Constraints on the global marine iron cycle from a simple inverse model

Abstract

A simple model of the global marine iron cycle is used to constrain the sources, sinks, and biological cycling of iron. The iron model is embedded in a data‐assimilated steady state circulation, with biological cycling driven by a prescribed, data‐constrained phosphate cycle. Biogeochemical parameters are determined by minimizing a suitably weighted quadratic mismatch with available dissolved iron (dFe) observations, including GEOTRACES transects. Because the effective iron sources and sinks overlap, current dFe observations cannot constrain sources and sinks independently. We therefore determine a family of optimal solutions for a range of the aeolian source strength σA from 0.3 to 6.1 Gmol/yr. We find that the dFe observations constrain the maximum Fe:P uptake ratio R0 to be proportional to σA, with a range that spans most available measurements. Thus, with either R0 or σA specified, a unique solution is determined. Global inventories of total and free iron are well constrained at (7.4 ± 0.2) × 1011 and (1.39 ± 0.05) × 1010 mol Fe, respectively. The dFe distributions are very similar across the family of solutions, with iron limitation in the known high‐nutrient low‐chlorophyll regions. Hydrothermal source strength ranges from 0.57 to 0.73 Gmol/yr and does not vary systematically with σA suggesting that the hydrothermal and aeolian parts of the iron cycle are largely decoupled. The hydrothermal dFe anomaly in the euphotic zone is ∼10% and concentrated in subpolar regions of iron limitation. Enhanced ligand concentrations in old waters and in hydrothermal plumes are necessary to capture key features of the dFe observations.

The age of iron and iron source attribution in the ocean

“…In the Southern Ocean the ∼20% hydrothermal and sedimentary contributions for σ A =3.5 Gmol/yr increase to ∼30% for σ A =1.1 Gmol/yr and decrease to ∼15% for σ A =6.1 Gmol/yr. (For σ A >6.1 Gmol/yr, the surface dFe field displays unrealistic features [ Frants et al , ].) Similarly, the maximum zonally averaged aeolian mean iron age in the Southern Ocean euphotic zone of ∼260 years for σ A =3.5 Gmol/yr increases to ∼360 years for σ A =1.1 Gmol/yr and decreases to ∼210 years for σ A =6.1 Gmol/yr.…”

“…We use the simple data‐constrained iron cycling model of Frants et al [] embedded in the steady data‐assimilated global circulation of Primeau et al []. The circulation has a horizontal resolution of 2° × 2° and 24 vertical levels, ranging in thickness from 36 m at the surface to ∼600 m in the deepest ocean.…”

“…The steady state total dFe concentration χ obeys where is the advective‐eddy‐diffusive transport operator and is an operator that assigns a fraction σ of the phosphate uptake J PO4 to remineralization at depth with a Martin curve (i.e., is the particle sinking and remineralization operator). The phosphate uptake rate per unit volume, J PO4 , was parameterized in terms of satellite estimates of net primary production, J NNP , as , with the coefficients α and η optimized to minimize the volume‐weighted quadratic misfit in the PO 4 concentration between a separate phosphorus cycling model and the observational climatology as described by Frants et al []. This parameterization of J PO4 was inspired by the work of Hansell et al [] and Teng et al [] and produces an export production with a realistic pattern.…”

Simulating natural carbon sequestration in the Southern Ocean: on uncertainties associated with eddy parameterizations and iron deposition

“…Among the reasons for this dire situation are challenges such as the following: (1) sources and sinks overlap spatially such that standing stocks of iron cannot constrain the (speed of the) iron cycling (Frants et al, 2016); (2) aeolian sources of iron are intermittent and thus hard to quantify, (e.g., Duggen et al, 2010;Olgun et al, 2011); (3) physicochemical stabilization is not well understood such that, for example, the importance of iron originating from hydrothermal vents remains uncertain (Resing et al, 2015); and (4) iron sinks such as scavenging and precipitation are not well understood (e.g., Tagliabue et al, 2014) and an explicit representation of the essential iron-binding ligand dynamics in models has only just begun (Völker and Tagliabue, 2015).…”

Iron fertilization efficiency and the number of past and future regenerations of iron in the ocean

“…The recyclable fraction, f rec , of DFe scavenged by opal and POP (f bSi and f POP in the notation of Pasquier and Holzer (2017)), was prescribed to be 90 % for all 20 scavenging particles. f rec is highly uncertain and in recent models its value has spanned the entire 0-100 % range (e.g., Moore and Braucher, 2008;Galbraith et al, 2010;Frants et al, 2016). We established the sensitivity of our results to the value of f rec by generating new state estimates from our typical state by prescribing different values of f rec ranging from 0 to 100 % and re-optimizing the scavenging and source parameters following the strategy of Pasquier and Holzer (2017) (see Appendix C for details).…”