Defining the Date of the Spring Phytoplankton Bloom
Although the concept of the spring phytoplankton bloom is widely accepted, there is no single accepted definition to determine the date at which this occurs. For this analysis, 3 different methods were used. They were arbitrarily named Metric 1, 2 and 3 for the purpose of this analysis.
These definitions and functions can be found here .
Metric 1, defined by Allen and Wolfe (2013), defines the spring bloom date as the peak surface phytoplankton concentration within four days of the surface nitrate concentration going below 0.5 micromolar for two consecutive days. [1] This threshold value was changed to 2.0 micromolar, for consistency with the nitrate half saturation constant used in the 201905 SalishSeaCast model run. The surface phytoplankton and nitrate concentrations are assessed as the average from the surface to three meters in depth.
Metric 2 is a definition from Olson et. al (2020) that describes the spring bloom time as the date of the first peak in which mean upper-3-m chlorophyll concentrations are above 5 micrograms/L for more than two consecutive days. [2]
Metric 3 is a method used by Suchy et. al in prep. For a given year, bloom initiation is identified from the week the chlorophyll concentration first reaches a threshold value (by looking at weekly averages), as long as one of the two following weeks was >70% of that threshold value. The threshold value is the median + 5% of the annual chlorophyll concentration.
While Metric 2 and 3 both represent early bloom conditions, Metric 1 is associated with nitrate depletion, a trigger of bloom decline.
A 2007-2019 time series of annual bloom dates at Station S3 according to each metric can be found in this notebook. Metric 2 and 3 bloom dates were consistently close to one another. Metric 3 bloom dates were not as consistent from year to year, as they ranged from occurring within a few days of Metric 1 and 2 to a difference of nearly 3 weeks. Metric 3 was not as heavily considered in the subsequent analyses.
The variables for calculating bloom timing were taken from daily model output (.ptrc files)
from February 15th to June 14th, a time frame chosen based on previous knowledge of spring
bloom timing in the region [3]. Pickle files holding these variables were created for each year
from 2007-2020, and are stored in /ocean/aisabell/MEOPAR/extracted_files. These files
were created using this code. The file names are in the format
springBloomTime_year_loc_modver.pkl, with loc being the location of
interest (eg. S3), modver being the model run (e.g. 201905) and year being the
year. In the same directory, there are also spring bloom time files containing variables from the
201812 model run from 2015-2019.
References: