REPORTS
                                                              8. C. Hilty, T. J. Lowery, D. E. Wemmer, A. Pines, Angew.
sors represents an additional dimension of          The HYPER-CEST technique is amenable to
                                                                Chem. Int. Ed. 45, 70 (2006).
sensitivity and specificity for molecular imag-      any type of MRI image acquisition methodology.
                                                              9. S.-I. Han et al., Anal. Chem. 77, 4008 (2005).
ing. The depletion process generating the image      We demonstrated CSI here, but faster acquisition         10. M. M. Spence et al., J. Am. Chem. Soc. 126, 15287 (2004).
contrast depends on several parameters, includ-      techniques that incorporate a frequency encoding         11. A. Bifone et al., Proc. Natl. Acad. Sci. U.S.A. 93, 12932
ing saturation power and time, sensor concen-       domain such as FLASH (fast low angle shot)              (1996).
                                                              12. K. M. Ward, A. H. Aletras, R. S. Balaban, J. Magn. Res.
tration, and ambient temperature. The latter       have been successfully used to acquire in vivo Xe
                                                                143, 79 (2000).
parameter provides another promising approach       tissue images (17).
                                                              13. A. A. Maudsley, S. K. Hilal, W. H. Perman, H. E. Simon,
to increase sensitivity even further, because the       The modular setup of the biosensor (i.e., the          J. Magn. Res. 51, 147 (1983).
exchange rate increases considerably when ap-       nuclei that are detected are not covalently bound        14. N. Goffeney, J. W. M. Bulte, J. Duyn, L. H. Bryant, P. C. M.
proaching 37-C (10). Characterization of the       to the targeting molecule) allows accumulation            van Zijl, J. Am. Chem. Soc. 123, 8628 (2001).
                                                              15. K. Knagge, J. Prange, D. Raftery, Chem. Phys. Lett. 397,
saturation dynamics is currently under way and      of the biosensor in the tissue for minutes to hours
                                                                11 (2004).
will reveal optimized parameters for future        before delivery of the hyperpolarized xenon           16. J. L. Mynar, T. J. Lowery, D. E. Wemmer, A. Pines,
applications.                       nuclei, which have much higher diffusivity. In            J. M. Frechet, J. Am. Chem. Soc. 128, 6334 (2006).
  The technique is also quite promising for       combination with the long spin-lattice relaxation        17. B. M. Goodson et al., Proc. Natl. Acad. Sci. U.S.A. 94,
                                                                14725 (1997).
biomedical imaging in vivo. A typical surface       time of Xe, this two-step process optimally pre-
                                                              18. G. Huber et al., J. Am. Chem. Soc. 128, 6239 (2006).
coil of 20 cm diameter detects a volume of ca.      serves the hyperpolarization before signal acqui-        19. J. P. Mugler et al., Magn. Res. Med. 37, 809 (1997).
2.1 liters, thus decreasing S/N for a (2.8 mm)3      sition. Biosensor cages that yield distinct xenon        20. S. D. Swanson, M. S. Rosen, K. P. Coulter, R. C. Welsh,
voxel by a factor of 27.2 compared with our        frequencies allow for multiplexing to detect             T. E. Chupp, Magn. Res. Med. 42, 1137 (1999).
                                                              21. This work was supported by the Director, Office of
setup. This loss is less than 50% of the gain for     simultaneously several different targets (18).
                                                                Science, Office of Basic Energy Sciences, Materials
an optimized system using 945% polarized         Also the serum- and tissue-specific Xe NMR
                                                                Sciences and Engineering Division, of the U.S.
isotopically enriched 129Xe. An isotropic reso-      signals (19, 20) arising after injection of the           Department of Energy under contract no. DE-AC03-
lution of 2 to 3 mm is feasible without signal      carrier medium can be used for perfusion studies           76SF00098. L.S. acknowledges support from the
averaging for a concentration of pure polarized      (Fig. 3B) in living tissue, making Xe-CSI a             Deutsche Forschungsgemeinschaft (SCHR 995/1-1)
                                                                through an Emmy Noether Fellowship. T.J.L.
129Xe that is È2 mM in tissue. This minimum        multimodal imaging technique.
                                                                acknowledges the Graduate Research and Education in
value is below those observed for direct                                            Adaptive bio-Technology (GREAT) Training Program of the
injection of Xe-carrying lipid solutions into rat                                       UC Systemwide Biotechnology Research and Education
                               References and Notes
muscle (70 mM) or for inhalation delivery for                                         Program (no. 2005-264), and C.H. acknowledges support
                              1. J. M. Tyszka, S. E. Fraser, R. E. Jacobs, Curr. Opin.
                                                                from the Schweizerischer Nationalfonds through a
                               Biotechnol. 16, 93 (2005).
brain tissue (8 mM) used in previous studies
                                                                postdoctoral fellowship.
                              2. A. Y. Louie et al., Nat. Biotechnol. 18, 321 (2000).
that demonstrated Xe tissue imaging in vivo
                              3. S. Aime, C. Carrera, D. Delli Castelli, S. Geninatti,
(17). Sensitive molecular imaging of the bio-                                       Supporting Online Material
                               E. Terreno, Angew. Chem. Int. Ed. 44, 1813 (2005).
sensor is therefore possible as long as the                                        www.sciencemag.org/cgi/content/full/314/5798/446/DC1
                              4. M. S. Albert et al., Nature 370, 199 (1994).
                                                              Materials and Methods
distribution of dissolved xenon can be imaged       5. M. M. Spence et al., Proc. Natl. Acad. Sci. U.S.A. 98,
                                                              Fig. S1
                               10654 (2001).
with sufficient S/N and the biosensor target is                                      References
                              6. T. J. Lowery et al., Magn. Reson. Imaging 21, 1235
not too dilute, because HYPER-CEST is based          (2003).
on the detection of the free Xe resonance, not       7. K. Bartik, M. Luhmer, J. P. Dutasta, A. Collet, J. Reisse,  28 June 2006; accepted 29 August 2006
direct detection of the biosensor resonance.         J. Am. Chem. Soc. 120, 784 (1998).              10.1126/science.1131847




Wetland Sedimentation from                                                 September, 2005, respectively, leaving behind a
                                                              devastated urban and rural landscape. Massive
                                                              amounts of water, salt, and sediments were re-
Hurricanes Katrina and Rita                                                distributed across the coastal zone within a few
                                                              hours as a storm surge of up to 5 m propagated in
R. Eugene Turner,1,2* Joseph J. Baustian,1,2 Erick M. Swenson,1,2 Jennifer S. Spicer2                   a northerly direction at the coastline south of New
                                                              Orleans, LA (Katrina), and near Sabine Pass,
                                                              Texas (TX) (Rita), inundating coastal wetlands in
More than 131 Â 106 metric tons (MT) of inorganic sediments accumulated in coastal wetlands
                                                              the region. A thick deposit of mud remained in
when Hurricanes Katrina and Rita crossed the Louisiana coast in 2005, plus another 281 Â 106 MT
                                                              these coastal wetlands after the storm waters re-
when accumulation was prorated for open water area. The annualized combined amount of
                                                              ceded (Fig. 1). We used this post-storm remnant
inorganic sediments per hurricane equals (i) 12% of the Mississippi River’s suspended load, (ii)
                                                              to learn about how coastal systems work.
5.5 times the inorganic load delivered by overbank flooding before flood protection levees were
                                                                The loss of LA_s coastal wetlands peaked be-
constructed, and (iii) 227 times the amount introduced by a river diversion built for wetland
                                                              tween 1955 and 1978 at 11,114 ha year –1 (1)
restoration. The accumulation from hurricanes is sufficient to account for all the inorganic
                                                              and declined to 2591 ha year –1 from 1990 to
sediments in healthy saltmarsh wetlands.
                                                              2000 (2). Coastal wetlands, barrier islands, and
   norganic sediments accumulating in coastal      sions. They may also arrive from offshore during         shallow waters are thought to provide some pro-

I  wetlands may be delivered from inland        tidal inundation or storm events. It is important to       tection from hurricanes, by increasing resistance
   sources via (i) unconstrained overbank flood-    know the quantities delivered by each pathway to         to storm surge propagation and by lowering hur-
ing, (ii) explosive releases through unintentional    understand how inorganic sediments contribute          ricane storm surge height (3). Restoring LA_s
breaks in constructed levees, and (iii) river diver-   to wetland stability and to spend wetland res-          wetlands has become a political priority, in part
                             toration funds effectively. Here we estimate the         because of this perceived wetland/storm surge
                             amount of inorganic sediments deposited on            connection. A major part of LA_s restoration
1
Coastal Ecology Institute, 2Department of Oceanography
                             wetlands of the microtidal Louisiana coast            effort is to divert part of the Mississippi River
and Coastal Sciences, Louisiana State University, Baton
                                                              into wetlands, and at considerable cost Eref.
                             during Hurricanes Katrina and Rita.
Rouge, LA 70803, USA.
                               Hurricanes Katrina and Rita passed through          (S1) in supporting online material (SOM)^.
*To whom correspondence should be addressed. E-mail:
                             the Louisiana (LA) coast on 29 August and 24           Widely adopted assumptions supporting this
euturne@lsu.edu


                                                                                               449
                     www.sciencemag.org       SCIENCE      VOL 314     20 OCTOBER 2006
REPORTS
                              2.23 g cm–2 (T1 SD 0 T 3.4; range, 0 to 28.6 g
   diversion are that flood protection levees have                             where water was withdrawn in a southerly
                              cm–2; n 0 169) and 2.25 g cm–2 in the deltaic
   eliminated overbank flooding, which has caused                              direction out of the wetlands (Fig. 2, D to F).
   diminished sediment accumulation and eventual    plain. The thickness of the newly deposited     The greatest deposition in the Deltatic Plain was
                              mud was 5.18 cm (T1 SD 0 T 7.7; range, 0 to
   wetland loss, and introducing sediments into                               in the Breton Sound estuary, on the east side of
                              68 g cm; n 0 186). The thickest newly de-
   estuaries via river diversions will enhance wet-                             the Mississippi River. The marshes within the
   land restoration.                  posited sediments were observed inland of the    4- longitude distance between the two hurri-
     If increasing inorganic sediment loading to   area of maximum bulk density in eastern LA     canes (approximately 300 km) had intermediate
   coastal wetlands is important for their restora-   but were coincidental with the sediments of     rates of deposition. The peak water level, but
   tion, then it is important to quantify the major   highest bulk density in western LA (Fig. 2C).    not sediment accumulation, was higher in west-
   sediment pathways. Hurricanes Katrina and      The annualized average sediment accumulation    ern Lake Pontchartrain during Hurricane Rita
   Rita obviously brought some inorganic sedi-     from one hurricane was 89% of the average      than during Hurricane Katrina because of these
   ment into the coastal wetlands, but how much,    accumulation in healthy saltmarsh wetlands in    differences in wind fields (11). The peak in bulk
                              the deltaic plain E0.166 g cm–2 year –1 (10)^.
   and where was it distributed? A few measure-                               density was highest on the eastern side of the
                                Sediment deposition (in grams per cm2) was
   ments of the inorganic sediments accumulating                              center of the storm track.
   from a hurricane have been made at specific     greatest near the center of the storm track (Fig.    There were peaks in sediment deposition
   sites along the coast (4–8) Eref. (S7) in SOM^,   2D). The highest values in the Chenier Plain    where navigation channels confined the incom-
   but until this study there were no coastwide     were on the east side of the hurricane path,    ing storm surge to a narrow area at Sabine Pass,
   data on the inorganic sediments accumulating     where counterclockwise winds brought a storm    TX, and in the industrial canal by Paris Road,
   from hurricanes. Here we show that the dom-     surge inland, and were least on the western side,  New Orleans, where the Intracoastal Waterway
   inant pathway of inorganic sediments into the
   microtidal LA coastal wetlands is from offshore
   to inshore during hurricanes, and not from over-
   bank flooding along the main channel of the
   Mississippi River, from smaller storm events, or
   from tidal inundations.
     We sampled from the shoreline to the on-
   shore limit of storm sediment deposition in
   wetlands (9) (Fig. 1A). We collected samples
   from all coastal watersheds in LA and at seven
   sites in eastern TX, using a helicopter and air-
   boat (145 samples) or by walking out 920 m
   into the wetland from access points reached by
   boat or car (53 samples). Freshly deposited mud
   was easily identifiable from the layers beneath
   on the basis of color, texture, and density, and
   by the absence of plant debris (Fig. 1E). At least
   one preliminary sampling was done at each
   location (often three or four) before the final
   sample was taken. Samples for sediment depth
   (in centimeters) and density (in grams per
   cm3) were taken only over the vegetation zone
   and not in rivulets between clumps of wetland
   plants.
     The average bulk density of the newly
   deposited material was 0.37 g cm–3 (T1 SD 0
   T 0.35; range, 0 to 1.78 g cm–3; n 0 170
   samples); it was highest near the coastline and
   decreased inland (Fig. 2B). The bulk density of
   material in these wetlands was determined by
   the amount of inorganic materials, not the
   organic content, and so bulk density multiplied
   by deposition height is an estimate of inorganic
   sediment deposition (9, 10). Sediment with a
   bulk density 91 was largely composed of sand
   (Fig. 1A) (9). There was sparse plant debris
   (stems, leaves, and roots) in the newly
   deposited sediments within a few kilometers
   from the shoreline. The unconfirmed hypoth-
   esis is that the source materials from offshore   Fig. 1. Examples of sediments deposited by Hurricanes Katrina and Rita. (A) Sand overwash on the
   came from where the bottom resistance to the     former location of the coastal community of Holly Beach, LA (photo taken 18 November 2005 by
   hurricane winds before landfall was the       R.E.T.). (B) Mud on the lawn of a St. Bernard Parish subdivision home (photo taken 27 September
   greatest: in the shallow water zone immedi-     2005 by M. Collins). (C) Mud on the marsh surface brought by Hurricanes Katrina and Rita (photo
   ately offshore of the deposition site.        taken 16 November 2005 by J.B.). (D) Recent mud deposit (10.5 cm) accumulated over a root mass
     The average dry weight accumulation of the    in the St. Bernard estuary (photo taken 16 November 2005 by J.B.). (E) Dried mud on the lawn of a
                              Chalmette, LA, subdivision home, September 2005 (photo taken by R. Richards).
   deposition layer at all sampled locations was


450                     20 OCTOBER 2006      VOL 314   SCIENCE    www.sciencemag.org
                                                                         REPORTS
meets the Mississippi River Gulf Outlet. These   annualized deposition from one hurricane       (Table 1) (13). The more frequent smaller storms
                          would be 8.3 Â 106 MT year–1 if all hurricanes
observations are consistent with results from                              not included in this analysis may also trans-
modeled storm surge velocities (12).        brought an equal amount of sediments to these    port substantial amounts of inorganic material
  The total amount of recently deposited wet-   wetlands. If sediments from these hurricanes are   (14, 15).
land sediments on the LA coast was calculated    deposited in open-water areas at the same rate as    The amount of sediments delivered to coast-
using information on the average sediment      in wetlands (9), then the pro rata deposition for  al wetlands by Hurricanes Katrina and Rita was
accretion and wetland area for each of four to   open-water areas is proportional to the open     greater than the estimated amounts once flow-
                          water/wetland area (1) and is equal to 17.8 Â
six subunits of four coastal regions. The min-                              ing through or over Mississippi River banks.
                          106 MT year –1, for a combined sediment
imum amount of inorganic sediment brought                                The suspended sediments overflowing uncon-
                          deposition of 26.1 Â 106 MT year –1.
in by these two hurricanes was estimated to be                              fined (natural) levees of the Mississippi River
131 Â 106 metric tons (MT) (9) (Table 1). The                              in the past century were 4.8 Â 106 MT year –1,
                            The sediment accumulations in wetlands and
                                                     and 1.7 Â 106 MT year –1 in a confined levee
average occurrence of a Category 3 or larger    open water were 4.0 and 8.5%, respectively, of
hurricane on this coast was every 7.88 years    the average annual suspended sediment load of    system with occasional crevasses (Table 1).
                          the Mississippi River (210 Â 106 MT year –1)
from 1879 to 2005 (9) (table S1). The                                  The Caernarvon Diversion, a restoration project
                                                     located downstream from New Orleans, LA,
                                                     delivered a 2-year average sediment load of
Fig. 2. Location of re-
                                                     0.115 Â 106 MT year –1 (16). One conclusion
cent sediment samples
                                                     to be drawn from these numbers is that the
and data arranged by
                                                     amount of sediment deposited on these wet-
longitude. (A) Sample lo-
                                                     lands from an average Category 3 or larger
cations (red dots) and
                                                     hurricane is 1.7 times the amount potentially
the distribution of coast-
                                                     available through unconfined overbank flood-
al wetlands in southern
                                                     ing, 4.6 times more than through crevasses in
LA (black background).
                                                     the unconfined channel, and 72 times more than
The vertical gray arrow
                                                     from this river diversion. The combined sedi-
is the crossing location
                                                     ment accumulation in wetlands and open water
of Hurricanes Rita (west-
ern LA) and Katrina (east-                                        resulting from an average Category 3 or larger
ern LA). (B) All samples                                         hurricane is 5.5 times larger than the material
(open circles) and sam-                                         delivered by unconfined flow in the Mississippi
ples with a bulk density                                         River and 227 times larger than that delivered
value 91.0 g cm–3 (red                                          by the Caernarvon Diversion.
dots). (C) All samples                                            However, these comparisons are conserva-
(open circles) and sam-                                         tive estimates. Not all of the inorganic sediment
ples with a vertical ac-                                         flowing from rivers and over or through levees
cretion 93 cm (red dots).                                        is deposited onto a wetland. Levees are higher
(D) Accumulation rela-                                          than the surrounding land because inorganics
tive to the longitude of
                                                     settle out onto the levees or within the nearby
sample collection (black
                                                     marshes. Also, the peak in river heights, and
circles). (E) Bulk density
                                                     hence in discharge, occurs in the spring when
relative to the longitude
                                                     water levels in the estuary are at their seasonal
of sample collection. (F)
                                                     low and wetlands are infrequently flooded (17).
Vertical accretion relative
                                                     Sediments that do accumulate are deposited
to the longitude of sam-
                                                     close to the diversion. For example, the
ple collection.
                                                     Caernarvon Diversion distributes about 50%
                                                     of its sediments into the wetlands for a
                                                     maximum distance of about 6 km, covering
                                                     about 15% of a direct path to the coastline
                                                     (Table 1) (18). Hurricanes, in contrast, are
                                                     much more democratic in that they flood the
                                                     entire coastal landscape with new sediments.
                                                       A coastwide perspective on sediment load-
                                                     ing to these wetlands, and perhaps to other mi-
Table 1. Estimates of the sediment source pathways for the Mississippi River deltaic plain in
                                                     crotidal coastal wetlands, is that most of the
Louisiana.
                                                     inorganics accumulating in them went down
                                                     the Mississippi_s birdfoot delta before they
                                       Amount (106 MT year –1 )
Sediment source pathways
                                                     were deposited during large storms. The es-
Mississippi River discharge into ocean (13)                      210
                                                     timates indicate that the amount of storm-
One hurricane every 7.88 years (table S1)
                                                     transported material is much greater than that
 Onto wetland only                                  8.3
                                                     introduced to wetlands from the historical over-
 Onto wetland and into open water (9)                         25.9
                                                     bank flow, from crevasses, or from river diver-
Overbank flooding (before flood protection levees) and into open water (22)      4.79
                                                     sions. In particular, hurricanes appear to be the
Crevasses through levees and into open water (22)                   1.81
                                                     overwhelming pathway for depositing new in-
Caernarvon Diversion
                                                     organic sediments in coastal wetlands in west-
 Into the estuary (16)                                0.115
                                                     ern LA, because the few riverine sources bring
 Onto wetland (18)                                  0.06
                                                     relatively trivial amounts of inorganic sedi-


                                                                               451
                   www.sciencemag.org     SCIENCE     VOL 314    20 OCTOBER 2006
REPORTS
                                        (contribution no. 58-3, Coastal Studies Institute, Louisiana  20. R. H. Kesel, Environ. Geol. Water Res. 11, 271 (1988).
   ments into the marsh. Because hurricanes are so
                                        State Univ., Baton Rouge, LA, 1958).              21. C. M. Belt Jr., Science 189, 681 (1975).
   important to the inorganic sediment budget, other
                                      8.  J. A. Nyman, R. D. DeLaune, H. H. Roberts, W. H. Patrick    22. Kesel (20) estimated the sediment load in the spring
   factors must be considered to understand how to              Jr., Mar. Ecol. Prog. Ser. 96, 269 (1992).             floods using water records from 1950 to 1983 and
   reduce wetland losses and further their restora-           9.  Materials and methods are available as supporting         determined the amount of sediment that would be
   tion. Changes in the in situ accumulation of               material on Science Online.                    available from unconfined overbank flooding if the levees
                                     10.  R. E. Turner, E. M. Swenson, C. S. Milan, in Concepts and     were not there. The estimate is actually an overestimate
   organics, rather than the reduction of inorganic
                                        Controversies in Tidal Marsh Ecology, M. Weinstein,        of the amount available, because the artificial constric-
   sediments arriving via overbank flooding, are               D. A. Kreeger, Eds. (Kluwer, Dordrecht, Netherlands,        tion of the river channel throughout the basin raised the
   implicated as a causal agent of wetland losses on             2001), pp. 583–595.                        flood stage for the same-sized discharge event (21).
   this coast. This is illustrated by the fact that the        11.  The peak water measured by U.S. Geological Survey       23. Supported by the NSF Division of Geomorphology and
                                        water-level gage 30174809020900, at Pass Manchac          Land-Use Dynamics (award EAR-061250), the National
   soil volume occupied by organic sediments plus
                                        Turtle Cove near Ponchatoula, LA, in the western Lake       Oceanic and Atmospheric Administration Coastal Ocean
   water in healthy saltmarsh wetlands is 990%                Pontchartrain watershed, was about 6 cm higher during       Program MULTISTRESS (award no. NA16OP2670), and a
   (10) and is certainly the same or higher in                Hurricane Rita than during Hurricane Katrina, even         Louisiana Board of Regents Fellowship (J.S.S.). We thank
   wetlands of lower salinity. This organic portion             though the hurricane path was over 300 km further away.      J. Gore for assistance in sampling from an airboat; E. Babin
                                     12.  Storm surge model results for Hurricanes Katrina and Rita     for assistance in finding photographs; H. Hampp, our deft,
   plays a major role in wetland soil stability and
                                        are available at the Louisiana State University Hurricane     safe, and considerate helicopter pilot; and B. Sen Gupta,
   hence in wetland ecosystem health (19).
                                        Center Web sites (http://hurricane.lsu.edu/floodprediction/    N. N. Rabalais, and three anonymous reviewers for
                                        katrina/deadly_funnel1.jpg and http://hurricane.lsu.edu/      constructive reviews of the manuscript.
     References and Notes                         floodprediction/rita24/images/adv24_SurgeTX_LA.jpg).
    1. R. H. Baumann, R. E. Turner, Environ. Geol. Water Res.     13.  G. J. Chakrapani, Curr. Sci. 88, 569 (2005).          Supporting Online Material
     15, 189 (1990).                         14.  R. H. Baumann, thesis, Louisiana State Univ., Baton      www.sciencemag.org/cgi/content/full/1129116/DC1
    2. T. A. Morton, J. C. Bernier, J. A. Barras, N. F. Ferina, USGS     Rouge, LA (1980).                       Materials and Methods
     Open-File Rep. 2005-1215 (2005).                15.  D. J. Reed, N. De Luca, A. L. Foote, Estuaries 20, 301     SOM Text
    3. G. W. Stone, X. Zhang, A. Sheremet, J. Coastal Res. 44,        (1997).                            Fig. S1
     40 (2005).                           16.  G. A. Snedden, J. E. Cable, C. Swarzenski, E. Swenson,     Table S1
    4. J. M. Rybczyk, D. R. Cahoon, Estuaries 25, 985 (2002).        Estuar. Coastal Shelf Sci., in press.             References and Notes
    5. M. L. Parsons, J. Coastal Res. 14, 939 (1998).         17.  R. E. Turner, Estuaries 14, 139 (1991).
    6. D. R. Cahoon et al., J. Coastal Res. 21 (special issue), 280  18.  K. W. Wheelock, thesis, Louisiana State Univ., Baton      24 April 2006; accepted 1 September 2006
     (1995).                                Rouge, LA (2003).                       Published online 21 September 2006;
    7. J. P. Morgan, L. G. Nichols, M. Wright, Morphological      19.  R. E. Turner, E. M. Swenson, C. S. Milan, J. M. Lee,      10.1126/science.1129116
     Effect of Hurricane Audrey on the Louisiana Coast           T. A. Oswald, Ecol. Res. 19, 29 (2004).            Include this information when citing this paper.




   A Combined Mitigation/Geoengineering                                                     Increased sulfate aerosol loading of the strato-
                                                                        sphere may present other risks, such as through its
                                                                        influence on stratospheric ozone. This particular
   Approach to Climate Stabilization                                                    risk, however, is likely to be small. The effect of
                                                                        sulfate aerosols depends on the chlorine loading
                                                                        (22–24). With current elevated chlorine loadings,
   T. M. L. Wigley
                                                                        ozone loss would be enhanced. This result would
   Projected anthropogenic warming and increases in CO2 concentration present a twofold threat, both                    delay the recovery of stratospheric ozone slightly
   from climate changes and from CO2 directly through increasing the acidity of the oceans. Future climate                 but only until anthropogenic chlorine loadings
   change may be reduced through mitigation (reductions in greenhouse gas emissions) or through                       returned to levels of the 1980s (which are ex-
   geoengineering. Most geoengineering approaches, however, do not address the problem of increasing                    pected to be reached by the late 2040s).
   ocean acidity. A combined mitigation/geoengineering strategy could remove this deficiency. Here we                      Figure 1 shows the projected effect of mul-
   consider the deliberate injection of sulfate aerosol precursors into the stratosphere. This action could                 tiple sequential eruptions of Mount Pinatubo
   substantially offset future warming and provide additional time to reduce human dependence on fossil                   every year, every 2 years, and every 4 years.
   fuels and stabilize CO2 concentrations cost-effectively at an acceptable level.                             The Pinatubo eruption–associated forcing that
                                                                        was used had a peak annual mean value of
                                                                        –2.97 W/m2 (20, 21). The climate simulations
     n the absence of policies to reduce the mag-          technological challenges faced by a mitigation-

   I  nitude of future climate change, the globe is          only approach.                            were carried out using an upwelling-diffusion
     expected to warm by È1- to 6-C over the                                               energy balance model EModel for the Assessment
                                       The geoengineering strategy examined here is
                                     the injection of aerosol or aerosol precursors Esuch
   21st century (1, 2). Estimated CO2 concentra-                                              of Greenhouse gas–Induced Climate Change
   tions in 2100 lie in the range from 540 to 970           as sulfur dioxide (SO2)^ into the stratosphere to          (MAGICC) (2, 25, 26)^ with a chosen climate
   parts per million, which is sufficient to cause           provide a negative forcing of the climate system           sensitivity of 3-C equilibrium warming for a CO2
                                                                        doubling (2 Â CO2). Figure 1 suggests that a
   substantial increases in ocean acidity (3–6).            and consequently offset part of the positive
                                                                        sustained stratospheric forcing of È–3 W/m2 (the
   Mitigation directed toward stabilizing CO2             forcing due to increasing greenhouse gas con-
   concentrations (7) addresses both problems but           centrations (18). Volcanic eruptions provide ideal          average asymptotic forcing for the biennial
   presents considerable economic and technologi-           experiments that can be used to assess the effects          eruption case) would be sufficient to offset much
   cal challenges (8, 9). Geoengineering (10–17)            of large anthropogenic emissions of SO2 on              of the anthropogenic warming expected over the
   could help reduce the future extent of climate           stratospheric aerosols and climate. We know, for           next century. Figure 1 also shows how rapidly the
                                     example, that the Mount Pinatubo eruption EJune
   change due to warming but does not address the                                              aerosol-induced cooling disappears once the in-
   problem of ocean acidity. Mitigation is therefore          1991 (19, 20)^ caused detectable short-term cool-          jection of material into the stratosphere stops, as
   necessary, but geoengineering could provide             ing (19–21) but did not seriously disrupt the            might become necessary should unexpected envi-
   additional time to address the economic and             climate system. Deliberately adding aerosols or           ronmental damages arise.
                                     aerosol precursors to the stratosphere, so that the           Three cases are considered to illustrate pos-
                                     loading is similar to the maximum loading from            sible options for the timing and duration of aerosol
   National Center for Atmospheric Research, Post Office Box
                                     the Mount Pinatubo eruption, should therefore            injections. In each case, the loading of the strato-
   3000, Boulder, CO 80307–3000, USA. E-mail: wigley@
                                     present minimal climate risks.                    sphere begins in 2010 and increases linearly to
   ucar.edu


452                           20 OCTOBER 2006         VOL 314      SCIENCE      www.sciencemag.org