Estimation of Upper Ocean Vertical Mixing from Surface Ocean Colour Observations

Abstract

The growth of marine phytoplankton is tightly connected to the strength of upper ocean vertical mixing and the depth of the mixed layer. The vertical mixing supplies nutrients from deeper ocean water to the ocean surface. Meanwhile, the mixing also moves phytoplankton cells in and out of the sunlit surface layer. Strong mixing and a deep mixed layer therefore lead to a fast nutrient supply while phytoplankton growth is likely to be light-limited. Weaker mixing and a shallow mixed layer loosen the light- limitation while the nutrient supply is reduced. Thanks to the chlorophyll-a (Chl a) pigments in the phytoplankton cells, changes in the phytoplankton concentration can be remotely tracked with satellite imagery. In this thesis the potentials of using ocean surface Chl a observations to estimate upper ocean vertical mixing are examined for the Northern Atlantic. The central tool in this thesis is a one-dimensional nutrient-phytoplankton (NP) model. To guarantee that the model represents the characteristic growth of the Northern Atlantic, biological model parameters are calibrated to in situ observations of vertical Chl a and nutrient profiles. A sensitivity analysis shows that the surface Chl a concentration is indeed very sensitive not only to the strength of the vertical mixing but also its vertical shape. It is thereby shown that the calibrated NP model is capable to build a bridge between surface Chl a concentrations and vertical mixing. Next, the NP model is coupled to the k − ε turbulence model of the General Ocean Turbulence Model (GOTM) framework. The k − ε model is a common choice to simulate upper ocean turbulent processes. Yet, a comparison of in situ and satellite observations with modelling results shows that the turbulence model overestimates the vertical mixing. As a consequence, modelled surface Chl a concentrations increase too late and too strongly compared to satellite observations. To improve the model results, the turbulence parameters of the k − ε model need to be calibrated. Usually these parameters are calibrated to laboratory and field observations. Since such observations are sparse, the novel approach in this thesis is to use surface Chl a concentrations for the calibration of the turbulence parameters. This novel calibration method is applied in two ways. First, identical twin experiments are set up to examine the robustness of the method. Second, the method is applied to satellite Chl a observations. Applying the calibration method to satellite Chl a observations leads to a model calibration that allows for an earlier and shorter intensification of the modelled surface Chl a. This surface signal very much resembles the early spring blooms common for the subtropical region and is also found in satellite Chl a observations. In comparison to the initial model results, the vertical mixing during winter is reduced by the new parameterisation, which also compares better to observations. It is therefore shown, to our knowledge for the first time, that surface Chl a observations can be used to calibrate parameters in a turbulence model and thereby to estimate upper ocean vertical mixing.