Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Corresponding author: J. K. Eccles, Division of Earth and Ocean Sciences, Nicholas School of the Environment, Duke University, Box 90227, Durham, NC 27708, USA. (email@example.com)
 One potential way to geologically sequester captured CO2 emissions is to inject them below the seafloor into marine sedimentary strata where pressures and temperatures would trap the CO2through “self-sealing” gravitational and hydrate-formation mechanisms. Here we map out the worldwide distribution and thicknesses of such self-sealing strata using a comprehensive, global dataset of deep-sea sediment cores in combination with digital grids of ocean floor heat flow, bathymetry, and sediment thickness. Based on our mapping, we estimate that the total bulk sediment volume of self-sealing strata is 63 million cubic kilometers, 0.8–1.4 km3 (or ∼1.3–2.7%) of which are sands with intrinsic permeability suitable for storing CO2. This is enough storage capacity to hold between 1,260–28,500 gigatonnes of CO2, or about 40–1,000 y of total global CO2 emissions. However, the storage capacity is unevenly distributed where it lies within the Exclusive Economic Zones (EEZ) of the world's largest CO2 emitting economies. The United States and India respectively release 16% and 62% of their annual CO2emissions (or 1 Bt/y and 800 Mt/y) within 500 km of self-sealing sands located in their EEZs, while only 6% of the annual emissions from China and the European Union (or 330 Mt/y and 250 Mt/y, respectively) occur within this distance.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Climate change caused by anthropogenic greenhouse gas (GHG) emissions could be mitigated with carbon capture and storage (CCS) [Intergovernmental Panel on Climate Change (IPCC), 2007], in which CO2 emissions from large industrial point sources, such as power plants, would be captured and stored securely in reservoirs including potential geologic reservoirs below the seafloor, the latter being one of the only forms of ocean storage permitted by the London Convention/London Protocol [Koide et al., 1997; IPCC, 2007]. House et al. proposed a particularly interesting form of sub-seafloor storage, which we call “self-sealing” storage, whereby CO2 is injected into marine sediments and is trapped under particular pressure and temperature (P/T) conditions [House et al., 2006]. The P/T conditions must be such that at equilibrium, CO2 in the trapping zone is denser than the surrounding pore fluid (a.k.a. the neutral buoyancy zone or NBZ) and will freeze into CO2 hydrates (a.k.a. the hydrate formation zone or HFZ), gravitationally trapping CO2 as well as infilling the sediment pore spaces and making the strata impermeable (Figure 1).
Levine et al. subsequently examined this idea and noted that conditions in deep-marine pelagic carbonates and clay-rich sediments often lack intrinsic permeability that would allow for sufficiently rapid injection of CO2 without unintended hydraulic fracturing. In addition, they caution that example P/T regimes in the ocean may cause the NBZ may vary substantially in thickness. Permeable marine sands, however, have successfully been used for CO2 storage in commercial projects (e.g., Sleipner, Snovhit, and PureGen) and continue to pursued in ocean settings [Eiken et al., 2011].
 We build on the work of House et al.  and Levine et al. by evaluating the storage potential of permeable marine sand layers in deep-ocean strata with P/T conditions for self-sealing trapping. Our approach is conceptually similar to that developed by Goldberg and others [Goldberg et al., 2008; Goldberg and Slagle, 2009; Slagle and Goldberg, 2011] who mapped out the storage potential for sequestering CO2in deep-ocean basalt formations, where injected CO2 would be gravitationally trapped in permeable ocean crust and sealed off by impermeable sediments above.
 We begin our analysis by expanding on the initial mapping by House et al. to create a worldwide map of self-sealing sedimentary strata with potential for CO2 injection and storage. We do this by mapping out the extent of the NBZ for CO2 in the world's oceans using global ocean datasets for bathymetry, and for pressure and temperature at and below the seafloor [Amante and Eakins, 2009; Boyer et al., 2009]. We then use additional global datasets for sediment thickness and geothermal gradient to map out the distribution and thicknesses of marine strata in which CO2 could be stored as a supercritical fluid below the HFZ [Pollack et al., 1993; Divins, 2003; Hasterok and Chapman, 2008]. The overlap between these two maps yields our map of self-sealing marine sedimentary strata, allowing us to identify the prevalence of permeable marine sands using lithology, salinity, and porosity analyses of cores from over 400 deep-sea sediment boreholes located within self-sealing regions from around the world [World Data Center for Marine Geology and Geophysics, 2000a, 2000b]. We then constrain what fraction of these strata could be sand layers suitable for rapid injection and storage of CO2 and conclude our resource evaluation by examining the access to this storage within the Exclusive Economic Zones of four of the largest CO2-emitting economies: the U.S., China, India and the E.U.
 We map the NBZ for CO2 based on the pressure and temperature at and below the seafloor. Seafloor depths worldwide are derived from global bathymetry data gridded at a resolution of 10′ of latitude by 10′ of longitude [Amante and Eakins, 2009]. Seafloor pressure at each point in the grid is calculated assuming hydrostatic conditions and a constant ocean water density of 1030 kg/m3. This is a conservative value for density over the range of depths we investigate [Boyer et al., 2009] but one that safeguards against overestimating storage potential. The seafloor temperature corresponding to each point in the bathymetry grid is determined by box-averaging the seafloor temperatures in the World Ocean Dataset [Boyer et al., 2009]. The resulting grids of seafloor pressure and temperature are then used in the equations of state for CO2 [Span and Wagner, 1996] to generate a corresponding grid of CO2 density at the seafloor. This grid is filtered for CO2 densities greater than 1040 kg/m3 to produce a map of where CO2 would be denser than the surrounding seawater and potentially form a hydrate, sinking to the seafloor (Figure 2a).
 We map where CO2 could be stored in marine sedimentary strata using the grids of global bathymetry [Amante and Eakins, 2009], global marine sediment thickness [Divins, 2003], and global heat flow [Pollack et al., 1993; Hasterok and Chapman, 2008] in a fashion similar to how Goldberg and Slagle used grids of global bathymetry and sediment thickness to screen for favorable storage sites in crustal basalts. The bathymetry and global heat flow grids are used here estimate the sub-seafloor depth at which pressure exceeds 7.1 MPa and temperature is >31.1°C. This is the critical pressure-temperature point beyond which CO2 becomes a supercritical fluid and would not form a hydrate [Span and Wagner, 1996], i.e., the sub-seafloor depth of this point occurs below the HFZ and NBZ. We calculate the formation fluid pressure assuming hydrostatic conditions and a constant density of 1030 kg/m3 for the saline pore waters. The temperature calculations are based on interpolated geothermal gradients from the Global Heat Flow Database [Pollack et al., 1993; Hasterok and Chapman, 2008]. We used ∼13,000 marine data points to create this interpolation, which was done at a horizontal resolution matching that of the sediment data (5′ × 5′). The average gradient for these data points over the entire ocean is 76.5°C/km. However, the gradient varies substantially with tectonic environment, so much so that use of the constant average geothermal gradient would have overestimated storage on passive continental margins and underestimated storage on relatively thin deep ocean crust.
 Following the conservative approach of House et al. , we take the deeper of the temperature and pressure thresholds as the depth at which the P/T regime would allow CO2 at equilibrium to be injected as a supercritical fluid. We then use the gridded data of world ocean sediment thickness to determine how much sediment, if any lies, below this depth (Figure 2b). Finally, we combine our map of the NBZ for CO2with our map of sediment thickness below the HFZ to produce a third map of strata thickness that meets both criteria. These are the self-sealing strata (Figure 3) that target a specific set of P/T conditions in marine sediments in regions beneath the NBZ and HFZ.
 It is important to note that our initial mapping of self-sealing strata does not discern whether the sediments are permeable enough to allow for rapid injection of large volumes of supercritical CO2. Most deep-sea pelagic sediments are fine-grained oozes, clays and muds with low permeability [e.g.,Levine et al., 2007]. However, permeable coarser-grained terrigenous sand beds are also found in deep-sea sedimentary strata in the form of turbidites [Bouma, 1964] and even sandy debris flows [Shanmugam, 1996], deposits laid down by gravity-driven sediment flows that carried the sands beyond the continental shelf and into the deep sea (see Discussion) [Shepard, 1981]. These sand beds are important offshore oil and gas reservoirs [Brainard and Martinez-Diaz, 2009], and are proving effective at storing CO2 at offshore injection pilot projects at Sleipner West and Snovhit [Eiken et al., 2011]. Permeable sand layers in self-sealing strata could be another type of reservoir for CO2 storage with large potential capacity.
 We use core data from the international Deep-Sea and Ocean Drilling Projects [World Data Center for Marine Geology and Geophysics, 2000a, 2000b] to estimate the average fraction of self-sealing strata that are sand layers with sufficient permeability and porosity to handle large-scale injection and storage of CO2 [Michael et al., 2009; Department of Energy (DOE), 2010]. This fraction, or net-to-gross ratio, represents the same type of factor used by the DOE and the USGS to estimate likely storage volume from bulk sediment volume in terrestrial strata [Burruss et al., 2009; DOE, 2010]. We calculate net-to-gross ratio as the summed thickness of all sand layers identified in cores divided by the summed thickness of all the cores.
 The cores are from 407 DSDP/ODP deep-sea sediment boreholes at 288 drilling sites located inside the self-sealing strata regions. Collectively the cores comprise a column of sediment >20 km long. We estimate the fraction of this total length that is sand primarily from the visual descriptions of the cores, using only those layers identified as “sand” and excluding layers with marginal descriptive terms such as “muddy sand”. Note that although the cores constitute relatively sparse samples of the boreholes from which they were extracted, all of the cores have been carefully measured and their lithology fully described (i.e., there is no disparity among cores in terms of sampling density). We also use two secondary measurements of the cores to check the validity of our analysis, these being bulk mineralogy data available for 186 (∼46%) of the cores and grain size data available for 206 (∼51%) of the cores. Sand layers are identified in these secondary data where the majority of sediments in the grain size data are sand-sized and the quartz flag in bulk mineralogy data indicates the presence of passive continental margin turbidites [McLennan et al., 1990]. The indicators and the size of the datasets are reported in Table 1; DSDP and ODP cores are separated because of different reporting methodology in the datasets. IODP data is not used in our analysis because visual descriptions are not available in machine-readable format at this time.
Table 1. Analysis of Net-to-Gross Ratio for Sedimentary Strata in Self-Sealing Regions
Visual Description (DSDP)
Sediment Lithology (ODP)
Number of Samples
Samples with >40% quartz
Samples with >75% sand
 We note that while we have analyzed cores from a large number of drilling sites, the spatial distribution of the boreholes is not dense relative to the vast expanse of the ocean floor. Additionally, while the boreholes do generally penetrate to depths below the HFZ/NBZ, they rarely if ever extend through the entire sediment column. As a result, our best estimate for net sand thickness is arrived at by multiplying our net-to-gross ratios by the sediment thickness at each borehole location to arrive at a total net sand thickness and thus CO2 storage potential at the site.
 Our estimates of net-to-gross ratio are conservative, but they remain probabilistic assessments of the likely sand volume within the self-sealing regions, assessments that are biased toward <1 km of sediments. Further research may reveal additional lateral and vertical variation in net-to-gross ratios beyond what we find, however, our aggregate estimates should be accurate if the cores are representative of likely storage volumes. Note that while there is considerable uncertainty in our estimates, we corroborate them using the more limited bulk mineralogy and grain size data available for a subset of the cores (see Discussion).
 We also examine P/T regimes in the deep ocean where pore waters in the sub-seafloor strata may remain denser than CO2, limiting the gravitational-trapping mechanism in the self-sealing trapping process [Levine et al., 2007]. To evaluate how widespread this scenario might be, we also examine the interstitial pore-water data from the DSDP/ODP cores to screen out any potential gravitational inversion zones where the NBZ might be compromised by extremely dense pore waters.
 Finally, we compute the average porosity from the ocean cores where it was available so our bulk sediment volume numbers can be meaningfully translated into CO2 storage volumes.
 The global map of the NBZ for CO2 (Figure 2a) constitutes a collective area of >282 M km2 or about 78% of the seafloor. For the most part, the zone lies below the 3000 m depth contour as House et al and others [House et al., 2006; Goldberg and Slagle, 2009] concluded. Figure 2b is the map of the thickness of marine sediments below the HFZ where CO2 could be stored as a supercritical fluid. Globally, 129 M km3 of ocean strata meet these conditions. Not surprisingly, the majority of such strata are found along passive continental margins (Figure 2b) where sediment accumulations are thickest. However, half of the strata lie above the NBZ for CO2 (Figure 2a). This shallower fraction is similar to onshore strata in that CO2 could only be housed safely where the strata possess geologic traps sealed by cap rocks, such as at the Sleipner West and Snøvhit injection projects [Michael et al., 2009].
 The overlap between the maps in Figures 2a and 2bis where self-sealing strata occur. Their distribution, which is extensive, is shown inFigure 3 along with the thickness of the HFZ and NBZ, which forms the trapping seal in these strata. The total bulk volume storage of these strata is 63 M km3. Again, because in theory the self-sealing strata do not require additional geologic traps, a good portion of this potential capacity might be utilized. However, the net-to-gross efficiency factor is needed to constrain the fraction of these sediments that could in fact have suitable permeability for CO2 injection and storage.
Table 1lists the net-to-gross ratios for the self-sealing regions that we estimate from the DSDP/ODP core data. The more abundant visual core descriptions yield ratios of 1.3% (DSDP) to 2.8% (ODP) (Table 1). These are bounded by the ratios we derive from the lesser number of bulk mineralogy and grain size measurements, which are 1.2% and 5.5%, respectively (Table 1). Since the measurement-based ratios are in line with the description-based ratios but potentially less accurate, we use the latter as the range for determining storage potential in the self-sealing strata. Multiplying the net-to-gross ratio of 1.3–2.8% by the sediment thicknesses at the borehole sites where the cores were recovered, we arrive at total sand thicknesses that can vary from <1 m to >100 m (SeeFigure 3). Note that these thicknesses are not for single sand layers, but rather estimated sums of the thicknesses of what would likely be multiple sand layers in a given core. When the fact that sand turbidites can extend up to 500 km laterally [Pilkey et al., 1980] is factored in, it becomes clear that the sand layers in the self-sealing regions could have substantial storage capacity.
 Finally, the numerous interstitial water analyses of the cores indicate that sub-seafloor regions of high salinity and thus high porewater density in self-sealing strata are likely to be rare. In fact, there are just six boreholes at three drilling sites with any measurement of density >1040 kg/m3in the top 200 m of sediment, and even at these sites the average density over these sub-seafloor depths remains <1040 kg/m3. Nonetheless, using all ∼200 boreholes for which interstitial water analyses were available, we estimate that there is a 3.3% chance that a given volume element of self-sealing strata would have the NBZ trapping component of the self-sealing mechanisms compromised. This leads us to slightly reduce (by 3.3%) our net-to-gross ratio to 1.3–2.7% in self-sealing regions. This final net-to-gross ratio thus represents the fraction of bulk sediment volume likely to be sand with suitable interstitial pore fluid density in the overlying sediment.
 Using these net-to-gross ratios, the in-situ-density of CO2, calculated average porosity, and a sweep efficiency (the fraction of pore volume that will be filled by CO2) of 0.5–5.5%, we estimate that self-sealing regions could have a volumetric storage capacity of between 1,260–28,500 gigatonnes of CO2. This is equivalent to 40–1,000 years of total global CO2 emissions.
 Access to the self-sealing strata for different nations, however, is not uniform. We have used the IEA database of global CO2 point sources [International Energy Agency Greenhouse Gas R&D Programme, 2002] to calculate the Euclidean distance from each point source location to the closest site where CO2could be disposed of in self-sealing strata; a critical step in evaluating self-sealing storage resource potential, as our analysis would be irrelevant if there was simply no access to these strata. Our results are compiled for the world as well as for the sources in the four largest CO2 emitting economies: China, India, the U.S. and the E.U. Approximately 12 of nearly 30 billion tonnes of global CO2 emissions in 2007 are emitted by point sources in the IEA database that include geocoordinates; another 2 billion tonnes are listed but cannot be mapped (>1 Gt of which are from China). The results for the large emitters are shown in Figure 3 (inset) ordered by distance and plotted as a cumulative emissions curve.
 Our global analysis reveals that almost 10 out of the 12 Gt CO2 emitted annually from all the point sources analyzed, or roughly a third of all global CO2emissions, are within 1,000 km of self-sealing strata; 4.5 Gt CO2 are emitted within 500 km, and nearly 500 Mt CO2 within 100 km. For comparison, CO2 is already being transported up to 808 km in the U.S., from natural geologic reservoirs in Colorado, Wyoming, Utah and New Mexico to the Permian Basin in west Texas where the CO2is re-injected underground to enhance oil recovery [Dooley et al., 2009]. Of the major economies we analyze for offshore CO2 storage (Figure 3, inset), India has the best initial access to self-sealing strata, with 800 Mt CO2 (62%) of its annual releases being within 500 km of such strata [Energy Information Administration, 2010]. Over this same distance, the U.S. releases 1Gt CO2/y (16%), while the EU and China, release only 250 Mt (6%) and 330 Mt (6%) of their respective CO2 emissions.
Levine et al.  identified three potential hazards that could be encountered in storing CO2in sands beneath self-sealing deep-marine sedimentary strata: (1) steep geothermal gradients, which would critically thin the NBZ where CO2 might be stored; (2) high saline pore waters in which dense, liquid CO2 would be buoyant and thus rise to the seafloor; and importantly, (3) the impermeable nature of many pelagic muds, clays and carbonate oozes, which would prevent rapid injection of CO2 in sufficient volumes without unintended hydraulic fracturing, compromising the seal integrity of the NBZ and HFZ trapping mechanisms. They considered the geothermal and salinity conditions in three potential scenarios, reflecting some areas in: (1) the North Atlantic, where the geothermal gradient is 35°C/km and the pore water salinity was ∼30 psu, (2) the Pacific, where the geothermal gradient may reach as high as 100°C/km; and (3) the Gulf of Mexico, where they indicate that salinity of sediment pore waters can be as high as 60 psu.
 Our global analysis of oceanographic data and deep-ocean sediment cores suggests that such salinity and temperature gradient concerns are rare in deep-marine, self-sealing strata. In particular, our results indicate that steep, limiting geothermal gradients are not widespread beneath self-sealing strata in the Altantic and Pacific oceans, and therefore, that the NBZ and HFZ does not vary substantially in thickness. Global heat flow data are integrated into the geospatial analysis of the NBZ and HFZ and we find that the strata that lie within both averages almost 200 m thick (Figure 3).
 Our global analysis of pore water salinity data in DSDP and ODP cores shows that >96% of the sampled pore waters are typically well below 60 psu, as noted for some areas in the Gulf of Mexico [Levine et al., 2007]. In fact, we found no instances where a salinity of 60 psu extends over more than a few meters of depth, and only one case where a core sample had pore waters with a salinity >60 psu within 10 m of the seafloor. We conclude that a gravity inversion that would actually leak CO2onto the seafloor is extremely unlikely. To illustrate the rarity of this scenario, we searched for drilling sites in the self-sealing regions mapped inFigure 3 with pore water samples within 200 meters of the seafloor that have densities >1040 kg/m3. Our search yielded just 6 holes in three locations that meet this threshold.
 Whether ocean sediments in self-sealing regions are permeable enough to store climatically significant volumes of CO2is a critical and complex question. Although low permeability pelagic oozes and clays are indeed the most common types of sediments found in deep-ocean settings, the targeted reservoirs for CO2injection are instead the underlying high-permeability sand layers, commonly identified within turbidite sequences, that often form the large, deep-water oil and gas fields now being exploited offshore all continents except Antarctica [Brantly, 2007]. The question we address is therefore whether such sands are volumetrically significant in self-sealing regions.
 Our analysis of sediment lithology descriptions from all DSDP and ODP cores (from 407 boreholes at 288 sites) suggests the answer is yes. Sand is present in many of the cores and occurs in significant amounts in certain self-sealing regions, especially off the passive continental margins identified inFigure 3as having the greatest self-sealing storage potential. The DSDP/ODP core data reveal the presence of extensive and in some cases thick sand bodies in these regions as well as numerous thinner sand layers (seeTable 1), both of which are viable storage options either in contiguous storage layers or “stacked” storage in multiple strata.
 In our analysis, we have extrapolated the net-to-gross sand ratio we calculate from DSDP/ODP cores to the entirety of the sediment column to arrive at our bulk storage capacity estimates. Such extrapolation is of course tenuous, but our resulting estimates are supported by other findings. For example, offshore of the Atlantic Margin and particularly seaward of the Baltimore Canyon and the Carolina Trough, we estimate that there may be thousands of meters of bulk sediment thickness with the right temperature and pressure conditions for self-sealing CO2storage. These strata include the mostly fine-grained Tertiary sediments contained in the cores we analyzed, plus deeper Lower Cretaceous sediments that are largely medium- to coarse-grained sands [Sheridan and Grow, 1988]. Thus in addition to the fraction of sands we find in the Tertiary sediments, there is considerably more sand towards the base of this self-sealing region into which CO2 could be injected.
 Similar additional potential exists in the U.S. Gulf of Mexico. The complex evolution of the strata within this basin has included the deposition of large sand bodies [Watkins and Drake, 1982], a number of which have been found to contain “giant” (500 million barrels of ultimately recoverable oil or gas equivalent [Halbouty, 2003]) oil and gas fields, such as the Thunderhorse and more recently Cascade and Chinook fields [Halbouty, 2003; Brantly, 2007; Masson et al., 2011].
 Offshore of India in the Bengal fan, DSDP Leg 218 cored numerous sandy turbidites. Coring and drilling during the leg was insufficient to confirm the location and disposition of these sands [Curray et al., 2002], but because the sediment discharges from both the Ganges and the Brahmaputra rivers are sandier than other major rivers, large volume of sand and other coarse sediment were inferred to have been carried to the Bengal Fan during Pleistocene sea level lowstands [Curray et al., 2002]. Note that these passive continental margins (i.e., the India margin and the combined U.S. Atlantic and Gulf of Mexico margins) have both the best self-sealing sediment potential and the closest proximity to onshore sources, and thus may be most attractive among the potential self-sealing storage reservoirs identified here.
 Our global analysis of comprehensive datasets leads us to conclude that the physical potential of self-sealing storage is very large. Whether self-sealing marine strata are also a technically and economically viable resource for storing CO2 though remains to be established. Technical requirements such as an extensive offshore pipeline system, possibly in combination with liquefied gas tankers to transport the supercritical CO2 into water depths >3000 m, could lead to total costs for CO2transport and storage that are significantly higher than those currently estimated for storage in onshore reservoirs. Furthermore, storage below the ocean is likely to be more expensive than its onshore counterpart and self-sealing storage in particular may be quite challenging and costly, from getting the CO2out to the self-sealing regions to building storage infrastructure roughly 3 km below the ocean surface [IPCC, 2005].
 Given the significant amount of relatively inexpensive CO2 storage potential on land, at least in the U.S. [Eccles et al., 2012], offshore storage would appear to make little economic sense. There are, however, significant advantages to offshore storage that could counterbalance its additional cost. These include potentially simpler geological requirements (i.e., no cap rock), a simplified regulatory and property rights environment, and no chance of underground aquifer contamination [Schrag, 2009; Goldberg et al., 2010]. Offshore storage may also not be as susceptible as land-based storage to minor earthquakes that could compromise storage integrity and lead to CO2 leaks [Zoback and Gorelick, 2012].
 There are certainly risks associated with storing CO2in self-sealing marine strata; this approach to emissions mitigation could help stabilize climate even as the world continues to predominantly rely on fossil fuels for energy, but could also create unintended and potentially threatening environmental problems. As the recent Deepwater-Horizon oil spill in the Gulf of Mexico shows, there are significant environmental risks associated with underwater drilling, and even self-sealing storage is unlikely to be immune from all possible modes of failure. If sufficient safeguards could be established, however, offshore CO2injection into self-sealing strata could obviate many of the risks associated not only with offshore but also onshore sequestration. Our maps of self-sealing strata provide nations with additional information for broadening their national-scale assessments of storage potential to include their EEZs in evaluating the feasibility of CCS as a climate mitigation strategy.
 We thank C. Moore, K. Logan, and P. Caldwell for research assistance. This research was supported by DOE award DE-FE0001934. We also thank the two anonymous reviewers whose suggestions significantly improved the original paper for publication. Finally, we offer special thanks to D. Goldberg for his help in finalizing the paper for publication.
 The Editor thanks two anonymous reviewers for assisting in the evaluation of this paper.