A Model to Forecast Spring Bloom

From  ”State of physical, biological, and selected fishery resources of Pacific Canadian marine ecosystems in 2012

Susan Allen at UBC, together with co-workers from UBC and Fisheries and Oceans Canada, has developed a computer model for the physics and lower trophic levels in the Strait of Georgia which can successfully hindcast the spring bloom (Collins et al., 2009). They are now looking to use the model to forecast the occurrence of the spring bloom.

The spring phytoplankton bloom in the Strait of Georgia develops in the late winter or early spring as light becomes more available to phytoplankton. Light becomes stronger because: 1) the season is changing, bringing longer days and the sun higher in the sky; 2) the mixing layer is becoming shallower because wind speeds tend to drop; 3) the cloud fraction in the sky decreases allowing more sunlight to reach the ocean. The bloom decreases when the near surface nutrients are consumed by the phytoplankton. The date of the maximum bloom (greatest phytoplankton biomass) is generally within a few days of when
surface nutrients are exhausted. This date lags the start date of the bloom at ocean surface, as  determined by satellite observations. The  spring bloom of phytoplankton is dominated by diatoms, which are large phytoplankton that are important food for large zooplankton. Because large zooplankton provide feed for much of the  food chain, timing and intensity of these blooms are believed to be critical for growth and survival of juveniles of many species.

Allen 1
Figure 1. Examples of 1990-2010 zooplankton time series from the Strait of Georgia (from Mackas et al. in press). Top panel shows total copepods, bottom panel shows total euphausiids. Squares indicate log-scale anomalies relative to average seasonal cycle. Grey circles indicate annual geomean dryweight biomass; small triangles are biomass in individual samples. 1996 had too few samples to report annual geomean and anomaly.

In the figure to the left, one can see the mixing associated with a storm about day 70; salinities sharply increased as deeper, high-salinity water was mixed into the surface waters and temperatures also sharply increased as deeper, warmer water was mixed upward. The top panel shows phytoplankton biomass (in dark red) and nitrate (in green); in grey is the cloud fraction averaged over the day. One can see the evidence of mixing due to storms such as the day 70 storm; nitrate rapidly increased and phytoplankton biomass decreased. One can also see the influence of low-wind, low cloud-fraction periods such as that around day 30 and again around day 80. Here phytoplankton biomass steadily increased and nitrate decreased. The 2012 spring bloom was late (Apr 7) because of the large number of storms and associated high-cloud fraction.

The model hindcast the peak biomass date as Apr 7 2012. The onset of the spring bloom (as opposed to the peak biomass date) is determined as that date when the fluorescence signal from surface chlorophyll equivalent measured by satellite is observed to be more than about 5 mg.m-3 over 30% or more of the surface of the Strait. In 2012, the onset of the spring bloom was observed to be late, on about March 31. These two dates (onset March 31, peak Apr 7) compared well with the peak of the bloom also estimated by satellite.

The model can also be run using current weather data in a nowcast mode. Prior to the date of the spring bloom, it is possible to hypothesize the future weather and estimate the day of that year’s peak bloom. To predict the date of the 2013 spring bloom, the authors used three different future weathers: 1) the average weather (1968-2010) with a small correction so that if it is used from the previous autumn, it does give the average spring bloom date; 2) the weather from 1999 which led to the latest spring bloom (14 April) in the 1968-2010 period; and 3) the weather from 1993 which led to the earliest spring bloom (24 February) in the same period. Their forecast can be seen at http://eos.ubc.ca/~sallen/SoG-bloomcast/results.html.For comparison, observational data from the BC ferries measured by VENUS can be found at http://venus.uvic.ca/data/data-plots/strait-of-georgia-plots/strait-of-georgia-bc-ferries-vancouver-nanaimo-today/