[1] Phytoplankton bloom phenology has important consequences for marine ecosystems and fisheries. Recent studies have used remotely sensed ocean color data to calculate metrics associated with the phenological cycle, such as the phytoplankton bloom initiation date, on regional and global scales. These metrics are often linked to physical or biological forcings. Most studies choose one of several common methods for calculating bloom initiation, leading to questions about whether bloom initiation dates calculated with different methods yield comparable results. Here we compare three methods for finding the date of phytoplankton bloom initiation in the North Atlantic: a biomass-based threshold method, a rate of change method, and a cumulative biomass-based threshold method. We use these methods to examine whether the onset of positive ocean-atmosphere heat fluxes coincides with subpolar bloom initiation. In several coherent locations, we find differences in the patterns of bloom initiation created by each method and differences in the synchrony between bloom initiation and positive heat fluxes, which likely indicate various physical processes at play in the study region. We also assess the effect of missing data on the chosen methods.Citation: Brody, S. R., M. S. Lozier, and J. P. Dunne (2013), A comparison of methods to determine phytoplankton bloom initiation,

24We compared foraminifera, flora and geochemical (δ 13 C , total organic content and 25 C:N) analyses to reconstruct the magnitude of coastal subsidence during the 26 AD1700 great megathrust earthquakes at the Cascadia subduction zone. Four 27 modern transects collected from three intertidal zones at Siletz Bay, Oregon, USA, 28 produced three elevation dependent groups in both the foraminifera and 29 geochemical datasets. Foraminiferal samples from the tidal flat and low marsh are 30 identified by M. fusca abundances of > 45%, middle and high marsh by M. fusca 31 abundances of < 45% and highest marsh by T. irregularis abundances > 25%. The 32 δ 13 C values from the geochemically defined groups decrease with increasing 33 elevation; -24.1 ± 1.7‰ in the tidal flat and low marsh; -27.3 ± 1.4‰ in the middle 34 and high marsh; and -29.6 ± 0.8‰ in the highest marsh samples. We applied these 35 modern foraminfera and geochemical distributions to a core that contained the AD 36 1700 earthquake. Both techniques produced similar results for the coseismic 37 subsidence (0.88 ± 0.39m and 0.71 ± 0.56m) suggesting that δ 13 C has potential as a 38 efficient proxy for use in paleoseismology. 39 40

Key Points:• We compare and synthesize current theories of bloom initiation • We develop a mixing length scale to predict the initiation of the bloom • Subpolar blooms are driven by a shift from buoyancy-to winddriven mixing Abstract Subpolar phytoplankton blooms have traditionally been attributed to changes in the depth of the ocean's seasonal thermocline: as the upper ocean warms and stratifies in the spring, phytoplankton reside within increasingly shallow depths where they experience higher light levels, and, as a result, begin to bloom. Recent studies have challenged this explanation, proposing instead that bloom initiation is driven either by the onset of positive heat fluxes, decreases in wind strength, decreases in grazing pressure, or by eddy-induced stratification. We compare traditional and recent ideas of bloom initiation and present a new argument that attributes the initiation to a decrease in the dominant mixing length scales in the upper ocean. From an examination of data across the subpolar North Atlantic, we find that decreases in this length scale are a better predictor of bloom initiation than current theories, thus providing a new explanation of bloom dynamics in a one-dimensional framework.

Since publication, the Sverdrup hypothesis, that phytoplankton are uniformly distributed within the ocean mixed layer and bloom once the ocean warms and stratifies in spring, has been the conventional explanation of subpolar phytoplankton spring bloom initiation. Recent studies have sought to differentiate between the actively mixing section of the upper ocean and the uniform-density mixed layer, arguing, as Sverdrup implied, that decreases in active mixing drive the spring bloom. In this study, we use in situ data to investigate the characteristics and depth of active mixing in both buoyancy- and wind-driven regimes and explore the idea that the shift from buoyancy-driven to wind-driven mixing in the late winter or early spring creates the conditions necessary for blooms to begin. We identify the bloom initiation based on net rates of biomass accumulation and relate changes in the depth of active mixing to changes in biomass depth profiles. These analyses support the idea that decreases in the depth of active mixing, a result of the transition from buoyancy-driven to wind-driven mixing, control the timing of the spring bloom.

The ability of Earth System Models to accurately simulate the seasonal cycle of the partial pressure of CO2 in surface water ( pCO2SW) has important implications for projecting future ocean carbon uptake. Here we develop objective model skill score metrics and assess the abilities of 18 CMIP5 models to simulate the seasonal mean, amplitude, and timing of pCO2SW in biogeographically defined ocean biomes. The models perform well at simulating the monthly timing of the seasonal minimum and maximum of pCO2SW, but perform somewhat worse at simulating the seasonal mean values, particularly in polar and equatorial regions. The results also illustrate that a single “best” model can be difficult to determine, despite an analysis restricted to the seasonality of a single variable. Nonetheless, groups of models tend to perform better than others, with significant regional differences. This suggests that particular models may be better suited for particular regions, though we find no evidence for model tuning. Timing and amplitude skill scores display a weak positive correlation with observational data density, while the seasonal mean scores display a weak negative correlation. Thus, additional mapped pCO2SW data may not directly increase model skill scores; however, improved knowledge of the dominant mechanisms may improve model skill. Lastly, we find skill score variability due to internal model variability to be much lower than variability within the CMIP5 intermodel spread, suggesting that mechanistic model differences are primarily responsible for differences in model skill scores.