca

Abstract

In this study, observations from aircraft, Doppler radar, and LANDSAT are used to better understand dynamical and microphysical characteristics of low-level Arctic clouds for climate change studies. Observations during the Beaufort and Arctic Storms Experiment were collected over the southern Beaufort Sea and the northern Mackenzie River Basin during 1 September–14 October 1994. Measurements from the cases of 8 September and 24–25 September are analyzed. In situ observations were made by instruments mounted on a Convair-580 research aircraft. Reflectivity and radial winds were obtained from an X-band Doppler radar located near Inuvik. The reflectivity field from LANDSAT observations concurrent with the aircraft and radar observations was also obtained. Dynamical activity, representing vertical air velocity (wa) and turbulent fluxes, is found to be larger in cloud regions. The sizes of coherent structures (e.g., cells) are from 0.1 to 15 km as determined by wavelet analysis and time series of aircraft data. This size is comparable with LANDSAT and Doppler radar–derived cell sizes. Reflectivity in embedded cells for the 8 September case was larger than that of single convective cells for the 24–25 September case. The effective radius for ice crystals (droplets) ranged from 37(7.5) μm to 70(9.5) μm for both cases. Using observations, parameterization of the ice crystal number concentration (Ni) is obtained from a heat budget equation. Results showed that Ni is a function of wa, radiative cooling, particle size, and supersaturation. The large-scale models may have large uncertainties related to microphysical and dynamical processes (e.g., particle size and vertical air velocity, respectively), which can directly or indirectly influence radiative processes. Overall, the results suggest that the microphysical and dynamical properties of Arctic clouds need to be further explored for climate change studies.

Corresponding author address: Dr. Ismail Gultepe, AES/ARMP Cloud Physics Research Division, 4905 Dufferin St., Toronto, ON M3H 5T4, Canada.

Email: ismail.gultepe@ec.gc.ca

1. Introduction

Arctic clouds are mainly composed of ice crystals and water droplets during the transition times between winter and summer. Clouds with particles of mixed phase can play an important role in climate change, due to changes in the vertical distribution of heat and moisture. During winter, these clouds consist of mostly ice crystals (Wilson et al. 1993). Liquid water in Arctic clouds was also common during winter (Curry et al. 1996).

Arctic clouds are usually believed to be stable and stratiform, because of the lack of convective energy for vertical development (Curry et al. 1988). Isaac and Stuart (1996) showed that stratocumulus clouds are the dominant precipitating and nonprecipitating cloud type during all seasons for the Mackenzie Valley–Beaufort Sea area. These clouds can form locally or in association with large-scale weather patterns. They can strongly affect the heat and moisture budget of the atmosphere. Curry and Herman (1985) studied in detail large-scale heat and moisture budgets, and the occurrence of Arctic clouds over the Beaufort Sea region. They stated that large amounts of low-level cloud cover during June are attributed primarily to the low-level advection of moisture, and a residual cooling due to radiation and boundary layer turbulence. They also emphasized the importance of large-scale heat and moisture transport for midlevel cloudiness. McInnes and Curry (1995), using a high-resolution 1D model with a second-order closure scheme for turbulence, studied multilayered boundary layer clouds. They found that vertical air velocities (wa) less than 1 cm s−1 were sufficient in maintaining cloud layers, and when the cloud is established, vertical air motion is not necessary for cloud maintenance. For large-scale subsidence, a vertical air velocity of about 1 cm s−1 resulted in an upper cloud deck dissipating in less than 5 h. Pinto et al. (1995) also used a similar model given in McInnes and Curry (1995). They showed that in the Arctic convective boundary layer above a large body of water, the maximum heat and moisture fluxes occur at the lowest model level (below 100 m). Both the heat and moisture fluxes decreased linearly with height above the surface layer. Sensible heat flux (SHF) and latent heat flux (LHF) at the lead surface were about 600 W m−2, and 0.08 g kg−1 m s−1, respectively. Pinto and Curry (1997) studied a low-level Arctic cloud and anticyclone development using a mesoscale model. They stated that the relationship between Arctic clouds and radiative processes are complicated by the interactions between the clouds and large-scale dynamical processes. All these studies indicated that dynamical parameters (e.g., vertical air velocity and turbulence fluxes) are important for modifying cloud microphysical and optical parameters.

The Arctic Ocean surface during transition times (e.g., fall and spring) can have leads, polynya, and open water surfaces. The temperature of the water surface in the Arctic Ocean is about −1.8°C. During transition times, the air–open-ocean interface is the major source of moisture and heat transfer to the atmosphere. If clouds form related to leads, the LHF and SHF may reach up to 50 and 200 W m−2, respectively (Schnell et al. 1989). It should be noted that the magnitude of these fluxes are related to environmental conditions, including wind direction, speed, and thermal gradients (Pinto et al. 1995; Pinto and Curry 1995). Andreas et al. (1979) estimated LHF and SHF from leads at about 100 and 400 W m−2, respectively. Their results were consistent with those of Schnell et al. (1989). They stated that because of an increase in longwave radiation at the surface resulting from clouds forming above and downwind from open leads, the role of the Arctic as a global heat sink may need to be reevaluated. LHF and SHF of about 20 and 40 W m−2, respectively, representing scales of about 1 km were found over polynya and leads in the Beaufort Sea (Gultepe et al. 2000, manuscript submitted to J. Climate). These studies indicated that the Arctic Ocean plays an important role in regulating the heat and moisture budget of the atmospheric boundary layer.

Extensive studies of turbulent fluxes together with heat and moisture budget in the Arctic boundary layer were carried out by Curry et al. (1988). They found that advection, radiation, and turbulent terms dominate moisture and heat budget values. They also suggested that internal cloud processes (e.g., buoyancy produced by diabatic processes) were responsible for turbulent kinetic energy. Curry et al. (1993) studied the effect of cloud microphysical parameters on the Arctic surface-energy budget. They found that there is a strong nonlinearity between cloud properties and surface radiative fluxes.

Variability in the vertical air velocity (wa) affects the cloud microphysical parameters (e.g., particle number concentration, water content, and size), and thus the individual point values of LHF and SHF. Zhang et al. (1997) using other studies also showed that under stable environmental conditions, 10 W m−2 SHF accounts for 10% of the energy used during the entire snow-melting season (1 mm day−1). Snow melting due to LHF is only about 0.5 mm day−1. Using measurements, they showed that SHF plus LHF cooling can easily delay snow melting at the surface when the net infrared radiation flux was about 2–10 W m−2. Ebert and Curry (1993a) using a 1D thermodynamic model found that an ice pack may completely melt as a result of increased cloud fraction and increased cloud optical depth during the wintertime. In turn, these increased values are function of dynamical parameters (e.g., vertical air velocity).

The change in cloud amount and optical parameters are a strong function of dynamical processes occurring in the cloud and boundary layer. Curry et al. (1996) suggested that the interannual variability in surface radiative fluxes arises from differences in cloud fraction and changes in the optical parameters that result from the variability in large-scale atmospheric dynamics. The shape and number concentration of ice crystals in clouds are strongly affected by wa and T (Heymsfield 1975). In addition, some general circulation models (GCMs), as shown by Hense and Heise (1984), use liquid water content (LWC), which is formulated as a function of relative humidity and wa. These later studies indicate that the microphysical and optical parameters are related to the size and magnitude of dynamical processes (e.g., convective cells) in Arctic clouds.

The shift in emphasis from cloud radiation to cloud dynamics problem topics is evident in working group activities (GEWEX News 1997) of the Global Energy and Water Experiment. For this reason, as well as to increase knowledge about Arctic cloud systems, this paper will focus on the dynamics and microphysics of Arctic cloud systems. Cloud dynamical structures and wa are investigated using aircraft, radar, and LANDSAT observations collected during the Beaufort and Arctic Storms Experiment (BASE) field project. In addition, relationships between cloud dynamical processes and microphysical parameters are discerned as a first step toward developing parameterizations of Arctic cloud processes for climate models.

2. Observations

Data used in this study were gathered from aircraft, Doppler radar, and LANDSAT observations collected during BASE, which took place in the fall of 1994 between 1 September and 13 October in the southern Beaufort Sea and northern Mackenzie River Basin. Hudak et al. (1995) describes the synoptic setting and operational plans of the project. Details of the intensive operational periods can be found in Hudak et al. (1996). The aircraft used was a National Research Council of Canada (NRCC) Convair-580 aircraft, equipped for cloud microphysical measurements by the Atmospheric Environment Service (AES). The following subsections summarize the observations collected for the 8 September and 24–25 September cases during BASE. Table 1 shows the platforms used in this study, the origin of cloud systems, and the temperature and height ranges during the aircraft flights. Note that turbulent fluxes are calculated for only the 8 and 24 September flights.

a. Aircraft observations

Observations such as temperature (T), dewpoint (Td), wind components (u, υ, and wa), LWC, total water content (TWC), droplet, ice crystal, and aerosol number concentrations (Nd, Ni, and Na) were obtained from the instruments mounted on the Convair-580 research aircraft. The aircraft was operated out of Inuvik in the Northwest Territories of Canada. The aircraft collected observations over the Beaufort Sea (∼71°N, 133°W) on the 8 September case to study the effects of a cold-air flow on cloud systems. For the 24–25 September case, observations were collected over the land (∼66°N, 136°W) to study cloud systems related to an airmass originating over the Pacific Ocean. Temperature was measured by a reverse flow temperature probe with an accuracy of ±0.5°C. The Td was measured by an EG&G dewpoint hygrometer. LWC was obtained from measurements of the King, FSSP-100, and Nevzorov probes (Korolev et al. 1998). The Nd, Ni, and Na were obtained from particle measuring systems FSSP-100, 2D-C, and PCASP probe measurements, respectively. The FSSP-100 probe was operated in the 5–95-μm range. FSSP measurements were corrected for probe dead time and coincidence (Baumgardner et al. 1985). Under most circumstances, the Nevzorov probe LWC and TWC measurements are accurate to 10%–15% (Korolev et al. 1998). The ice water content (IWC) measurements by the Nevzorov probe in the mixed phase are unverified, but wind tunnel tests of small frozen ice spheres indicate an accuracy of 10%–20%. Measurements by the 2D-C probe were made in the range of 25–800 μm. Details of the 2D probe measurements and image processing techniques can be found in Heymsfield and Parrish (1978). Here Na measurements were made in the range of 0.3- to 3-μm diameter. Wind measurements were made using a Rosemount 858 pressure probe and a Litton LTN-90-100 inertial reference system (MacPherson 1993). The rms error for wa is about 0.15 m s−1. The flight speed for the Convair was approximately 85 m s−1. The latitude and longitude were obtained by a Global Positioning System of Marconi and Northstar, mounted on the Convair. The absolute accuracy of this system is about 10 m.

b. Doppler radar observations

The radar used in this study was the University of Toronto X-band (3.2 cm) Doppler radar (Hudak and Nissen 1996; Thomson and List 1996). It was located at the upper air station at Inuvik, Northwest Territories. Volume scans consisting of 13 elevation angles from 1° to 67° were taken typically every 10 min. Interspersed among the volume scans were range height indicator scans (RHI) for selected azimuths. In both modes, reflectivity (Ze) and radial velocity (VR) were measured with a range resolution of 250 m. The volume scans were used to determine the horizontal distribution of precipitation intensity and to estimate the wind field. RHIs were used to examine the details of the vertical structure of precipitation. On occasion, these scans were interrupted to operate the radar in a vertical pointing mode. In this case, the full Doppler spectrum was collected with a range resolution of 125 m. With typical horizontal winds, the space resolution of the data in the horizontal was about 10 m. This information was used to examine the microphysical properties of the cloud systems.

c. LANDSAT observations

The LANDSAT overpasses at 1951 UTC for the 8 and 24–25 September cases were nearly concurrent with the aircraft flights. The aircraft sampling time was approximately 45 min later than the LANDSAT overpass time for the 8 September case, and concurrent with the LANDSAT overpass for the 24 September case. Radiation observations as digital counts (DC) from LANDSAT were collected at six solar reflectance channels from 0.45 up to 2.35 μm of the thematic mapper with 30-m resolution over a 185 km2 area. The 0.83- and 11.5-μm channels were used for dynamical structure analysis and blackbody temperature calculation, respectively. The IR channel had a field of view of 114 m and it was used to remove the cirrus effect when reflectance at 0.83 μm is obtained. Winds observed by the aircraft and rawinsonde were used to adjust the time differences and match an area approximately covering 40 km2 around the aircraft flight path for which LANDSAT raw data (as digital counts) were used.

3. Large-scale meteorological characteristics

Cloud systems in the Inuvik area during the autumn are related to two distinguishing weather patterns. The first is characterized by storms originating over the Arctic Ocean where the upper-level flow is either zonal or from the northwest. The second is characterized by disturbances originating over the Pacific Ocean (Gulf of Alaska) with a southwesterly upper flow (Hudak et al. 1995). The 8 September case is an example of clouds from an Arctic disturbance. In this instance, the 500-mb flow with a maximum speed of 25 m s−1 was from the northwest near Inuvik (Fig. 1a). The pattern of upper-level winds indicates that an upper trough from the high Arctic extended southward through the Inuvik area. A 1005-mb low pressure system was located south of Inuvik at 0000 UTC 9 September 1994. Surface air temperatures were near 0°C. Figure 1b shows the synoptic patterns at 0000 UTC 25 September 1994. This is an example of a Pacific cloud system with a persistent upper flow from the southwest near Inuvik. There was a strong low pressure system, central pressure 986 mb, well north of Inuvik over the Arctic Ocean and a second strong low over the Gulf of Alaska (Fig. 1b). Surface temperatures were somewhat colder in this case than the previous case, being around −5°C near Inuvik.

Profiles of temperature and dewpoint from rawinsonde measurements at Inuvik are shown in Figs. 2a and 2b for the 8 September (1200 UTC) and 25 September (0000 UTC) cases, respectively. The 8 September case (Fig. 2a) is an example of an airmass in an Arctic disturbance. There is a shallow unstable layer to 900 mb, a frontal inversion from 900 to 800 mb, then a marginally stable layer up to about 500 mb. The cloud layer extends to 500 mb. The 25 September case (Fig. 2b) is an example of an airmass in a Pacific disturbance. There is a shallow elevated inversion from 950 to 930 mb above the surface-based neutral layer. A conditionally unstable layer extending from 930 to about 550 mb is seen, then a strong inversion from 550 to 450 mb. The shallow elevated inversion marks the lower boundary of the warmer Pacific air aloft. In this case, the moisture extends up to about 450 mb.

4. Method

Dynamical structures are important for transferring heat and moisture in cloud systems. Their size and timescales are not well known for Arctic cloud systems. Appropriate model time and space steps to represent the dynamical processes are crucial for studying climate change, and they play an important role in atmospheric budget calculations. For this reason, time series, spectral analysis, and wavelet analysis are used to study the sizes of dynamical structures. Sensible heat and latent heat fluxes are also estimated from the aircraft measurements.

The size and magnitude of dynamical structures cannot be obtained using only time series of raw data from one platform. A combination of observations and analysis of data from different platforms are needed to better specify the time and space characteristics of dynamical processes. A convective cell throughout the paper is defined when magnitude of the interested parameter (e.g., wa, reflectance) exceeded its sum of the mean and standard deviation (std dev) values. Using this definition and wavelet analysis, and images from the Doppler radar and LANDSAT, the characteristics of dynamical structures are obtained for both time- and space scales.

Aircraft observations of wind and temperature were collected at a sampling rate of 8 Hz. The sampling rate of the water vapor mixing ratio (qυ) was reduced to 8 Hz from 10 Hz. Time series of fluctuations of these observations were first obtained. Then, using 200-s moving averages (∼20 km), high-frequency fluctuations are filtered. Before smoothing, the mean and trend are removed from the observations. Then, individual fluctuations and covariance (e.g., fluxes) were obtained at constant altitude levels. Here qυ values from the Russian Lyman-α were not available for the 24 September case. The spectral energy for the wind components was calculated using a method similar to that of Gultepe and Starr (1995). A significant change in the slope of the energy spectra is indicative of the size of the dynamical structures dominating the cloud system.

A wavelet analysis is applied to measurements of wind, and digital counts from LANDSAT, to obtain frequency versus time information to analyze the dynamical structures. This information cannot be obtained by using moving averages in the time series. Results from the wavelet analysis show the magnitude of different frequencies versus time or spatial scales, representing dynamical structures, for example, waves, cells, and turbulence. Wavelet transforms can be used to study dynamical processes by localizing their properties in both time and frequency (Liu et al. 1995). The wavelet transform, Ws(a, b), given by Liu et al. (1995) is applied here as
i1520-0442-13-7-1225-e1
where s(t) is the time series and w* the wavelet function. A Morlet Gaussian wavelet is chosen as the structure function to analyze the aircraft and LANDSAT observations. Because it has orthogonal characteristics, it is the best way to represent data in a time-frequency domain, and is given by Liu et al. (1995) as
i1520-0442-13-7-1225-e2
where the wavelet function is dilated by a factor a and shifted by b. Here ω is the frequency. Aircraft wind observations and LANDSAT reflectance values (digital counts) are used in the wavelet analysis to estimate the size of dynamical structures in the cloud systems.

5. Results

In this section, results obtained from the time series and profiles from aircraft, from rawinsondes at Inuvik, the plan position indicators (PPIs) and RHIs from radar, and the reflectances from LANDSAT will be discussed. Embedded cells within the stratiform cloud system characterize the 8 September case. Individual convective cells mostly characterize the 24–25 September cases. In the following subsections, results related to Doppler radar, LANDSAT, and aircraft observations will be summarized sequentially.

a. Radar observations

1) The 8 September case

Doppler radar observations provided a large horizontal coverage of precipitation echoes with embedded cells. Several examples at different times are given to represent the cloud horizontal plan-view and vertical cross section. Figure 3a shows a PPI scan at 1840 UTC. The elevation angle is 6.3° and the large rings are at a 5-km spacing. The range resolution for the reflectivity factor (Ze) is 250 m. In this case, the size of the large reflectivity regions is assumed to correlate with that of the embedded cores (convective cells). As seen from Fig. 3a, their sizes vary from a few hundred meters up to 2 km, and the maximum Ze is about 21 dBZ. For a better understanding of the cloud dynamical structure, a vertical cross section of the Ze field is informative. Figure 3b shows a RHI scan at 1946 UTC along the 200° azimuth. The maximum Ze is about 23 dBZ. However, larger Ze values are seen from the surface to 0.5 km in the vertical where the melting layer is situated. The top of this layer matches well with the 0°C level obtained from rawinsonde observations. The echo tops were very irregular ranging from 1 to 3 km, indicating the presence of embedded convection. A vertically sloped feature between horizontal distances 2 and 20 km from the radar is seen. This feature is likely related to moist air advection from the south. From rawinsonde observations, the vertical shear of the horizontal wind between 0.5 km and 1.5 km is estimated as approximately 10 m s−1 km−1 with a directional change of 100°. The size of cells between 1- and 20-km range can be seen in Fig. 3b. These characteristics were representative of the whole cloud system.

2) The 24–25 September case

Figure 4 shows data from the 24–25 September case in the same format and sequence as those presented for the 8 September case. Figure 4a shows a PPI scan taken at 0344 UTC 25 September. The elevation angle is 3.0° and the maximum range is 30 km. Echo coverage on the image was scattered, but there were high reflectivity regions (∼24 dBZ) with horizontal dimensions from a few hundred meters up to 5 km. These regions are likely responsible for large heat and moisture transfer into upper levels. Figure 4b shows a vertical cross section of the reflectivity field (RHI) within the clouds systems at 0500 UTC. The azimuth angle is 225° and the horizontal size of each convective cloud varies from 0.2 to 4 km with tops about 2 km. Although cloud-top heights in Fig. 4b are less than 2 km, they developed to about 6–7.5 km later in time (not shown). Aircraft data were not available for comparisons in this specific time period and location.

Temporal and spatial development of the cloud system started at 2340 UTC 24 September (not shown), and it ended late morning on 25 September. Shallow clouds developed at about 2340 UTC. They started to develop horizontally and vertically and their thickness soon reached 4 km. Maximum cloud tops were up to 7.5 km. Particle fall speed (wp) at about 1 km was approximately −1 m s−1 at 2340 UTC. Note that the increased fall speeds indicate the development of a melting layer by 0224 UTC at 1 km. At about 0224 UTC, wp reached about −3.5 m s−1 with Ze at about 15 dBZ (not shown). It is likely that a melting layer developed between 0.5- and 1.5-km height.

b. LANDSAT observations

1) The 8 September case

Using a 20 km × 20 km area specified within the LANDSAT image (∼100 km × 100 km), the size of convective cells is obtained from the DC at 0.83 μm (DC0.83). The area chosen did not have high-level clouds and this was verified by using the 11.5-μm channel. Large contributions to DC0.83 come from the upper part of the clouds. It is assumed that the cloud top had ice crystals (see Figs. 8 and 14), and the number concentration is fixed. The shape effect is not considered. The uncertainty can be very large if the errors in all these assumptions are considered at the same time.

Figure 5a shows DC0.83 values in a three-dimensional plot. The horizontal resolution in this figure is about 30 m. The figure indicates that the cloud field was not homogeneous, presumably due to mixing processes resulting from convective lifting. DC0.83 is related to the number concentration (assumed constant) of ice particles, particle size, and IWC. Since particle size increases with increasing IWC, the larger DC0.83 can be related to larger IWC values, assuming that both mixing and convection are occurring at the same time. Spikes of large DC0.83 values representing a scale of about 30 m are seen in the banded sections. Large DC0.83 values have a cell size ranging between 100 m and 10 km. The mean and std dev values of DC0.83 are found to be about 91.2 and 15.3, respectively. The probability of DC0.83 values greater than the mean + std dev is estimated at about 10%. It is used as a lower limit for cell definition. Figure 5b shows the distance versus digital counts after removing the trends and means from the raw data in the N–S direction of the entire image. The darkest line shows a 200-point running average. The lighter line is for data after removing a 20-point running average data from the original observations. The light line is obtained after removing a 10-point running average. The three different curves represent the scales from 30 m up to 60 km. The magnitudes for all three scales are comparable to each other, but larger-scale processes have much higher peak values relative to the other scales. A wavelet analysis of the digital counts from the 0.83-μm LANDSAT channel is shown in Fig. 5c. The colored regions show the intensity of wavelets (real part of the data). These values contribute to the total magnitude. Colors close to red are for high magnitude and colors close to blue are for low magnitude. It is seen that scales from small (0.5–1 s−1) to large (0.001 s−1) sizes play an important role in deriving large-scale reflectance values. The reason for this scaling behavior is not properly known, but largely dynamical processes (e.g., waves or cells) may be responsible.

2) The 24 September case

LANDSAT observations indicated that convective processes on this day were not very active on scales less than 1 km. Figure 6a shows the three-dimensional view of DC0.83 in the z axis. Structures from 2–3 km up to 10 km are more common than small-scale structures. The cloud field was more homogeneous on this day compared to the 8 September case. The mean and std dev values of DC0.83 are about 100.4 and 4.6, respectively. The probability of DC0.83 values greater than the mean + std dev (∼105 DC0.83) is estimated at about 21%. This threshold is used as a lower limit for cell definition.

Figure 6b shows digital counts versus distance along the N–S direction of the entire image (not shown). This figure is obtained in a similar manner as Fig. 5b. The 20-point running average shows a peak at about the 5-km horizontal scale. The large-scale motion has a wavelength of about 30 km. Smaller-scale processes are very turbulent (e.g., high-frequency fluctuations), and they are likely generated by small-scale instabilities resulted from mixing, condensation, and radiative processes. Figure 6c shows a wavelet analysis of the digital counts at 0.83 μm. This figure is obtained using a segment about 35 km long. Mesoscale processes have a scale size at about 1 km. High-frequency fluctuations with low intensifies are in the upper part of the figure.

c. Aircraft observations

1) Vertical profiles

In this section, the vertical profiles of wa, horizontal wind (Uh), wind direction (D), equivalent potential temperature (θe), vapor mixing ratio (qυ), and aerosol number concentration (Na) are shown in Figs. 7a–c for the 8, 24, and 25 September flights, respectively. Note that these profiles are not made at the same time as the horizontal legs. It should also be noted that wind components from the profiles can only be used for qualitative discussions when instrument wetting and sharp turns occur. In Fig. 7, the up-arrow and down-arrow represent profiles taken during aircraft ascent and descent, respectively. Profiling up in Fig. 7b and profiling down in Fig. 7c were close to the Inuvik airport, and the others were made over the southern part of the Yukon. The double-sided arrow indicates the cloud tops estimated when Ni becomes negligible and Na becomes larger. Because of icing of the instruments, profiles of the wind components in Fig. 7b are not used. The extreme values are considered as artifacts. The Uh shear in the vertical in Fig. 7a2 was strong in the shallow layers. A sharp change in Uh at about 3 km in Fig. 7a2 indicates that dynamical instability was strong. In the upper layers, the difference between the two soundings for the same location indicates convergence/divergence over about a 4-h time period. Convergence in the upper layers of Fig. 7c2 was larger than that of Fig. 7a2. Small-scale fluctuations during the ascent at upper levels in Fig. 7c2 were also strong as compared to those of Fig. 7a2. Wind direction differences between the two profiles in Fig. 7a3 did not change significantly. Winds aloft were mostly westerly while at low levels they were northerly. The winds near the surface in the descending profile of Fig. 7c3 were mostly northerly and southwesterly in the ascending profile, while, in the upper levels above 5 km, they were southerly. The cold-air advection aloft at low levels (descending profile) likely moved over the warmer ocean surface, favoring convective cell formation as observed by the Doppler radar. Here θe profiles show that the 8 September case had colder values as compared to the 25 September case. Small-scale convective instabilities are seen at higher levels in Figs. 7b4 and 7c4. Both Figs. 7a5 and 7b5 show that qυ increased over about a 4-hour time period. The increase in Fig. 7b5 was larger than that in Fig. 7a5. An increase in Fig. 7a5 is likely due to open water to the north while in Fig. 7b5 it is likely due to warm and moist advection from the Pacific Ocean over the region. The possibility of conditional instability was also larger for the 24–25 September case. The Na profiles indicate that the aerosol concentration (∼500 cm−3) in the boundary layer was larger in Figs. 7b6 and 7c6 as compared to Fig. 7a6. This shows that continental flow (24–25 September case) brought in more aerosols as compared to the 8 September case. Overall, the characteristics of both the 8 and 24–25 September cases indicated that convective and dynamical instabilities for the 24–25 September case were more intense.

2) Time series

(i) The 8 September case

Time series are used to understand the characteristics of the cloud microphysical and dynamical processes. Time series of relative humidity with respect to liquid water (RHw), T, altitude (z), the Nevzorov LWC (LWCN), FSSP − 100LWC (LWCf), total ice crystal number concentration (Ni) of the 2D-C probe, and Nevzorov TWC (TWCN) are shown in Figs. 8a–g, respectively. When the RHw time series reaches 100%, the FSSP likely measured droplets and ice crystals at cold T’s. The horizontal line in Fig. 8a indicates the region of RHw greater than 95%. Droplet regions are found during descent and ascent when T ≥ −15°C. A maximum LWC of about 0.2–0.3 g m−3 is found at about −5°C (500 s). The 2D-C image analysis indicated that some frozen droplets (ice spheres) or liquid droplets are also found at about 5000 s (−10°C). Figures 8d and 8g show that the IWC reached about 0.3 g m−3 at 2300 s. It is possible that droplets may contribute to this large value of IWC. IWC is obtained from either the TWC − LWC or TWC as measured by the Nevzorov hot wire probe. The possible contribution of droplets to reflectance values at cloud top can be seen as increased values in LWC, but this contribution is neglected because the droplets were found at low concentrations. The wa fluctuations (wa) between 1000 and 2000 s (not shown) are found to be about 0.5–2 m s−1 (with 40% uncertainty). This uncertainty is due to wetting and a large noise in the mean wa.

Arctic clouds can play an important role in transferring heat and moisture. To demonstrate this, 3D wind components obtained from aircraft observations in the cloudy air and clear air are spectrally analyzed. Figure 9 shows the turbulent kinetic energy (TKE) spectra of the wind components at 2.8 km for the cloudy segment between 1000 and 2000 s. All wind components have a similar slope. A change in the slope from −5/3 to −3 occurs at about 100 m where the dissipation of TKE becomes dominant. This approximately agrees with sizes estimated from the LANDSAT and radar data. It is found that the magnitude of turbulent motions (wa = waowa) within the cloud were 3D (waU′ ≈ V′). The observed and mean wa are represented by wao and wa, respectively.

Figure 10 is the same as Fig. 9, but for a clear air segment between 8000 and 9000 s on 24 September. This figure shows that the TKE of the vertical component of wind at 8 km is smaller than the TKE of U and V. This implies that TKE in the cloud-free regions of the Arctic is likely transferred by the horizontal 2D turbulent eddies. This is probably due to a stronger inversion caused by radiative cooling at cloud top under the clear skies. The TKE can be transferred to the upper troposphere mostly by wa within the cloud systems. TKE transfer into the upper troposphere within the cloud is related to the magnitude of wa fluctuations, which are larger within the cloud as compared to clear air.

A time series of wa was used to estimate the size of coherent structures (e.g., cells). Figure 11a shows wa versus time along a constant altitude flight leg within the cloud. The dark solid line is obtained by filtering out scales less than 20 km from the observations using a 200-s running average. The thin line is for the results after filtering scales larger than 20 km. The wa in both scales have a maximum near 3 m s−1, but the sizes of structures in the horizontal are different. The extent of the horizontal distance for upward motion regions is larger as compared to those of downward motions. This indicates that the cloud system is in the developing stage at low levels. Note that in the case of wetting and sharp turns, uncertainty in wa can be as much as 40%–50%. Figure 11b shows a wavelet analysis of wa along a constant altitude flight leg for a 128-s time period. The space scale is obtained using the aircraft true air speed of about 85 m s−1. In this plot, two distinct regions are seen: 1) microscale structures and 2) mesoscale structures. Microscale structures have sizes of less than 100–200 m. Mesoscale structures (e.g., cells and gravity waves) have a horizontal extent of up to 6 km. However, all these various scales interact with each other to sustain the cloud system. This is seen with various frequencies at a given time.

As indicated earlier, Arctic clouds (both stratiform and cumuliform types) are not only a sink for moisture and heat fluxes; they can play a very important role for transferring heat and moisture. A histogram representing the number of points versus SHF over a 135-s time interval, along a constant altitude flight leg at about 0.4 km (close to the cloud base of 0.2 km), is shown in Fig. 12a. In this particular leg, most points represent downward fluxes. Figure 12b shows the cumulative probability (cp) values together with individual probability values versus binned SHF (solid line). The median SHF value is about −50 W m−2. The single largest probability of occurrence of the SHF value in a bin is found near +20 W m−2. The median value helps describe the distribution of data points whereas mean values can be distorted by a few points with large values. Figure 12c shows the time series of SHF. The top panel in this figure shows four possible combinations of wa and T′. They represent upward and downward motion for wa, and cold and warm air for T. The maximum downward SHF is about −400 W m−2. The upward SHF has a maximum at about 300 W m−2. The ratio of warm to cold downdrafts is about 2. The fraction of warm to cold updrafts is approximately 40%, and the intensity of fluxes for cold updrafts is larger than that of warm updrafts. This type of analysis can be more useful for climate studies if longer time periods for various cases are used.

The role of LHF in Arctic clouds is not well known because of the lack of accurate values of the dewpoint temperature. Here, vapor-mixing ratio values obtained from a Russian L-α hygrometer (Mezrin 1997) are used to calculate LHF. Figure 13 is similar to Fig. 12 except LHF replaces SHF. Probability values of LHF in Fig. 13a shows a near-normal distribution. The median value is slightly smaller than zero (Fig. 13b). A time series of LHF is shown in Fig 13c. The top panel in this figure shows four possible combinations of wa and qυ; they represent upward and downward motion for wa, and moist and dry air for qυ. Single point values of LHF are found between +500 and −500 W m−2. Approximately 60% of the LHF values have a negative value. The ratio of moist downdrafts to dry downdrafts is about 0.5, and there were 25% more moist updrafts than dry updrafts. It should be noted that the uncertainty in the LHF, based on a visual analysis of wa and qυ, can be as high as 30%–40% (Gultepe and Rao 1993). The horizontal extent for both upward and downward LHF is about 0.5–2 km.

(ii) The 24–25 September case

1. The 24 September flight: Time series of RHw, T, z, LWCN, LWCf, Ni, and TWCN (similar to Fig. 8) are shown in Figs. 14a–g, respectively. In the beginning of the flight (T = 0°C), all LWC probes indicated the presence of liquid water of at least 0.1 g m−3. After t = 2000 s, ice crystals were clearly present between −20° and −30°C. The Ni was between 50 and 600 L−1. In the later time period, after 15 000 s, LWC regions of about 0.2 g m−3 are also found, and they may coexist with ice particles. LWC and IWC regions cannot be distinguished if small particles (less than 100 μm) are not classified properly into liquid or ice. The wa observations had a large uncertainty because of wetting problems during this flight. Therefore, the fluxes can be used only with caution.

The time series of wa (4000–5000 s) just below 8 km, close to cloud top, is shown in Fig. 15a and a wavelet analysis applied to observations after 4000 s is shown in Fig. 15b. The thin line in Fig. 15a represents scales less than 20 km and the thick solid line is for larger scales. It can be seen that the magnitude of wa is much smaller at high altitudes as compared to the low altitudes of the 8 September case. The magnitude of wa (±0.3 m s−1) for both small and larger scales are similar to each other. Figure 15b shows that the sizes of mesoscale structures are about 1 km.

High-frequency moisture measurements are not available for this case, and only SHF at about 8-km height (2800 s in Fig. 14) within the cloud are calculated from wa and T fluctuations. Figures 16a–c show a histogram of the number of points used in the SHF calculation, the probability and cp values, and a time series of SHF, respectively. The median value of the SHF is about 5–10 W m−2 (Fig. 16b). In Fig. 16c, at about 45-s elapsed time (corresponding to 2800 s in Fig. 14), the SHF is about 80 W m−2. Downward SHF values are much smaller than upward SHF. At about 60 s, warm air was transferred downward with eddies. Many positive and negative heat fluxes are concentrated in the range of −20 to +20 W m−2 (Fig. 16c). The ratio of SHF for a warm updraft to that of a cold updraft is about 2.5. The ratio of cold air to warm air in the downdrafts is about 4.

2. The 25 September flight: Aircraft observations on this day were made approximately from 1600 to 1900 LST. The aircraft performed a profile prior to the constant altitude flight and collected data in another profile before landing at Inuvik airport. Figure 17 displays the same parameters as Fig. 8. Both profiles showed regions where droplets were present at low altitudes and ice crystals were farther aloft. The liquid water and ice regions are seen at about −10°C at t = 2500 s. At t < 2500 s, Nd reached about 100 cm−3 (not shown), and LWC had a maximum near 0.25 g m−3. RHw during the profiles reached 100% in the phase change regions. TWC is dominated by droplets when T is warmer than about −10°C, and dominated by ice crystals elsewhere. Note that observations were collected over slightly different locations over the region. LWC reached a maximum of 0.3 g m−3 before landing (11 000 s) and Ni reached a maximum of 600 L−1 at about 2500 s.

(iii) Microphysical characteristics of Arctic clouds
The mean and std dev values of the LWC, TWC, IWC, Nd, Ni, effective radius of droplets (reffd) and ice (reffi) versus 5°C temperature intervals are shown in Tables 2, 3, and 4 for the 8 September, 24 September, and 25 September flights, respectively. Mean and std dev values of 2D-C probe data (Ni) are obtained for about twenty 5-min flight segments. Mixed phase conditions are not included in the analysis. Regions of ice and liquid water are found using measurements from the FSSP, Nevzorov probes, and RHw. A large uncertainty (∼30%–50%) in reffi is possible in this case because of the unknown phase of particles when the size is less than 100 μm. A decrease in reff due to the large number of ice particles in the smaller size range, and an increase in reff due to the contribution of the particles with sizes greater than 800 μm are possible. These uncertainties using a theoretical particle size distribution (0–1200 μm) are about 30% and 10%, respectively. Here, reffd and reffi are, respectively, obtained from FSSP and 2D-C probe measurements as
i1520-0442-13-7-1225-e3
where reqi is calculated assuming that an ice crystal volume is equal to that of a droplet.

The results for the 8 September case are shown in Table 2. The maximum value (60 μm) of reffi occurs at about −17.5°C and the minimum value (44 μm) at about −7.5°C. Here Nd values are found to be around 45–65 cm−3. The IWC value is about 0.02–0.09 g m−3. LWC ranges from 0.03 g m−3 to 0.12 g m−3. Using the particle shape recognition method given by Heymsfield and Parrish (1978), the mean and std dev values of the probability of occurrence (po) of “recognized” dendrite, column, plate, and aggregates (poly crystals) are estimated as 4 ± 5%, 8 ± 5%, 4 ± 14%, and 84 ± 14% for the entire flight period, respectively. Most of the particle shapes with sizes less than 100 μm are defined as unknown and not included in the shape recognition analysis. The T and RHw are shown in Fig. 8. The ratio of Ni with particle sizes less than 100 μm to total Ni is about 35%–50%. The shapes of ∼50% of the particles with sizes greater than 100 μm are also not known properly (Korolev et al. 1999). The mean and std dev of reffd are about 8.3–9.7 μm and 1.5–3.0 μm, respectively. The mean Ni is less than 210 L−1 at all T’s. Here reffi ranges between 44 and 60 μm.

The results for the 24 September flight are shown in Table 3. For the entire time period, the po values for corresponding particle shapes with sizes greater than 100 μm are 6 ± 4%, 5 ± 2%, 1 ± 0.2%, and 88 ± 3%, respectively, again the number concentration of unknown or irregular particles can be very large. The T and RHw are given in Fig. 14. TWC (IWC) did not change significantly with T. The Ni gradually increased with T from about −7.5°C to −27.5°C. The values of reffi ranged between 47 and 70 μm. Observations corresponding to this table showed that LWC is about 0.3 g m−3 for a short time period at the beginning of the flight. For this reason, it is not shown in Table 3. The rest of the observations represent only ice or mixed phase.

The results for the 25 September flight are shown in Table 4. The po values for the corresponding particle shapes are 12 ± 14%, 5 ± 5%, 2 ± 5%, and 81 ± 15%, respectively. At the same temperature range, LWC values for the 25 September case are comparable with those of the 8 September flight. In general, IWC estimates are found between 0.03 g m−3 and 0.34 g m−3, with greater variability for T < −10°C. The mean and std dev of Nd at −2.5(2.5)°C are 132(81) cm−3 and 43(54) cm−3, respectively. The reffd is equal to ∼7.5 μm. The Ni is about 250–400 L−1 between −10°C and 0°C, decreasing to 31 L−1 at about −17.5°C. For comparison, observations (Fig. 8) corresponding to Table 4 showed that the LWC is about 0.3 g m−3 at the beginning of the flight. The mean Ni was about 600 L−1 at −25°C at 4000 s.

The averaged reffi in the interval of −20°–0°C is 59 μm for the 24 September flight and 50 μm for the 25 September flight as compared to 51 μm for the 8 September case. Although these values seem to be within the error range of the 2D-C probe, because of density differences between liquid water and ice, and the ice crystal length used in the calculations, a change in reffi of about 10 μm could be important. This indicates that the effective radius of ice crystals change significantly from case to case. The larger po values of aggregates (as shown by the aircraft data) indicate that the magnitude of the dynamical activity (e.g., wa) for the 8 September case was larger than the other case. Strong wa results in more aggregates and multibranched ice crystals (Heymsfield 1975). The direct use of mean wa is avoided because of the large uncertainty in mean wa from aircraft measurements. The maximum Ni for the 8 September case is about 200 L−1 less than 600 L−1 found in the 24–25 September case (Figs. 14–17).

6. Discussion

In this section, the importance of dynamical processes for cloud development, the size of structures, and their effect on climate change will be discussed. In addition, comparisons of cases based on microphysics and dynamics will be summarized.

a. Importance of dynamical processes and parameterization of ice crystal number concentration

Dynamical processes within the clouds are very important for the vertical transport of heat and moisture. They are also responsible for distributing TKE from 2D eddies into 3D eddies in the atmosphere. The vertical air velocities (wa) are still not measured accurately from instruments mounted on aircraft (Gultepe and Starr 1995).

A method to study the importance of wa on the cloud microphysical parameters (e.g., ice crystal number concentration, Ni) is obtained using the growth rate (i/dt) of the mass of the ice crystal population by diffusion (Fletcher 1962) as
i1520-0442-13-7-1225-e5
where Si is the supersaturation with respect to ice, Gi the modified diffusivity, fj the ventilation coefficient, Cj the shape factor for the specific size L, ρj the ice crystal density, and m the number of bins in the ice crystal size spectra. Units for each parameter are given in appendix A. Note that only ice crystals are used in the calculations because of the small percentage (∼10%) of droplets compared to ice particles. In Eq. (5), the diffusional growth of ice particles is considered, and other processes, for example, aggregation and riming, are neglected.
Using the ice crystal growth equation and assuming that the horizontal advection, turbulent flux, and local change terms are small compared to other terms, the heat budget equation (Gultepe and Rao 1993; Gultepe et al. 1990) can be written as
i1520-0442-13-7-1225-e6
where θ is the potential temperature, z the altitude, Ls the latent heat of sublimation, cp the specific heat at constant pressure, ρa the air density, and t the time. Here IR and SW represent infrared and shortwave radiative fluxes. The term I is the vertical heat advection, the term II the latent heat release due to sublimation, and the term III the radiative heating/cooling rate (Qr) due to absorption/emission of radiation. Using a unimodal size distribution, and knowing θ and Si, Ni is obtained from the above equations as follows:
i1520-0442-13-7-1225-e7
This equation is used to calculate Ni as a function of ice crystal size L, Qr, Si, and wa. Results from this method emphasize the importance of cloud dynamics and radiative cooling on the microphysical parameters (e.g., Ni).

The assumptions of the unimodal ice crystal size distribution—no nucleation or aggregation, no turbulent flux term, and spherical particle shape—are used in Eq. (7) to obtain Ni. These assumptions to obtain Ni are reasonable compared to what is used in the present large-scale models. Present GCMs assume constant Ni and shape, or they calculate Ni from NiT relationships (Meyers et al. 1992). In Eq. (7), Ni is a function of θ, Qr, wa, temperature gradient (∂θ/∂z), and Si. These are the most important parameters affecting Ni, under the given assumptions. A problem may arise due to subgrid-scale variation of wa. The wa can be divided into mean and fluctuation components. Fluctuations of wa for a particular scale are a function of dynamical processes, for example, cells or small-scale turbulence. Then, wa can be obtained either as a function of space scale, or using probability distribution curves.

Four cases with different reffi, Si, Qr, and temperature gradients shown in Table 5 are applied to Eq. (7) and the results are shown in Fig. 18. Case 1 represents observations of both cases used in this study. The Qr about −20°C day−1 is assumed (Tsay al. 1989; Pinto et al. 1995). The Qr was not available during BASE. Case 2 is chosen for a small particle size equal to 25 μm. In this case, it is assumed that Si is small and Qr is large compared to case 1. This is true when particle size becomes smaller, resulting in a large radiative effect in the climate models. The temperature gradient (∂T/∂z) is taken approximately as −0.01°C m−1 for both cases from observations. For case 3, a large particle size is assumed when compared to the other two 2 cases. The Si is approximately taken equal to 0.10 from observations assuming saturation with respect to liquid water. The Qr = −40°C day−1 (Tsay et al. 1989) is chosen as an upper limit with a temperature gradient of −0.001°C m−1, representing stable conditions in the environment. Case 4 represents unstable environmental conditions with a particle size of 100 μm. The Ni calculated from Eq. (7) using Table 5 and changing vertical air velocity from 0.1 m s−1 to 1 m s−1 (representing various dynamical conditions) is shown in Fig. 18a. The horizontal dotted line in Fig. 18a is for the maximum ice crystal number concentration (Nimax) observed during BASE. Note that the results from case 3 (stable case) are shown with a line at Ni ∼ 4 L−1 in this figure. The reason is that the product of a large positive value of ∂θ/∂z and a constant small value of wa (∼0 m s−1) in a stable environment does not change with wa, resulting in a constant value of Ni.

The total number concentration of ice crystals, obtained from the 2-DC probe strobe counts, versus temperature for the entire BASE data is shown in Fig. 18b. It is seen that the maximum Ni is less than 2000 L−1 for a given T value. This figure indicates that Ni may be a function of other parameters (e.g., wa and Si) besides T. The results show that Ni is strongly dependent on wa and reff (or L). Real observations can be found anywhere below the case 2 and Nimax, lines on Fig. 18 depending on the environmental conditions. For a given time period, the values assumed to be constant in Table 5 may change significantly from case to case because of complex interactions involving physical, dynamical, and radiative processes. It is seen for cases 1 and 2 that Ni changes quickly at small values (<0.20 m s−1) of vertical air velocity. This indicates that wa should be included in the microphysical parameterization used in GCMs/climate models.

b. Size of dynamical structures

Clouds in the Arctic atmosphere can include structures with sizes up to 30–40 km, depending on the environmental conditions. Stratiform clouds in the Arctic are believed to be homogeneous, but this study indicates that wa on a 100-m scale can easily exceed 0.5 m s−1. Embedded cells within the cloud systems are also responsible for extending cloud lifetime (Gultepe and Rao 1993; Gultepe et al. 1995). Convective cells may persist in the higher latitudes, depending on the thermodynamical and dynamical conditions of the cloud systems. Table 6 shows that the sizes of cells from all platforms were between 0.1 and 15 km. Note that results from LANDSAT are only good for a column-averaged value that does not see the individual cells at different altitudes. Cell definition in this case is given as when a reflectance value is greater than mean + std dev value over the region of interest. Their magnitude cannot be as large as those seen in midlatitude deep-convective clouds, but their relative strength and their frequency of occurrence cannot be neglected. For modeling purposes, accurate information on the dynamical structures and their sizes are important for choosing grid intervals for both time and space. Arctic clouds have different boundary conditions compared to midlatitude clouds. This makes the present study unique because of the shortage of statistically meaningful values for dynamical processes.

c. Comparisons of cases

Comparisons of observations from two cases indicated that Arctic clouds are dynamically active in transferring heat and moisture into upper levels. The most important difference between the two cases is that the clouds for 8 September developed in an air mass originating over the Arctic Ocean, whereas for the 24–25 September case, the air mass originated over the Pacific Ocean. Although only two cases are studied here representing air masses originating from the Arctic and Pacific Oceans, the 13 days with air masses of Arctic Ocean origin and the 16 days of Pacific Ocean origin had distinctive microphysical and dynamical characteristics. As was determined in the same geographical area using First International Satellite Cloud Climatology Project Regional Experiment-Arctic Cloud Experiment (FIRE–ACE) data (Gultepe and Isaac 1999), cloud microphysical and dynamical characteristics were similar when their origin was the same.

1) Rawinsonde observations

The surface temperature over land for the 8 September case was about 0°C. On this day, cooler air moved from the Arctic Ocean over the project area, resulting in embedded convective elements within the cloud. Temperature gradients are not as strong as the 24–25 September case, because the air was from relatively colder ocean surfaces. For the 24–25 September case, when warmer air moved over the cold surface in the project area, individual convective elements were seen on this day compared to the embedded cells of the 8 September case. Figures 1a and 1b show that the surface T for the 24–25 September case was cooler than the 8 September case.

2) Doppler radar observations

Based on Doppler radar observations, the reflectivity values for the 8 September case during two time periods (1439–1552 and 1630–2245 UTC) were most often near 11 dBZ (Fig. 19a), and for the 25 September case (0313–0430 and 1240–1341 UTC), they were about 6 dBZ (Fig. 19b). The large number of values (10%) beyond 20 dBZ in Fig. 19a was primarily in the melting layer (see Fig. 3). For the 25 September case, the number of points for the same criteria is negligible. The median reflectivity on the 8 and 25 September cases was approximately 10 and 5 dBZ, close to the aforementioned modal values, respectively. Overall, the percentage of significant reflectivity values in Arctic stratiform clouds indicates that their internal structure can be important for climate studies. Cooling due to melting process results in large T gradients in the vertical, and this situation can promote stronger SHF and LHF. Eventually, melting results in stabilization of the atmosphere. It is estimated that overall 5 out of 13 Intensive Observation Period cases (∼35 ± 15%) during BASE had a melting layer. Thick melting layers can be due to 0°C values in a deep mixed boundary layer. The present melting-layer thickness (∼1 km) compared to those of midlatitude cloud systems can be explained by the well-mixed deep boundary layer with T ∼ 0°C. This thickness of the melting layer is comparable with a 0.1–1-km thickness found in midlatitudes (Houze 1981; Houze et al. 1981; Stewart et al. 1996). Zerr (1997) showed that the melting layer increases with decreasing stability around 0°C and increasing ice particle mass.

Horizontal and vertical coverage of convective cells is shown using the number of points (Np) versus size (L)–height (h) plots. PPI volume scans transformed to a Cartesian grid of 0.5-km resolution are used in the analysis, but only two examples are discussed. Figure 19c is for the 8 September case (1840 UTC). In this figure, Np is obtained using a criteria of Ze > 10 dBZ. This value is chosen based on the median value of Ze for the 8 September case. Figure 19d is for the 25 September case (1250 UTC). The total Np values for Figs. 19c and 19d were 502 and 1018, respectively. Figure 19c shows that the size of cells varied from 0.5 to 5 km in the horizontal (Np > 20). Their vertical extent varied from 1 to 2 km within the stratiform cloud. Figure 19d shows that the size of cells varied from 0.5 to 3 km in the horizontal. In this case, their vertical extent varied from 0.3 to 1 km. It should be noted that the size of horizontal structures was found close to 15–20 km for some other time periods. For the 8 September case, some cell tops even reached 3.3-km height. These results indicate that a statistical analysis of dynamical structures can help us to better understand the convective cell size distribution and underlying formation mechanisms.

3) LANDSAT observations

Examples from LANDSAT images (Figs. 5 and 6) indicated that the size and intensity of convective cells for the 8 September case were different than those of the 24–25 September case. The later case had structures with a large horizontal extent (about 10–20 km) compared to a few km for the 8 September case. It should be noted that the uncertainty in cloud reflectance can be large if the optical thickness of the cloud becomes smaller. The visual and radar observations indicated that the clouds were physically (optically) thick enough to reduce the effect of surface albedo. Therefore, this uncertainty in the calculations is neglected. Reflectance values from satellites are a function of Ni, reff, and IWC. But, assuming ice crystal length and IWC change significantly with wa in convective cells, reflectance from LANDSAT can provide us some information on the size of structures. Because of limited insitu observations and the high uncertainty in wa, the verification for this assumption is left to a new dataset from the FIRE–ACE.

4) Aircraft observations

Aircraft observations for the 8 September case showed that the magnitude of the dynamical processes (e.g., cells, turbulence) were large compared to these of the earlier studies in the Arctic and midlatitudes. For the 24–25 September case, the magnitude of wa at about 8 km was also comparable with values found in midlatitudes. The maximum wa for cells for the 8 September case (Fig. 11) was about ±2 m s−1. The wa equal to ±0.3 m s−1 at about 8 km for the 24–25 September case is found to be significant (Fig. 15). The maximum LWC value was about 0.3 g m−3. In general, the maximum Ni for the 8 September case is found to be smaller than for the 24–25 September case (∼400 vs 600 cm−3, Fig. 14), perhaps indicating the effect of aerosols with a different origin and T. The Na for the 8 September case had values less than 200 cm−3 compared to other cases of more than 400 cm−3 in the boundary layer. The maximum IWC was larger for the 8 September case (∼0.3 g m−3) than the 24–25 September case (0.1 g m−3), explaining the importance of air masses with different physical and dynamical characteristics. Profiles made during ascents and descents showed that wa, Uh, D, θe, qυ, and Na were significantly larger for the 8 September case compared to the 24–25 September case. The difference between two soundings for a given parameter was also much larger for the 8 September case. Figures 7b and 7c are used together for the estimation of differences. However, the advection of the moisture and heat (temperature) in the horizontal for the 24–25 September case was very strong. It is likely that this resulted in larger mesoscale structures seen in the radar and LANDSAT images (see Figs. 3, 4, 5, and 6).

The effective size of ice crystals (reffi) was calculated using different assumptions. The difference among them was significant, and can easily exceed 20 μm. In the calculations, if particle sizes less than 100 μm were included, the difference can be much larger than 20 μm. Another important issue related to the effective size calculation of ice crystals is the assumption of columnar shape for all particles, as used in GCMs (Kiehl et al. 1996; Ebert and Curry 1993b). This study showed that less than 10% of the ice crystals have a columnar shape. Unless these issues are properly studied, the calculations from both GCMs and satellite retrievals of cloud microphysical parameters may include large uncertainties when addressing climate change variability.

The SHF and LHF calculations indicated that Arctic clouds transfer significant amounts of moisture and heat to the upper troposphere by means of eddies in micro- and mesoscales. Although individual points can be upward or downward, their median (or mean) values are found positive. A detailed analysis of fluxes averaged over longer time periods and several levels needs to be considered for a better understanding of climate change. The accuracy of the observations cannot be quantified for the 24–25 September case because of icing problems in the wind measurement instruments. The values of LHF and SHF within the cloud were found comparable to those obtained from Arctic leads by Andreas et al. (1979). They are also found comparable with SHF up to 80 and LHF = −20 to +20 W m−2 calculated in midlatitude boundary layer clouds (Albrecht et al. 1985).

7. Conclusions

In this study, observations from aircraft, Doppler radar, LANDSAT, and rawinsondes are used to better understand microphysical and dynamical characteristics of Arctic clouds. The results suggest that dynamical structures and their relationship to microphysical parameters within Arctic clouds need to be explored in further detail. Developing a better parameterization for model applications and remote sensing studies is strongly dependent on an improved understanding of both microphysical and dynamical processes. The effects of these processes in the calculation of optical parameters are important for climate change studies.

Some conclusions from the present study can be summarized as follows:

  1. Developments in the latest observational technology (e.g., TWC probe measurements), together with better observational methods, can lead to more accurate measurements of IWC within the Arctic clouds.
  2. Dynamical parameters (e.g., wa, LHF, and SHF) in size and magnitude are similar to turbulent fluxes measured within midlatitude cloud systems (Belair et al. 1998; Albrecht et al. 1985), including frontal clouds and boundary layer clouds.
  3. Arctic clouds are not stable and homogeneous when they are related to large-scale synoptic systems (e.g., cyclonic systems). The cases studied likely have characteristics related to the region and time of the year. Note that clouds related to anticyclonic circulations can have different microphysical and dynamical characteristics.
  4. Doppler radar and LANDSAT observations can be used to study the size of dynamical structures and their space variability.
  5. Vertical air velocity is important for IWC and Ni parameterizations. But, it cannot be measured accurately with the current technology.
  6. Ice microphysics related to IWC, effective size, and particle shape needs to be better described and quantified for both GCMs (Fowler et al. 1996; Ghan et al. 1997; Gultepe et al. 1998) and remote sensing applications (Francis et al. 1998). The present GCMs use a columnar shape for ice crystals, and the number concentration is only a function of temperature (Fletcher 1962) or both temperature and supersaturation with respect to ice (Meyers et al. 1992).
  7. Reflectivity fields from Doppler radar indicate that Arctic cloud systems formed in air masses from the Arctic Ocean have more intense Ze values at low levels than those of cloud systems originating from the Pacific Ocean. This can be explained by cold-air surges from the Arctic over open water surfaces, resulting in convection that transports heat and moisture from the boundary layer to the free atmosphere.
  8. Melting layers (∼35 ± 15% of time) during transition seasons can be important in climate studies. These layers can generate a strong temperature gradient and wind shear in a layer up to 1 km thick, resulting in an increase in LHF and SHF.
  9. Model time and space steps should consider the scale of the observed structures in the Arctic region.

The observations from Arctic clouds are still too sparse to be climatologically significant. The case studies here suggest that dynamical processes can have comparable importance to those at midlatitude locations. More observations for other cases from BASE and FIRE–ACE could be used for better parameterization of IWC/TWC versus T and wa. Equation (7) can be modified to obtain Ni using parameters from GCMs. In this case, wa is an important variable on the subgrid-scale that must be considered for a better understanding of climate change in the Arctic region.

Doppler radar observations combined with the LANDSAT data should also be utilized for the study of dynamical processes, and their effects on cloud physical processes. This will assist the users of climate and mesoscale models to better validate their model results. The wavelet analysis should also be used for a 2D analysis of convective elements within the cloud systems.

The new observations from FIRE–ACE, which were collected over the Arctic region during April–July 1998, can be used for further validation of the results from earlier field projects. Accurate observations of small ice crystal habit, dewpoint temperature, and mixing ratio are needed for microphysical parameterizations.

Acknowledgments

The authors would thank to Dr. A. K. Liu of NASA Goddard Space Flight Center, Greenbelt, Maryland, for the wavelet analysis related discussions. The authors also thank Dr. M. Y. Mezrin of the Central Aerological Observatory, Dolgoprudny, Moskow, Russia, for the atmospheric moisture measurements. K. Sung, M. Couture, and D. Jahani of the Cloud Physics Research Division of AES are also thanked for their assistance in the data analysis. The aircraft wind data analysis was helped by discussions with Ian MacPherson of NRCC.

REFERENCES

  • Albrecht, B. A., R. S. Penc, and W. H. Schubert, 1985: An observational study of cloud-topped mixed layers. J. Atmos. Sci.,42, 800–822.

  • Andreas, E. L., C. A. Paulson, R. M. Williams, R. W. Lindsay, and J. A. Businger, 1979: The turbulent heat flux from Arctic leads. Bound.-Layer. Meteor.,17, 57–91.

  • Baumgardner, D., J. W. Strapp, and J. E. Dye, 1985: Evaluation of the forward scattering spectrometer probe. Part II: Corrections for coincidence and dead-time losses. J. Atmos. Oceanic Technol.,2, 626–632.

  • Belair, S., P. Lacarrere, J. Noilhan, V. Masson, and J. Stein, 1998: High-resolution of surface and turbulent fluxes during HAPEX-MOBILHY. Mon. Wea. Rev.,126, 2234–2253.

  • Curry, J. A., and G. F. Herman, 1985: Relationship between large-scale heat and moisture budgets and the occurrence of Arctic stratus clouds. Mon. Wea. Rev.,113, 1441–1457.

  • ——, E. E. Ebert, and G. F. Herman, 1988: Mean and turbulence structure of the summertime Arctic cloudy boundary layer. Quart. J. Roy. Meteor. Soc.,114, 715–746.

  • ——, J. L. Schramm, and E. E. Ebert, 1993: Impact of clouds on the surface radiation balance of the Arctic Ocean. Meteor. Atmos. Phys.,51, 197–217.

  • ——, W. B. Rossow, D. Randall, and J. L. Schramm, 1996: Overview of Arctic Cloud and radiation characteristics. J. Climate,9, 1731–1764.

  • Ebert, E. E., and J. A. Curry, 1993a: An intermediate one dimensional thermodynamic sea ice model for investigating ice–atmosphere interactions. J. Geophys. Res.,98 (D4), 10 085–10 109.

  • ——, and ——, 1993b: A parameterization of ice cloud optical properties for climate models. J. Geophys. Res.,97 (D4), 3831–3836.

  • Fletcher, N. H., 1962: The Physics of Rainclouds. Cambridge University Press, 386 pp.

  • Fowler, D. L., D. A. Randall, and S. A. Rutledge, 1996: Liquid and ice cloud microphysics in the CSU general circulation model. Part I: Model description and simulated microphysical processes. J. Climate,9, 489–529.

  • Francis, P. N., P. Hignett, and A. Macke, 1998: The retrieval of cirrus cloud properties from aircraft multi-spectral reflectance measurements during EUCREX ’93. Quart. J. Roy. Meteor. Soc.,124, 1273–1291.

  • GEWEX News, 1997: Satellite cloud data for regional studies. GEWEX News,7, 1–3.

  • Ghan, S. J., L. R. Leing, and Q. Hu, 1997: Application of cloud microphysics to NCAR community climate models. J. Geophys. Res.,102 (D14), 16 507–16 527.

  • Gultepe, I., and G. V. Rao, 1993: Moisture and heat budgets of a cirrus cloud from aircraft measurements during FIRE. Quart. J. Roy. Meteor. Soc.,119, 957–974.

  • ——, and D. O’C. Starr, 1995: Dynamical structure and turbulence in cirrus clouds: Aircraft observations during FIRE. J. Atmos. Sci.,52, 4159–4182.

  • ——, and G. A. Isaac, 1999: The influence of air mass origin on the microphysical structure of Arctic boundary layer clouds. 33d CMOS Congress, Montreal, PQ, Canada, CMOS, 22.

  • ——, A. J. Heymsfield, and D. H. Lenschow, 1990: A comparison of vertical velocity in cirrus obtained from aircraft and lidar measurements. J. Atmos. Oceanic Technol.,7, 58–67.

  • ——, D. O’C. Starr, A. J. Heymsfield, T. Uttal, T. P. Ackerman, and D. L. Westphal, 1995: Dynamical characteristics of cirrus clouds from aircraft and radar observations in micro and meso-gamma scales. J. Atmos. Sci.,52, 4060–4078.

  • ——, G. A. Isaac, A. Korolev, S. Cober, and J. W. Strapp, 1998: Effects of ice crystal shape on the radiative characteristics of low level ice clouds. Preprints, 14th Conf. on Cloud Physics, Everett, WA, Amer. Meteor. Soc., 372–375.

  • Hense, A., and E. Heise, 1984: A sensitivity study of cloud parameterizations in general circulation models. Beitr. Phys. Atmos.,57, 240–258.

  • Heymsfield, A. J., 1975: Cirrus uncinus generating cells and the evolution of cirriform clouds. Part II: The structure and circulations of the cirrus uncinus generating head. J. Atmos. Sci.,32, 809–819.

  • ——, and J. L. Parrish, 1978: A computational technique for increasing the effective sampling volume of the PMS two-dimensional particle size spectrometer. J. Appl. Meteor.,17, 1566–1572.

  • Houze, R. A., 1981: Structure of atmospheric precipitating systems:A global survey. Radio Sci.,16, 671–689.

  • ——, S. A. Rutledge, T. J. Matejka, and P. V. Hobbs, 1981: The mesoscale and microscale and organization of clouds and precipitation in midlatitude cyclones. Part III: Air motions and precipitation growth in a warm-frontal rainband. J. Atmos. Sci.38, 639–649.

  • Hudak, D. R., and R. Nissen, 1996: Doppler radar applications in major winter snow storms. Atmos. Res.,41, 109–130.

  • ——, R. E. Stewart, G. W. K. Moore, and E. T. Hudson, 1995: Synoptic considerations of storms in the southern Beaufort Sea—expectations for BASE. Proc. Fourth Conf. on Polar Meteor. Oceanogr., Dallas, TX, Amer. Meteor. Soc., 234–237.

  • ——, R. List, and T. Krauss, 1996: Microphysical properties of Arctic cyclones. Preprints, 12th Int. Conf. on Clouds and Precipitation, Zurich, Switzerland, International Commission on Clouds and Precipitation, 158–161.

  • Isaac, G. A., and R. A. Stuart, 1996: Relationships between cloud type and amount, precipitation and surface temperature in the Mackenzie River valley–Beaufort Sea area. J. Climate,9, 1921–1941.

  • Kiehl, J. T., J. J. Hack, G. B. Bonan, B. A. Boville, B. P. Briegleb, D. L. Williamson, and P. J. Rasch, 1996: Description of the NCAR Community Climate Model (CCM3). NCAR Tech. Note NCAR/TN-420+STR, National Center for Atmospheric Research, 152 pp. [Available from NCAR, P.O. Box 3000, Boulder, CO 80307-3000.].

  • Korolev, A. V., J. W. Strapp, G. A. Isaac, and A. Nevzorov, 1998: The Nevzorov airborne hot wire LWC/TWC probe: Principles of operation and performance characteristics. J Atmos. Oceanic Technol.,15, 1496–1511.

  • ——, G. A. Isaac, and J. Hallett, 1999: Ice particle habits in Arctic clouds. Geophys. Res. Lett.,26, 1299–1302.

  • Liu, A. K., C. Y. Peng, B. Chapron, E. M. Christensen, and N. E. Huang, 1995: Direction and magnitude of wind stress over wave groups observed during SWADE. Global Atmos. Ocean. Sys.,3, 175–194.

  • MacPherson, J. I., 1993: Use of a wing-mounted airflow pod for airborne wind and flux measurements. Preprints, Eighth Symp. on Meteor. Obser. and Instrumentation, Anaheim, CA, Amer. Meteor. Soc., 169–174.

  • McInnes, K. L., and J. A. Curry, 1995: Modelling the mean and turbulence structure of the summertime Arctic cloud boundary layer. Bound.-Layer Meteor.,73, 125–143.

  • Meyers, M. P., P. J. DeMott, and W. R. Cotton, 1992: New primary ice-nucleation parameterizations in an explicit cloud model. J. Appl. Meteor.,31, 708–721.

  • Mezrin, M. Yu, 1997: Humidity measurements from aircraft. Atmos. Res.,44, 53–59.

  • Pinto, J. O., and J. A. Curry, 1995: Atmospheric convective plumes emanating from leads. Part II: Microphysical and radiative processes. J. Geophys. Res.,100 (C3), 4633–4642.

  • ——, and ——, 1997: Role of radiative transfer in the modeled meso-scale development of summertime Arctic stratus. J. Geophys. Res.,102 (C3), 13 861–13 872.

  • ——, ——, and K. L. McInnes, 1995: Atmospheric plumes emanating from leads. Part I: Thermodynamic structure. J. Geophys. Res.,100 (C3), 4621–4631.

  • Schnell, R. C., R. G. Barry, M. W. Miles, E. L. Andreas, L. F. Radke, C. A. Brock, M. P. McCormick, and J. L. Moore, 1989: Lidar detection of leads in Arctic Sea ice. Lett. Nature,339, 530–532.

  • Stewart, R. E., R. W. Crawford, K. K. Szeto, and D. R. Hudak, 1996:Horizontal aircraft passes across 0°C regions within winter storms. Atmos.-Ocean,34, 133–159.

  • Thomson, A. D., and R. List, 1996: Raindrop spectra and updraft determination by combining Doppler radar and distrometer. J. Atmos. Oceanic Technol.,13, 465–476.

  • Tsay., S.-C., K. Stamnes, and K. Jayaweera, 1989: Radiative energy budget in the cloudy and hazy Arctic. J. Atmos. Sci.,46, 1002–1018.

  • Wilson, L. D., J. A. Curry, and T. P. Ackerman, 1993: Satellite retrieval of lower-tropospheric ice crystal clouds in the polar regions. J. Climate,6, 1467–1472.

  • Zerr, R. J., 1997: Freezing rain: An observational and theoretical study. J. Appl. Meteor.,36, 1647–1661.

  • Zhang, T., S. A. Bowling, and K. Stamnes, 1997: Impact of the atmosphere on surface radiative fluxes and snowmelt in the Arctic and subArctic. J. Geophys. Res.,102 (D4), 4287–4302.

APPENDIX

Symbols Used in Deriving Eq. (7)

  • Ci  (μm):   Shape factor.
  • cp   (J K−1 kg−1):  Specific heat at constant pressure.
  • i/dt  (kg m−3 s−1):  Growth rate of ice crystal population.
  • Fi:   Ventilation coefficient (here equal to 1.5).
  • Gi  (cm2 s−1):   Modified diffusion coefficient.
  • IR and SW:   Infrared and SW fluxes.
  • Ls  (J kg−1):   Latent heat of sublimation.
  • L  (μm):   Ice crystal size.
  • m:   Number of channels (here equal to 1).
  • Ni  (L−1):   Ice crystal number concentration.
  • Si:   Super saturation with respect to ice.
  • T  (°C):   Temperature.
  • wa  (m s−1):   Vertical air velocity.
  • z  (m):   Heights.
  • ρa  (kg m−3):   Air density.
  • ρi  (kg m−3):   Ice crystal density.
  • θ  (K):   Potential temperature.

Fig. 1.
Fig. 1.

(a) Large-scale meteorological conditions for 0000 UTC 9 Sep 1994. Arrows are for horizontal wind (m s−1) at the 500-mb level. Dashed and solid lines are for temperature (°C) and sea level pressure (mb), respectively. (b) Large-scale meteorological conditions for 0000 UTC 25 Sep. Arrows are for horizontal wind (m s−1)at the 500-mb pressure level. Dashed and solid lines are for temperature (°C) and sea level pressure (mb), respectively.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 2.
Fig. 2.

(a) Rawinsonde observations for 1200 UTC 8 Sep. (b) Rawinsonde observations for 0000 UTC 25 Sep.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 3.
Fig. 3.

(a) PPI scan for 1840 UTC and (b) RHI scan for 1946 UTC for the 8 Sep case. The color bar shows the reflectivity scale in dBZ units. In (b), the vertical axis is height and the horizontal axis is distance from the radar.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 4.
Fig. 4.

(a) PPI scan for 0344 UTC and (b) RHI scan for 0500 UTC for the 25 Sep case. The color bar shows the reflectivity field in dBZ units. In (b), the vertical axis is height and the horizontal axis is distance from the radar.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 5.
Fig. 5.

LANDSAT image of the DC0.83 for an area of 20 km × 20 km for the 8 Sep case: (a) for 3D image with vertical axis representing DC0.83, where x and y axes represent pixel number; (b) digital counts vs NS distance from LANDSAT observations after the mean is taken out. Three different scales are seen (see text for details); and (c) the wavelet analysis of the observations obtained for a segment about 35 km in length, with frequency in the y axis and distance in the x axis.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 6.
Fig. 6.

The same as Fig. 5 except for the 24 Sep case.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 7.
Fig. 7.

Vertical profiles of wa, horizontal wind (Uh), wind direction (D), equivalent potential temperature (θe) vapor mixing ratio (qυ), and aerosol number concentration (Na) for (a) 8 Sep, (b) 24 Sep, (c) 25 Sep, are shown in boxes 1–6, respectively. The arrow shows aircraft ascent/descent. Note that winds in (b) not reliable due to instrument wetting; therefore, arrows are not shown.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 8.
Fig. 8.

Time series of aircraft observations of RHw, T, z, LWCN, LWCf, Ni, and TWCN for the 8 Sep case. Horizontal lines show the regions of RHw > 95%. The aircraft speed is approximately 85 m s−1.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 9.
Fig. 9.

Spectral energy density vs wavenumber for the in-cloud wind observations (U, V, and wa) at about 2.8 km (1000–2000 s) for the 8 Sep case. Dashed lines are for −3 and −5/3 slopes.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 10.
Fig. 10.

Same as Fig. 9 but for clear air wind observations at about 8 km (8000–9000 s) for the 24 Sep case.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 11.
Fig. 11.

(a) Time series of wa vs time for the 8 Sep case. The thick solid line represents scales at about 20 km. The combination of thick and thin solid lines gives the original time series from aircraft. (b) Wavelet analysis of wa.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 12.
Fig. 12.

For the 8 Sep flight at about 0.4 km, (a) the number of points vs binned SHF values, (b) probability and cumulative probability values of SHF, and (c) time series of SHF. See text for other symbols.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 13.
Fig. 13.

The same as Fig. 12 except for LHF.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 14.
Fig. 14.

The same as Fig. 8 except for the 24 Sep flight.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 15.
Fig. 15.

(a) Time series of wa vs time for the 24 Sep flight. The thick solid line represents scales at about 20 km. A combination of thick and thin solid lines gives the original time series from the aircraft. (b) Wavelet analysis of wa.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 16.
Fig. 16.

For the 24 Sep flight at about 8 km, (a) the number of points vs binned SHF values, (b) probability and cumulative probability values of SHF, and (c) time series of SHF. See text for other symbols.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 17.
Fig. 17.

The same as Fig. 8 except for the 25 Sep flight.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 18.
Fig. 18.

(a) Ice crystal concentration (Ni) vs vertical air velocity. See Table 5 and text for the assumptions made in the use of Eq. (7) to obtain ice crystal number concentration. Straight dark solid line is for the maximum Ni value observed during BASE. (b) Ice crystal number concentration vs temperature for the 2D-C data collected for the entire BASE project. Filled circles are averaged Ni for 5°C intervals.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Fig. 19.
Fig. 19.

The number of points estimated in the 0.5-dBZ intervals for radar observations for the 8 and 25 Sep cases are shown in (a) for 1439–1552 and 1630–2245 UTC and in (b) for 0313–0430 and 1240–1341 UTC, respectively. Observations used in (a) are collected along six volume scans at 12 elevation angles from 3.0° to 67.2°. Observations in (b) are collected along seven volume scans at 13 elevation angles from 1.0° to 67.2°. Number of points (Np) vs cell size (L) and height (h) for (c) the 8 Sep case on 1840 UTC and for (d) the 25 Sep case on 1250 UTC. The Np is obtained using Ze > 10 dBZ.

Citation: Journal of Climate 13, 7; 10.1175/1520-0442(2000)013<1225:DAMCOA>2.0.CO;2

Table 1.

Time periods and characteristics of aircraft, radar, and LANDSAT observations collected during the BASE 1994 Arctic field project.

Table 1.
Table 2.

Mean and std dev values of LWC, TWC, IWC, Nd, Ni, reffd, and reffi calculated at 5°C temperature intervals for the 8 Sepflight.

Table 2.
Table 3.

Mean and std dev values of LWC, TWC, IWC, Nd, Ni, reffd, and reffi calculated at 5°C temperature intervals for the 24 Sep flight.

Table 3.
Table 4.

Mean and std dev values of LWC, TWC, IWC, Nd, Ni, reffd, and reffi calculated at 5°C temperature intervals for the 25 Sep flight.

Table 4.
Table 5.

Cases used to obtain Ni from Eq. (7). See text and appendix for symbols.

Table 5.
Table 6.

The size of cells obtained using observations from various platforms. The cell size is obtained when the size is larger than mean plus standard deviation.

Table 6.
Save