In
this study we look beyond the previously studied effects of oceanic
An unexpected feature is the effect of the model's internal variability of
deep-water formation in the Southern Ocean, which, in some model runs, causes
additional oceanic carbon uptake after injection termination relative to a
control run without injection and therefore with slightly different
atmospheric
Anthropogenic
Marchetti (1977) proposed directly injecting
Over millennial timescales carbon from direct injection can simply be viewed
as “delayed” emissions, in terms of its climatic effect and fate, since the
carbon cycle will eventually reach a chemical equilibrium (mainly an
equilibrium between the oceanic and atmospheric carbon reservoirs, although
carbonate compensation and weathering feedbacks start acting on timescales
longer than 5000 years; e.g., Zeebe, 2012). However, on decadal to centennial
timescales, carbon that is sequestered via direct injection cannot simply be
treated as delayed emissions because the injected carbon must take
fundamentally different pathways than those of carbon that is emitted
directly into the atmosphere. Since these pathways operate on many different
timescales and are partially controlled by climate feedbacks, it takes a
considerable amount of time until the carbon cycle and climate reach the same
state as if the emissions had just been delayed. This is because injecting
Because direct injection of
However, a more comprehensive assessment of the carbon sequestration and climate mitigation potential of direct injection also requires accounting for the changes in all ambient carbon fluxes resulting from carbon cycle and climate feedbacks (Mueller et al., 2004; Vichi et al., 2013).
In this study, which follows Orr et al. (2001) in the configuration of the
However, since the future strength of terrestrial carbon cycle feedbacks,
such as the
The model used is version 2.9 of the University of Victoria Earth System
Climate Model (UVic ESCM). It consists of four dynamically coupled
components: a three-dimensional general circulation ocean model (Pacanowski,
1996), a dynamic–thermodynamic sea-ice model (Bitz and Lipscomb, 1999), a
terrestrial model (Meissner et al., 2003) and a one-layer atmospheric
energy–moisture balance model (based on Fanning and Weaver, 1996). All
components have a common horizontal resolution of
3.6
The model has been spun-up for 10 000 years under preindustrial atmospheric
and astronomical boundary conditions and run from 1765 to 2005 using
historical fossil-fuel and land-use carbon emissions (Keller et al., 2014).
From the year 2006 to 2100 the model is forced with
Continental ice sheets, volcanic forcing and astronomical boundary conditions are held constant to facilitate the experimental setting and analyses (e.g., to prevent confounding feedback effects) (Keller et al., 2014). Parameterized geostrophic wind anomalies, which are a first-order approximation of dynamical feedbacks associated with changing winds in a changing climate (Weaver et al., 2001), are also applied.
Simulated
To track the physical transport of the injected
In all of our injection simulations we subtract the amount of injected
Following previous studies (e.g., Jain and Cao, 2005; Ridgwell et al., 2011), additional simulations are conducted to investigate how climate-change-induced feedbacks affect the fate of injected
Absolute changes in oceanic and land carbon between I-3000 and the RCP 8.5 control run (I-3000 simulation minus RCP 8.5 control run) at the end of the injection period (year 2120). The black rectangles represent the locations of the seven injection sites, where the injections occurred in the center of the black rectangles.
To determine how long the injected carbon stays in the ocean, we follow the
IPCC (2005) and calculate a fraction retained (
To assess the global carbon cycle response to the injections, we use another
metric, the net fraction stored (
To investigate if the targeted atmospheric carbon reductions in the
With Emissions simulations differ from what would happen if
As mentioned in the introduction, this modeling study of direct
Overview of all conducted simulations and their anthropogenic forcing. “X” denotes that the respective forcing is applied. WE: With Emissions; CM: Complete Mitigation.
Hereafter, the perturbed control runs are referred to as RCP 8.5
control
An overview of all conducted simulations with their anthropogenic forcing is shown in Table 1.
The physical climate and biogeochemical cycles of the Earth system during the
RCP 8.5 control simulation are in the same state as described in Keller et
al. (2014). Here, we briefly describe global carbon cycling during the
control simulation so that comparisons can be made to the With Emissions simulations (Sect. 3.4). Subsequently, we briefly outline the
global carbon cycling of the perturbed control runs RCP 8.5
control
By the end of the simulation in the year 3020, about 6000
By the end of the extended RCP 8.5 control run about 58 % of the emitted
Globally integrated carbon of the RCP 8.5 control run, the RCP 8.5
control
Simulated terrestrial carbon uptake is initially high as well, but then
declines rapidly, with the terrestrial reservoir becoming a source for
atmospheric carbon in the year 2139 before leveling off at very little net
exchange between the terrestrial reservoir and the atmosphere after about
the year 2280 (Fig. 2g). The initial increase in total land carbon uptake is due
to the simulated
As expected, simulated terrestrial carbon uptake is higher in the RCP 8.5
control
Accordingly, the atmospheric carbon concentration in the RCP 8.5
control
Global carbon cycling in the RCP 8.5
control
Here, we compare the With Emissions simulations to the RCP 8.5
control run to assess injection-related seawater chemistry changes. By the
final year of the injection period (year 2119), a total of 10
Simulated ocean surface
Here, we assess to which extent the simulated
By comparing the With Emissions and Complete Mitigation simulations at all depths, we can determine how climate change affects FR. As in previous studies, our results show that FR is enhanced by climate change (Jain and Cao, 2005; Ridgwell et al., 2011). In the With Emissions simulations, values of FR are always higher than in the Complete Mitigation runs (Table 2). For I-800 and I-1500, the FR increase due to climate change is largest in the Pacific, whereas for I-3000, Atlantic sites show the highest FR increase due to a larger ocean response to climate change (Table 2). However, in all simulations more of the injected carbon is retained in the Pacific compared to injections in other ocean basins.
We also assess whether the enhanced FRs in our With Emissions
simulations are affected by changes in the Atlantic Meridional Overturning
Circulation (AMOC). Relative to the preindustrial period, which has a maximum AMOC
intensity of 15.98
Comparison of fractions retained (FR) between Orr et al. (2001), Orr (2004) (full range of their global efficiency, which is the same as the FR defined in Sect. 2.2 and is based on seven ocean general circulation models (OGCMs) and one zonally averaged model result) and our Complete Mitigation (CM) and With Emissions (WE) simulations for all injection sites (Global) and on an inter-basin level for the Atlantic sites (Bay of Biscay, New York, Rio de Janeiro), the Pacific sites (San Francisco, Tokyo) and the Indian sites (Jakarta, Mumbai). The FR values (%) are given for the last year of the injections (2119), 500 years after the simulations started (2519) and for the last year of the simulations (3019). For each entry of the table, numbers to the left of the vertical bars denote results of the CM runs, numbers to the right results of the WE runs. Note that the illustrated years refer to our simulations, ranging from the year 2020 until the year 3020. The GOSAC–OCMIP simulations started in the year 2000 and ended in the year 2500 (Orr et al., 2001).
Model-predicted FR (Table 2) refers to the injected
These results show the importance of accounting for carbon cycle feedbacks
when assessing the effectiveness of marine
Here we first briefly show how the atmospheric carbon reduction, relative to
the RCP8.5 control run (see Sect. 3.1), differs between With Emissions simulations and the Direct Air Capture run. Subsequently,
we investigate how carbon cycle and climate feedbacks affect the distribution
of carbon between different reservoirs upon injection of
In the With Emissions simulations and the Direct Air Capture run, the “globally injected carbon” denotes the targeted
atmospheric carbon reduction. The globally injected carbon – in the
absence of leakage and backfluxes – equals the oceanic carbon addition or
atmospheric
Absolute change in atmospheric carbon in the Direct Air Capture run (DAC) and in the With Emissions simulations, relative to the RCP 8.5 control run. The black dashed line denotes the globally injected carbon (GIC), which is subtracted from the emission forcing (see Sect. 2.2).
This is explained by injected carbon leaking from the ocean back to the
atmosphere and the response of atmosphere-to-land and atmosphere-to-ocean
fluxes to the reduction in atmospheric carbon. The rapid divergence even for
the deepest injection points where FR is high, points to carbon cycle and
climate feedbacks, which are directly related to changes in atmospheric
While ocean feedbacks in response to
By the last year of the injection period (year 2119), I-800 shows the highest
divergence from globally injected carbon (Fig. 4c) with an
atmospheric carbon reduction of only 48
Roughly similar patterns are found for injection simulations I-1500 and
I-3000 during the injection period, although with less outgassing occurring
for the deeper injections (Fig. 4c), which led to a slightly larger reduction
in terrestrial carbon uptake by the last year of the injection. Thus, the
largest reduction in total atmospheric carbon with 60
Our results suggest that the terrestrial response due to the atmospheric
carbon reduction is mainly governed by the reduced
Feedbacks from the terrestrial system to atmospheric
The neglected effect of the
After the injections are stopped (end of year 2119), I-800 shows a continuous
outgassing of about 40
Absolute changes between the WE simulations and the RCP 8.5 control
run for
Unlike I-800, I-3000 actually gets closer to the globally injected
carbon trajectory after the end of the injection period until the year 2199,
with about 64
For I-1500, an unexpected oceanic carbon uptake event is observed from the
last year of the injection period (Fig. 4c, d). This is caused by a large
temporary carbon flux from the atmosphere into the ocean (Fig. 4d), with a
total of
Recurring open-ocean deep convection in the Southern Ocean has been found in many CMIP5 models (De Lavergne et al., 2014) and also in the Kiel Climate Model, for which the driving mechanism could be linked to internal climate variability (Martin et al., 2013). Although the modeled deep convection events feature similarities to processes associated with the Weddell Polyna of the 1970s (Martin et al., 2013), uncertainty remains regarding their realism. An important model constraint in this respect is a coarse grid resolution, which hinders, for instance, the correct representation of bottom water formation processes on the continental shelf and instead might favor open-ocean deep convection (Bernardello et al., 2014).
It is intriguing that among 19 millennial-scale simulations performed
for this study, a deep convection event occurred only in 3 simulations: the I-1500, an injection run with a 10-year injection period (not shown) and
the Direct Air Capture run. Apparently, small internal variability
combined with certain
Here we show how varying the
As illustrated by the error bars in Fig. 6c, varying the
Absolute changes in terrestrial land carbon uptake and total land carbon show
the largest sensitivities to the scaled
The magnitude of the responses that can be seen in the perturbed injection
runs I-3000
Although the above response is informative, the future strength of the
We use an Earth system model of intermediate complexity to
simulate direct
The response of the carbon cycle during and after the injections is dominated
by the partial outgassing of injected
Further, we find that varying the
Furthermore, the influence of the highly uncertain carbon-cycle and climate
feedbacks in our findings, in addition to the sporadic deep convection event
in I-1500, illustrates the difficulty of quantitatively detecting,
attributing and eventually accounting for carbon storage and carbon fluxes
generated by individual carbon sequestration measures even in relatively
coarse-resolution models with little internal climate variability
(“noise”). Nevertheless, our findings point to the importance of accounting
for all carbon fluxes in the carbon cycle and not only for those of the
manipulated reservoir to obtain a comprehensive assessment of direct oceanic
The model data used to generate the table and the figures is available online
at
The Deutsche Forschungsgemeinschaft (DFG) supported financially this study via the Priority Program 1689. We thank Torge Martin, Wolfgang Koeve, Nadine Mengis, Julia Getzlaff, Levin Nickelsen, Peter Vandromme, Markus Pahlow, Wilfried Rickels and Ell Yuming Feng for their thoughtful discussions and advice. Edited by: M. Heimann Reviewed by: C. Heinze and one anonymous referee