1. Introduction
Global carbon dioxide concentration in Earth’s atmosphere continues to rise (now exceeding 400 ppmv) largely because of increasing industrialization and associated fossil fuel combustion. Surface air temperature of Earth has experienced nearly 1°C of global warming in the past 100 years, which has probably contributed to episodes of extreme weather such as heat waves and drought, storms, and floods, as well as cryosphere depletion and sea level rise (IPCC 2013). Climate model projections through the twenty-first century suggest further increases in global surface air temperature of an additional 2°–5°C depending in part on the different scenarios of increasing atmospheric CO2 concentrations (IPCC 2013). Consideration of potential geophysical and economic consequences of such warming can be further examined in IPCC (2013), which also states that most components of climate change could continue even if future CO2 increases are controlled (because of the long residence time of CO2 gas in the troposphere). Different residence times range from land uptake through photosynthesis and respiration (up to 100 years) to ocean invasion (up to 1000 years). Local solutions for limiting the CO2 being released into the atmosphere are essential and reasonable approaches for reducing atmospheric CO2; however, a global-scale reduction of atmospheric CO2 should also be considered (see McNutt et al. 2015). It is noted that 1 Gt of carbon (i.e., 3.7 Gt of CO2) sequestered annually from Earth’s atmosphere equates to 0.5 ppmv of CO2 reduction (see Trenberth 1981).
A global-scale direct air capture (DAC) approach could be a viable method [as discussed by Keith (2009)], which has helped to motivate this laboratory study. Possible solutions for capturing ambient CO2 range from gas-scrubbing technologies to the use of molecular sieves (e.g., see McLaren 2014). It is noted that the separation of CO2 from a gas through phase-change deposition is possible by either cooling or pressure change or by a combination of the two processes.
Dry ice manufacturing is the most well-known operation for converting CO2 from a gaseous to solid state (noting that dry ice sublimates at 195 K). Dry ice plants are typically placed next to ethanol plants, where they can capture the flue gas coming from the ethanol plants (which is ~99% CO2). The ethanol plant’s flue gas is piped directly to the dry ice plant and placed under 20–40 bars of pressure to liquefy the CO2, and then passed through an expansion valve. As the liquid encounters expansion, approximately 50% of the CO2 evaporates (producing cooling) and 50% snows out (which is turned into dry ice when compacted). Another process of separating CO2 from a gas is found in commercial cryogenic air separators. Cryogenic air separators use ambient air to convert the major atmospheric constituents of N2, O2, and Ar from the gaseous phase into the liquid phase through a process involving pressurized compression and expansion. Neither liquid CO2 nor solid CO2 is a final commercial product from cryogenic air separation. It is noted that commercial dry ice deposition chooses pressurization (i.e., cooling at 1 bar of pressure is not an operational choice). CO2 deposition through pressurization versus cooling at constant pressure requires more energy, especially for global-scale CO2 sequestration that could possibly produce climate change mitigation. Specifically, a global-scale reduction of CO2 from Earth’s atmosphere through CO2 snow deposition at 1 bar of pressure (in the naturally cold climate of Antarctica) is a choice that supports the laboratory investigation presented in this study. This laboratory CO2 sequestration approach is also consistent with the global DAC in Antarctica as proposed by Agee et al. (2013), hereinafter AOR). The envisioned approach is to build and test a series of laboratory DAC prototypes (with increasing specifications) to demonstrate that this process has potential for real-world application for constructing and operating a prototype sequestration facility in Antarctica.
2. Design and demonstration of laboratory cooling for DAC
The AOR approach to DAC is to cool ambient air to 135 K at 1 bar of pressure through the use of a closed-loop liquid nitrogen refrigeration system and heat exchanger. The first small-scale laboratory experiment to test this approach is an open system prototype design consisting of a glass cylindrical sequestration chamber capable of withstanding cryogenic temperatures. Insulation is required to reach and maintain chamber temperatures below 135 K, with a top-mounted container providing the liquid nitrogen (LN2) as the refrigerant (with top-down cooling). This small-scale laboratory prototype is designed to evaluate the efficacy of the top-down convective mixing and the ability to achieve CO2 sequestration throughout the chamber.
a. Design and considerations for laboratory prototype
To design and employ the top-down cooling method (through the use of LN2), a number of considerations were taken in selecting the appropriate materials to design this initial DAC prototype. This design has three main components: the LN2 steel container and aluminum base plate interface, the Pyrex glass (code 7740) cylinder, and two layers of insulation (polyethylene and polyisocyanurate). Figure 1a shows the laboratory prototype assembled (without its polyisocyanurate exterior insulation). The CO2 depletion from ambient air occurs inside the Pyrex glass cylinder, which has a volume of ~26.5 L (referred to as the sequestration chamber). The steel LN2 container is mounted on top of the aluminum plate interface, and this interface (with an “O-ring” seal) rests on top of the sequestration chamber. Figure 1b shows the 14 entry ports in the circular aluminum base that are located around and outside of the LN2 container to allow instrument access into the sequestration chamber below.
(a) The LN2 container (2.5 L) mounted on an aluminum plate interface adjoining a Pyrex glass cylinder (26.5 L) assembled with polyethylene insulation. The Pyrex cylinder’s polyethylene shield contains viewing ports that can be matched with the additional exterior polyisocyanurate insulation shield (not depicted here). (b) View of bottom of aluminum plate (30.5-cm diameter) with 14 instrumentation ports that provide access to the chamber below.
Citation: Journal of Applied Meteorology and Climatology 55, 8; 10.1175/JAMC-D-16-0135.1
The use of a steel container to hold the LN2 (mounted on the aluminum plate interface) is to more effectively chill the air within the sequestration chamber from above. Steel and aluminum are highly conductive materials (15.9 W m−1 K−1 and 205 W m−1 K−1, respectively) while the Pyrex glass cylinder is a low-conductive material (1.005 W m−1 K−1) with a high specific heat capacity (750 J kg−1 K−1). Using these conductive materials to cool the ambient air in a chamber that has low conductivity/high specific heat capacity is appropriate for cooling, while maintaining the integrity of the sequestration chamber.
To limit the heat transfer into the entire apparatus, two layers of different insulation were applied (polyethylene and polyisocyanurate). The inner layer of insulation is polyethylene, which is wrapped around both the Pyrex glass cylinder and the LN2 container as shown in Fig. 1a. Polyethylene is an intermediate buffer (which insulates to 205 K) but primarily serves as a moisture barrier. As cooling ensues, the polyethylene shields condensation on the outer walls of the Pyrex glass and steel container from the polyisocyanurate. The second more effective thermal insulation designed for the apparatus was constructed to specifications by ITW Insulation Systems. Figure 2 shows this TRYMER 2500 polyisocyanurate “clamshell” that wraps around the entire laboratory prototype, including the top and bottom. Polyisocyanurate can insulate temperatures down to 90 K (as recommended by AOR for potential CO2 dry ice storage in Antarctica).
The TRYMER 2500 polyisocyanurate clamshell that encompasses the entire DAC prototype. Cut out viewing ports line up with the polyethylene viewing ports to see inside the chamber (see Fig. 3). This insulation is effective to 90 K (45 K below the CO2 deposition temperature).
Citation: Journal of Applied Meteorology and Climatology 55, 8; 10.1175/JAMC-D-16-0135.1
During experimentation, the entire volume of the sequestration chamber achieved nearly uniform temperatures ≤135 K because of convective mixing of the cold thermal plumes resulting from the top-down cooling design of the prototype apparatus (see Fig. 3). Once the cooling process starts, the critical upper depth of the chamber cools faster and becomes colder than the bottom of the chamber. As the cooling progresses, colder air plumes sink and the relatively warmer plumes rise from the bottom of the chamber to the top (where cooling continues). This upside-down convection is known as Rayleigh–Taylor instability, which results from a more dense fluid over a less dense fluid leading to convective mixing throughout the entire chamber.
The schematic of the DAC prototype (not depicting the aluminum plate or insulation) showing cold plumes sinking and compensating warmer plumes rising in thermal convection that can be attributed to Rayleigh–Taylor instability.
Citation: Journal of Applied Meteorology and Climatology 55, 8; 10.1175/JAMC-D-16-0135.1
Heat transfer is initiated where the LN2 interfaces the highly conductive aluminum plate, producing uniform cooling at the top of the sequestration chamber. The Pyrex glass is a poor heat conductor, which will support additional cooling of the interior chamber air. Through convective mixing (as shown later with experimental data), the bottom will eventually become the coldest as the warmer plumes move to the top of the chamber. Cooling is maintained at the top of the chamber through a continuous supply of LN2, which continuously supports convective overturning resulting in quasi-uniform temperatures throughout the sequestration chamber at deposition values ≤135 K.
b. Methodology to small-scale laboratory demonstration
The methodology in this laboratory experiment, in addition to cooling, also requires measurements of CO2 concentrations in the sequestration chamber air for both precooling and postcooling. As noted previously, the DAC prototype contains 14 entry ports surrounding the LN2 container that allows access into the sequestration chamber. These ports are used for instrumentation measurements consisting of thermocouples for measuring the sequestration chamber air temperature at different levels and to also extract the sequestration chamber air for the LI-7000 to take CO2 concentration measurements.
The basic procedure for the refrigeration process consists of maintaining the level of LN2 in an insulated container while recording chamber air temperature at three levels: 1) level 1 at 5.08 cm (2 in.) from the top of the sequestration chamber, 2) level 2 at 20.32 cm (8 in.) from the top of the chamber, and 3) level 3 at 35.56 cm (14 in.) of depth into the chamber (10.16 cm or 4 in. from the bottom) with type K thermocouples. Temperatures were taken in short time increments (10–15 min) in the first hour because of rapid cooling, and then every 20–30 min as the sequestration chamber approached the deposition temperature. Once all three thermocouples measured the sequestration chamber air temperature at 135 K or lower, the postcooling CO2 concentration measurement was taken and sampled through the LI-7000 CO2/H2O gas analyzer.
The laboratory prototype had air leaks through the ports and aluminum plate interface, resulting in a nonairtight apparatus. For this reason, a small amount of room air was slowly entrained into the CO2 sequestration chamber with only minor effects (yet leaving residual CO2 values). Depending on the number of people and size of a room, air in buildings can easily achieve values up to 600 ppmv. It is also noted that the small amount of room air pulled through the leaks in the apparatus provides a level of safety to avoid implosion of an airtight system. To measure CO2 concentrations within the sequestration chamber, air from the chamber was sampled through the LI-7000 by an external pump before the refrigeration process started, and subsequently after the entire depth of the sequestration chamber reached 135 K or lower. The objective of this procedure was to determine the amount of CO2 depletion after the sequestration chamber reached values of ≤135 K. Figure 4 shows the configuration for the CO2 concentration measurements.
Flowchart showing the experimental setup for CO2 sequestration measurements. Air pulled from the sequestration chamber is taken to cell B of the LI-7000 and exits (after measurement) through an external pump into the laboratory. The air cylinder is used as a reference for the LI-7000 and is inserted through cell A with a controlled flow rate (MFC).
Citation: Journal of Applied Meteorology and Climatology 55, 8; 10.1175/JAMC-D-16-0135.1
c. Convective overturning and temperature measurements
The first experiment conducted (labeled experiment I) to achieve the required cooling and CO2 deposition was also repeated on a different day (labeled experiment II). The sequestration chamber for experiment I required approximately 4.5 h of cooling to achieve the first reading of 135 K, and approximately an additional hour for all levels of the sequestration chamber to reach 135 K (or colder). The coldest temperature achieved in the two experiments conducted was 124.8 K, located at the top level. The bottom level was nearly as cold, with the middle the warmest at ~135 K for any level (see Table 1).
This table summarizes experiment I and II cooling results: the coldest temperature out of three levels, the amount of time it took to reach deposition temperature of 135 K (−138°C), and the amount of LN2 required to cool the chamber.
Figures 5 and 6 show the time elapsed during chamber cooling for experiment I and experiment II, respectively. For both experiments, all levels read the same start temperature, and the top cooled the fastest and the bottom the slowest. Both of these plots show that the midlevel thermocouple had a slightly warmer temperature (yet always ≤ 135 K).
Time plots of direct thermocouple readings from the three different levels within the sequestration chamber for experiment I. The midlevel measured slightly warmer, while the bottom level measured the coldest (at the end of the experiment). However, there was a small temperature gradient between the top and bottom levels.
Citation: Journal of Applied Meteorology and Climatology 55, 8; 10.1175/JAMC-D-16-0135.1
As in Fig. 5, but for experiment II. The midlevel measured slightly warmer while the top level measured the coldest (at the end of the experiment). Again, there was a small temperature gradient between the top and bottom levels.
Citation: Journal of Applied Meteorology and Climatology 55, 8; 10.1175/JAMC-D-16-0135.1
It was observed that the design of the system could withstand the required cold temperatures, as the integrity of the apparatus was not compromised by the cooling process. This stage of experimentation completed the first goal of the DAC prototype design, namely, the ability to achieve CO2 deposition temperature at 1 bar of pressure.
3. CO2 sequestration in the laboratory
The second goal of the experimentation was to effectively deplete the CO2 from inside the glass chamber. Sequestration results are presented and discussed for experiments I and II as described above. Table 2 shows the CO2 concentrations before and after the chamber cooling, as well as the lowest chamber temperature reached during each experiment. Both experiments were conducted by the method described in section 2b. The initial LI-7000 postcooling CO2 measurements dropped rapidly in the first minute, and after 3–5 min the minimum values of 35–50 ppmv were achieved. After obtaining these minimum values, continuous sampling for 10 more minutes showed the CO2 concentration increasing to ~100 ppmv. When air was being pulled out of the sequestration chamber by the combined effort of the LI-7000 and external pump, a small amount of air from the laboratory was being pulled back into the sequestration chamber because of the pressure leaks in the instrument ports and around the aluminum plate interface. These experiments still showed 89%–93% CO2 depletion after chamber cooling to ≤135 K (however, a sealed apparatus should produce complete sequestration).
Results from experiment I and experiment II sequestration experiments showing precooling and postcooling CO2 concentrations, as well as the lowest temperature reached and the percent CO2 depletion from the precooling to postcooling.
The experimental design produced a 125-K temperature as the lowest value measured during the experiments. The lowest recorded CO2 concentration measured postcooling was 35.8 ppmv. These results accomplished the second goal of the DAC apparatus and experimentation, which was to observe measurable CO2 depletion. The CO2 sequestered by this experimental method establishes deposition as the physical explanation for depletion (either by frost on surfaces and/or deposition on aerosols). The amount of CO2 mass depleted was 21 mg in experiment I and 23 mg in experiment II. Without higher CO2 concentrations (as discussed in section 5), it was not possible to observe the presence of CO2 deposition or cloud formation.
4. CO2 sequestration in Antarctica
As noted in the introduction, this laboratory study has implications for removing CO2 from the atmosphere in Antarctica based on a methodology proposed by AOR. A reasonable location for global-scale sequestration and storage of atmospheric CO2 would be a cold, isolated area that provides a natural environment with resources that support DAC. Antarctica is well suited for DAC because of its cold temperatures, which allow for reduced power requirements when changing CO2 from gas to solid through closed-loop LN2 refrigeration. Heat extraction during CO2 sequestration can also be supplied to support adjacent buildings and operational facilities. Katabatic wind energy is abundantly available for wind farms and there is vast space for numerous sequestration facilities and storage. International collaboration also seems logical to address the global-scale CO2 problem and there is an existing Antarctic Treaty that provides governance for the participating nations that conduct scientific operations on that continent.
It is noted, however, that these advantages are also accompanied by disadvantages. The remote location adds challenges to building and maintaining operational facilities as well as permanent storage facilities. The extreme cold and extended darkness also preclude a year-round operation. Turbines and blades used in wind farms are subject to stress and metal fatigue, especially at Antarctic temperatures. Operational efficiency of refrigeration units is an unknown, as well as the nuisance of frost formation and H2O snow and ice deposition. The availability of a large energy source that would allow appreciable reduction of ambient CO2 is also a concern. Further discussion of the advantages and disadvantages of each of the components of DAC in Antarctica (based on AOR) is provided below.
The entire methodology for this DAC approach requires more detailed consideration of the challenges confronted and the investigations required to establish proof of concept, leading to the construction of a prototype facility.
a. Antarctica is cold and dry
The coldest and driest continent on Earth is Antarctica, making it a candidate location for CO2 deposition to minimize the energy and insulation requirements for sequestration and storage. According to Miao et al. (2001), based on satellite observations, the annual mean of the total water vapor mass of the entire volume over the continent of Antarctica is ~20 × 1012 kg. Using this information, it is estimated (calculations not shown) that the atmospheric CO2 mass is on the average 7 times the H2O vapor mass. Therefore, water ice can be viewed as a minor constituent of the total snow/ice sequestration and storage. It could be simply stored with the solid CO2, yet it remains as a by-product of refrigeration that requires consideration.
b. Antarctica wind energy and energy requirements for sequestration
1) Wind energy
Antarctica is known for its katabatic winds resulting from cold dense air on an elevated ice cap that accelerates downward by gravitational force. Coastal regions frequently receive these katabatic winds and are reasonable locations for the installation of wind farms. Such locations are also the most accessible for building and operating the sequestration facilities, yet remain subject to the seasonal difficulties of Antarctic sea ice. It is noted that the West Antarctic Ice Sheet is one area that would not be considered a good location. Wind turbine technology is continuing to improve to offer better design for increasing power output and to allow operation at cold temperatures, as well as turbine threshold cutoff for operational safety in extreme wind.
One small wind farm (three 1-MW turbines powered by katabatic winds) on New Zealand’s Antarctica Scott Base has been in operation since January 2010. Solar energy could also seasonally supplement the energy provided by the wind farms.
2) Energy requirements
Energy requirements are substantial for a project that calls for the removal of 1 Gt of carbon annually from the atmosphere as proposed and discussed in AOR. Calculations in that study suggest that the equivalent of sixteen 1200-MW wind farms are required to sequester 1 Gt of carbon annually from the atmosphere (leading to a reduction of 0.5 ppmv of CO2). Larger reductions would require the appropriate multiple number of wind farms to achieve the larger sequestration (e.g., an annual reduction of 5 ppmv would require 160 such wind farms amounting to 192 000 MW). This can be compared with the 75 000 MW of wind energy capacity in the United States in 2015 (American Wind Energy Association; http://www.awea.org/MediaCenter/pressrelease.aspx?ItemNumber=8463), which is slightly less than the amounts produced in both China and in the European Union. These three totals combined yield a production in excess of 225 000 MW. Energy equivalent to that amount exceeds that required (192 000 MW) to sequester 5 ppmv CO2 per year (as noted above). Ten years of sequestration removes 50 ppmv (or doubling the number of wind farms would require only 5 years). Although such amounts are seen to be within the realm of reality, a thorough examination of the potential for wind energy production in Antarctica is required.
c. Antarctica deposition and storage
The CO2 dry ice accumulation from the DAC deposition process can be stored in insulated landfills per the AOR concept. Refrigeration chambers would require cleaning and maintenance. Landfills would be insulated with polyisocyanurate (effective for temperatures as low as 90 K) to maintain temperatures less than 195 K (the dry ice sublimation temperature at 1 bar of pressure). Storage units are at risk to cracking in the extreme cold or cracking as a result of the shifting of ice/snow formations, resulting in the unwanted sublimation of dry ice into the ambient air.
From the increased sequestration amounts in section 4b (above) the increased requirements for storage can be calculated. Again, using the values presented in AOR, the total volume required for storing the equivalent of 0.5 ppmv of dry ice is 2 km × 2 km × 160 m. Removal and storage of 5 ppmv of dry ice equivalent would amount to 10 such volumes. For a scale comparison, this total collective volume is equivalent to the size of Lake Okeechobee in Florida, which has an average water depth of 3 m.
d. Antarctica Treaty
The Antarctic Treaty has the necessary infrastructure that commands international governance and scientific cooperation (see http://www.nsf.gov/od/opp/antarct/anttrty.jsp). Climate change due to anthropogenic CO2 is global, and so are the emissions into the one atmosphere that all nations share. The construction of CO2 deposition plants, wind farms, and landfills provides opportunity for international collaboration to curtail the effects of anthropogenic global warming. It is recommended that expertise from different countries contribute to the construction of sequestration and storage facilities because of the magnitude of this global problem.
5. Summary and conclusions
A simple DAC laboratory prototype has been designed, constructed, and implemented to achieve the CO2 deposition temperatures of ≤135 K (with top-down cooling through Rayleigh–Taylor mixing) and CO2 sequestration from ambient air at 1 bar of pressure. The primary objectives of the laboratory prototype experiment were met, and they were accomplished with a low chamber temperature of 125 K and CO2 depletion of ~90% (leaks accounted for measuring less than complete CO2 depletion). The experiment was repeated on a different day producing similar results. Cooling of the 26.5-L sequestration chamber required approximately 5 h to achieve the theoretical deposition temperature. Three thermocouple probes strategically placed inside the chamber documented the nature of the cooling process, demonstrating the top-down cooling and the effectiveness of the associated Rayleigh–Taylor mixing.
Future considerations should focus on whether the CO2 deposition process involves frosting on the top conducting plate and sides of the chamber and/or CO2 snow precipitation from a CO2 cloud formation. Neither was visible in this experiment because of the small amount of solid CO2 mass. The objective behind investigating whether CO2 frost or cloud/snow is forming within the sequestration chamber will help determine CO2 aerosol sequestration efficiency for a larger-scale apparatus. Pursuit of this objective will require a prototype system that sequesters 300 000 ppmv of CO2 (through a continuous supply process) yielding approximately 10 g of solid CO2. This amount would provide better potential for observing a CO2 snow cloud formation (with the aid of photography). The findings and experience from the preliminary laboratory experiments provided here offer considerable guidance in building a better and larger DAC laboratory prototype system (prior to consideration of a prototype closed-loop refrigeration facility in Antarctica as proposed by AOR). Further, it is noted that a heat exchanger would need to be designed and constructed for the sequestration process to help quantify more accurately the energy requirements for the larger prototype system.
Last, it is noted in general that this area of experimental research on the production of CO2 snow clouds has important ramifications for aerosol cloud chemistry and the comparative study of water snowflakes and CO2 snowflakes. Planetary scientists have research interests and activity in this area of study as well (e.g., see Hayne et al. 2012; Cziczo et al. 2013; Hu et al. 2012). A typical Earth snowflake is a hexagon, which is about 100 times the size of a typical CO2 snowflake (which has a cuboctahedral shape). Distinction in the types and spectra of aerosol populations (and their effectiveness for deposition) has important ramifications for weather processes in both Earth and Mars planetary atmospheres.
Acknowledgments
The authors extend their gratitude to ITW Insulation Systems for providing the custom-made TRYMER 2500 polyisocyanurate insulation shield. Special thanks are also extended to Dr. R. Michael Everly and the Jonathan Amy Facility for Chemical Instrumentation in Purdue University’s Department of Chemistry for contributions in the design and construction of the laboratory apparatus. A very special acknowledgement is extended to Professor Paul Shepson of Purdue University for his research guidance and for providing space in his chemistry laboratory for the experimentation. Similarly, special thanks are extended to Professor Robert Jacko of Purdue University’s School of Civil Engineering for his insight into experimentation, the engineering design, and general support provided throughout this research project.
REFERENCES
Agee, E., A. Orton, and J. Rogers, 2013: CO2 snow deposition in Antarctica to curtail anthropogenic global warming. J. Appl. Meteor. Climatol., 52, 281–288, doi:10.1175/JAMC-D-12-0110.1.
Cziczo, D. J., and Coauthors, 2013: Ice nucleation by surrogates of Martian mineral dust: What can we learn about Mars without leaving Earth? J. Geophys. Res., 118, 1945–1954, doi:10.1002/jgre.20155.
Hayne, P. O., D. A. Paige, J. T. Schofield, D. M. Kass, A. Kleniböhl, N. G. Heavens, and D. J. McCleese, 2012: Carbon dioxide snow clouds on Mars: South Polar winter observations by the Mars Climate Sounder. J. Geophys. Res., 117, EO8014, doi:10.1029/2011JE004040.
Hu, R., K. Cahoy, and M. T. Zuber, 2012: Mars atmospheric CO2 condensation above the North and South Poles as revealed by radio occultation, climate sounder, and laser ranging observations. J. Geophys. Res., 117, EO7002, doi:10.1029/2012JE004087.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., doi:10.1017/CBO9781107415324.
Keith, D. W., 2009: Why capture CO2 from the atmosphere? Science, 325, 1654–1655, doi:10.1126/science.1175680.
McLaren, D., 2014: Capturing the imagination: Prospects for direct air capture as a climate measure. Case Study, Geoengineering Our Climate, Working Paper and Opinion Article Series. [Available online at http://wp.me/p2zsRk-9Z.]
McNutt, M. K., and Coauthors, 2015: Climate Intervention: Carbon Dioxide Removal and Reliable Sequestration. The National Academies Press, 154 pp.
Miao, J., K. Kunzi, G. Heygster, T. A. Lachlan-Cope, and J. Turner, 2001: Atmospheric water vapor over Antarctica derived from Special Sensor Microwave/Temperature 2 data. J. Geophys. Res., 106, 10 187–10 203, doi:10.1029/2000JD900811.
Trenberth, K. E., 1981: Seasonal variations in global sea level pressure and the total mass of the atmosphere. J. Geophys. Res., 86, 5238–5246, doi:10.1029/JC086iC06p05238.
