The use of imaging flow cytobots (IFCBs) for plankton research is increasing worldwide and coordinated IFCB networks are being developed to monitor harmful algal blooms (HABs) in several coastal regions. Monitoring programs with IFCBs designed to run continuously can generate up to 70 samples per day creating a wealth of image data. Ideally, data streams are monitored daily (real-time) for data quality assurance and quality control (QA/QC). However, front end data QA/QC can be cumbersome for personnel and is often left for a later date once thousands of data files have accumulated. Particle size distribution (PSD) is used to inform food web dynamics, calculate total community biomass, and calculate radiative transfer in ocean remote sensing. PSD can be generated from equivalent spherical diameter (ESD), a measure derived from IFCB image processing, and in previous work, anomalous IFCB generated PSDs identified bloom events in San Francisco Bay, CA. We propose that variations in PSDs also reveal “bad” data to allow for some automation in backend QA/QC procedures. As more and larger IFCB networks come online worldwide, the use of automated data QA/QC is prudent to increase the efficiency of working with these datasets. While full automation of IFCB data QA/QC is unlikely, using PSD to automatically flag data allows users to focus their efforts on a reduced number of data to determine whether they are questionable or reflect shifts in community structure.
Hayashi et al. (Fri,) studied this question.