Quantitative modelling of spatial variability in the north Atlantic spring phytoplankton bloom

• Main Author: Hemmings, J.C.P.
• Format: Text
• Language: English
• Published: 1999
• Subjects: North Atlantic
• Online Access: http://nora.nerc.ac.uk/id/eprint/142095/; https://nora.nerc.ac.uk/id/eprint/142095/1/0000350.pdf

The effects of variability in the physical environment on the development of the spring phytoplankton bloom are investigated using a physically forced model of the annual plankton cycle in the ocean mixed layer. The model is optimised to fit survey data from the eastern North Atlantic, collected over a 1500 x 1500 km area between 39N and 54N, from April-June 1991, establishing the feasibility of using spatially distributed point-in-time data in model calibration. Measurements made below the seasonal pycnocline show the existence of an empirical relationship between preformed nitrate and salinity in this area, allowing salinity-based estimates of pre-bloom mixed layer nitrate concentration to be made. These estimates provide important additional constraints for the model. The observed spatio-temporal patterns, at scales between 36 km and 1500 km, in nutrients, chlorophyll and measures of bloom progression derived from these data with reference to pre-bloom nitrate are discussed, together with the corresponding patterns in seasonal stratification. During the spring bloom, when biogeochemical concentrations vary rapidly in response to the developing stratification, absence of data defining its history limits the value of comparison between point-in-time observations and model results. Predictions of variation in stratification at the seasonal time-scale from general circulation models (GCMs) can be used in place of observational data to force ecosystem models. However, the degree to which observations are used to constrain the model solutions should allow for both model error in stratification and misrepresentation of the seasonal development of stratification by the observations. The latter occurs due to sampling error associated with short-term fluctuations. It can be corrected for if a suitable contemporary sea surface temperature data set is available to define the variation of mixed layer temperature at the seasonal time-scale. It is shown that the accuracy of the GCM predictions can be improved by the ...