CytoSense high-frequency monitoring data meets machine learning

22.03.2018  by  Lucyna Wlodarczyk


Monitoring platform on lake Greifensee in Switzerland is the place where the published study was performed.

Results of the experiment running on lake Greifensee in summer-fall 2014 and 2015. The data was acquired at 6 depths with a high frequency of 4 hours.

How will environmental changes influence phytoplankton communities? A recently published study by Mridul Thomas, Francesco Pomati and co-workers poses an important step towards answering this question. The researchers analysed the phytoplankton community in lake Greifensee, Switzerland every 4 hours at 6 different depths using a CytoSense flow cytometer. Within two years they collected over 7000 measurements. This tremendous amount of phytoplankton monitoring data served as a base to quantify the predictability of phytoplankton cell density using machine learning algorithms. One of the very promising findings is that the environmental dependencies of cyanobacteria and eukaryotic algae growth were in accordance with findings in controlled laboratory studies and this allowed the machine learning approach to make accurate forecasts of cell densities.

This work is now published in Ecology Letters under the title: “The predictability of a lake phytoplankton community, over time‐scales of hours to years”. The full text is available here.

 

Congratulations to the authors for this outstanding work!