104 Results from the 2018 University of British Columbia Workshop on Evaluation of Cloud Probe Processing Software

Monday, 9 July 2018
Regency A/B/C (Hyatt Regency Vancouver)
Greg McFarquhar, CIMMS, Norman, OK; and D. Baumgardner, A. Bansemer, J. Crosier, R. Dupuy, P. D. Rosenberg, D. Delene, A. Korolev, J. R. French, C. gurganus, P. Lawson, A. J. Heymsfield, M. Krämer, C. Voigt, A. Afchine, Y. Boose, Y. Wang, and I. Heckman

Immediately before the AMS Cloud Physics Conference, a workshop on the evaluation of cloud probe processing software is being held. This represents the 4th workshop of a series initiated in 2014 with the Data Analysis Workshop at the Massachusetts Institute of Technology, followed by the 2016 Workshop on Data Processing, Analysis and Presentation Software in Manchester and the 2017 Training Workshop on Processing of Cloud Particle Measurements held in Oberpfaffenhofen, Germany. The 2014 workshop summarized the software algorithms developed for processing data from instruments that make airborne measurements of cloud microphysical properties, whereas the 2016 workshop introduced software packages that implement these algorithms. The 2016 workshop set the stage for the 2017 workshop that provided hands-on training to students and early career scientists with the software packages that are most widely used by the scientific community.

The focus of the 2018 workshop is to resolve known discrepancies in how the most frequently used software packages implement processing algorithms. These discrepancies were identified during the 2014 workshop, discussed as part of the 2016 and 2017 workshops, whereas they remain a major impediment when comparing measurements from similar probes but from projects at different times and locations. In the months prior to the workshop, stepwise implementation of various correction algorithms was applied to a synthetic data set simulating data generated by a Cloud Imaging Probe, Precipitation Imaging Probe, Two-Dimensional Stereo Probe and High Volume Precipitation Sampler-3. The sequence of steps included comparing the number and dimension (length, width and area) of complete and partial images that are extracted, before any corrections, then after corrections for out of focus particles, followed by comparisons of number distribution functions before corrections for particle size, after corrections for particle size, after removal of shattered artifacts, and after reacceptance of spuriously removed particles. The results of the different algorithms were compared at each step in order to determine which algorithms and which variables in the algorithms produce discrepancies. For example, it was determined that various algorithms treated particles at the beginning or end of buffers differently, causing some discrepancies in the total particles counts.

The expected outcome from this meeting is to identify the basis for which algorithms produce discrepancies between the processing algorithms, and what aspects of those algorithms produce those discrepancies. This presentation will summarize the most important processing algorithms and the proper implementation of them when applied to Optical Array Probe measurements, outlining the uncertainties in the processing algorithms. Based on this workshop, a procedure for establishing a reference library of processing algorithms that includes those algorithms accepted as state-of-art by the community through either publication in the refereed literature or through passing some mutually agreed upon internal vetting process will be established. Thus, this work helps the cloud physics community move towards the development of community software for processing cloud probe measurements and towards the establishment of a consensus on the processing of cloud probe measurements.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner