NCAR/ACD FLEXPART backtrajectories for SAS NOMADSS C130 flights
2013-11-07
Christoph Knote, ACD, NCAR
3450 Mitchell Lane
Boulder, CO, 80301
USA
+1 303 497 1874
knote@ucar.edu
1.0) Overview
This dataset contains output of FLEXPART, a Lagrangian Particle Dispersion Model
(LPDM). It provides information on the history of air sampled along the C130
flight track by releasing "particles" (infinitesimally small parcels of air) from
the plane location and following its path backwards in time. Advection,
turbulence and convection are the processes considered. A particle itself is
"inert", processes like deposition, emission or chemical conversion are not
considered. Thereby the result could be called "airmass history", as the
particles behave like air. Its location back in time and space since released
is called "backtrajectory". The chaotic nature of the atmosphere and
uncertainties in model estimates of wind vector, turbulence and convection
requires a statistical approach to estimate all probable backtrajectories.
Hence not only one, but a large number of particles (1e5 - 1e6) are released
over a short time period (1 hr) and followed back in time. While the exact
trajectory for each particle is used internally, the main model result is the
number of particles and the time they spent within a 3D lat/lon/altitude grid,
called a "sensitivity" or "residence time" field. This is calculated at
discrete intervals (every hour) in time since release. These fields are the
raw output of the model calculations. For ease of use, deterministic
backtrajectories as mass weighted mean of the plume at each hour, and 5 plume
cluster centroids (Stohl et al., Atm. Env., 2002) are calculated as well.
2.0) Model description
The FLEXPART model (http://flexpart.eu) in version 9.02 has been used. FLEXPART
has been originally developed by A. Stohl (NILU, Norway) and is continually
improved by an international team of developers. The model has been modified to
allow the output of NetCDF data instead of the native binary format. This code
has been kindly provided by Stephan Henne / Dominik Brunner of Empa,
Duebendorf, Switzerland.
3.0) Model setup
The main FLEXPART configuration file can be found in Appendix A.1.
Meteorological forcing data from analyses every 6 hours (interlaced with the 3
hour forecast fields) of the Global Forecasting System (GFS) of the National
Center for Environmental Prediction (NCEP)
(http://www.nco.ncep.noaa.gov/pmb/products/gfs/) have been used. These are
provided on a 0.5 x 0.5 degrees regular latitude/longitude grid.
Release points have been defined every five minutes along the GPS flight track
of the C130 for each research flight. Between 15.000 and 30.000 particles per
point are released instantaneously and followed back in time for approximately
72 hours. The resulting footprint is hence representative for air sampled at the
start date (given as filename and in header information / plots).
The sampling grid used in model output is at 0.2 x 0.2 degrees resolution with
18 vertical levels (see A.2), roughly covering the Northern Hemisphere from the
date line to Greenwich.
4.0) Data format
File name conventions: bt--.(png|ict|nc) with
YYYYmmdd_HHMM the start time of the particles.
bt-RF16-20130708_1503.png is hence the plot output of particles released during
research flight 16 at 15:03 UTC on July 8th 2013.
Three distinct outputs are generated:
* NetCDF output: CF-compliant NetCDF version 4 (using standard HDF compression)
files that contain a 2D (lat, lon) field for model orography ("oro"), and a 4D
(time, level, lat, lon) field of the "residence time" (or sensitivity) at each
point in time in space ("conc", in units "s m3 kg-1"). Stohl et al., ACP, 2005
discuss this fields in detail.
Variables:
time - seconds since start time of simulation (coincides with end of
particle release interval)
lon/lat - longitude / latitude of output grid cell centers
level - height above ground of output grid levels (level top)
oro - surface altitude in meters
conc - residence time / sensitivity field for particles released
* ICARTT output: a text file containing backtrajectories in ICARTT/NASA AMES
format. Given are for each hour the mass-weighted center trajectory locations
(lon, lat, altitude) and the centroids (lat, lon, altitude) for 5 clusters
including uncertainty (root mean square) and fraction of total particle number
pertaining to each cluster. The methods to derive these trajectories are
described in Stohl et al., Atm. Env., 2002.
* Plot output: an overview plot showing a horizontal integration (s m2 kg-1)
and vertical average (s m3 kg-1) of the "conc" field in the NetCDF file, and
additionally the trajectories (center / clusters) given in the ICARTT output.
5.0) Data remarks
What to use when?
The atmosphere is chaotic - a deterministic backtrajectory can be very
misleading (see e.g. Stohl et al., Atm. Env., 2002) as it does not capture
stochastic processes like turbulence or convection. For all quantitative
assessments the user is strongly urged to use the residence time / sensitivity
fields in the NetCDF output instead of the trajectories in the ICARTT file.
Higher time/space resolution?
More detailed calculations with higher spatial and temporal resolutions are
possible by using output of a regional model (e.g. WRF). This can easily be
done on a per request basis - just let me know.
Why NetCDF v4?
NetCDF version 4 is used as sparse matrix-type compression is available there.
Backtrajectories are usually very spares (lots of "empty cells"), hence such a
compression algorithm is very efficient here: a NetCDF v4 backtrajectory file
uses ~25 MB, where as the same file as NetCDF v3 needs ~2.1 GB (sic). All
standard tools (nco, ncl, cdo, IDL, R, MatLab, IGOR) can work with NetCDF v4,
but some older installations might need to be re-installed / re-compiled.
6.0) References
Stohl, A., Eckhardt, S., Forster, C., James, P., Spichtinger, N., & Seibert, P.
(2002). A replacement for simple back trajectory calculations in the
interpretation of atmospheric trace substance measurements. Atmospheric
Environment, 36(29), 4635-4648.
Stohl, A., Forster, C., Frank, A., Seibert, P., & Wotawa, G. (2005). Technical
note: The Lagrangian particle dispersion model FLEXPART version 6.2.
Atmospheric Chemistry and Physics, 5(9), 2461-2474.
A.1) COMMAND file
+++++++++++++ HEADER +++++++++++++++++
+++++++++++++ HEADER +++++++++++++++++
+++++++++++++ HEADER +++++++++++++++++
+++++++++++++ HEADER +++++++++++++++++
+++++++++++++ HEADER +++++++++++++++++
+++++++++++++ HEADER +++++++++++++++++
+++++++++++++ HEADER +++++++++++++++++
-1
3600
3600
300
9999999
300 SYNC
5.0 CTL
4 IFINE
13 IOUT
0 IPOUT
1 LSUBGRID
1 LCONVECTION
0 LAGESPECTRA
0 IPIN
1 IOFR
0 IFLUX
0 MDOMAINFILL
1 IND_SOURCE
2 IND_RECEPTOR
0 MQUASILAG
0 NESTED_OUTPUT
2 LINIT_COND INITIAL COND. FOR BW RUNS: 0=NO,1=MASS UNIT,2=MASS MIXING RATIO UNIT
A.2) OUTGRID file
********************************************************************************
* *
* Input file for the Lagrangian particle dispersion model FLEXPART *
* Please specify your output grid *
* *
********************************************************************************
1. ------.---- 4X,F11.4
-178.1000 GEOGRAFICAL LONGITUDE OF LOWER LEFT CORNER OF OUTPUT GRID
OUTLONLEFT (left boundary of the first grid cell - not its centre)
2. ------.---- 4X,F11.4
-20.1000 GEOGRAFICAL LATITUDE OF LOWER LEFT CORNER OF OUTPUT GRID
OUTLATLOWER (lower boundary of the first grid cell - not its centre)
3. ----- 4X,I5
900 NUMBER OF GRID POINTS IN X DIRECTION (= No. of cells + 1)
NUMXGRID
4. ----- 4X,I5
500 NUMBER OF GRID POINTS IN Y DIRECTION (= No. of cells + 1)
NUMYGRID
5. ------.--- 4X,F10.3
0.200 GRID DISTANCE IN X DIRECTION
DXOUTLON
6. ------.--- 4X,F10.3
0.200 GRID DISTANCE IN Y DIRECTION
DYOUTLAT
10. -----.- 4X, F7.1
100.0
LEVEL 1 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
250.0
LEVEL 2 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
500.0
LEVEL 2 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
1000.0
LEVEL 2 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
1500.0
LEVEL 3 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
2000.0
LEVEL 3 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
3000.0
LEVEL 4 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
4000.0
LEVEL 5 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
5000.0
LEVEL 6 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
6000.0
LEVEL 7 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
7000.0
LEVEL 7 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
8000.0
LEVEL 7 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
9000.0
LEVEL 8 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
10000.0
LEVEL 8 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
12500.0
LEVEL 8 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
15000.0
LEVEL 8 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
20000.0
LEVEL 8 HEIGHT OF LEVEL (UPPER BOUNDARY)
10. -----.- 4X, F7.1
50000.0
LEVEL 9 HEIGHT OF LEVEL (UPPER BOUNDARY)