radtunfold − time-wise unfolding filter
radtunfold [−b BadWidth] [−h] [−k] [−n MinNum] [−r MaxRatio] [−w HalfWidth] [−t dbzMin] <input >output
radtunfold was created as an attempt to automate the process that would otherwise be manually accomplished by radedit. It operates on the assumption that there is time continuity in the wind velocities --ray-to-ray.
Radtunfold is designed to compare a Doppler velocity with its background, which is defined by the surrounding velocities with the same range. If the given velocity is not within the Nyquist interval centered at the median of its background, then the program add a multiple of the Nyquist interval to the velocity to bring it into that interval. If the the background is not well defined, the program takes note of that and either skips or kills the gate according to the command-line option [-k]. The final percent of skipped/killed gates is printed to stderr. The background is considered bad if there are not enough gates contributing to it or if its median is not representative of the velocities. The representativeness is measured in terms of the ratio of background velocities outside the Nyquist interval centered at the median to velocities of the background inside that interval. The minimum value for this ratio can be changed with the option [−r MaxRatio ], and the minimum number of rays required for the background with the option [−m MinNum ]. The number of rays used for the background can be controlled with the option [−w HalfWidth ]. When the background’s median is not representative, the program makes a final attempt to provide a background median to perform the unfolding. This is only tried if there is a neighbor gate that was successfully unfolded and the current gate is not already within the Nyquist interval centered at that gate. In this scenario, the program looks ahead a number of rays according to option [−b BadWidth ]. If the last good gate of this larger group has a velocity within the Nyquist interval centered at the last previously unfolded velocity then, the program reports the value of that unfolded velocity as the background median and bad otherwise.
Input data is expected on the standard input and radial velocities must have been previously unfolded along the ray. Typically the output from hrdrad2 with unfolding activated will be the input to radtunfold.
All input data are passed to the standard output, with radial velocities modified as described above.
−b BadWidth Number of rays to look over bad chunks. The default number is 25.
−h print a usage message.
−k
gates whose background is not well defined are set to
bad.
The default is to skip them.
−n MinNum Minimum number of velocities to define the background. The default number is 3.
−r MaxRatio Maximum ratio of background velocities outside the Nyquist interval centered at the median to velocities of the background inside that interval. The default value is 0.3 --If this ratio is exceeded, the background is deemed not good enough for the unfolding.
−w HalfWidth Number of rays to left and right used to define the background of a given ray. The default number is 5.
−t dbzMin Sets the minimum reflectivity threshold. All data beneath this threshold is set to badlim. If omitted, a value of −10 is assumed.
cdftunfold −t −10 < infile > outfile
hrdrad2, cdfrayrv , radedit
Situations where there is not enough good data surrounding a bad ray, or the unfolding along the ray messes up the majority of the rays, then radtunfold will smear the bad data out.
When you run
the program using −k it will report fewer
skipped/killed gates than if you omit it. This is because
after killing a gate, there is one less gate to use in the
comparisons. Consequently, a gate that would have been
flagged as skipped might no longer have enough data around
it to make a solid comparison, and so it will be skipped,
but no record of this skip will be kept.
However, if you omit −k the inconsistent data
will still be there for the comparison to be made. This
means that you will get a more complete record of skipped
data, but you also run the risk of this data having a
negative impact on good data.
If there is a section where at least 55% of the data is
wrong, the program will unfold the correct data instead.
This program is
a modification of the original radedit code
written by Dave Raymond. The modification was written
by Physics-major student Dave Wooten during summer of
2006.