Julia Lober Undergraduate, Tufts University, School of Arts and Sciences, Earth and Ocean Sciences
COLLABORATOR Pan Irene Chen MLA Candidate, Rhode Island School of Design, Landscape Architecture
This project aims to explore the diversity and variation of phytoplankton through images collected during research cruises off the east coast of the United States, totaling over 10 million images. 50 images from Rhode Island samples were randomly selected from each sample location and stitched together using a machine-learning generative adversarial networks (GAN) process. The colors vary based on the size of the cells in each frame, mirroring the effect that morphological differences in phytoplankton have on the optical signatures. Looking at the three videos side-to-side, we can begin to see the variation that exists even in just a small slice of the ocean.
“As a researcher with the Mouw Lab, I work to connect the ocean’s optical signature with the populations of phytoplankton. Though small, phytoplankton lives at such great scales in the ocean that they affect the way that we see the water. We can even register this optical effect from space using specialized satellites.”
The decoding process for optical signatures is especially complicated in the coastal ocean of the Northeastern US. Fortunately, there is a collection of optical and phytoplankton data collected on a number of cruises traversing these waters. This data gives us a special look into a microscopic community that we normally see only from far away.
Julia & Irene trained machine learning GAN (generative adversarial network) models on phytoplankton populations sampled at three different points along the coastal ocean of the Northeastern US. The GAN models find similarities between phytoplankton cells, generating composite images that highlight their unifying characteristics.
Working and talking with collaborators and critics, Julia and Irene arrived at a middle ground. Allowing the colors to vary based on the shape and size, without drawing any lines to specific colors. While the colors may be fictional, the fundamental idea that color is influenced by the size and shape of the cell is very much grounded in scientific truth.
“The biggest challenge for me during this process was balancing artistic representation and scientific truth. For this project, the topic of color really challenged these tensions. The optical signatures that I work with cannot be boiled down to a single color, and I was faced with constant decisions about what assumptions or generalizations to make, without implying untrue information that could be construed as scientific truth.”
The colors in the kaleidoscope are determined by the area and density of the cell in the image, reflecting the way shape and size influence the phytoplankton's interactions with light in the ocean.
“What is my personal relationship to this topic? How do I feel about the things that I am trying to communicate? Who am I trying to tell it to? Most importantly, why am I trying to tell this story? When I develop scientific materials, I am more focused on how to effectively communicate my information and less so on why. Some of these questions are still unanswered for me. Though I could point to an audience for my project, people who may know nothing about phytoplankton, it was hard to delve into my feelings about this topic. What is certain, is that I will continue to approach my projects, both scientific and artistic, from a mindset that includes these why questions as much as the how."
Imaging FlowCytoBot, IFCB After Effects Runway
© Phytoplankton Kaleidoscope, 2021
This material is based upon work supported in part by the National Science Foundation under EPSCoR Cooperative Agreement #OIA-1655221.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.