process2d

EOL Facility: 
Research Aviation Facility (RAF)
Maintenance Status: 
Stable
Developers/Maintainers: 
RAF
Used by these EOL facilities: 
RAF
Application: 
OAP (2D) data processor. (Process 2D image and produce quantitative data.)
Status: 
Active & Maintained
Level of Support: 
Supported

Stakeholders:

Operating System: 
Mac
Linux

Processor to process Fast2D, 2DS, CIP, and 3V-CPI data.  Generates histogram, size distribution and other derived quantitative values from OAP image data.  Reads data stored in the generic OAP file format.  C++ code base, no GUI so portable to other systems.

 

OAP file specifications and Format description: http://www.eol.ucar.edu/raf/Software/OAPfiles.html

 

Processing Details


The Fast-2DC software processes and writes data with two particle populations, round particles and all particles.  The “round particle” population is intended to represent liquid water particles.  The “all particle” population follows the more traditional method of processing 2D image data, placing both round and irregularly shaped particles together into the same particle size distribution.  Both of these populations are processed for the entire duration of the raw data file.  The applicability of these populations will change based on many factors, and the decision of which population is most appropriate is left to the discretion of the end-user.

Part 1:  Particle Sizing and Sample Area

Particles can be measured by three methods, circle-fit, sizing across the array (x-size) and sizing with the airflow (y-size).  

The circle-fit method is the default sizing method.  It simply fits the smallest possible circle around a particle image and uses the diameter of that circle as the diameter of the particle.  This method is used for its computational efficiency, as well as its ability to produce a clean comparison of the area of particle to the area of the circle.  This area ratio is used for subsequent particle rejection, roundness detection, and may also be used for computing such parameters as fall velocity and optical extinction.

The x-size and y-size methods measure the maximum difference between shaded pixels in their respective directions.   X-size may be useful for spinning disc calibrations, or for any time where the probe's timing did not match the particle speed resulting in distorted images.   

In the case of particles flagged as “round” a sizing correction is applied following Korolev (2007).  This correction is based on the size of the Poisson spot seen when imaging liquid particles, and indicates magnification of a particle due to its position in the depth of field.  If a Poisson spot is detected, its area is measured and compared to the area of the complete particle.  The ratio of these two areas is used to find a correction factor, which reduces the size measurement to its expected pre-magnification value.  

In all sizing methods, partially imaged particles which touch either or both ends of the diode array are allowed by default.   The sample area of the probe is computed following the “reconstruction” method in Equation 17 of Heymsfield and Parrish (1978).  If the user elects to reject partially imaged particles, the sample area is computed following Equation 4 of the same reference.

Part 2:  Shattering Corrections

Large particles that impact on the forward surface of a probe arm can break into many pieces and then be imaged by the Fast-2DC probe.  This results in an overestimate of the concentration of small particles.   Since these small particles appear in clusters, the time between neighboring particles, or interarrival time, may be used to detect suspected shattering events.  The Fast-2DC software corrects for shattering events using the methods described in Field, et al. (2006), which are briefly described below.

The Fast-2DC software maintains a circular buffer of the last 400 interarrival times measured.  For each 1Hz time period, a histogram of these interarrival times is compiled.  If there are very few shattering events, the histogram will resemble a Poisson distribution.  If there are many shattering events the distribution will have a double-Poisson shape with two distinct peaks, one for the natural particle population and one for the shattered particle population.  A non-linear least squares double-Poisson fit is made for each of these distributions.  The interarrival time of each peak of the double-Poisson shape is then found, and an appropriate cutoff is determined to distinguish between the natural population and the shattered population.  At this point all particles, and their preceding neighbor, with an interarrival time below this cutoff time are rejected.  This method also rejects naturally-occurring particles that may have short interarrival times, so a statistical correction is made to account for these particles.

Shattering corrections may be turned off at the command line if the user wishes not to use them.

Part 3:  Particle Rejection Criteria

The particle rejection criteria in the Fast-2DC software serve two purposes, to distinguish between “round” and “all” particles, and to remove image artifacts.  Image artifact rejection is simply based on the ratio of the measured area of a particle (after holes are filled) to the area of   the smallest circle that can enclose that particle.  If this ratio falls below a certain threshold, the particle is rejected.  Distinguishing between “round” and “all” particles is done in a similar manner, with the area ratio requirement raised to eliminate particles that do not meet a certain roundness.   The rejection criteria details are as follows:

“All” particles rejected if:
Area ratio < 0.1
Particle size outside of size-bin range

“Round” particles rejected if:
Area ratio < 0.4
Area ratio < 0.5 for particles 10 pixels or larger
Size greater than 6mm
Corrected particle size outside of size-bin range

File Output

After processing is complete, a new netCDF file will be created or variables will be added to an existing netCDF file.  Units, descriptions, and other metadata may be found in the netCDF file itself.  The variables created by the Fast-2DC software are:

A2DCA         Fast 2DC Corrected Counts per Channel, All Particles
A2DCR         Fast 2DC Corrected Counts per Channel, Round Particles
C2DCA         Fast 2DC Concentration per Channel, All Particles
C2DCR         Fast 2DC Concentration per Channel, Round Particles
I2DCA         Interarrival Time Counts, All Particles Including Rejections
CONC2DCA        Total Fast 2DC Concentration, All Particles
CONC2DCR        Total Fast 2DC Concentration, Round Particles
PLWC2DCR        Fast 2DC Liquid Water Content, Round Particles
PLWC2DCA        Fast 2DC Liquid Water Content, All Particles
DBAR2DCR        Fast 2DC Mean Particle Diameter, Round Particles
DBAR2DCA        Fast 2DC Mean Particle Diameter, All Particles
DISP2DCR        Fast 2DC Dispersion, Round Particles
DISP2DCA        Fast 2DC Dispersion, All Particles
DBZ2DCR         Fast 2DC Calculated Reflectivity, Round Particles
DBZ2DCA          Fast 2DC Calculated Reflectivity, All Particles
REFF2DCR          Fast 2DC Effective Radius, Round Particles
REFF2DCA          Fast 2DC Effective Radius, All Particles
NACCEPT2DCR      Number of Particles Accepted, Round Particles
NACCEPT2DCA      Number of Particles Accepted, All Particles
NREJECT2DCR      Number of Particles Rejected, Round Particles
NREJECT2DCA      Number of Particles Rejected, All Particles
poisson_coeff1     Interarrival Time Fit Coefficient 1
poisson_coeff2     Interarrival Time Fit Coefficient 2
poisson_coeff3     Interarrival Time Fit Coefficient 3
poisson_cutoff      Interarrival Time Lower Limit
poisson_correction      Count/Concentration Correction Factor for Interarrival Rejection
SA              Sample area per channel
bin_endpoints    Size bin endpoints (microns)
bin_midpoints    Size bin midpoints (microns)
interarrival_endpoints    Interarrival bin endpoints (seconds)

References

Field, P. R., A. J. Heymsfield, A. Bansemer, 2006: Shattering and Particle Interarrival Times Measured by Optical Array Probes in Ice Clouds. J. Atmos. Oceanic Technol., 23, 1357–1371.
doi: 10.1175/JTECH1922.1

Heymsfield, Andrew J., Joanne L. Parrish, 1986: An Interactive System for Processing PMS Two-Dimensional Imaging Probe Data. J. Atmos. Oceanic Technol., 3, 734–736.
doi: 10.1175/1520-0426(1986)003<0734:AISFPP>2.0.CO;2

Korolev, Alexei, 2007: Reconstruction of the Sizes of Spherical Particles from Their Shadow Images. Part I: Theoretical Considerations. J. Atmos. Oceanic Technol., 24, 376–389.
doi: 10.1175/JTECH1980.1