• Brümmer, B., 1985: Structure, dynamics and energetics of boundary layer roll vortices from KonTur aircaft observations. Contrib. Atmos. Phys., 58 , 237254.

    • Search Google Scholar
    • Export Citation
  • Department of Energy, 1996: Science Plan for the Atmospheric Radiation Measurement Program (ARM), DOE/ER-0670T, U.S. Dept. of Energy, Office of Energy Research, Washington, DC, 50 pp.

  • De Souza, R. L., 1972: A study of atmospheric flow over a tropical island. Ph.D. thesis, Florida State University, 203 pp.

  • Etling, D., , and R. A. Brown, 1993: Roll vortices in the planetary boundary layer. Bound.-Layer Meteor., 65 , 215248.

  • Fairall, C. W., , A. B. White, , J. B. Edson, , and J. E. Hare, 1997: Integrated shipboard measurements of the marine boundary layer. J. Atmos. Oceanic Technol., 14 , 338359.

    • Search Google Scholar
    • Export Citation
  • French, J. R., , G. Vali, , and R. D. Kelly, 1999: Evolution of small cumulus clouds in Florida: Observations of pulsating growth. Atmos. Res., 52 , 143165.

    • Search Google Scholar
    • Export Citation
  • Garratt, J. R., 1990: The internal boundary layer—A review. Bound.-Layer Meteor., 50 , 171203.

  • Grachev, A. A., , J. E. Hare, , and C. W. Fairall, 2001: Observation of 4–5 day meridional wind and surface stress oscillations during Nauru ’99. Proc. 11th ARM Science Team Meeting, Atlanta, GA, U.S. Department of Energy. [Available online at http://www.arm.gov/publications/proceedings/conf11/abstracts/grachev-aa.pdf].

  • Hsu, H-M., 1987: Mesoscale lake-effect snowstorms in the vicinity of Lake Michigan: Linear theory and numerical simulations. J. Atmos. Sci., 44 , 10191040.

    • Search Google Scholar
    • Export Citation
  • Japan Marine Science and Technology Center, 1999: The R/V. MIRAI cruise report “Nauru ’99”. MR99-K03, JAMSTEC, 121 pp. [Available from JAMSTEC, 2-15 Natsushima, Yokosuka 237-0061, Japan].

    • Search Google Scholar
    • Export Citation
  • Kang, S-D., , and F. Kimura, 1997: A numerical study of the mechanism of clouds-street formation in the lee of an isolated mountain near a coast. J. Meteor. Soc. Japan, 75 , 955968.

    • Search Google Scholar
    • Export Citation
  • Keenan, T., and Coauthors, 2001: The Maritime Continent Thunderstorm Experiment (MCTEX): Overview and some results. Bull. Amer. Meteor. Soc., 81 , 24332455.

    • Search Google Scholar
    • Export Citation
  • LeMone, M. A., , and W. T. Pennell, 1976: The relationship of trade wind cumulus distrbution to subcloud layer fluxes and structure. Mon. Wea. Rev., 104 , 524539.

    • Search Google Scholar
    • Export Citation
  • Long, C. N., , and T. P. Ackerman, 2000: Identification of clear skies from broadband pyranometer measurements and calculation of downwelling shortwave cloud effects. J. Geophys. Res., 105 , 1560915629.

    • Search Google Scholar
    • Export Citation
  • Mahrer, Y., , and R. A. Pielke, 1976: Numerical simulation of airflow over Barbados. Mon. Wea. Rev., 104 , 13921402.

  • Malkus, J. S., 1955: The effects of a large island upon the trade-wind air stream. Quart. J. Roy. Meteor. Soc., 81 , 538550.

  • Malkus, J. S., 1963: Tropical rain induced by a small heat source. J. Appl. Meteor., 2 , 547556.

  • Malkus, J. S., , and A. E. Bunker, 1952: Observational studies of the air flow over Nantucket island during the summer of 1950. Papers in Physical Oceanography and Meteorology, XII, No. 2, Massachusetts Institute of Technology and Woods Hole Oceanographic Institution, 50 pp.

  • Matthews, S., , J. M. Hacker, , and D. G. Pethick, 2000: The flights of the ARA Cessna 404 ‘Investigator 2’ during Naura ’99. Airborne Research Australia Tech. Rep. 03/2000, 61 pp. [Available from Airborne Research Australia, P.O. Box 335, Salisbury South, SA 5106, Austalia.].

  • Matthews, S., , P. Schwerdtfeger, , and J. M. Hacker, 2002: Use of albedo modelling and aircraft measurements to examine the albedo of Naura. Aust. Meteor. Mag., 51 , 229236.

    • Search Google Scholar
    • Export Citation
  • McFarlane, S. A., , C. N. Long, , and D. M. Flynn, 2005: Impact of island-induced clouds on surface measurements: Analysis of the ARM Naura Island Effect Study data. J. Appl. Meteor., 44 , 10451065.

    • Search Google Scholar
    • Export Citation
  • Nordeen, M. L., , P. Minnis, , D. R. Doelling, , D. G. Pethick, , and L. Nguyen, 2001: Satellite observations of cloud plumes generated by Nauru. Geophys. Res. Lett., 28 , 631634.

    • Search Google Scholar
    • Export Citation
  • Savijärvi, H., , and S. Matthews, 2004: Flow over small heat islands: A numerical sensitivity study. J. Atmos. Sci., 61 , 859868.

  • Smith, R. B., , and V. Grubišic, 1993: Aerial observations of Hawaii’s wake. J. Atmos. Sci., 50 , 37283750.

  • Smith, R. B., , A. C. Gleason, , P. A. Gluhosky, , and V. Grubišic, 1997: The wake of St. Vincent. J. Atmos. Sci., 54 , 606623.

  • Stull, R. B., 1985: A fair-weather cumulus cloud classification scheme for mixed-layer studies. J. Climate Appl. Meteor., 24 , 4956.

  • Takemi, T., 2000: Diurnal variability of the fair weather boundary layer over the western tropical equatorial Pacific. Umi To Sora (Sea and Sky), 76 , 7378.

    • Search Google Scholar
    • Export Citation
  • Taylor, P. A., 1970: A model of airflow above changes in surface heat flux, temperature and roughness for neutral and unstable conditions. Bound.-Layer Meteor., 1 , 1839.

    • Search Google Scholar
    • Export Citation
  • Thomson, R. E., , J. F. R. Gower, , and N. W. Bowker, 1977: Vortex streets in the wake of the Aleutian Islands. Mon. Wea. Rev., 105 , 873884.

    • Search Google Scholar
    • Export Citation
  • Yoneyama, K., , and M. Katsumata, 2000: Overview of the R/V Mirai Nauru’99 cruise. Umi To Sora (Sea and Sky), 76 , 5963.

  • View in gallery

    Map of the locations of the ships during the (main panel) small triangle phase with aircraft traverses of the island circulation and cloud street and (inset) big triangle phase. All distances are in km from the ARCS, located at 0°16′S, 166°54′E.

  • View in gallery

    Comparison of ARCS and Ronald H. Brown sensors on the night of 4 Jul 1999. The ship was anchored 500 m downwind of the ARCS. A trace of rain (<0.2 mm) fell around 0100 LT. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling longwave irradiance (Ldn), and surface temperature (Tsurface), respectively.

  • View in gallery

    Side by side comparison of ship sensors on 3 Jul 1999. During the first half of the comparison, the ships traveled into the wind. In the second half, both ships were maneuvering, disturbing the wind measurements. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling longwave irradiance (Ldn), and SST.

  • View in gallery

    Estimates of clear-sky global solar irradiance. Curves are the deviation of the model of Long and Ackerman (2000) fitted using only global irradiance from the same model fitted using both global and diffuse irradiance to determine clear-sky conditions.

  • View in gallery

    Comparison of island surface temperature measured at the ARCS and the average of aircraft traverses of the island. Vertical bars are 1 standard deviation. In addition to the low-level passes shown in Fig. 1, passes of the island made during big triangle flights have been included to increase the sample size.

  • View in gallery

    Time series of radiation and surface meteorology from the three stations on Nauru. Downwelling longwave irradiance and rainfall were measured at the ARCS only. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling shortwave irradiance (Sdn), downwelling longwave irradiance (Ldn), surface temperature (Tsurface), net radiation (Rnet), and rainfall amount.

  • View in gallery

    Time series of radiation and surface meteorology from the Mirai and Ronald H. Brown. The big triangle phase (Fig. 1) lasted from 24 to 30 Jun 1999. The small triangle phase was from 1 to 5 Jul 1999. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling shortwave irradiance (Sdn), downwelling longwave irradiance (Ldn), SST, net radiation (Rnet), and rainfall amount.

  • View in gallery

    Comparison of diurnal cycles of radiation and surface meteorology over the ocean and island. Curves are averages of all days divided into 1-h segments. Two days when a cloud street did not form have been excluded. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling shortwave irradiance (Sdn), downwelling longwave irradiance (Ldn), surface temperature (Tsurface), and net radiation (Rnet).

  • View in gallery

    The (top) temperature and (bottom) humidity of the mixed layer above and downwind of Nauru. Curves are the average of two sets of traverses interpolated onto a common axis. Results are offset for clarity, with dashed lines indicating the mean mixed-layer potential temperature, 27.8°C, and specific humidity, 17.0 g kg−1. (top)–(bottom) Flight altitudes for each were 710, 360, 180, and 90 m. Successive temperature traces are offset by 1 K and humidity traces are offset by 2 g kg−1, except the 710-m trace, which is offset by 5 g kg−1. Distances are measured from the ARCS. The location of Nauru is indicated by the thick line.

  • View in gallery

    Cross section of the mixed layer below the cloud street at 400-m altitude. Distance is measured relative to the ARCS. The center of the cloud street was located at −7 km. (top) Potential temperature (θ), (middle) specific humidity (q), and (bottom) virtual potential temperature (θυ).

  • View in gallery

    WSI analysis of cloud cover above the ARCS on 29 Jun 1999. Curves are cloud cover in each of four quadrants centered on the cardinal points.

  • View in gallery

    Radiosonde-measured wind profile in the marine boundary layer on the afternoon of 2 Jul 1999. Wind components are aligned along and across the axis of the cloud plume.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 85 85 25
PDF Downloads 33 33 9

Modification of the Atmospheric Boundary Layer by a Small Island: Observations from Nauru

View More View Less
  • 1 Ensis—Forest Biosecurity and Protection, CSIRO, Yarralumla, Australia
  • | 2 Airborne Research Australia, Salisbury South, Australia
  • | 3 The Pennsylvania State University, University Park, Pennsylvania
  • | 4 NOAA/Environmental Technology Laboratory, Boulder, Colorado
  • | 5 Pacific Northwest National Laboratory, Richland, Washington
  • | 6 Brookhaven National Laboratory, Upton, New York
© Get Permissions
Full access

Abstract

Nauru, a small island in the tropical Pacific, generates cloud plumes that may grow to over 100-km lengths. This study uses observations to examine the mesoscale disturbance of the marine atmospheric boundary layer by the island that produces these cloud plumes. Observations of the surface layer were made from two ships in the vicinity of Nauru and from instruments on the island. The structure of the atmospheric boundary layer over the island was investigated using aircraft flights. Cloud production over Nauru was examined using remote sensing instruments. The diurnal cycles of surface meteorology and radiation are characterized at a point near the west (downwind) coast of Nauru. The spatial variation of surface meteorology and radiation are also examined using surface and aircraft measurements. During the day, the island surface layer is warmer than the marine surface layer and wind speed is lower than over the ocean. Surface heating forces the growth of a thermal internal boundary layer, within which a plume of cumulus clouds forms. Cloud production begins early in the morning over the ocean near the island’s lee shore; as heating intensifies during the day, cloud production moves upwind over Nauru. These clouds form a plume that may extend over 100 km downwind of Nauru. Aircraft observations showed that a plume of warm, dry air develops over the island that extends 15–20 km downwind before dissipating. Limited observations suggest that the cloud plume may be sustained farther downwind of Nauru by a pair of convective rolls. Suggestions for further investigation of the cloud plume are made.

Corresponding author address: Dr. Stuart Matthews, Ensis—Forest Biosecurity and Protection, CSIRO, Locked Bag 17, Granville, NSW 2142, Australia. Email: stuart.matthews@ensisjv.com

Abstract

Nauru, a small island in the tropical Pacific, generates cloud plumes that may grow to over 100-km lengths. This study uses observations to examine the mesoscale disturbance of the marine atmospheric boundary layer by the island that produces these cloud plumes. Observations of the surface layer were made from two ships in the vicinity of Nauru and from instruments on the island. The structure of the atmospheric boundary layer over the island was investigated using aircraft flights. Cloud production over Nauru was examined using remote sensing instruments. The diurnal cycles of surface meteorology and radiation are characterized at a point near the west (downwind) coast of Nauru. The spatial variation of surface meteorology and radiation are also examined using surface and aircraft measurements. During the day, the island surface layer is warmer than the marine surface layer and wind speed is lower than over the ocean. Surface heating forces the growth of a thermal internal boundary layer, within which a plume of cumulus clouds forms. Cloud production begins early in the morning over the ocean near the island’s lee shore; as heating intensifies during the day, cloud production moves upwind over Nauru. These clouds form a plume that may extend over 100 km downwind of Nauru. Aircraft observations showed that a plume of warm, dry air develops over the island that extends 15–20 km downwind before dissipating. Limited observations suggest that the cloud plume may be sustained farther downwind of Nauru by a pair of convective rolls. Suggestions for further investigation of the cloud plume are made.

Corresponding author address: Dr. Stuart Matthews, Ensis—Forest Biosecurity and Protection, CSIRO, Locked Bag 17, Granville, NSW 2142, Australia. Email: stuart.matthews@ensisjv.com

1. Introduction

The Atmospheric Radiation Measurement program (ARM) aims to improve the parameterization of clouds and radiation in climate models (Department of Energy 1996). The observational component of ARM comprises long-term, routine measurements of clouds, radiation, and meteorology with additional intensive field campaigns to address specific questions. Routine measurements are made in three locations: the southern Great Plains in the United States, the North Slope of Alaska, and the tropical western Pacific Ocean (TWP). The TWP sites, known as Atmospheric Radiation and Cloud Stations (ARCSs), house radiometers, surface meteorology instruments, cloud remote sensing equipment, and atmospheric profiling systems. Because of the quantity and nature of the instrumentation required, it was necessary to place the ARM sites on land rather than on, for example, buoys. In the TWP, three sites are maintained—at Manus Island, Papua New Guinea (opened 1996), at Nauru (opened 1998), and at Darwin, Australia (opened 2002).

Nauru, a small, flat island in the western Pacific Ocean, is the world’s smallest republic (Fig. 1). Most of the land area of Nauru is the central plateau, known as “Topside,” surrounded by a narrow coastal strip. The ARCS is situated on the west (downwind) coast of Nauru. Although Nauru is a very small island (5 km in diameter, with most of the island below a 30-m altitude), it was anticipated that it would have some effect on the ARCS measurements. To examine this “island effect” and to relate the island-based measurements to data collected over the ocean, the Nauru99 field campaign was conducted in June and July 1999. The field campaign included instruments on the island itself, two ships stationed near the island, and measurements made by Airborne Research Australia’s Cessna 404 aircraft. This paper integrates observations from Nauru99 to characterize the mesoscale disturbance to the marine flow caused by the island. These results will be of use to studies that characterize the effects of the island-generated circulation on ARCS measurements and seek to compensate for these effects.

Nauru’s island circulation is also an interesting meteorological phenomenon in its own right. Three types of atmospheric circulations associated with islands can be identified in the literature. The first is generated by tall islands that are mechanical obstacles to the flow and generate a wake; examples are the Aleutian Islands (Thomson et al. 1977), Hawaii (Smith and Grubišic 1993), and St. Vincent (Smith et al. 1997). The second is produced by large islands in low-wind speed regimes that generate mesoscale convection over the island, driven by the convergence of coastal sea breezes; examples are the Tiwi Islands (Keenan et al. 2000) and Puerto Rico (Malkus 1955). The third type is associated with small islands that do not generate proper sea breezes yet may produce convective clouds that may form long plumes; examples are Anegada, British Virgin Islands, (Malkus 1963), Nantucket, Massachusetts (Malkus and Bunker 1952), Barbados (De Souza 1972), and, it shall be shown, Nauru.

Both Anegada (Malkus 1963) and Nantucket (Malkus and Bunker 1952) produce cumulus clouds that form a cloud plume up to 30 km in length. Plumes of heated turbulent air were observed over and downwind of the islands. Specific humidity in the boundary layer above Anegada was observed to be higher than over the surrounding ocean, while results from Nantucket were mixed. A plume of warm, dry air was observed over and 30 km downwind of Barbados in aircraft measurements reported by De Souza (1972). An analysis of pilot balloon trajectories indicates a mesoscale “island circulation” with downward motion over Barbados and upward motion downwind of the island during the day, with the reverse at night. Mahrer and Pielke (1976) have argued that De Souza neglected to account for terrain effect in divergence calculations but on the basis of model results predicted that this island circulation should be observed over a flat island. The effect of topography in flow over a small island is discussed further by Savijärvi and Matthews (2004). Nordeen et al. (2001) observed a plume of cumulus clouds downwind of Nauru using satellite imagery. Cloud plumes were first observed shortly after sunrise and grew to a mean length of 125 km by 1630 local time (LT). The plumes were aligned with the mean wind direction and grew at one-third of the mean cloud-level wind speed.

The present study builds on prior observations of island effects using a more extensive suite of surface-based observations, in addition to extending aircraft measurements similar to those made by Malkus and Bunker (1952), Malkus (1963), and De Souza (1972) to a new location.

2. Observations

The Nauru99 field experiment was conducted from 17 June to 17 July 1999. The main platforms participating in Nauru99 were the ARM Atmospheric Radiation and Cloud Station (ARCS) on Nauru, the research vessels Mirai and Ronald H. Brown, and Airborne Research Australia’s Cessna 404 aircraft. The observations used in this study were from two phases of Nauru99: the big triangle phase from 24 to 30 June and the small triangle phase from 1 to 4 July. During the big triangle phase, the two ships were stationed 250 km from Nauru, forming a triangle with the island (Fig. 1). During the small triangle phase, the ships moved to within 50 km of the island. Throughout both phases, the aircraft made measurement flights over the ocean between the three surface platforms and also over Nauru. All data used in this study were obtained from the ARM data archive (see online at www.archive.arm.gov).

a. Instruments on Nauru

Measurements were made at three locations on Nauru during Nauru99: at the ARCS installation and at two locations in the interior of the island called Topside 1 and 2 (Fig. 1). The ARCS installation is a permanent facility on the west coast of Nauru, collecting climatological data. The ARCS incorporates instruments measuring surface meteorology, broad- and narrowband irradiances, atmospheric profiles, and cloud properties. Measurements from the ARCS are processed and sorted by ARM into several data packages. This study uses measurements from four of the ARCS data packages:

  • Sky radiation—downwelling long- and shortwave irradiance
  • Ground radiation—upwelling long- and shortwave irradiance and radiometric surface temperature
  • Surface meteorology—air temperature, relative humidity, and air pressure at 2-m altitude and wind speed and direction at a height of 10 m
  • Whole sky imager (WSI)—cloud amount over the whole sky and in nine subareas: a 10° circle at the zenith, four quadrants from the 0° to 45° zenith angle centered at the cardinal points, and four quadrants from the 45° to 80° zenith angle
All data were sampled at a 1-min resolution, except for WSI, which was sampled at 10-min intervals.

The two weather stations in the island interior measured air temperature, relative humidity, air pressure, wind speed and direction, and upwelling longwave irradiance and downwelling shortwave irradiance at 1-min intervals. Radiometers were mounted at 2 m AGL and the remaining instruments were mounted at 3 m AGL. The towers were located at 30 m MSL. Data were recorded on self-contained dataloggers and downloaded manually.

b. Ships

Both ships deployed during Nauru99 were equipped to make a broad range of meteorological and oceanographic measurements of which this study uses only a small subset. Measurements used from the Mirai were

  • air temperature, relative humidity, wind speed and direction, air pressure, rainfall, and downwelling long- and shortwave irradiance from Brookhaven National Laboratory instruments mounted above the bow of the ship at 30 m ASL,
  • sea surface temperature measured by the Rutherford Appleton Laboratory Scanning Infrared SST Radiometer, and
  • vertical profiles of temperature, relative humidity, and winds from Väisälä radiosondes.
Time series measurements were sampled at 1-min intervals and radiosondes were launched every 3 h. Full details of instrumentation and data processing are given in Japan Marine Science and Technology Center (1999).

Measurements used from the Ronald H. Brown were

  • air temperature, relative humidity, wind speed and direction, air pressure, rainfall, and downwelling long- and shortwave irradiance from Environmental Technology Laboratory (ETL) instruments mounted on a staff on the bow of the ship at 18 m ASL, and
  • SST measured by ETL’s “sea snake,” a thermometer towed behind the ship at a depth of 5 cm.
All measurements were obtained as 10-min averages. Fairall et al. (1997) describe the ETL instruments.

c. Aircraft

Measurements in the ABL were made using Airborne Research Australia’s Cessna 404 aircraft, Investigator 2. The aircraft was equipped with instruments for measuring air temperature, humidity, the three-dimensional wind vector, radiometric surface temperature, up- and downwelling long- and shortwave irradiance, and cloud liquid water content. Meteorological instruments were mounted on the nosecone or had air intakes in the nose. Radiometers were mounted on the roof and in a pod underneath the aircraft. Full details of the instruments and data processing are given in Matthews et al. (2000).

Eight data gathering flights were made during Nauru99. Only those flights used in this paper are described here; a complete description of all flights is given in Matthews et al. (2000). During the big triangle phase of Nauru99, eight triangles were flown from Nauru to the Mirai, to the Ronald H. Brown, and back to Nauru (Fig. 1). Between corners of the triangle, the aircraft maintained an altitude of 30 m. During the small triangle phase, a variety of flight patterns were flown near Nauru. On 2 July, three stacks of triangles following the same pattern as the big triangles were flown. Each stack consisted of up to four triangles at altitudes of 30, 400, 1200, and 3000 m. The MiraiRonald H. Brown legs of the third stack were extended to pass under and through the cloud plume. Three crosswind passes over Nauru at an altitude of 150 m were also flown on 2 July. On 4 July, three stacks of alongwind traverses of Nauru were made. Up to five traverses were made for each stack, at altitudes ranging from 40 to 1100 m. The aircraft inertial navigation system failed on this day, so no wind measurements were available for 4 July.

3. Results

a. Preliminary investigations

Two questions were addressed before using the measurements from the ships and ARCS to investigate the effect of Nauru on the marine ABL: were the measurements from the platforms consistent, and what did the point measurements from the ships and ARCS represent?

The first question was addressed using comparisons between measurement platforms. Side by side comparisons were used among the island stations, between the ships, and between the ARCS and the Ronald H. Brown. Because side by side comparisons of downwelling solar irradiance Sdn were not possible, an examination of clear-sky Sdn was made. Comparisons with aircraft data were made only for surface temperature and albedo [section 3a(2)], as no other mean quantities from the aircraft were used when contrasting the island and ocean measurements. Spatial variation of radiation and meteorology over the island and ocean were examined and used to construct representative “ocean” and “island” datasets.

1) Platform comparisons

The two weather stations were relocated to the ARCS compound on 22 July 1999 for a 12-day comparison with the ARCS instruments. Measurements from this period were used to remove air temperature–dependent differences of up to 2 K in air temperature and up to 15% in relative humidity measurements from the Topside weather station measurements. Upwelling longwave irradiance measurements were scaled to remove 0.5% and 0.7% excesses from the weather station measurements. No other adjustments were required.

The Ronald H. Brown was stationed 500 m downwind of the ARCS on the night of 4 July. Direct comparisons between the ship and island-based instruments were not possible because of local effects due to the differences in physical properties between land and water. The two measurements not expected to be affected, wind direction and downwelling longwave irradiance, agreed to within instrumental accuracy (Fig. 2). Specific humidity was 1 g kg−1 lower at the ARCS than at the Ronald H. Brown. With the information available, it was not possible to determine whether this was a physical effect or was due to differences in calibration. Prior to a rain shower at 0100 LT, surface temperatures Tsurface differed by less than 1°C. At this time, potential temperature θ also differed by less than 1°C, indicating that the two sensors were in agreement. The different roughness lengths of the land and ocean meant that ARCS wind speed was lower than at the Ronald H. Brown and it was not possible to test for any differences in calibrations.

On 3 July, the Mirai and Ronald H. Brown cruised side by side to allow for a comparison of the ships’ instruments (Fig. 3). With the exception of downwelling longwave irradiance, all measurements agreed to within instrumental errors, although wind measurements were disturbed during the second half of the comparison when the ships were maneuvering. A 5 W m−2 difference in longwave irradiance was corrected by increasing measurements from the Mirai.

Downwelling shortwave irradiance Sdn measurements were compared by examining clear-sky irradiance for each platform. Long and Ackerman (2000) developed a method to identify clear-sky conditions from global and diffuse shortwave irradiance using four tests. The tests are based on the magnitude and rate of change of global shortwave irradiance and on the magnitude and variance of diffuse irradiance. Clear-sky Sdn may be compared between locations by fitting a model of Sdn to those measurements identified as clear sky:
i1520-0493-135-3-891-e1
where X is the cosine of the solar zenith angle and a and b are regression coefficients. Because measurements of diffuse shortwave irradiance were not available from the ships, the full method could not be applied and the two global irradiance tests were only relied on to identify clear skies. Omission of the diffuse irradiance tests meant that some periods when small, scattered clouds that did not obscure the solar beam were present were erroneously identified as clear (Long and Ackerman 2000).

The omission of the diffuse irradiance tests from the ARCS measurements resulted in overestimation of Sdn by up to 50 W m−2 at zenith angles below 45° and underestimation by up to 30 W m−2 at zenith angles greater than 45° (Fig. 4). A visual inspection of data series indicated that the overestimation was due to samples with enhanced Sdn because of scattering from cloud being marked as clear, while the underestimation was an artifact of fitting Eq. (1) to these exaggerated values.

Estimates of clear-sky Sdn from the ships differed by less than 1 W m−2 at all solar zenith angles (Fig. 4) but were 30 W m−2 below Sdn at the ARCS at solar noon. The differences between land and sea may have been due to differences in instrument calibration, clear-sky Sdn, or the amount of cloud not detected by the algorithm. It was not possible to separate these effects, although the differences were larger than would be expected from typical differences in instrument calibration. For comparisons of mean Sdn between land and sea, differences less than 30 W m−2 were not considered significant.

2) Spatial variation on Nauru

The Sdn in the presence of a cloud plume has been investigated for the Nauru99 period by McFarlane et al. (2005). They found that Sdn was up to 10% lower on the lee side of the island than on the windward side. Downwelling longwave irradance was 5–10 W m−2 higher on the lee side of the island than on the windward side.

The ground facing radiometers at the ARCS viewed the gravel surface of the ARCS enclosure. The albedo of this surface was 40%. Albedo measurements from the aircraft, reported in Matthews et al. (2002), found that Nauru’s area-averaged albedo was much lower, around 18%. Similarly, aircraft measurements of radiometric surface temperature showed the majority of the island surface to be cooler during the day than the ground viewed by the ARCS radiometers (Fig. 5). ARCS surface temperature measurements have been scaled using the least squares line of best fit in this comparison to give a more representative value for surface temperature. No measurements were made for ARCS temperatures in the range of 25°–30°C (Fig. 5), and hence there is some uncertainty about the accuracy of the scaling in this range.

From internal boundary layer theory, it would be expected that the surface layer would adjust to the change in surface forcing that the island represents, resulting in alongwind (i.e., east–west) gradients in wind speed, air temperature, and humidity until a new equilibrium is reached (Garratt 1990). Temperature, wind speed, and humidity were similar at the ARCS and at Topside 1, while at Topside 2, the wind speed was higher, the air cooler, and the specific humidity similar to the ARCS (Fig. 6). It is unclear whether the differences between the two Topside stations were due to local effects or by differences in exposure to the marine airflow. However, it is likely that the ARCS measurements were representative of at least the western half of the island, with cooler and windier conditions near the windward coast.

For simplicity, the ARCS measurements, with modified albedo and surface temperature, are used in the remainder of this paper and are referred to as the “island dataset.”

3) Spatial variation over the ocean

Satellite imagery suggested that atmospheric conditions were homogenous throughout the area of the Nauru99 experiment, so the ship measurements could serve as an estimate of the flow incident on Nauru. During the 5 days of big triangle flights, the only measurement to show a systematic, spatial gradient was SST (Fig. 7). A pool of warmer water extending 35–40 km north and east from the Ronald H. Brown was observed in aircraft measurements. The uniformity of all other measurements was reflected in the similarity of the ship data (SST and rain events excepted) throughout the experiment. When the ships moved closer together for the small triangle phase, the difference in SST closed and all measurements were in agreement. On the basis of this uniformity, it was concluded that the surface measurements from the ships were representative of the air incident on Nauru (except for local events such as rain and SST from Ronald H. Brown). The Mirai dataset was used as the “marine” dataset. In constructing the net radiation budget, albedo was assumed to be 5%, and upwelling longwave irradiance was calculated as the sum of emitted irradiance (from surface temperature using the Stefan–Bolzmann law) and reflected irradiance (from downwelling longwave irradiance using an emissivity of 0.97).

b. The atmosphere over the ocean

Nauru99 was characterized by suppressed convective conditions. Significant cloud clusters were not observed over the experimental area but were seen farther west (Yoneyama and Katsumata 2000). Balloon soundings from the Mirai showed a shallow mixed layer of 600–650-m depth with a stable atmosphere above (Savijärvi and Matthews 2004). The free stream trade wind at Nauru averaged 10 m s−1 at 2 km ASL. Clouds were mostly small fair-weather cumulus (Takemi 2000), and the largest rain event was 9 mm at the Ronald H. Brown on 28 June (Fig. 7). The mean low- cloud fraction derived from ceilometer measurements on the Mirai was estimated to be 0.142 and the cumulus fraction to be 0.122 (Takemi 2000). The median cloud-base height was 750 m, 100–150 m above the mixed-layer top and coincident with the lifting condensation level.

Surface layer conditions during Nauru99 were quite uniform, exceptions being the rain showers at both ships on 27–28 June. Although most measurements showed some variability on time scales longer than one day, there were not large changes, and one day was much like another with the exception of wind speed, which increased steadily in the earlier part of the experiment (Fig. 7). To better reveal diurnal cycles, mean series were constructed as the average of measurements from all days except 28 June [for which no downwelling solar radiation measurements were available from the Mirai (Fig. 7)] divided into 1-h segments (Fig. 8).

The most significant variation in potential temperature was the diurnal cycle, which had an average amplitude of 0.5°C. A spectral analysis by Takemi (2000) also revealed smaller temperature variations on longer time scales, with a peak at 3.5 days.

Specific humidity increased before, and decreased after, the rain showers at both ships. Smaller variations were also observed on time scales longer than 1 day. A weak semidiurnal cycle in q was observed, with peaks at midday and midnight (Fig. 8).

Wind speed increased from <2 m s−1 to a peak of 9 m s−1 during the period of 24–29 June. Thereafter, wind speed varied between 5 and 9 m s−1 with an average strength of 6.5 m s−1. Also, a diurnal cycle with its peak at around 0800 LT was observed in the composite measurements. During Nauru99, wind direction oscillated from ENE to ESE with a period of 4–5 days (Grachev et al. 2001). No significant diurnal cycle was observed in wind direction.

A small upward trend in SST at the Mirai was seen during the big triangle phase; Yoneyama and Katsumata (2000) report that the 28°C isotherm deepened during this time. The diurnal cycle had an average amplitude of 0.3°C.

Downwelling shortwave radiation deviated from a simple solar zenith angle dependence only in the presence of the ubiquitous cumulus clouds. The uniformity of the marine atmosphere was also reflected in downwelling longwave irradiance, Ldn, with day to day variations of less than 10 W m−2. Increases in the magnitude of Ldn due to the presence of cloud were observed throughout the field campaign, with amplitudes of up to 30 W m−2.

c. Surface meteorology of Nauru

Nauru presented a disturbance to the marine ABL. The surface meteorology of Nauru is thus the result of the modification of the marine atmosphere by this disturbance, and the surface layer meteorology on Nauru must be considered in relation to marine conditions.

Time series of the island and marine datasets are shown in Figs. 6 and 7. The examination of satellite images revealed 2 days when a cloud plume did not form, 24 and 30 June. These days have been excluded when examining diurnal cycles. Diurnal curves of radiation and surface meteorology were constructed as the averages of measurements from the remaining 8 days divided into 1-h segments (Fig. 8).

In the mean diurnal data, Sdn over Nauru was lower than over the ocean during the afternoon. Until 1300 LT the differences were less than 20 W m−2, but in the late afternoon Sdn at Nauru was up to 100 W m−2 below the ocean measurements. Averaged over the course of the day, the difference was 14 W m−2. Although a difference in Sdn was observed only during the afternoon, clouds were produced downwind of Nauru earlier in the day but did not significantly affect Sdn measurements, as they did not obscure the sun (section 3e).

Mean Ldn was of similar magnitude over both the island and ocean, but a different diurnal pattern was observed in each location. Over the ocean there were peaks in the early morning and late afternoon with a minimum at midday. This pattern broadly matched the diurnal cycle of low cloud fraction observed from the Mirai (Takemi 2000). At Nauru, Ldn was at a maximum around midday and at a minimum before dawn. Two factors acted together to influence Ldn over Nauru: increased low cloud fraction during the day and the heating of the air column above the island.

Surface temperature on Nauru exhibited a strong cycle of daytime heating and nighttime cooling with an amplitude of up to 17°C. In the mean cycle, island surface temperature Tsurface was at a minimum (2.5°C below SST) shortly before sunrise. During the morning, Tsurface climbed to a maximum of 36.8°C, which was sustained for several hours. During the afternoon the surface cooled, and Tsurface dropped below SST shortly after sunset, at which time air temperature also dropped below the marine air temperature. Maximum surface temperatures increased during the first 4 days of Nauru99 but dropped by 8°C after the rainfall on 28 June. Maxima then increased steadily for the remainder of the experiment. Minimum surface temperatures were less strongly affected by rain but showed similar behavior.

As a consequence of Nauru’s higher surface temperature and albedo, net radiation at the island surface was lower than at the ocean surface during the day, by 160 W m−2 at midday. Small day to day variations in net radiation were driven largely by variation in surface temperature on days when a cloud plume formed and by changes in Sdn on the days when a cloud plume did not form but clouds were present over the island.

The 2-m air temperature record from Nauru was dominated by a cycle of daytime heating and nighttime cooling, forced by changes in surface temperature. Throughout Nauru99, daily maxima ranged from 1° to 3°C above the marine air temperature, while minima were up to 5°C cooler. On average, air temperature increased shortly after sunrise from a nighttime minimum of 25.9°C and exceeded the marine air temperature after 0830 LT. A maximum temperature of 29.4°C was achieved at midday. Thereafter, air temperature dropped and was equal with the marine air temperature at sunset (1915 LT), then continued cooling throughout the night

Specific humidity on Nauru was consistently 1 g kg−1 lower than over the ocean but showed day to day trends similar to the marine record. As has been described above, it was not possible to perform satisfactory interplatform comparisons of ship and island humidity sensors. The observed 1 g kg−1 difference was small enough that it may have either been due to calibration differences or of physical origin (e.g., adjustment of the surface layer to the dry surface of the island).

Wind speed over Nauru was lower than over the ocean at all times and had a diurnal cycle with its maximum in the early afternoon. The average cycle had an amplitude of 2.1 m s−1, increasing from a minimum of 2 m s−1 prior to dawn to a maximum of 4.1 m s−1 around noon, then decaying in the evening. During the earlier part of the experiment, in which wind speed increased over the ocean, maximum wind speed over Nauru also increased but more slowly. The difference between the daily maximum wind speed at Nauru and the marine wind speed increased from <1 m s−1 on 24 June to around 4 m s−1 on 28 June. The difference in maxima remained at 2–3 m s−1 throughout the rest of Nauru99. Wind direction on Nauru closely tracked wind direction over the ocean but was systematically 7°–8° less on Nauru than over the ocean.

d. Atmospheric boundary layer structure

Aircraft flights to examine the structure of the boundary layer above Nauru were made on 4 July. Between 1230 and 1430 LT, two stacks of four flight legs each were flown over and downwind of Nauru (Fig. 1) to examine the structure of the boundary layer. Flight legs were at 90, 180, 360, and 710 m above sea level. During this time, cumulus clouds were forming over Nauru. A cloud plume approximately 5-km wide extended 15–20 km downwind of the island. The cloud base was at 750 m. Beyond this, a thinner plume extended farther 20–30 km. To better reveal the mean ABL structure, measurements from the two stacks have been interpolated to a common axis and the average calculated. A third stack with legs at 40, 470, and 1000 m was flown between 1040 and 1120 LT but is not shown here, as measurements were very similar to those of the stacks with more complete coverage.

A plume of warmer air extended from the surface of Nauru through the depth of the mixed layer to 20 km downwind of Nauru (Fig. 9). The core of this region, at 0–15 km downwind of Nauru, was 0.3°C above the marine air temperature with a region of stronger heating near the surface of the island. The leading edge of the mixed-layer plume moved downwind with increasing altitude. This is typical of a developing thermal internal boundary layer (Garratt 1990). Closer to 20 km downwind, temperatures returned to the marine value as the plume dissipated. The mixed-layer plume was generated by a strong sensible heat flux from the surface of the island. Although flux measurements from 4 July were not available, values were available from passes over the island made on 2 July. These showed sensible heat fluxes of 55–295 W m−2 over Nauru, compared with <5 W m−2 over the ocean. In the leg at 710 m, an upward trend in temperature west of Nauru was observed, although the trace was dominated by fluctuations. It is not possible to say from these measurements whether this subcloud heating was due to the upward transport of heat from the island surface or enhanced downward mixing of warmer air in the wake of the island.

The region of the mixed-layer plume downwind of Nauru was also drier than the marine air. A region with specific humidity up to 1 g kg−1 below the marine value extended from Nauru to 15 km downwind of the island. This drying was not as clearly defined as the heating, due to the larger variance of the specific humidity. On 2 July, the latent heat flux at the surface of Nauru was 160–245 W m−2. This was larger than over the ocean, enhanced by increased turbulence. Without flux measurements, it was not possible to identify the cause of the drying. However, a probable explanation is enhanced mixing between the mixed layer and subcloud layers due to increased turbulence over Nauru. As with the temperature measurements, humidity measurements in the subcloud layer were dominated by fluctuations, but there was not a well-defined trend.

The boundary layer below the cloud plume, which had cloud bases at 750 m, was examined using two sets of crosswind passes flown on 2 July (Fig. 1). The aircraft flew across the cloud plume at altitudes of 30, 400, and 1100 m. The lowest and highest passes did not show temperature or humidity deviating from the undisturbed marine values. The pass at 400 m, through the middle of the mixed layer, displayed two striking features. On either side of the cloud plume, areas of increased temperature and decreased humidity extended 6–7 km from the center of the cloud plume (Fig. 10). Potential temperature was as much as 0.7°C above values in the nearby atmosphere, while specific humidity was as much as 3 g kg−1 lower. While variations in virtual potential temperature were smaller, the air below the cloud plume was more buoyant than the air on either side of the plume (Fig. 10).

This pattern below the cloud plume suggests the downward mixing of warm, dry air on either side of the cloud plume. A similar pattern was observed in temperature and humidity measurements in large fields of convective rolls by Brümmer (1985) and LeMone and Pennell (1976). Further confirmation of the presence or absence of a pair of convective rolls would have best been provided by means of vertical wind measurements. However, to reveal the mean vertical motion of convective rolls, it is necessary to average over a large number of measurements, which was not possible here with only two passes under the cloud plume.

e. Clouds

The cumulus clouds produced over Nauru were observed in WSI images and aircraft video. These clouds formed as small, individual cells separated by clear air. This type of cloud formation is typical of forced convection, whereby clouds form when thermals in the boundary layer reach the lifting condensation level (Stull 1985). Weather conditions on 29 June provided a good opportunity to examine the diurnal cycle of cloud production over Nauru. On this day, clouds did not form over the ocean in the vicinity of Nauru, and thus all clouds recorded by remote sensing instruments were due to the presence of the island. The visual inspection of WSI images showed that cloud production began at 0935 LT, with clouds forming in two locations—to the north and south of the ARCS approximately 5 km downwind of the island’s lee coast. During the day, cloud production intensified and moved eastward over the ARCS, then over the interior of Nauru. During this time, the WSI images did not show a preferred location for cloud formation. In the late afternoon, cloud production decreased, retreated beyond the lee coast, and ceased shortly before sunset.

Whole sky imager analyses of cloud cover provide a quantitative description of cloud production. Cloud amounts for quadrants of the sky centered on the cardinal directions are shown in Fig. 11. The sky was clear until 0935 LT. The apparent cloudiness in the east prior to 0935 LT was a misclassification of the sun’s aureole as cloud. This problem also exaggerated cloud amounts in the north quadrant by up to 10% between 1200 and 1430 LT and in the west quadrant by up to 10% from 1430 LT to sunset. In spite of these difficulties, a clear asymmetry was observed in the cloud field. From 1200 to 1700 LT cloud cover in the west ranged from 20% to 50%, while in the east and south cloud cover was below 20%, except for a brief period around 1600 LT. Cloud cover in the north was more variable because of the ESE wind resulting in some island-produced clouds located in the north quadrant. In the late afternoon cloud production collapsed, with only a very few clouds detected in the period from 1700 LT to sunset.

4. Discussion

The Nauru island circulation was driven by the change in surface properties encountered by marine air as it was advected over the island. Nauru differed from the ocean surface in two important respects: the surface was rougher and was hotter during the day.

The island surface was heated by the absorption of solar radiation. Although the albedo of Nauru was larger than that of the ocean, and hence net radiation was lower, the effective heat capacity of the soil was lower than that of the ocean mixed layer, thus allowing the surface to heat to an average 10° above SST at midday.

The heating of air over the island initiated the development of a thermal internal boundary layer (TIBL). The TIBL grew rapidly into the neutral mixed layer, reaching the middle of the mixed layer over the lee coast of Nauru. The TIBL was characterized by increased temperature, strong turbulent mixing, and a large upward sensible heat flux at the surface. Downwind of Nauru, a plume of warm air persisted for 20 km before dissipating. This plume was also drier than the marine air. Cross sections flown beneath the cloud plume suggested that it was sustained by a pair of convective rolls. If this was the case, then there must have been a transition from the heat island circulation (characterized in our observations by warming and drying of the mixed layer) to the roll circulation. Although no observations were available in the present study to investigate the transition, the modeling results of Hsu (1987) and Kang and Kimura (1997) suggest that the interaction of the mean wind with convergence in the updraft portion of the heat island circulation initiates the roll circulation.

Specific humidity in the surface layer was lower on Nauru than over the ocean. Unfortunately it was not possible to determine whether the difference was due to drying of the surface layer air or differences in instrument calibration. The adjustment of the surface layer air to the dryer surface of the island is a plausible physical explanation for the difference.

The roughness of the island acted to decrease the wind speed in the surface layer, while convective mixing above the island acted to increase the wind speed, a process described in the TIBL context by Taylor (1970). The varying intensity of surface heating during the day modulated this increase, resulting in a diurnally varying wind speed. The net result was a diurnal cycle of 10-m wind speed that was in phase with surface temperature, with both mean and maximum speeds below the wind speed over the ocean.

Fair-weather cumulus clouds formed over Nauru during the day. Although there were insufficient measurements to investigate the vertical velocity structure of the TIBL, the nature of the cloud formation seen in whole sky imager observations implied condensation in rising thermals (Stull 1985). In the morning and late afternoon, when surface heating was less than during the middle of the day, cloud formation often occurred over the ocean downwind of Nauru. During the middle of the day, cloud formation moved upwind over the island. Clouds advected away from Nauru, forming a cloud plume. The cloud feedback on the radiation budget was small. Because the clouds formed over the ocean or over the western part of the island, the sun was only obscured at the ARCS during the later part of the day when it had moved into the western part of the sky.

Models of flow over small heat islands predict the formation of a mesoscale circulation with a cell of downward motion over the upwind side of the island and a cell of upward motion downwind of the island (Savijärvi and Matthews 2004). Unfortunately, suitable wind measurements were not made during Nauru99 to determine whether this circulation exists over Nauru. However, Hsu (1987) used a three-dimensional model to show that the updraft cell of the circulation generated by an elongated island should have maxima in vertical velocity near the ends of the island. These areas of enhanced upward motion would be favorable for cloud formation. Sky camera imagery of clouds showed preferred locations for cloud formation to the north and south of the ARCS during the morning and evening, when heating was too weak for widespread cloud formation, consistent with Hsu’s results. Hence, a 3D numerical simulation of the Nauru case may provide interesting insights into this type of circulation.

Nordeen et al. (2001) observed that cloud plumes developed to an average length of 125 km and that these plumes grew at one-third of the mean cloud-level wind speed, 35 km h−1. The time required for an air parcel to travel the length of the cloud plume was thus much longer than the typical 30-min life of trade cumulus clouds (French et al. 1999), so the length of the cloud plume requires explanation. Kang and Kimura (1997) demonstrated that long cloud plumes could be generated in the lee of coastal mountains by a pair of convective rolls originating in the convergence zone behind the mountain. These rolls were sustained by a positive buoyancy flux from the ocean surface. Transverse aircraft passes below Nauru’s cloud plume showed temperature and humidity traces characteristic of convective rolls, although conclusive evidence was not available. Also, the region downwind of a heat island of finite width is a convergence zone (Hsu 1987). So it is possible that a pair of convective rolls was generated by Nauru and that these maintained the cloud plume.

However, the maintenance of the postulated rolls requires explanation. The convective roll energy budget includes three terms: production of roll energy from the kinetic energy of the mean wind (shear production), production or loss of roll energy by buoyancy, and dissipation (Brümmer 1985). Since dissipation is always an energy sink, either or both the shear and buoyancy term must balance losses in the other terms when integrated over the depth of the rolls.

The determination of the roll energy budget requires measurements at a sufficient range of heights to adequately resolve vertical variations in individual terms. Brümmer’s (1985) measurements suggest that observations are required near the ocean surface, in the middle of the mixed layer, near cloud base, in mid-cloud, and at cloud top. Clearly our limited observations did not meet this criteria. However, some qualitative observations can be made. Shear production occurs when there is an inflection point in the wind profile (Etling and Brown 1993). Profiles of along- and across-roll wind speed from radiosondes did not show an inflection point at the time when the aircraft traversed the cloud plume (Fig. 12), suggesting that the shear production term was zero on 2 July 1999.

The temperature and humidity measurements made in the mixed layer below the cloud plume (Fig. 10) show warm, dry air moving downward on either side of the plume. Virtual potential temperature in the mixed layer (Fig. 10) shows that the air posited to be moving upward below the cloud plume was less dense than the air descending on either side, making the circulation locally thermally direct. The circulation was also thermally direct at the ocean surface, with upward sensible and latent heat fluxes. However, in cases where Brümmer (1985) observed buoyancy sources at the surface and in the mixed layer, a buoyancy sink was also observed higher in the mixed layer and in cloud. So, without traverses at the above suggested heights, it was not possible to determine whether the buoyancy term in the energy budget was positive or negative when integrated over the depth of the roll.

Of the two energy sources available to maintain convective rolls, shear and buoyant production, our limited observations suggest that the shear term was zero though we were unable to determine the buoyancy term. If the buoyancy term was negative, then our supposition that the cloud plume was maintained by convective rolls must be false. In any future observations of island cloud plumes, an attempt should be made to confirm the existence of convective rolls beneath the plumes and to calculate the roll energy budget over the depth of the rolls. These objectives could be met with roll traverses near the ocean surface, in the middle of mixed layer, at cloud base, in mid-cloud, and at cloud top.

5. Summary

Modification of the marine ABL by Nauru has been investigated using observations from surface-based instruments and an aircraft. Measurements were made during a period of suppressed convection, with a shallow mixed layer under scattered cumulus clouds.

The daily cycle of solar radiation forced a diurnal cycle of 3.5°C in air temperature, in response to a diurnal cycle of 10.8°C in surface temperature. The combined effects of mechanical deceleration and acceleration during the daytime due to heating of the surface produced a diurnal cycle in wind speed of 2.1 m s−1.

The heating of the island’s surface during the day generated a thermal internal boundary layer (TIBL) over Nauru within the 600-m-deep mixed layer. The warmer air was advected away from the island and formed a plume 15–20 km long and up to 0.3°C warmer than the undisturbed mixed layer. This plume of warm air was also up to 1 g kg−1 drier than the undisturbed mixed layer. Enhanced convection in the TIBL produced cumulus clouds over Nauru. These clouds, with bases at 750 m, formed long plumes, up to several hundred km long in some instances. The time required for an air parcel to travel the length of the cloud plume was longer than the lifetime of a typical cumulus cloud. Limited observations suggest that the cloud plume may have been maintained by a pair of convective rolls, but further investigation would be required to confirm this. These characteristic features of Nauru’s island circulation (a warm, dry plume and the production of cumulus clouds) were similar to those previously observed over Anegada (Malkus 1963), Nantucket (Malkus and Bunker 1952), and Barbados (De Souza 1972). All four island circulations appear to be driven by the same mechanism—enhanced convection over the island due to surface heating.

The modification of the boundary layer by Nauru, most notably in enhanced cloud production, will affect climate measurements made by ARM on the island. Studies such as those being pioneered by McFarlane et al. (2005) will be required to correct for these island effects.

Acknowledgments

The ARM data were obtained from the Atmospheric Radiation Measurement Program, sponsored by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, Environmental Sciences Division. Airborne Research Australia was established with funding from the Australian Commonwealth’s Major National Research Facilities Scheme. Investigator 2 was piloted by Noel Roediger and the aircraft radiometers were maintained by David Pethick. Dr. Long acknowledges the support of the Climate Change Research Division of the U.S. Department of Energy as part of the ARM Program. Comments by an anonymous reviewer helped to improve the discussion of convective roll dynamics.

REFERENCES

  • Brümmer, B., 1985: Structure, dynamics and energetics of boundary layer roll vortices from KonTur aircaft observations. Contrib. Atmos. Phys., 58 , 237254.

    • Search Google Scholar
    • Export Citation
  • Department of Energy, 1996: Science Plan for the Atmospheric Radiation Measurement Program (ARM), DOE/ER-0670T, U.S. Dept. of Energy, Office of Energy Research, Washington, DC, 50 pp.

  • De Souza, R. L., 1972: A study of atmospheric flow over a tropical island. Ph.D. thesis, Florida State University, 203 pp.

  • Etling, D., , and R. A. Brown, 1993: Roll vortices in the planetary boundary layer. Bound.-Layer Meteor., 65 , 215248.

  • Fairall, C. W., , A. B. White, , J. B. Edson, , and J. E. Hare, 1997: Integrated shipboard measurements of the marine boundary layer. J. Atmos. Oceanic Technol., 14 , 338359.

    • Search Google Scholar
    • Export Citation
  • French, J. R., , G. Vali, , and R. D. Kelly, 1999: Evolution of small cumulus clouds in Florida: Observations of pulsating growth. Atmos. Res., 52 , 143165.

    • Search Google Scholar
    • Export Citation
  • Garratt, J. R., 1990: The internal boundary layer—A review. Bound.-Layer Meteor., 50 , 171203.

  • Grachev, A. A., , J. E. Hare, , and C. W. Fairall, 2001: Observation of 4–5 day meridional wind and surface stress oscillations during Nauru ’99. Proc. 11th ARM Science Team Meeting, Atlanta, GA, U.S. Department of Energy. [Available online at http://www.arm.gov/publications/proceedings/conf11/abstracts/grachev-aa.pdf].

  • Hsu, H-M., 1987: Mesoscale lake-effect snowstorms in the vicinity of Lake Michigan: Linear theory and numerical simulations. J. Atmos. Sci., 44 , 10191040.

    • Search Google Scholar
    • Export Citation
  • Japan Marine Science and Technology Center, 1999: The R/V. MIRAI cruise report “Nauru ’99”. MR99-K03, JAMSTEC, 121 pp. [Available from JAMSTEC, 2-15 Natsushima, Yokosuka 237-0061, Japan].

    • Search Google Scholar
    • Export Citation
  • Kang, S-D., , and F. Kimura, 1997: A numerical study of the mechanism of clouds-street formation in the lee of an isolated mountain near a coast. J. Meteor. Soc. Japan, 75 , 955968.

    • Search Google Scholar
    • Export Citation
  • Keenan, T., and Coauthors, 2001: The Maritime Continent Thunderstorm Experiment (MCTEX): Overview and some results. Bull. Amer. Meteor. Soc., 81 , 24332455.

    • Search Google Scholar
    • Export Citation
  • LeMone, M. A., , and W. T. Pennell, 1976: The relationship of trade wind cumulus distrbution to subcloud layer fluxes and structure. Mon. Wea. Rev., 104 , 524539.

    • Search Google Scholar
    • Export Citation
  • Long, C. N., , and T. P. Ackerman, 2000: Identification of clear skies from broadband pyranometer measurements and calculation of downwelling shortwave cloud effects. J. Geophys. Res., 105 , 1560915629.

    • Search Google Scholar
    • Export Citation
  • Mahrer, Y., , and R. A. Pielke, 1976: Numerical simulation of airflow over Barbados. Mon. Wea. Rev., 104 , 13921402.

  • Malkus, J. S., 1955: The effects of a large island upon the trade-wind air stream. Quart. J. Roy. Meteor. Soc., 81 , 538550.

  • Malkus, J. S., 1963: Tropical rain induced by a small heat source. J. Appl. Meteor., 2 , 547556.

  • Malkus, J. S., , and A. E. Bunker, 1952: Observational studies of the air flow over Nantucket island during the summer of 1950. Papers in Physical Oceanography and Meteorology, XII, No. 2, Massachusetts Institute of Technology and Woods Hole Oceanographic Institution, 50 pp.

  • Matthews, S., , J. M. Hacker, , and D. G. Pethick, 2000: The flights of the ARA Cessna 404 ‘Investigator 2’ during Naura ’99. Airborne Research Australia Tech. Rep. 03/2000, 61 pp. [Available from Airborne Research Australia, P.O. Box 335, Salisbury South, SA 5106, Austalia.].

  • Matthews, S., , P. Schwerdtfeger, , and J. M. Hacker, 2002: Use of albedo modelling and aircraft measurements to examine the albedo of Naura. Aust. Meteor. Mag., 51 , 229236.

    • Search Google Scholar
    • Export Citation
  • McFarlane, S. A., , C. N. Long, , and D. M. Flynn, 2005: Impact of island-induced clouds on surface measurements: Analysis of the ARM Naura Island Effect Study data. J. Appl. Meteor., 44 , 10451065.

    • Search Google Scholar
    • Export Citation
  • Nordeen, M. L., , P. Minnis, , D. R. Doelling, , D. G. Pethick, , and L. Nguyen, 2001: Satellite observations of cloud plumes generated by Nauru. Geophys. Res. Lett., 28 , 631634.

    • Search Google Scholar
    • Export Citation
  • Savijärvi, H., , and S. Matthews, 2004: Flow over small heat islands: A numerical sensitivity study. J. Atmos. Sci., 61 , 859868.

  • Smith, R. B., , and V. Grubišic, 1993: Aerial observations of Hawaii’s wake. J. Atmos. Sci., 50 , 37283750.

  • Smith, R. B., , A. C. Gleason, , P. A. Gluhosky, , and V. Grubišic, 1997: The wake of St. Vincent. J. Atmos. Sci., 54 , 606623.

  • Stull, R. B., 1985: A fair-weather cumulus cloud classification scheme for mixed-layer studies. J. Climate Appl. Meteor., 24 , 4956.

  • Takemi, T., 2000: Diurnal variability of the fair weather boundary layer over the western tropical equatorial Pacific. Umi To Sora (Sea and Sky), 76 , 7378.

    • Search Google Scholar
    • Export Citation
  • Taylor, P. A., 1970: A model of airflow above changes in surface heat flux, temperature and roughness for neutral and unstable conditions. Bound.-Layer Meteor., 1 , 1839.

    • Search Google Scholar
    • Export Citation
  • Thomson, R. E., , J. F. R. Gower, , and N. W. Bowker, 1977: Vortex streets in the wake of the Aleutian Islands. Mon. Wea. Rev., 105 , 873884.

    • Search Google Scholar
    • Export Citation
  • Yoneyama, K., , and M. Katsumata, 2000: Overview of the R/V Mirai Nauru’99 cruise. Umi To Sora (Sea and Sky), 76 , 5963.

Fig. 1.
Fig. 1.

Map of the locations of the ships during the (main panel) small triangle phase with aircraft traverses of the island circulation and cloud street and (inset) big triangle phase. All distances are in km from the ARCS, located at 0°16′S, 166°54′E.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 2.
Fig. 2.

Comparison of ARCS and Ronald H. Brown sensors on the night of 4 Jul 1999. The ship was anchored 500 m downwind of the ARCS. A trace of rain (<0.2 mm) fell around 0100 LT. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling longwave irradiance (Ldn), and surface temperature (Tsurface), respectively.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 3.
Fig. 3.

Side by side comparison of ship sensors on 3 Jul 1999. During the first half of the comparison, the ships traveled into the wind. In the second half, both ships were maneuvering, disturbing the wind measurements. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling longwave irradiance (Ldn), and SST.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 4.
Fig. 4.

Estimates of clear-sky global solar irradiance. Curves are the deviation of the model of Long and Ackerman (2000) fitted using only global irradiance from the same model fitted using both global and diffuse irradiance to determine clear-sky conditions.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 5.
Fig. 5.

Comparison of island surface temperature measured at the ARCS and the average of aircraft traverses of the island. Vertical bars are 1 standard deviation. In addition to the low-level passes shown in Fig. 1, passes of the island made during big triangle flights have been included to increase the sample size.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 6.
Fig. 6.

Time series of radiation and surface meteorology from the three stations on Nauru. Downwelling longwave irradiance and rainfall were measured at the ARCS only. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling shortwave irradiance (Sdn), downwelling longwave irradiance (Ldn), surface temperature (Tsurface), net radiation (Rnet), and rainfall amount.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 7.
Fig. 7.

Time series of radiation and surface meteorology from the Mirai and Ronald H. Brown. The big triangle phase (Fig. 1) lasted from 24 to 30 Jun 1999. The small triangle phase was from 1 to 5 Jul 1999. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling shortwave irradiance (Sdn), downwelling longwave irradiance (Ldn), SST, net radiation (Rnet), and rainfall amount.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 8.
Fig. 8.

Comparison of diurnal cycles of radiation and surface meteorology over the ocean and island. Curves are averages of all days divided into 1-h segments. Two days when a cloud street did not form have been excluded. (top)–(bottom) Series of potential temperature (θ), specific humidity (q), wind speed, wind direction, downwelling shortwave irradiance (Sdn), downwelling longwave irradiance (Ldn), surface temperature (Tsurface), and net radiation (Rnet).

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 9.
Fig. 9.

The (top) temperature and (bottom) humidity of the mixed layer above and downwind of Nauru. Curves are the average of two sets of traverses interpolated onto a common axis. Results are offset for clarity, with dashed lines indicating the mean mixed-layer potential temperature, 27.8°C, and specific humidity, 17.0 g kg−1. (top)–(bottom) Flight altitudes for each were 710, 360, 180, and 90 m. Successive temperature traces are offset by 1 K and humidity traces are offset by 2 g kg−1, except the 710-m trace, which is offset by 5 g kg−1. Distances are measured from the ARCS. The location of Nauru is indicated by the thick line.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 10.
Fig. 10.

Cross section of the mixed layer below the cloud street at 400-m altitude. Distance is measured relative to the ARCS. The center of the cloud street was located at −7 km. (top) Potential temperature (θ), (middle) specific humidity (q), and (bottom) virtual potential temperature (θυ).

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 11.
Fig. 11.

WSI analysis of cloud cover above the ARCS on 29 Jun 1999. Curves are cloud cover in each of four quadrants centered on the cardinal points.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Fig. 12.
Fig. 12.

Radiosonde-measured wind profile in the marine boundary layer on the afternoon of 2 Jul 1999. Wind components are aligned along and across the axis of the cloud plume.

Citation: Monthly Weather Review 135, 3; 10.1175/MWR3319.1

Save