• Angevine, W. M., A. B. White, and S. K. Avery, 1994: Boundary-layer depth and entrainment zone characterization with a boundary-layer profiler. Bound.-Layer Meteor.,68, 375–385.

    • Crossref
    • Export Citation
  • Carter, D. A., W. L. Ecklund, K. S. Gage, M. Spowart, H. L. Cole, E. F. Chamberlain, W. F. Dabberdt, and J. Wilson, 1992: First test of a shipboard wind profiler. Bull. Amer. Meteor. Soc.,73, 1587–1592.

    • Crossref
    • Export Citation
  • ——, K. S. Gage, W. L. Ecklund, W. M. Angevine, P. E. Johnston, A. C. Riddle, J. Wilson, and C. R. Williams, 1995: Developments in UHF lower tropospheric wind profiling at NOAA’s Aeronomy Laboratory. Radio Sci.,30, 977–1001.

    • Crossref
    • Export Citation
  • Frank, W. M., H. Wang, and J. L. McBride, 1996: Rawinsonde budget analyses during the TOGA COARE IOP. J. Atmos. Sci.,53, 1761–1780.

    • Crossref
    • Export Citation
  • Gutzler, D. S., and L. M. Hartten, 1995: Daily variability of lower tropospheric winds over the tropical western Pacific. J. Geophys. Res.,100, 22999–23008.

    • Crossref
    • Export Citation
  • ——, G. N. Kiladis, G. A. Meehl, K. M. Weickmann, and M. Wheeler, 1994: The global climate of December 1992–February 1993. Part II: Large-scale variability across the tropical western Pacific during TOGA COARE. J. Climate,7, 1606–1622.

  • Hashiguchi, H., S. Fukao, T. Tsuda, M. D. Yamanaka, D. L. Tobing, T. Sribimawati, S. W. B. Harijono, and H. Wiryosumarto, 1995:Observations of the planetary boundary layer over equatorial Indonesia with an L band clear-air Doppler radar: Initial results. Radio Sci.,30, 1043–1054.

    • Crossref
    • Export Citation
  • Lin, X., and R. H. Johnson, 1996: Kinematic and thermodynamic characteristics of the flow over the western Pacific warm pool during TOGA COARE. J. Atmos. Sci.,53, 695–715.

    • Crossref
    • Export Citation
  • McPhaden, M. J., 1993: TOGA–TAO and the 1991–93 El Niño–Southern Oscillation Event. Oceanogr.,6, 36–44.

    • Crossref
    • Export Citation
  • Miller, E. R., and A. C. Riddle, 1994: TOGA COARE Integrated Sounding Station Data Rep. Volume IA Revised Edition, 87 pp. [Available from National Center for Atmospheric Research, ATD/SSSF, P.O. Box 3000, Boulder, CO 80303-3000.].

  • Parsons, D., and Coauthors, 1994: The Integrated Sounding System:Description and Preliminary Observations from TOGA COARE. Bull. Amer. Meteor. Soc.,75, 553–567.

    • Crossref
    • Export Citation
  • Riddle, A. C., W. M. Angevine, W. L. Ecklund, E. R. Miller, D. B. Parsons, D. A. Carter, and K. S. Gage, 1996: In situ and remotely sensed horizontal winds and temperature intercomparisons obtained using Integrated Sounding Systems during TOGA COARE. Contrib. Atmos. Phys.,69, 49–61.

  • Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer, 666 pp.

    • Crossref
    • Export Citation
  • Webster, P. J., and R. Lukas, 1992: TOGA COARE: The Coupled Ocean Atmosphere Response Experiment. Bull. Amer. Meteor. Soc.,73, 1377–1417.

    • Crossref
    • Export Citation
  • View in gallery
    Fig. 1.

    Map of the ISS sites operating during the COARE IOP. Kavieng, Kapingamarangi, R/V Kexue #1, and R/V Shiyan #3 delineate the polygonal intensive flux array (IFA).

  • View in gallery
    Fig. 2.

    Daily scalar means and standard deviations of surface winds at RV3 during the IOP. Wind speeds are plotted in the left column, wind directions in the right column. The top row shows NOAA/AL-processed winds, AL, while the bottom row shows the difference between NOAA/AL-processed and filtered NCAR/ATD-processed winds, AL-ATDf. Statistics are presented to the right of the plots. See text for details of processing methods.

  • View in gallery
    Fig. 3.

    Schematic illustration of the procedure used to merge the surface and profiler ISS data. Squares indicate approximate heights of ISS measurements, while diamonds indicate approximate heights at which “filled” values were computed. See text for details.

  • View in gallery
    Fig. 4.

    Monthly mean wind profiles taken from the boundary layer dataset (solid) and the linearly interpolated dataset (dashed). (a) Wind speed at RV3 in December, (b) wind direction at RV3 in December, (c) wind speed at MAN in December, and (d) wind direction at MAN in December. Note that only a 60° range is plotted in (b) and (d).

  • View in gallery
    Fig. 5.

    Monthly mean wind profiles taken from the boundary layer dataset (solid) and the linearly interpolated dataset (dashed). (a) Wind speed at RV3 in January, (b) wind direction at RV3 in January, (c) wind speed at MAN in November, and (d) wind direction at MAN in November. Note that a 120° range is plotted in (b) and (d).

  • View in gallery
    Fig. 6.

    Scatterplots of half-hourly surface wind speeds vs wind speeds aloft at an island ISS (KAV) and a shipboard ISS (RV3). The profiler winds are from the high-resolution (low mode) dataset and were included only if they met the criteria set forth in section 4a.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 148 13 1
PDF Downloads 21 10 0

Reconciliation of Surface and Profiler Winds at ISS Sites

Leslie M. HarttenCooperative Institute for Research in Environmental Sciences/NOAA Aeronomy Lab, University of Colorado, Boulder, Colorado

Search for other papers by Leslie M. Hartten in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Integrated Sounding Systems (ISSs), which combine surface-based, balloon-borne, and radar observation capabilities, were deployed for the first time during the Intensive Observing Period (IOP) of the recent Coupled Ocean–Atmosphere Response Experiment. This note addresses efforts to synthesize the data from these disparate platforms as well as avenues for future research that were opened in the process.

The collaborative nature of the ISSs led to the application of different processing methods to the surface winds than were used with the winds measured by 915-MHz wind profilers. A new set of ship-based surface winds that are more directly analogous to the profiler winds has been developed. The statistical properties of these “AL-processed ISS surface winds” are shown to be similar to those of the land-based ISS surface winds, to the low-level profiler winds, and to surface winds measured at nearby buoys. A method of combining the surface and profiler winds from an ISS into one coherent dataset is also presented here; it involves assuming a logarithmic wind profile over a surface layer whose depth is invariant over the course of the IOP. While there are some obvious oversimplifications to this method, it is the most sophisticated option available from currently collected ISS data, it is more physically reasonable than a simple linear interpolation between the surface and higher-altitude winds, and it yields wind profiles that are acceptable for many applications. Both of these new datasets are now available to the community.

The process of combining the two sets of measurements not only led to a reconsideration of the postprocessing of the shipboard surface winds but also led to renewed interest in the effects of sea clutter on profiler winds. Further work is now under way in the profiler community to address the issue of sea clutter on ship-based and near-sea profiler installations.

Corresponding author address: Dr. Leslie M. Hartten, CIRES, University of Colorado, Campus Box 216, Boulder, CO 80309-0216.

Email: lhartten@al.noaa.gov

Abstract

Integrated Sounding Systems (ISSs), which combine surface-based, balloon-borne, and radar observation capabilities, were deployed for the first time during the Intensive Observing Period (IOP) of the recent Coupled Ocean–Atmosphere Response Experiment. This note addresses efforts to synthesize the data from these disparate platforms as well as avenues for future research that were opened in the process.

The collaborative nature of the ISSs led to the application of different processing methods to the surface winds than were used with the winds measured by 915-MHz wind profilers. A new set of ship-based surface winds that are more directly analogous to the profiler winds has been developed. The statistical properties of these “AL-processed ISS surface winds” are shown to be similar to those of the land-based ISS surface winds, to the low-level profiler winds, and to surface winds measured at nearby buoys. A method of combining the surface and profiler winds from an ISS into one coherent dataset is also presented here; it involves assuming a logarithmic wind profile over a surface layer whose depth is invariant over the course of the IOP. While there are some obvious oversimplifications to this method, it is the most sophisticated option available from currently collected ISS data, it is more physically reasonable than a simple linear interpolation between the surface and higher-altitude winds, and it yields wind profiles that are acceptable for many applications. Both of these new datasets are now available to the community.

The process of combining the two sets of measurements not only led to a reconsideration of the postprocessing of the shipboard surface winds but also led to renewed interest in the effects of sea clutter on profiler winds. Further work is now under way in the profiler community to address the issue of sea clutter on ship-based and near-sea profiler installations.

Corresponding author address: Dr. Leslie M. Hartten, CIRES, University of Colorado, Campus Box 216, Boulder, CO 80309-0216.

Email: lhartten@al.noaa.gov

1. Introduction

The Intensive Observing Period (IOP) of TOGA COARE (Tropical Ocean Global Atmosphere Coupled Ocean–Atmosphere Response Experiment) was conducted from November 1992 through February 1993 over the western equatorial Pacific (Webster and Lukas 1992). Integrated Sounding Systems (ISSs) were operated at six locations during the IOP; four were on islands, while two were on the Chinese ships R/Vs Kexue #1 and Shiyan #3 (Fig. 1). Each ISS consisted of an Omega balloon sounding system, a 915-MHz Doppler wind profiler, a Radio Acoustic Sounding System (RASS), and a surface observing station (Parsons et al. 1994). This was the first deployment of ISSs; they have since been deployed in several tropical and midlatitude field campaigns (D. Parsons 1996, personal communication; W. Angevine 1997, personal communication). The idea behind the ISSs was that a more complete picture of the atmosphere could be obtained by combining the measurements taken by several complementary platforms, but the process of actually integrating these measurements is just beginning. This note describes efforts taken in that direction.

In order to combine meteorological data collected from different platforms, possible incompatabilities in data sampling and processing, as well as gaps in data coverage, must be addressed. In the case of the ISS data, the two processing centers dealt with ship position and movement uncertainties in two different but reasonable manners (Miller and Riddle 1994; Riddle et al. 1996). As a result, intercomparisons between the shipboard low-level Omegasonde winds (which are dependent on the surface winds) and the profiler winds are not straightforward (Riddle et al. 1996); neither are attempts to diagnose boundary layer structure or large-scale profiles of dynamical quantities using both the shipboard surface and profiler winds. Further complicating matters is the 300–700-m gap in coverage between the surface measurement and the first profiler gate.

This note describes recent attempts to reprocess the surface ship data so that they are compatable with the postprocessed profiler data and presents one method that may be used to combine the lower tropospheric dynamical data collected at the ISS sites. Section 2 discusses the dynamical ISS data and the various processing methods applied to the ship-based winds, while section 3 describes the results of reprocessing of the surface winds. Section 4 shows how the reprocessed surface winds and the profiler winds may be combined into one dataset and discusses some of the results of this process, including the discovery of sea-clutter contamination in the profiler winds at some sites. Section 5 contains concluding remarks, and data availability is given in the appendix.

2. ISS data from TOGA COARE

The components of the ISSs deployed during COARE are discussed in detail in Parsons et al. (1994) and the references therein. The postprocessing of the ISS data reflected the collaborative nature of the systems; Omega sounding and surface data were handled by NCAR’s Atmospheric Technology Division (NCAR/ATD), while the profiler data were postprocessed by NOAA’s Aeronomy Lab (NOAA/AL). The deployment configuration was similar to that which has since been used at other sites; any differences are usually found in the balloon tracking system, the frequency of balloon launches, and/or the profiler range gate settings, although additional platforms such as lidars are also sometimes included. For COARE, Omega balloon soundings were made every 6 h centered on 0000 UTC. The 915-MHz Doppler wind profilers (Carter et al. 1992; Carter et al. 1995) were configured to repeatedly sample winds along three beams, one vertical and two oblique, in sequence for 25 min of each half-hour, alternating between “low” and “high” altitude mode throughout that time. The vertical extents of these data are approximately 3 and 6 km, respectively. Algorithms to remove the effects of ground clutter, birds, insects, and hydrometeors were applied to the collected spectra and then half-hourly averages were computed from all the resulting “clean” triads. The surface stations measured a variety of meteorological variables, including 1-min winds from 10-m towers on land and from 16-m towers on the second decks of the ships. Ship position, speed, and course information were provided by a GPS receiver, while the direction of the ship’s bow was determined by a magnetometer. Table 1 presents some basic information about the stations.

The GPS receivers were of a simple, nondifferential type that can display nonzero speed when stationary; although this can be improved with time averaging (P. Johnston 1995, personal communication), unaveraged GPS position was sampled on the ISS ships. During postprocessing at NOAA/AL, time series plots showed occasional GPS drift; the position information would very quickly go to unreasonable values, causing the GPS-reported ship speed and direction to do likewise (T. Riddle 1995, personal communication). One likely source of this GPS drift is poor satellite geometry, for example, having an insufficient number of satellites visible or having satellites inappropriately spaced in the visible sky. The Aeronomy Lab attempted to “fix” the GPS information during those times by both objective and subjective means, and some missing position information from RV1 during 11–28 January was filled in by assuming that the ship held a fixed position.

Also suspect were the magnetometer data; the difference between the ship’s corrected velocity vector direction (from GPS data) and the ship’s bow direction (from magnetometer data) for cases when ship speed exceeded 4.0 m s−1 occasionally grew to 10°–30° in the span of a few minutes. This is not physically reasonable;while the ship might reasonably be headed in a direction slightly off “dead ahead” at speeds greater than 4.0 m s−1, it should not be 10° or more off, nor should it to be twisting rapidly as it moves forward. In spite of this problem, the magnetometer data (bow direction) were more steady than the cleaned GPS data (velocity direction) when the ship was under way. (When the ship was not under way, the velocity direction could indicate the direction of a current carrying the ship and be unrelated to the bow direction.)

The procedure developed at the Aeronomy Lab to derive the profiler winds used the magnetometer to determine the direction of the wind, as the profiler measures winds relative to the bow of the ship. However, a constant “correction” was added because the aforementioned comparison of the magnetometer direction and ship course (when the ship was under way) suggested that the magnetometer had a bias. The actual bias was probably variable, but the correction, 12° for RV3 and 22° for RV1, was a constant and implicitly included the magnetic declination. This correction was applied at all time periods, regardless of ship speed. The profiler velocity vector was then corrected for ship motion (derived from cleaned GPS data) if the ship was moving at least 1.0 m s−1. At slower speeds ship velocity looked more like a random quantity than a true measure of speed; the integrated velocity did not correspond to the position changes when ships speeds were low. The essence of these corrections is documented in Miller and Riddle (1994) and Riddle et al. (1996).

While NCAR/ATD was aware of these concerns regarding the ship magnetometer and GPS data, they decided not to manipulate these datasets in any way (Miller and Riddle 1994). NCAR/ATD’s earth-relative surface winds, therefore, were constructed by adjusting the shiprelative winds based on the magnetometer readings and then subtracting the recorded GPS ship motion from the resultant vector wind. If one wanted to merge the originally released ship-based surface and profiler winds, there would be inherent differences in the data due to processing methods that could obscure the lower-tropospheric dynamics.

3. Reprocessed surface ISS data

A new set of postprocessed surface winds has been derived at NOAA/AL from the ship-relative winds collected on the 16-m towers, using the same methods applied to the shipboard profiler winds. The magnetometer readings from RV3 (RV1) were corrected for an apparent bias of 12° (22°) before the 1-min ship-relative wind components were converted into vector wind components (u, υ). Vector ship motion, computed from the cleaned GPS data, was subtracted from that vector wind if the ship speed was greater than 1.0 m s−1. Since the GPS data were only cleaned during those periods in which the profilers were operating, vector ship motion and therefore the reprocessed surface winds are also only available at those times. The appendix gives information regarding data availability. The reasonableness of these new shipboard surface winds is evaluated below by comparing various statistical properties of the new and original shipboard winds, the land-based ISS winds, and independent surface measurements.

Three different time series of ship surface winds have been examined: daily means of the new NOAA/AL-processed winds, AL; daily means computed from all the original NCAR/ATD-processed winds with speeds less than 100.0 m s−1, ATD1; and daily means computed from all original NCAR/ATD-processed winds within 5σ of μ, ATDf, where μ is the IOP mean and σ is the root mean variance from the ATD time series. For this phase of the work, all means and standard deviations are scalar, not vector, statistics because of the nature of the magnetometer correction and its expected effects. Figure 2 shows daily mean wind speeds and directions at RV3 from the AL time series, with error bars showing ±1 standard deviation for each day; results for RV1 are qualitatively similar and are not shown. The IOP means and standard deviations of the daily-mean values, together with the square root of the IOP-mean variance (root mean variance, σ), are to the right of the plots. The daily standard deviations of the wind directions were computed by summing the squares of the differences, on a 360° dial, between mean and individual directions. Also presented are differences between the NOAA/AL-processed and the filtered NCAR/ATD-processed winds, AL–ATDf.

The IOP means and root mean variances of the ATD wind speeds at RV3 and RV1 (not shown) are much larger than those from the AL-processed set (shown in Fig. 2 for RV3 only). Applying a 5σ despiking filter to the ATD-processed winds decreases the mean wind speed at RV3 and the root mean variance at both ships. The resultant ATDf time series have means and root mean variances very similar to the AL-processed time series, except for the still large root mean variance at RV3. Comparison with time series of daily-mean wind speeds and directions at the four land ISS sites (not shown) indicates that both the AL and ATDf speeds are stronger than those over land, as would be expected; the AL time series at both ships and the ATDf time series at RV1 have root mean variances similar to those of the land winds.

Applying a filter to the ATD winds has little effect on the means or root mean variances of the direction time series. The resulting daily-mean filtered ATD-processed directions and the AL-processed wind directions from both ships differ from each other irregularly (Fig. 2). The median differences over the IOP, +5° at RV3 and +17° at RV1, are in the correct sense but somewhat smaller than the magnetometer corrections applied in the NOAA/AL processing. At RV3 (RV1), 21% (20%) of the daily-mean differences are within 2° of the magnetometer corrections and 69% (71%) are within 10° of the corrections. IOP-mean surface wind directions are rotated clockwise (counterclockwise) at Northern (Southern) Hemisphere land stations compared to the Southern Hemisphere AL-processed ship values, as expected from simple frictional arguments.

Half-hourly vector-averaged ISS surface winds have also been constructed from the 1-min data and compared with an independent set of open-ocean surface winds. At Kapingamarangi, Nauru, and the two ships, these have been vector averaged into daily means and compared with daily-mean winds constructed from hourly values measured at nearby TAO buoys (McPhaden 1993). The locations of the buoys are shown in Fig. 1;the distances between the buoys and the nearest ISS stations are 150–225 km. Table 2 shows some statistics for the mean wind speeds and directions. In the IOP mean, the wind speeds at the two ISS land sites are less than half of those at the nearest buoys, while the speeds of the AL winds at the ships are slightly stronger than at the buoy at 2°S, 165°E and those of the ATDf winds are slightly weaker. The correlation coefficients of the daily-mean time series for the ISS sites and the nearest buoys range from 0.62 at KAP to 0.77 at RV3 (AL version) and are significant at the 0.5% level.2 Looking at the wind direction, in the IOP mean the land ISS winds are coming from a more clockwise direction than the TAO winds, while the ship ISS winds are coming from a more counterclockwise direction. The magnitude of the differences between the means ranges from 4° to 40°, with correlations ranging from 0.30 at RV1 (AL version) to 0.87 at NAU; all the correlations except that at RV1 (AL version) are significant at the 10% level or higher. Further testing indicates that AL and ATDf speeds and directions at RV1 and RV3 are equally well correlated with the TAO buoy.

4. Combining surface and profiler ISS data

a. Procedure

The premise behind the ISS was that the strengths of each platform would compensate for the weaknesses of the others (Parsons et al. 1994) and that the data from the various platforms could be integrated, that is, combined. Once a set of ship-based ISS surface winds processed in the same manner as the profiler winds had been developed, an attempt was made, in the spirit of the original concept, to merge the ISS surface and profiler winds. The simplest way to do this would be to fit a straight line between the surface and lowest profiler wind, but this ignores almost everything we think we know about near-surface dynamics and therefore can be considered incompatable with such high-resolution data. The most rigorous way to combine the data would be to have half-hourly estimates of several boundary layer parameters at each site and then estimate a surface-layer depth (as a function of boundary layer type, depth, and stability) and fit a theoretical wind profile between the surface and a low-altitude profiler level. Unfortunately, the ISSs as currently configured do not measure fluxes or quantities from which one can derive flux estimates, so stability arguments cannot be employed and different types of boundary layers cannot be identified on half-hour timescales, although it is sometimes possible to determine boundary layer height, zi, from peaks in the range-corrected signal-to-noise ratio (RCSNR) (Angevine et al. 1994) or reflectivity (Hashiguchi et al. 1995). Therefore, an intermediate path has been taken here.

The schematic in Fig. 3 illustrates the general procedure that has been used to merge the surface and profiler winds, although the altitude of the lowest profiler data varies from station to station (see Table 1). Estimates of zi from RCSNR at all six stations over the entire IOP indicate that 0.5 km is a reasonable first guess of mean boundary layer depth; this implies a mean surface-layer depth of 50 m (Stull 1988). A roughness length z0 of 0.01 m has been assumed for all of the sites (Stull 1988), as has neutral stability. This stability assumption is surely incorrect many times at many or all of the sites but is reasonable in the absence of any reliable method of estimating half-hourly stability at all sites during the 4-month-long IOP. The surface layer has been judged to start at ground level at KAV, MAN, and NAU and at mean sea level at KAP and the two ships. By assuming that the wind varies logarithmically with height in the surface layer and linearly above that to the level of the first profiler observation, intermediate-level winds have been computed. The resultant profiles will be referred to as the “boundary layer” set. These are vertical averages over the surface layer and at least three other layers above that (see Fig. 3 for approximate spacing). The vertical averaging makes these “filled data” more analogous to the profiler data, which represent averages over scanned volumes rather than measurements at specific altitudes. For comparative purposes a “linearly interpolated” set has also been created;in it, winds at the intermediate levels have been computed by linearly interpolating u and υ between the surface and the lowest profiler level. In both the boundary layer and linearly interpolated sets, profiler winds were only used if there were at least two triads of scans per half-hour, if the mean spectral widths from both oblique beams were less than 3.0 m s−1, and if the quality flag was less than 4 (Riddle et al. 1996). The high- and low-resolution profiler winds were blended together via a weighted average at approximately 1500 m, with the low-resolution profiler winds used exclusively below that level and high-resolution observations above. Data availability is discussed in the appendix.

b. Results

Differences between profiles produced by the boundary layer (BL) set versus the linearly interpolated vary considerably at any given station over time. Long-term mean profiles give a better indication of the types of differences that are seen and of the larger-scale effects of the BL procedure. Figures 4 and 5 compare some monthly mean wind profiles from the boundary layer set and from the linearly interpolated set at RV3 and MAN during December, when mean winds at both stations were strong (Fig. 4), and during January and November, respectively, when winds were light (Fig. 5). The vertical variation in wind speeds and directions above about 1 km at RV3 is due to the rejection of varying amounts of data at those heights (Miller and Riddle 1994); the number of observations going into each monthly average at 1100 m is 778 in December and 809 in January, while at 1300 m it is 859 and 874, respectively. The sharp change in the vertical gradient of the wind speed between approximately 25 and approximately 125 m seen in the RV3 plots (Figs. 4a and 5a) occurs frequently in individual profiles and is in some senses an artifact of the sampling used in the filling procedure; it arises when the difference between the assumed logarithmic wind at the top of the surface layer and the first profiler wind is such that both the mean surface layer wind speed, assigned to 25 m, and the mean wind speed in the layer from 100 to 150 m, assigned to 125 m, are less than the mean speed between the top of the surface layer and 100 m. In this situation the maximum low-level wind theoretically occurs at the top of the surface layer but is aliased upward by the sampling method. A similar discontinuity is occasionally seen in the direction profiles (e.g., Fig. 5b).

During December, wind speeds at both sites were 4.0–6.0 m s−1 throughout most of the lower 2 km of the troposphere (Figs. 4a and 4c). The low-level wind speeds at RV3 were up to 0.31 m s−1 higher in the BL dataset than in the linearly interpolated set, while at MAN the differences were negligible. Low-level wind directions at both sites (Figs. 4b and 4d) were rotated 1° clockwise (less westerly) in the BL data. In January at RV3 and in November at MAN mean wind speeds through the lower 2 km of the atmosphere were on the order of 1.0 m s−1 (Figs. 5a and 5c), with the BL speeds about 0.1 m s−1 greater than the linearly interpolated ones. The low-level BL directions were rotated about 10° counterclockwise (more southerly) at MAN (Fig. 5d) relative to the linearly interpolated data, while at RV3 there was at most a 1° difference between the two sets (Fig. 5b).

Much of the initial research using ISS data from COARE used data from only one of the three platforms (e.g., Gutzler and Hartten 1995; Frank et al. 1996; Lin and Johnson 1996), although Gutzler et al. (1994) juxtaposed quick-look profiler and Omegasonde winds collected during the IOP. The desire to combine the surface and profiler data led to a reevaluation of the shipboard surface winds and the creation of the new surface dataset described in section 3. The process of actually combining the data resulted in a merged dataset and also led to important discoveries regarding the profiler winds. When long-term mean profiles of wind speeds were first constructed as described above, using profiler data at and above 300 m at all sites, dramatic speed minima were often found near 300 m in profiles from KAP, RV1, and RV3. Scatterplots of surface wind speeds versus wind speeds at 300–800 m during the entire IOP, such as those in Fig. 6, showed that while wind speeds aloft were usually greater than surface speeds at the three island sites, as expected, low-level profiler wind speeds were slower than surface speeds up to 85% of the time at the two ships and 23% of the time at KAP, with the problem extending through the sixth range gate (approximately 600 m) at the ships but confined to the third range gate (approximately 300 m) at KAP. Analysis of several of the individual profiler beam scans from which the half-hourly profiler winds were constructed indicated that contamination by sea clutter was the cause of these unreasonable values, and so the work presented here has used data only at and above the levels that appear to be unaffected by clutter (cf. Table 1). Further work is under way at NOAA’s Aeronomy Lab and Environmental Technology Lab (AL and ETL) and at NCAR to prevent sea clutter contamination whenever possible, to identify it when it does occur, and to extract as much atmospheric signal as possible from observations in which it is present.

5. Discussion

ISSs were developed to draw upon the strengths of several platforms and collect a more complete picture of the troposphere than could be obtained by one platform in isolation and were first deployed in the field during TOGA COARE. As a step toward fulfilling that original intent, surface and 915-MHz profiler winds from ISS sites have here been combined into one dataset for the first time. This process led to the creation of a second version of the shipboard surface winds, to the development of a simple theoretically based algorithm that can be used to merge surface and profiler winds from any ISS dataset, and to the discovery of sea-clutter contamination in the profiler winds.

A new version of COARE surface wind data has been constructed for the two ship-based ISS sites, RV1 and RV3. These “AL-processed ISS surface winds” were processed in the same way as were the ship-based profiler winds (Riddle et al. 1996); spikes in GPS ship speeds were removed and magnetometer readings were adjusted for an apparent bias before the ship-relative winds were converted to earth-relative winds. The resultant winds have statistics that are more similar to those of the land ISS stations than are the statistics of either the original NCAR/ATD winds or a heavily filtered version of those data. These AL-processed ship surface winds are also much more compatible with the profiler winds because the ship position and motion have been accounted for in the identical manner and are therefore recommended for use with those ship-based profiler winds. The appendix gives information on obtaining these data.

One possible means of merging the surface and profiler ISS winds has been explored; the procedure involves assuming that, to a first approximation, the boundary layer is 500 m deep and that both u and υ vary logarithmically within the 50-m-deep surface layer and linearly above it to the lowest reliable profiler data, located at 300–700 m during COARE. While the boundary layer depth and stability assumptions are very simple, they enable the implementation of an algorithm that allows for the known effects of surface friction in some rough sense; the only other algorithm possible with currently gathered ISS data, a linear interpolation between the surface data and the first reliable profiler data, does not. Individual and long-term mean profiles look reasonable at the ISS sites, although the layer depths over which the filled data are averaged can cause a wind speed or direction relative maximum at the top of the surface layer to be aliased to a slightly higher altitude. Wind profiles constructed from the boundary layer set of winds are only slightly different from profiles based on the linearly interpolated data; the difference most often involves a slight increase in low-level wind speeds in the boundary layer set. Estimates of large-scale mean winds and second-order dynamical quantities are now being computed using these merged data; comparison of those results with other such estimates from different platforms will reveal more about the suitability of this procedure for large-scale analyses. It is possible that those interested in doing short-term case studies using ISS data from COARE or other field experiments may want to try more sophisticated methods of combining the data and may also find enough ancillary information to do so. For example, researchers focusing on the midlatitude daytime convective boundary layer or on a mesoscale convective complex during COARE would probably be justified discarding the assumption of neutral stability made here.

The act of combining collocated surface and profiler winds led to the discovery of sea-clutter contamination in the profiler data and provided one means of making bulk decisions about the veracity of the data. The issue of sea clutter in profiler winds is now an area of active research in the profiler community, and considerable attention to the issue is being paid during current deployments of profilers, whether as part of ISSs or not.

Analaysis of the lower troposphere using measurements taken by ISSs is only just getting under way, since they have only been deployed during the last 4 years, but initial results are encouraging. The integration of the measurements collected by the various platforms could perhaps be enhanced at future ISS sites by either including flux measurements or a second set of near-surface thermodynamic observations at a height different than that of the surface observations so that more sophisticated methods could be used to bridge the gap between surface and profiler winds.

Acknowledgments

This research was supported by the U.S. TOGA Project Office through a grant to the NOAA Aeronomy Laboratory. The TAO data and information on their use were provided by the TOGA–TAO Project Office, Dr. Michael J. McPhaden, director. Thanks to Paul Johnston, Tony Riddle, and John Wilson of the Tropical Dynamics and Climate group at the Aeronomy Lab for discussions regarding the 915-MHz profiler data; to Erik Miller and Dave Parsons (of NCAR/ATD) for information regarding the original surface winds; to Wayne Angevine for comments on determining boundary layer depth from profiler measurements; and to Dave Gutzler for many helpful discussions regarding the research presented. Useful comments on earlier versions of this manuscript were provided by Wayne Angevine, Warner Ecklund, Ken Gage, Dave Gutzler, Paul Johnston, George Kiladis, and Peggy LeMone.

REFERENCES

  • Angevine, W. M., A. B. White, and S. K. Avery, 1994: Boundary-layer depth and entrainment zone characterization with a boundary-layer profiler. Bound.-Layer Meteor.,68, 375–385.

    • Crossref
    • Export Citation
  • Carter, D. A., W. L. Ecklund, K. S. Gage, M. Spowart, H. L. Cole, E. F. Chamberlain, W. F. Dabberdt, and J. Wilson, 1992: First test of a shipboard wind profiler. Bull. Amer. Meteor. Soc.,73, 1587–1592.

    • Crossref
    • Export Citation
  • ——, K. S. Gage, W. L. Ecklund, W. M. Angevine, P. E. Johnston, A. C. Riddle, J. Wilson, and C. R. Williams, 1995: Developments in UHF lower tropospheric wind profiling at NOAA’s Aeronomy Laboratory. Radio Sci.,30, 977–1001.

    • Crossref
    • Export Citation
  • Frank, W. M., H. Wang, and J. L. McBride, 1996: Rawinsonde budget analyses during the TOGA COARE IOP. J. Atmos. Sci.,53, 1761–1780.

    • Crossref
    • Export Citation
  • Gutzler, D. S., and L. M. Hartten, 1995: Daily variability of lower tropospheric winds over the tropical western Pacific. J. Geophys. Res.,100, 22999–23008.

    • Crossref
    • Export Citation
  • ——, G. N. Kiladis, G. A. Meehl, K. M. Weickmann, and M. Wheeler, 1994: The global climate of December 1992–February 1993. Part II: Large-scale variability across the tropical western Pacific during TOGA COARE. J. Climate,7, 1606–1622.

  • Hashiguchi, H., S. Fukao, T. Tsuda, M. D. Yamanaka, D. L. Tobing, T. Sribimawati, S. W. B. Harijono, and H. Wiryosumarto, 1995:Observations of the planetary boundary layer over equatorial Indonesia with an L band clear-air Doppler radar: Initial results. Radio Sci.,30, 1043–1054.

    • Crossref
    • Export Citation
  • Lin, X., and R. H. Johnson, 1996: Kinematic and thermodynamic characteristics of the flow over the western Pacific warm pool during TOGA COARE. J. Atmos. Sci.,53, 695–715.

    • Crossref
    • Export Citation
  • McPhaden, M. J., 1993: TOGA–TAO and the 1991–93 El Niño–Southern Oscillation Event. Oceanogr.,6, 36–44.

    • Crossref
    • Export Citation
  • Miller, E. R., and A. C. Riddle, 1994: TOGA COARE Integrated Sounding Station Data Rep. Volume IA Revised Edition, 87 pp. [Available from National Center for Atmospheric Research, ATD/SSSF, P.O. Box 3000, Boulder, CO 80303-3000.].

  • Parsons, D., and Coauthors, 1994: The Integrated Sounding System:Description and Preliminary Observations from TOGA COARE. Bull. Amer. Meteor. Soc.,75, 553–567.

    • Crossref
    • Export Citation
  • Riddle, A. C., W. M. Angevine, W. L. Ecklund, E. R. Miller, D. B. Parsons, D. A. Carter, and K. S. Gage, 1996: In situ and remotely sensed horizontal winds and temperature intercomparisons obtained using Integrated Sounding Systems during TOGA COARE. Contrib. Atmos. Phys.,69, 49–61.

  • Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer, 666 pp.

    • Crossref
    • Export Citation
  • Webster, P. J., and R. Lukas, 1992: TOGA COARE: The Coupled Ocean Atmosphere Response Experiment. Bull. Amer. Meteor. Soc.,73, 1377–1417.

    • Crossref
    • Export Citation

APPENDIX

Data Availability

Corrected GPS information, available only for those times at which there were profiler data, are found in the files gps.1.fixed and gps.2.fixed. These were prepared by Tony Riddle of the NOAA Aeronomy Lab, and are available by anonymous ftp to aeolus.al.noaa.gov in directory pub/toga/sfc_ISS. “Ship 1” is R/V Kexue #1 and “Ship 2” is R/V Shiyan #3. The “AL-processed ISS surface winds” are also available at the same location. A README file describes the format and organization of the data, as well as many of the technical details presented here. These surface wind data may also be obtained from the Center for Ocean Atmospheric Prediction Studies (COAPS) at The Florida State University; for more information, see http://www.coaps.fsu.edu/coare/coaremet.html.

The boundary layer set of merged surface and low-mode profiler winds is available by anonymous ftp to aeolus.al.noaa.gov in directory pub/toga/BLwinds. The format and organization of the data, as well as technical details, are described in a README file.

Fig. 1.
Fig. 1.

Map of the ISS sites operating during the COARE IOP. Kavieng, Kapingamarangi, R/V Kexue #1, and R/V Shiyan #3 delineate the polygonal intensive flux array (IFA).

Citation: Journal of Atmospheric and Oceanic Technology 15, 3; 10.1175/1520-0426(1998)015<0826:ROSAPW>2.0.CO;2

Fig. 2.
Fig. 2.

Daily scalar means and standard deviations of surface winds at RV3 during the IOP. Wind speeds are plotted in the left column, wind directions in the right column. The top row shows NOAA/AL-processed winds, AL, while the bottom row shows the difference between NOAA/AL-processed and filtered NCAR/ATD-processed winds, AL-ATDf. Statistics are presented to the right of the plots. See text for details of processing methods.

Citation: Journal of Atmospheric and Oceanic Technology 15, 3; 10.1175/1520-0426(1998)015<0826:ROSAPW>2.0.CO;2

Fig. 3.
Fig. 3.

Schematic illustration of the procedure used to merge the surface and profiler ISS data. Squares indicate approximate heights of ISS measurements, while diamonds indicate approximate heights at which “filled” values were computed. See text for details.

Citation: Journal of Atmospheric and Oceanic Technology 15, 3; 10.1175/1520-0426(1998)015<0826:ROSAPW>2.0.CO;2

Fig. 4.
Fig. 4.

Monthly mean wind profiles taken from the boundary layer dataset (solid) and the linearly interpolated dataset (dashed). (a) Wind speed at RV3 in December, (b) wind direction at RV3 in December, (c) wind speed at MAN in December, and (d) wind direction at MAN in December. Note that only a 60° range is plotted in (b) and (d).

Citation: Journal of Atmospheric and Oceanic Technology 15, 3; 10.1175/1520-0426(1998)015<0826:ROSAPW>2.0.CO;2

Fig. 5.
Fig. 5.

Monthly mean wind profiles taken from the boundary layer dataset (solid) and the linearly interpolated dataset (dashed). (a) Wind speed at RV3 in January, (b) wind direction at RV3 in January, (c) wind speed at MAN in November, and (d) wind direction at MAN in November. Note that a 120° range is plotted in (b) and (d).

Citation: Journal of Atmospheric and Oceanic Technology 15, 3; 10.1175/1520-0426(1998)015<0826:ROSAPW>2.0.CO;2

Fig. 6.
Fig. 6.

Scatterplots of half-hourly surface wind speeds vs wind speeds aloft at an island ISS (KAV) and a shipboard ISS (RV3). The profiler winds are from the high-resolution (low mode) dataset and were included only if they met the criteria set forth in section 4a.

Citation: Journal of Atmospheric and Oceanic Technology 15, 3; 10.1175/1520-0426(1998)015<0826:ROSAPW>2.0.CO;2

Table 1.

The ISS sites operating during the COARE IOP. The listed elevations are of the surfaces on which the ISS instruments were located. Profiler measurements are taken relative to the top of the unit, which is 2 m above the surface, while surface winds are measured atop 10-m (land except KAP), 16-m (KAP), or 16-m (ship) towers.

Table 1.
Table 2.

Statistics of surface wind time series from ISS sites and TAO moored buoys during the COARE IOP. The buoys are identified by their position, given to the nearest degree. “Mean difference” is the mean of the difference time series formed by subtracting the TAO wind speeds and directions from those at the ISS sites. All statistics are computed only over days during which both sites in an ISS/TAO pair had data. Correlations significant at the 0.5% level are marked with an “*,” while those significant at the 5% and 10% levels are marked with “+” and “++,” respectively.

Table 2.

1

Speeds greater than or equal to 100.0 m s−1 arose from unreasonable GPS-derived ship speeds.

2

Significance was evaluated using a Student’s t distribution and degrees of freedom equal to the total record length divided by the number of days required for wind speeds to become decorrelated.

Save