Augmenting the Biological Pump THE SHORTCOMINGS OF GEOENGINEERED UPWELLING

Author(s): Bauman, SJ; Costa, MT; Fong, MB; House, BM; Perez, EM; Tan, MH; Thornton, AE; Franks, PJS

C O M M E N TA R Y e ocean is the largest reservoir of mobile carbon over decadal to centennial time scales, absorbing approximately 41% of cumulative anthropogenic CO 2 emissions (Sabine and Tanhua, 2010).
Various geoengineering solutions seek to exploit this uptake capacity (see Vaughan and Lenton, 2011, for a review), including CO 2 injection (Marchetti, 1977), iron fertilization (Martin et al., 1994), and arti cial upwelling (Lovelock and Rapley, 2007). e ubiquity of social mediaallowing anyone to "self-publish"-and funding from crowd-sources and private foundations have allowed some proposals to gain traction outside of the peer-reviewed scienti c literature.

ARTIFICIAL UPWELLING PROPOSALS
Arti cial upwelling aims to stimulate primary production by bringing nutrient-rich, sub-euphotic water up to the surface. e upwelled nutrients would fuel enhanced xation of inorganic carbon into organic carbon, thus removing dissolved CO 2 from surface waters and potentially increasing the ux of atmospheric CO 2 into the ocean. Sinking of particulate organic carbon (POC) could then sequester that carbon in deep ocean waters for decades or centuries.
Models have also been used to explore the utility of arti cial upwelling in sequestering carbon (Dutreuil et al., 2009;Yool et al., 2009;Oschlies et al., 2010) and its potentially undesirable global e ects. As we describe below, these studies have clari ed the constraints, limitations, and consequences of geoengineered global carbon sequestration. Calvin's (2013) (Stegen et al., 1993).
Millennial-scale sequestration requires pumping carbon to an isopycnal that will remain out of contact with the atmosphere for at least 1,000 years. De Vries and Primeau (2011)

Nutrient Supply and Horizontal Advection
Maintaining upwelling nutrient plumes requires a deep nutrient resupply, which depends on regionally heterogeneous advection and remineralization.
Simulations show that surface nutrient plumes disperse quickly; in one arti cial upwelling model, nutrients were diluted to less than 2% of initial plume concentrations 10 m downstream from the injection site (Williamson et al., 2009).
Blooms are typically apparent ve to seven days a er nutrient injection to the euphotic zone, a er which phytoplankton begin to sink or are grazed (Boyd et al., 2007). At a canonical horizontal speed of 10 cm s -1 , a bloom travels 50 km in ve days. To capture the bloom, Calvin's down-pumps would need to be ~ 50 km to 100 km downstream of the up-pumps.
However, de ning "downstream" would be challenging in regions of even moderate mesoscale current variability.

PHYSICAL CONSIDER ATIONS
Moving water against a density gradient requires an input of kinetic energy. We estimated lower-bound energy requirements using temperature and salinity data from a typical Southern those of tidal energy are ~ 3.5 TW. us, our estimate of energy required represents more than all global wind energy and over one-third of the entire internal energy budget of the world ocean.
Although upwelling rates of ~ 45 m 3 h -1 have been achieved with conventional pumping methods, they are unsustainable for longer periods of time (White et al., 2010). Even novel techniques like airli ing operate with low (~ 15%) energy e ciency at high ow rates (White et al., 2010;Fan et al., 2013).
Pumps driven by ocean salinity gradients require no permanent energy input but only produce ow rates of around 46 m 3 day -1 (Tsubaki et al., 2007), substantially increasing the required 0.8 billion 1 m diameter pipes estimated by Yool et al. (2009)

CARBON CYCLE DYNAMICS
e biological and solubility pumps transport carbon into the ocean interior, creating a surface-to-deep dissolved inorganic carbon (DIC) gradient. e biological pump, driven by particulate carbon export, generates 90% of this gradient, and 10% is generated by the solubility pump, driven by temperature-dependent CO 2 solubility, water mass formation, and biologically generated air-sea pCO 2 gradients (Sarmiento and Gruber, 2006).
Arti cial upwelling might increase oceanic CO 2 uptake if phytoplankton blooms were to enhance export production.
However, increased primary production does not always result in increased export production, and export production is not a straightforward predictor for air-sea gas exchange (Oschlies and Kahler, 2004).
Because > 95% of organic carbon is remineralized within the upper 1,000 m (Yool et al., 2009), the increase in oceanic CO 2 uptake relative to primary production (i.e., e ciency) would be low. surface-to-deep DIC di erence (Millero, 2007), this pumping would bring 29 GtC yr -1 to the surface as DIC.  (Fuhrman and Capone, 1991;Jin and Gruber, 2003). In addition to generating potent greenhouse gases, expansion of denitri cation zones would increase the loss of nitrate, which fuels new production and is a limiting nutrient for global ocean primary productivity (Codispoti et al., 2001;Gruber, 2004)

ECOLOGICAL IMPACTS Eutrophication
Planktonic biomass and community structure depend on nutrient uxes; where surface nutrient uxes are low, Figure 3. Contours of power required for up-and down-pumping of water as a function of volumetric pumping rate and depth of the 1,000-year horizon. Water is assumed to be pumped up from 200 m (typically below the mixed layer depth and nutricline) to a depth of 50 m within the euphotic zone. Temperature, salinity, and pressure data are from a high-resolution conductivity-temperature-depth (CTD) cast (WOD Unique Cast Number 15561184) during the 2011 Southern Ocean cruise operated by Scripps Institution of Oceanography. e lack of a significant pycnocline at this location ensures these energy estimates are conservative.  (Aure et al., 2007).
Larger phytoplankton at the base of a marine food web result in greater biomass at all trophic levels, including commercially valuable species. Unfortunately, upwelling enhancement of sheries will reduce the e cacy of carbon sequestration because of the respiration of the xed organic carbon.
In some systems, upwelling leads to harmful algal blooms (HABs) (Ryan et al., 2009;though see McClimans et al., 2010), causing sh and marine mammal mortality (Flewelling et al., 2005) as well as economic and health concerns (Jin et al., 2008 (Cullen and Boyd, 2008;Lampitt et al., 2008;Law, 2008 to thank an anonymous reviewer and John Cullen, whose comments led to a much improved manuscript.