The ASTEX Lagrangians: An Hourly Dataset for Modellers

Compiled by Chris Bretherton, last revised July 1996, links revived June 2004.

Note: All files are referred to are in the UW anonymous ftp archive, and are identified relative to the path pub/astex/lagr on anonymous ftp on Click here to download Lagrangian data files from this archive.

The two ASTEX Lagrangian experiments were comprehensive studies
of the evolution of a boundary layer airmass over roughly 36 hour
periods.  In each Lagrangian, aircraft observations were taken an
continuously as possible following the mean motion of the air in
the marine boundary layer (MBL).

The information in this directory is mainly oriented toward
providing an effective comparison of the two ASTEX Lagrangian
experiments with column models of the cloud-topped marine
boundary layer.  The data presented here are discussed in three
papers, the first two of which are available via anonymous ftp
(see README for parent directory ASTEX for more information).
The papers discussing the data are:

   I   Bretherton, C. S. and R. Pincus, 1995: Cloudiness and
                    Marine Boundary Layer Dynamics in the ASTEX
                    Lagrangian Experiments.  Part I:  Synoptic
                    setting and vertical structure.
                    J. Atmos. Sci., 52, 2707-2723  

   II  Bretherton, C. S., P. A. Austin and S. T. Siems, 1995:
                    Cloudiness and Marine Boundary Layer Dynamics
                    in the ASTEX Lagrangian Experiments.  Part
                    II:  Cloudiness, drizzle, surface fluxes and
                    J. Atmos. Sci., 52, 2724-2735.

   III Austin, P. A., and C. S. Bretherton, 1995:  Broadband
                    radiative fluxes during the ASTEX Lagrangian
                    experiments.   Available from Phil Austin as
                    a UBC tech report.

The hourly data for the two Lagrangians are in the subdirectory
/hourly of /lagr1 and /lagr2, respectively.  The /hourly
subdirectory contains a a subdirectory /ECMWF (described below),
a file named 'scalars', and a set of files named 'lagr[n]_h[mm]'
containing hourly composite soundings for the periods:

16 UTC 12 June-10 UTC 14 June, 1992 for Lagr. 1
22 UTC 18 June-14 UTC 20 June, 1992 for Lagr. 2

In the hourly file name, [n] = 1 or 2 is the Lagrangian number,
and [mm] is the hour number, starting from 00 at the first hour.
Each file has a header with hourly interpolated values of:

      June day   decimal day of June
      lat, lon   trajectory 'best guess' latitude (N) and
                 longitude (W) [I]
      SST        best fit SST (K)  [II,  figs. 1a  and  1b]
      psurf      surface pressure (hPa) [II, fig 2]
      Div        avg. horizontal divergence over the MBL depth
                 (10^-6 s^-1) [II,section 3d].  Mean vertical
                 velocity = -D*z in the MBL.
      LW_dn      downwelling broadband longwave flux at 700 hPa 
                 (W/m^2) [III, table 1]. To adjust this 
                 observation to a different pressure p, add 
                 0.4*(p - 700) W/m^2.  Except for days 19.45-
                 19.7, the downgoing longwave flux above the
                 inversion closely approximates the clear sky 
                 value obtained from the ECMWF column sounding
                 (which is used for the sounding at pressures 
                 less than 700 mb).  On days 19.45-19.7,
                 overlying cirrus raised the downgoing longwave
                 flux by about 50 W/m^2.
      N          Mean in-cloud droplet concentration (cm^-3)
                 derived by hourly interpolation from the in-
                 cloud segments of horizontal legs in which 30%
                 or more of the leg was in-cloud [II,fig. 5].
                 This N is based on measurements entirely inde-
                 pendent of the N's tabulated in the soundings.

Together with the sounding information, the header information in
each hourly file is sufficient to completely specify boundary
conditions and initial conditions for a Lagrangian column model
of the marine boundary layer started at an arbitrary hour into
either Lagrangian.  The file 'scalars' tabulates the hourly
values of the above quantities for the entire Lagrangian, along
with the inversion-base pressure pib (hPa) for each hour.

In each hourly file, after the header follows a composite
sounding with 50-100 mb resolution from 100 mb-700 mb and 10 mb
resolution from 700 mb down to the surface.  The columns ug and
vg are the geostrophic wind components interpolated from ECMWF
height fields, qv and ql are water vapor and liquid water mixing
ratios and N is the mean droplet concentration.  The remaining
variables are standard.

Note that unlike in the header, the N tabulated in the sounding
is an average over both cloudy and non-cloudy segments of several
aircraft soundings.  If N exceeds 50% of the mean in-cloud drop
concentration given in the sounding header, it is a fair guess
that most of the average is over cloud.  In addition, qv may not
match the saturation mixing ratio even in mainly cloudy segments
due to small measurement biases, cloud holes, etc.  For using a
composite sounding to initialize a model run, I recommend that
you replace the vapor mixing ratio by the saturation mixing ratio
wherever either N exceeds 50% of the mean in-cloud drop
concentration, or ql exceeds 0.05 g/kg.

Note also that since the geostrophic winds are from a large-scale
analysis, they may have significant errors and are not always
entirely consistent with the observed wind field.  Caveat emptor!

     The hourly soundings were derived as follows:

(i)  From each of the 17 2mb averaged aircraft soundings in each
     Lagrangian, a corresponded blended sounding was constructed
     by assuming:
     (a)  Between the lowest sounding pressure and the surface
          pressure (anywhere from 1-5 mb higher), theta, qv, u, 
          v are uniform and equal to the lowest sounding value.
     (b)  More than 50 mb above the sounding top, and everywhere
          above 700 mb, ECMWF data interpolated to the sounding
          time and position are used.
     (c)  At the sounding top, the difference of the aircraft
          and ECMWF data is found.  Between 0 and 50 mb above the
          sounding top, this difference is linearly ramped to
          zero and added to the ECMWF data.
     (d)  Geostrophic winds from ECMWF height fields are 
          tabulated at all pressures
     (e)  A column itype is included in the sounding which 
          indicates the type of data used at each pressure:
          1 - aircraft
          2 - aircraft, with ECMWF winds (soundings 9,10 of
                                               Lagr. 1 only)
          3 - ECMWF
          4 - Ramped between ECMWF and aircraft
          5 - Extrapolated below lowest aircraft level
     The 2mb aircraft and blended soundings can be found in 
     subdirectories soundings/2mb and soundings/blended of /lagr1
     and /lagr2.

(ii) From the blended soundings, an hourly composite sounding was
     created by using a weighted average of all aircraft 
     soundings within 3 hours of the compositing time.  Soundings
     more than 3 hours away were used if necessary to make a
     temporally unbiased average.  The compositing was done so as
     to preserve the inversion sharpness by vertically scaling
     each sounding to have the same inversion base height and
     surface pressure. A different scaling is used above the
     inversion base to bring the scaled pressure back to its true
     value at 700 mb for each blended sounding.

The subdirectory /ECMWF should not be necessary to look at, but
     List of the 21 pressure levels for the ECMWF-derived (.ec)
     fields, which run from 1010 hPa to 100 hPa.,,,,,
     The hourly ECMWF theta (K), qv (g/kg), u, v, ug, vg (m/s).
     Each field is stored in order of increasing hour, with each
     of the 21 pressure levels in given sequentially
     for each hour. Written in (f7.2,9f8.2) format with 10 values
     per row.

                  Model Verification

The following data in lagr/lagr[n] may be useful for model

(1)  The hourly composite soundings within the boundary layer.
(2)  The drizzle and cloud fraction data in the subdirectory

Other data such as turbulence profiles have been analyzed by
other investigators.  We hope to add such data to this archive if
possible.  If you have a dataset for the Lagrangians that might
be useful for comparison with dynamical or chemical column models
of the MBL, and if you would like to add your dataset to this
archive, send email to Chris Bretherton at

Plots of Data

Hourly data from lagr[n]/hourly

Lagrangian 1 trajectory, theta, qv, ql, N, u and v, ug and vg.

Lagrangian 2 trajectory, theta, qv, ql, N, u and v, ug and vg.

Data from lagr[n]/legmeans

Figures 1a and 1b of paper II are a plot of SST for Lagrangians 1 and 2, respectively. The four data sources are tabulated in anonymous ftp in lagr[n]/sst and discussed in README.sst. Each data source has been corrected for assumed biases, and is plotted along with the best fit SST used in the hourly dataset.

Figure 4 of paper II is a plot of the cloud fraction derived from FSSP in-situ measurements and upward pointing PRT5 Electra radiomenter measurements. Satellite derived cloud fraction (fig 10 of paper I) also is included for comparison.

Figure 5 of paper II is a plot of the in-cloud average droplet concentration from horizontal aircraft legs with at least 30% of leg in cloud

Figures 6 and 7 of II are plots of drizzle rate and drizzle fraction for the two Lagrangians.