Evaluation of modeling ecosystem seasonality in the University of Victoria Earth System climate model

Abstract Investigating climate model ability to simulate ecosystem seasonality, for instance causes and dynamics of phytoplankton blooms in the North Atlantic, is of major importance, because interannual and seasonal variations in bloom timing, duration and intensity caused by anthropogenic induced global climate change, can lead to species shifts and alterations in the trophic food web and biogeochemical cycles, which otherwise may remain undetected on an annual scale. The seasonal variability of the North Atlantic spring bloom is highly related to surface pCO2 (Takahashi et al., 2002; Corbiere et al., 2007; Signorini et al., 2012) and bloom dynamics have a substantial role in carbon sequestration (Watson et al., 1991; Townsend et al., 1994). Hence it is necessary to improve predictions of seasonal variability in ecosystem models in order to simulate future global warming more precisely. To evaluate the accuracy of the University of Victoria Earth System climate model (UVic model) simulations of timing, forcing factors and limitations (e.g. mixed layer depth, temperature, irradiance, nutrients, zooplankton grazing pressure) to the North Atlantic spring bloom, model output has been compared to observations from MODIS satellite images, WOA09 data, IFREMER records and PAP measurements. The results showed that the UVic model simulated phytoplankton growth rates inaccurately and estimated the spring bloom start approximately one month too late. The model consistently underestimated actual temperature values, but temperature changes were significantly correlated between observations and predictions. The mixed layer shallowed earlier, more and patchier in practice than in the model. The simulated bloom is limited by solar radiance in early spring, while nutrient limitations are pivotal during summer. At specific study sites temperature was detected as driving factor for bloom formation. The role of the mixed layer depth on bloom dynamics was not simulated adequately by UVic. Therefore especially simulations of the mixed layer depth and associated shoaling processes need to be optimized, maybe by improved parameterization of eddies and wind stress, to achieve accurate predictions of bloom dynamics and related forcing factors. This is important to provide precise simulations of phytoplankton bloom dynamics in the North Atlantic region for solid predictions about CO2 sequestration, biological pump magnitude and other biological and physical interaction processes with respect to seasonal variability.