• Open Access

Can marine cloud brightening reduce coral bleaching?



Increases in coral bleaching events over the last few decades have been largely caused by rising sea surface temperatures (SST), and continued warming is expected to cause even greater increases through this century. We use a Global Climate Model to examine the potential of marine cloud brightening (MCB) to cool oceanic surface waters in three coral reef provinces. Our simulations indicate that under doubled CO2 conditions, the substantial increases in coral bleaching conditions from current values in three reef regions (Caribbean, French Polynesia, and the Great Barrier Reef) were eliminated when MCB was applied, which reduced the SSTs at these sites roughly to their original values.

1. Introduction

The continued increase in CO2 emissions into the atmosphere has generated greater interest in engineering strategies to ameliorate the effects of climate change (Shepherd et al., 2009; SRMGI, 2011). Some of these involve solar radiation management (SRM), i.e. ‘shading’ the planet from incoming sunlight and thus slowing the rate of warming at the Earth's surface. One such SRM technique is marine cloud brightening (MCB): (Latham, 1990, 2002; Bower et al., 2006; Latham et al., 2008; Jones et al., 2009, 2011; Rasch et al., 2009; Korhonen et al., 2010; Bala et al., 2011; Latham et al., 2012a, 2012b). MCB geoengineering is designed to produce a cooling that in principle can maintain the Earth's average surface temperature and polar sea–ice cover at roughly current values in the face of increasing atmospheric CO2 concentrations, at least up to the 2×CO2 point. MCB involves seeding low-level marine stratocumulus (ice-free) clouds with submicrometer-sized seawater droplets. The particles have sufficiently high-salt mass to act as cloud condensation nuclei (CCN), thereby increasing the cloud droplet number concentration (CDNC) and the cloud optical thickness. The resulting effect is an increase in cloud albedo for incoming shortwave radiation (Twomey, 1977). Also, smaller cloud droplets coagulate more slowly, thus suppressing precipitation development and increasing cloud lifetime (Albrecht, 1989). The engineering design for MCB includes deployment of ocean-based vessels strategically placed to release seawater CCN into the turbulent boundary layer beneath marine stratocumulus clouds.

MCB was originally developed as a means of counteracting warming on a global scale. However, MCB in principle could also be used to target subglobal regions of particular interest. That is, seeding sites could be selected:

  • to maximize a desired effect (e.g. reduction in warming);
  • to minimize adverse effects (e.g. reduction in precipitation).

Some previous modeling studies have illustrated a cooling effect by seeding the three regions of most extensive marine stratocumulus clouds: off the west coasts of Africa, North America, and South America (Jones et al., 2009, Latham et al., 2012a; 2012b, 2011). Seeding these three regions, e.g. produced a significant reduction in sea surface temperatures (SST) where hurricanes traditionally develop in the Atlantic, raising the possibility of weakening them (Latham et al., 2012b). Mostly, the three seeding regions covered less than 10% of the marine stratocumulus regions but some larger regions are considered in some of the articles.

The influence of SRM on marine ecosystems has recently been addressed by Russell et al. (2012). In this article, we evaluate computationally the potential for MCB seeding in these same three regions to significantly reduce tropical SSTs in three major coral reef provinces—the Caribbean, French Polynesia, and the Great Barrier Reef—and thereby reduce the rates of coral bleaching.

2. Methods

2.1. Simulations of SST with and without MCB

SST values were extracted from model calculations conducted with the Hadley Centre Global Environment Model (HadGEM1), version 6.1 of the UK Meteorological Office Unified Model. Further details of this model are provided in Latham et al. (2012a, 2012b).

HadGEM1 was modified to have a fixed CDNC in the three aforementioned regions of low-level (about 1 km high) marine stratocumulus clouds. The normal (natural) value of CDNC in HadGEM1 is about 60 cm−3. For the MCB simulations, the CDNC was given a value of 375 cm−3 at all model levels between 0 and 3 km, which is consistent with the treatment used in Jones et al. (2009, 2011), Latham et al. (2008, 2012a, 2012b) and Parkes et al. (2012).

Three simulations were completed to determine the effects of MCB on SST values. They were (1) a Control simulation forced by an atmospheric CO2 concentration of 440 ppm, (2) a 2×CO2 simulation forced by an atmospheric CO2 concentration of 560 ppm, and (3) a 2×CO2 MCB simulation that includes MCB in the three regions of marine stratocumulus. For the 2×CO2 simulations, the model was run from 2020 to ˜2045 with increasing CO2 until 2045, and then held stable at double CO2 values until 2090. For each 70-year simulation, the 10-day averaged SSTs of the final 20 years (2070–2090) were used for the coral bleaching analysis.

2.2. Coral bleaching calculations

For each simulation, the SSTs were used to calculate coral heat stress and bleaching. Heat stress in corals occurs once the temperature exceeds some predetermined threshold above the climatological maximum. This threshold is traditionally set at 1 °C (Liu et al., 2003), or alternatively, is allowed to vary regionally as a function of the natural variability of the year-to-year maxima. In the latter case, the threshold is typically determined as n × SDmax, where n is an empirically determined factor, and SDmax is the standard deviation of the annual maxima over the climatological period. We apply a threshold of 2.45 × SDmax following Donner (2009).

Coral bleaching occurs when heat stress accumulates over a period of time. This is typically calculated as degree heating weeks (DHW), which is the accumulation of the heat stress over a 12-week period (Liu et al., 2003). When weekly data are used, mild bleaching occurs once the DHW > 4, and severe bleaching occurs once DHW > 8.

SSTs of the final 20-year Control case were used to calculate the climatological SST maximum for every ocean cell within each region, including those that do not include reefs. In both the Caribbean and the Great Barrier Reef regions, many of the coastal grid cells were not resolved in the model; here, we assume that changes in offshore temperatures are largely representative of those affecting adjacent reefs. The climatological values derived from the Control run were used in the DHW calculations for each ocean cell for the three regions, for both the 2×CO2 and 2×CO2 + MCB cases. To adjust for the 10-day average SSTs of the model output (vs weekly averages), a bleaching event was designated as mild when DHW = 3–5, and severe when DHW ≥ 6. These values are slightly lower than the 7-day DHW of 4 and 8, respectively. Bleaching events were calculated separately for the 1 °C and 2.45 × SDmax heat-stress thresholds (Figure 1).

Figure 1.

(a) Regional average sea surface temperature (SST) for French Polynesia, comparing SST for the three cases. (b) SST for a single cell within the French Polynesia region, showing the climatological maximum (thick black line), and the 1 °C (thin black line) and 2.45 × SDmax (thin red line) thresholds above the climatological maximum. (c) The degree heating weeks (DHW) based on the 1 °C (thin black line) and 2.45 × SDmax (thin red line) thresholds.

3. Results

3.1. Global SST

The changes in SST between the Control and 2×CO2 simulations (Figure 2(a)) are consistent with patterns produced in other coupled GCM studies mentioned earlier. Temperature increases between 0.5 and 2 K occurred over much of the tropical area (Figure 2(a)). The effects of MCB on SSTs in the 2×CO2 simulations more than cancel these temperature increases (Figure 2(b)). In case of three-patch MCB in the 2×CO2 atmosphere, the global temperatures decrease by 0.12 K. Full-area seeding at 2×CO2 leads to much greater cooling over the entire climate system and reduces tropical temperatures by more than 5 K, which is too large to be beneficial and could cause cold-temperature stress in corals.

Figure 2.

(a) Change in annual average sea surface temperature (°C) between the 2×CO2 and control simulations. (b) Change in annual average sea surface temperature (°C) between the control and 2×CO2 + MCB simulations. The dashed black boxes in both panels represent the three coral reef regions.

3.2. Temperatures in coral reef regions

In all three coral reef regions, the doubling of atmospheric CO2 raised the annual average temperatures by at least 0.5 K relative to the Control scenario, whereas MCB restored SSTs to near the Control values (Table 1). The greater cooling in French Polynesia (Figure 2(b)) illustrates the cooling effect of MCB in waters off Peru that are then transported by ocean currents westward across the Pacific.

Table 1. Changes in sea surface temperatures between the 2×CO2 experiments and the Control, as described in Section 'Global SST' (all values in °C). The three regions, Caribbean, Great Barrier Reef, and Polynesia are shown in black rectangles in Figure 2
Region2×CO2 − Control2×CO2 + MCB − Control
  1. 2×CO2 + MCB all clouds – control: global temperature change = −6.5 °C
Great Barrier Reef0.550.13

Coral bleaching events were rare based on SSTs from the Control case (Table 2); only French Polynesia exhibited bleaching for either heat-stress threshold. The number of bleaching events in the 2×CO2 case was dramatically higher for all three regions. For the 2×CO2 + MCB case, however, bleaching events were almost entirely eliminated in all three regions.

Table 2. The number of mild and severe bleaching events over a 20-year period for the three simulations: Control, 2×CO2, and 2×CO2+MCB. Bleaching events are calculated based on two heat-stress thresholds: 1 °C above the climatological maximum, and 2.45 × SDmax above the climatological maximum. Climatological maxima were calculated based on the Control. Note that the Control provides the background number of bleaching events expected under normal conditions. Numbers in parentheses are the number of ocean cells within each reef region
RegionControl2×CO22×CO2 + MCB
Caribbean (439)
1 °C0029410400
2.45 × SDmax0065517730
Great Barrier Reef (179)
1 °C003237100
2.45 × SDmax003113600
French Polynesia (899)
1 °C95210561884320
2.45 × SDmax30850150220

4. Discussion

The possible utilization of MCB for examining the subglobal scale topic of coral bleaching amelioration possesses a number of positive and negative attributes. On the positive side, our model-based analysis indicates that MCB seeding in key patches of marine stratocumulus would not only lower temperatures over all three major reef regions studied, but could also restore temperatures to the Control levels. Coral reef ecosystems may thus be beneficiaries of MCB geoengineering. Other subglobal locations for which calculations predict that cooling due to MCB can compensate for the warming produced by fossil fuel burning are the Arctic (Parkes et al., 2012) and the hurricane-generating waters of the tropical Atlantic (Latham et al., 2012b). MCB has additional advantages such as: (1) it does not require a long spin-up time (about 1 year) to achieve a cooling effect, (2) once halted, the effects of MCB on clouds are easily and quickly reversed [the life-time of salt particles in the atmosphere is about 10 days (Salter et al., 2008)], and (3) the utilization of mobile MCB spray vessels could provide a high degree of control, given the relative ease of adjusting their positioning, as well as the flux of sea salt spray particles that act as CCN. The regions where MCB would be most effective are those where marine stratocumulus clouds are frequently present or where surface currents deliver waters that have been cooled by MCB. The flexibility in the location of MCB seeding may permit fine tuning to maximize its effectiveness and to minimize side effects.

Latham et al. (2012a) present a detailed three-stage plan for field-testing MCB, should this be authorized, on a spatial scale of about 100 × 100 km, which seems very likely to be too small to produce any significant climate effects. It would be based on—but on a much smaller scale than—the successful VOCALS field experiment (Wood et al., 2011) which involved exhaustive studies of marine stratocumulus clouds. A field study of the impact of MCB on coral bleaching at a selected site would utilize some but by no means all of the equipment and procedures required for the three-stage field experiment.

However, there are a number of possible disadvantages and limitations of MCB seeding designed to reduce coral bleaching: (1) use of MCB may reduce the photosynthetically available radiation (PAR) reaching the surface, which would affect both the depth of light penetration (which affects the depth limits of corals and seagrass beds), as well as primary production in the water column, (2) Although blocking shortwave radiation leads to a reduction in ocean temperatures in these reef regions, it does not reduce the process of ocean acidification, which is a direct consequence of rising atmospheric CO2, and is detrimental to coral growth and reef development, and (3) as with any SRM geoengineering technique, the MCB process would have to be continued indefinitely. Cessation of MCB, particularly if atmospheric CO2 concentrations continue to rise, could result in very rapid warming, which for coral reefs is the most common condition leading to coral bleaching. The estimated annual cost of deploying MCB to inhibit coral bleaching around the major reefs is about $40M. However, social, ethical, and political costs related to compensation are extremely hard to estimate, and may dominate the costs.

As mentioned earlier, a primary requirement of SRM research is to fully examine all possible adverse consequences of deployment, and to abandon this work if significant ones cannot be remedied. Previous modeling analyses of the potential adverse effects of MCB have largely been restricted to global applications of MCB and mostly to the possibility that rainfall would be reduced in particular regions. Jones et al. (2009), e.g. found that MCB reduced rainfall significantly in the Amazonian region, but later (Jones et al. 2011) found that altering the locations of seeding largely eliminated this problem. Bala et al. (2011) also found that MCB seeding would cause a substantial reduction in rainfall, but that virtually all of this loss was over the oceans, with no net loss over land. Whether or not appreciable rainfall reduction occurs specifically over Amazonia appears to depend on the choice of seeding site(s). Thus, there exists a useful element of flexibility regarding the unforeseen consequences issue. Latham et al. (2012a, 2012b) provide a more detailed account of research into this issue.

Another major consideration relates to governing the research associated with geoengineering, including MCB. This is not a trivial consideration, particularly for the large-scale deployments necessary to achieve the experimental ends suggested by the modeling in this article. A governance framework—the processes, mechanisms, institutions, and individuals guiding ordered rule and collective action (Folke et al., 2005)—will be necessary to guide and inform MCB research and will almost certainly need to be developed in advance of field testing (Shepherd et al., 2009; Keith et al., 2010). Establishing such a framework will assist scientists, decision makers, and the public with evaluating and managing both known and as-yet-unknown risks and benefits of MCB.

Among the governance questions to consider are: How can ongoing, independently run assessments of geoengineering research and its outcomes be achieved? How may newly acquired knowledge be best incorporated into existing governance and research structures? Also, means of allowing public scrutiny and input, and research and decision transparency will be important to incorporate when developing SRM or MCB governance. Issues of liability and equity with regard to MCB decision-making and application will need to address how to attend to the needs of the voiceless and those left worse off from MCB use. In essence, effective governance will frame ways in which MCB and other geoengineering efforts are best regulated as a ‘public good’ (Rayner et al., 2009). Governance will benefit from integrating existing research protocols and lessons learned with geoengineering-specific governance guidelines and ideas such as those generated by the Oxford Principles (Rayner et al., 2009) and groups like the Solar Radiation Management Initiative (SRMGI, 2011).

We conclude that MCB seeding would likely lower tropical SSTs and thus lower the risk of coral reef bleaching for several decades. However, various technological and governance hurdles remain before a fully operational MCB system can be put in place. As with any SRM geoengineering technique, much more work is also required before we can fully gauge whether MCB will cause adverse consequences that cannot be remedied—in which case it should not be utilized. Given the current rate of CO2 increase in the atmosphere, however, MCB should continue to be evaluated as a measure to prevent particularly dangerous aspects of climate change.


We are grateful for the use of NERC, NCAS, HECToR supercomputer resources. Support for elements of this research was provided by the Fund for Innovative Climate and Energy Research, FICER, at the University of Calgary. This does not constitute endorsement of deployment in any form of cloud albedo modification by the funding agency. Part of this work was undertaken while working on the End-to-end quantification of uncertainty for impacts prediction project (EQUIP)funded by NERC (Grant number NE/H003525/1). NCAR is sponsored by the National Science Foundation.