vp) from a polar volume (pvol) fileR/calculate_vp.R
calculate_vp.RdCalculates a vertical profile of biological scatterers (vp) from a polar
volume (pvol) file using the algorithm
vol2bird (Dokter et al.
2011 doi:10.1098/rsif.2010.0116
). Requires a running
Docker daemon, unless a local installation of
vol2bird is specified with local_install.
calculate_vp(
file,
vpfile = "",
pvolfile_out = "",
autoconf = FALSE,
verbose = FALSE,
warnings = TRUE,
mount,
sd_vvp_threshold,
rcs = 11,
dual_pol = TRUE,
rho_hv = 0.95,
elev_min = 0,
elev_max = 90,
azim_min = 0,
azim_max = 360,
range_min = 5000,
range_max = 35000,
n_layer = 20,
h_layer = 200,
dealias = TRUE,
nyquist_min = if (dealias) 5 else 25,
dbz_quantity = "DBZH",
mistnet = FALSE,
mistnet_elevations = c(0.5, 1.5, 2.5, 3.5, 4.5),
local_install,
local_mistnet
)Character (vector). Either a path to a single radar polar volume
(pvol) file containing multiple scans/sweeps, or multiple paths to scan
files containing a single scan/sweep. Or a single pvol object. The file data format should be either 1)
ODIM
format, which is the implementation of the OPERA data information model in
the HDF5 format, 2) a format
supported by the RSL library or 3) Vaisala
IRIS (IRIS RAW) format.
Character. File name. When provided, writes a vertical profile
file (vpfile) in the ODIM HDF5 format to disk.
Character. File name. When provided, writes a polar
volume (pvol) file in the ODIM HDF5 format to disk. Useful for converting
RSL formats to ODIM.
Logical. When TRUE, default optimal configuration settings
are selected automatically and other user settings are ignored.
Logical. When TRUE, Docker stdout is piped to the R
console. Always TRUE on Windows.
Logical. When TRUE, vol2bird warnings are piped to the R
console.
Character. Directory path of the mount point for the Docker container.
Numeric. Lower threshold for the radial velocity
standard deviation (profile quantity sd_vvp) in m/s. Biological signals
with sd_vvp < sd_vvp_threshold are set to zero. Defaults to 2 m/s for
C-band radars and 1 m/s for S-band radars.
Numeric. Radar cross section per bird to use, in cm^2.
Logical. When TRUE, uses dual-pol mode, in which
meteorological echoes are filtered using the correlation coefficient
rho_hv. When FALSE, uses single polarization mode based only on
reflectivity and radial velocity quantities.
Numeric. Lower threshold in correlation coefficient to use for filtering meteorological scattering.
Numeric. Minimum elevation angle to include, in degrees.
Numeric. Maximum elevation angle to include, in degrees.
Numeric. Minimum azimuth to include, in degrees clockwise from north.
Numeric. Maximum azimuth to include, in degrees clockwise from north.
Numeric. Minimum range to include, in m.
Numeric. Maximum range to include, in m.
Numeric. Number of altitude layers to use in generated profile.
Numeric. Width of altitude layers to use in generated profile, in m.
Logical. Whether to dealias radial velocities. This should typically be done when the scans in the polar volume have low Nyquist velocities (below 25 m/s).
Numeric. Minimum Nyquist velocity of scans to include, in m/s.
Name of the available reflectivity factor to use if not
DBZH (e.g. DBZV, TH, TV).
Logical. Whether to use the MistNet segmentation model.
Numeric vector of length 5. Elevation angles to feed to the MistNet segmentation model, which expects exactly 5 elevation scans at 0.5, 1.5, 2.5, 3.5 and 4.5 degrees. Specifying different elevation angles may compromise segmentation results.
Character. Path to local vol2bird installation (e.g.
your/vol2bird_install_directory/vol2bird/bin/vol2bird.sh).
Character. Path to local MistNet segmentation model in
PyTorch format (e.g. /your/path/mistnet_nexrad.pt).
A vertical profile object of class vp. When defined, output files
vpfile and pvolfile_out are saved to disk.
Common arguments set by users are file, vpfile, autoconf and mount.
Turn on autoconf to automatically select the optimal parameters for a given
radar file. The default for C-band data is to apply rain-filtering in single
polarization mode and dual polarization mode when available. The default for
S-band data is to apply precipitation filtering in dual-polarization mode
only.
Arguments that sometimes require non-default values are: rcs,
sd_vvp_threshold, range_max, dual_pol, dealias. Other arguments are
typically left at their defaults.
On repeated calls of calculate_vp(), the Docker container mount can be
recycled from one call to the next if subsequent calls share the same mount
argument. Re-mounting a Docker container takes time, therefore it is advised
to choose a mount point that is a parent directory of all volume files to be
processed, such that calculate_vp() calls are as fast as possible.
For altitude layers with a VVP-retrieved radial velocity standard deviation
value below the threshold sd_vvp_threshold, the bird density dens is set
to zero (see vertical profile vp class). This threshold
might be dependent on radar processing settings. Results from validation
campaigns so far indicate that 2 m/s is the best choice for this parameter
for most C-band weather radars, which is used as the C-band default. For
S-band, the default threshold is 1 m/s.
The default radar cross section (rcs) (11 cm^2) corresponds to the average
value found by Dokter et al. (2011) in a calibration campaign of a full
migration autumn season in western Europe at C-band. Its value may depend on
radar wavelength. rcs will scale approximately \(M^{2/3}\) with M the
bird's mass.
azim_min and azim_max only affects reflectivity-derived estimates in the
profile (DBZH, eta, dens), not radial-velocity derived estimates (u,
v, w, ff, dd, sd_vvp), which are estimated on all azimuths at all
times. azim_min, azim_max may be set to exclude an angular sector with
high ground clutter.
Using default values of range_min and range_max is recommended. Ranges
closer than 5 km tend to be contaminated by ground clutter, while range gates
beyond 35 km become too wide to resolve the default altitude layer width of
200 meter (see beam_width()). range_max may be extended up to 40 km
(40000) for volumes with low elevations only, in order to extend coverage
to higher altitudes.
The algorithm has been tested and developed for altitude layers with h_layer = 200m. Smaller widths than 100 m are not recommended as they may cause
instabilities of the volume velocity profiling (VVP) and dealiasing routines,
and effectively lead to pseudo-replicated altitude data, since altitudinal
patterns smaller than the beam width cannot be resolved.
You can bypass the Docker container and speed up processing by installing
vol2bird locally (not on Windows). Point local_install to the path of your
local vol2bird executable, e.g.
/your/vol2bird_install_directory/vol2bird/bin/vol2bird. Your local vol2bird
executable will be called through a bash login shell. LD_LIBRARY_PATH
(Linux) or DYLD_LIBRARY_PATH (Mac) should be correctly specified in your
.bashrc or .bash_profile file and contain all the required shared
libraries by vol2bird. See vol2bird installation pages on
GitHub for details.
When using MistNet with a local vol2bird installation, also point parameter
local_mistnet to your local download of the MistNet segmentation model in
PyTorch format, e.g. /your/path/mistnet_nexrad.pt. The MistNet model can
be downloaded at https://s3.amazonaws.com/mistnet/mistnet_nexrad.pt.
Dokter et al. (2011) is the main reference for the profiling algorithm
(vol2bird) underlying this function. When using the mistnet option, please
also cite Lin et al. (2019). When dealiasing data (dealias), please also
cite Haase et al. (2004).
Dokter AM, Liechti F, Stark H, Delobbe L,Tabary P, Holleman I (2011) Bird migration flight altitudes studied by a network of operational weather radars, Journal of the Royal Society Interface 8 (54), pp. 30-43. doi:10.1098/rsif.2010.0116
Haase G & Landelius T (2004) Dealiasing of Doppler radar velocities using a torus mapping. Journal of Atmospheric and Oceanic Technology 21(10), pp. 1566-1573. doi:10.1175/1520-0426(2004)021<1566:DODRVU>2.0.CO;2
Lin T-Y, Winner K, Bernstein G, Mittal A, Dokter AM, Horton KG, Nilsson C, Van Doren BM, Farnsworth A, La Sorte FA, Maji S, Sheldon D (2019) MistNet: Measuring historical bird migration in the US using archived weather radar data and convolutional neural networks. Methods in Ecology and Evolution 10 (11), pp. 1908-22. doi:10.1111/2041-210X.13280
# \dontrun{
# Locate and read the polar volume example file
pvolfile <- system.file("extdata", "volume.h5", package = "bioRad")
# Copy the file to a home directory with read/write permissions
file.copy(pvolfile, "~/volume.h5")
#> [1] TRUE
# Calculate the profile
vp <- calculate_vp("~/volume.h5")
# Get summary info
vp
#> Vertical profile (class vp)
#>
#> radar: seang
#> source: WMO:02606,RAD:SE50,PLC:Angelholm,NOD:seang,ORG:82,CTY:643,CMT:Swedish radar
#> nominal time: 2015-10-18 18:00:00
#> generated by: vol2bird 0.5.0.9169
# Clean up
file.remove("~/volume.h5")
#> [1] TRUE
# }