Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 83
5
Projections of Sea-Level Change
O
bservations provide unequivocal evidence that contribution cannot yet be modeled satisfactorily.
global mean sea level has been rising over the Given this shortcoming, some investigators use cur-
past century, but that the rate of sea-level rise rent observations to extrapolate the future behavior
has significant regional variability. The key question for of the cryosphere. Another approach to projections,
planners is how much sea level will rise in their region called the "semi-empirical" approach, is based on the
in an increasingly warm future world. Most projections observed relationship between sea-level change and
are based on knowledge of the current contributions to global temperature change, and takes no account of the
sea-level change and assumptions about future warm- individual contributions to sea-level rise or their physi-
ing and the behavior of key geophysical processes. cal constraints. Recent projections of global sea-level
This chapter describes methods for projecting global rise from these different approaches are summarized
and regional sea-level rise, summarizes recent results, below.
and presents the committee's own projections for the
years 2030, 2050, and 2100, relative to year 2000. The Models of Physical Processes
chapter begins with a discussion of the global projec-
tions, then describes how these are adjusted using local The Intergovernmental Panel on Climate Change
and regional information from the U.S. west coast to (IPCC) Fourth Assessment Report projected global
develop projections for California, Oregon, and Wash- sea-level rise to 2100, relative to the year 2000,
ington. The chapter concludes with a discussion of rare using numerical models forced by different emis-
extreme events that could induce a large, rapid change sion scenarios, as well as simplified climate models.
in sea level along the west coast of the United States. The scenarios represent a range of driving forces
and emissions developed using different assumptions
RECENT GLOBAL SEA-LEVEL about demographic, social, economic, technologi-
PROJECTIONS cal, and environmental developments (Box 5.1). The
IPCC (2007) projected the individual contributions
Projections of future global sea-level change are of steric changes and melting of glaciers and ice caps,
commonly made using models of the primary processes the Greenland Ice Sheet, and the Antarctic Ice Sheet
that contribute to global sea-level change--the trans- to future sea-level change for each emission scenario,
fer of fresh water from the melting cryosphere to the then summed the contributions. Changes in water
oceans, and changes in water density (steric changes) stored in other land reservoirs or extracted from the
arising mainly from the thermal expansion of ocean ground or aquifers were considered too uncertain to
water as it warms. Although the steric contribution project. The IPCC (2007) projections are given in
can be computed from ocean models, the cryospheric Table 5.1 and are discussed below.
83
OCR for page 84
84 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
BOX 5.1
IPCC (2000) Emission Scenarios
The IPCC Fourth Assessment Report projected global sea levels over the next 100 years based on 6 families of emission scenarios described in
IPCC (2000). The A1 scenario family assumes high economic growth, low population growth that peaks mid century, and the rapid introduction of more
efficient technologies. Within this family are scenarios designated as A1FI (fossil fuel intensive), A1B (balanced fuel), and A1T (predominantly nonfossil
fuels). The A2 scenario family assumes slower economic growth and technological change, but high population growth. The B1 scenario family assumes
the same low population growth as the A1 scenarios, but a shift toward a lower-emission service and information economy and cleaner technologies.
Finally, the B2 scenario family assumes moderate population growth, intermediate economic growth, and slower and more diverse technological change
than in the B1 and A1 scenarios. The A1FI scenario yields the highest carbon dioxide (CO2) emissions by 2100, and the B1 scenario yields the lowest
CO2 emissions.
TABLE 5.1 IPCC (2007) Projected Contributions to Global Sea-Level Change, Relative to 2000
Projections for 2100 (cm)
Term B1 B2 A1B A1T A2 A1FI
Thermal expansion 1024 1228 1332 1230 1435 1741
Glaciers and ice caps 714 715 815 815 816 817
Greenland Ice Sheet SMB 15 16 18 17 18 212
Antarctica Ice Sheet SMB -10 -2 -11 -2 -12 -2 -12 -2 -12 -3 -14 -3
Sea-level rise 1838 2043 2148 2045 2351 2659
Scaled-up ice sheet discharge 09 011 -113 -113 -113 -117
SOURCE: Adapted from Table 10.7 in Meehl et al. (2007).
NOTE: SMB = surface mass balance.
Steric Contributions in Greenland and Antarctica and empirical models of
the mass balance response of glaciers and ice caps to
The IPCC Fourth Assessment Report used general temperature and precipitation forcing. They projected
circulation models (GCMs) of the ocean and atmo- that glaciers and ice caps would be the largest source of
sphere to estimate global steric response. Because the new water to the oceans throughout the 21st century
GCM simulations were only available for three emis- (Table 5.1). The ice sheets were projected to contrib-
sion scenarios, simplified climate models were used for ute less new water than glaciers and ice caps, mainly
the other three scenarios. Global ocean models com- because the Antarctic contribution was expected to be
pute both temperature and salinity, so their outputs can negative (i.e., mass gained because of increased snow-
be used directly to calculate changes in sea level due to fall would withdraw water from the ocean). However,
thermal expansion (thermosteric changes) and changes recent observations of Antarctica show the opposite--a
in salinity (halosteric changes). Thermosteric contribu- growing Antarctic contribution to sea-level rise due
tions from the ocean general circulation models used to the rapid transfer of ice from land to the ocean by
in the IPCC Fourth Assessment Report are shown in glacier flow and iceberg calving, referred to here as
Figure 5.1 (Meehl et al., 2007). Note that the model "rapid dynamic response" (see "Glaciers, Ice Caps, and
results vary with time and emission scenario. The IPCC Ice Sheets" in Chapter 3).
(2007) projected that thermal expansion would account At the time data were synthesized for the IPCC
for 5569 percent of sea-level rise in 2100 (Table 5.1). Fourth Assessment Report (until mid-2006), rapid
transfers of ice at a global level were only beginning to
Cryospheric Contributions be observed. In addition, the relatively simple treatment
of land ice dynamics in the climate models precluded
The IPCC (2007) estimated the cryosphere re- simulation of rapid dynamics. Although stand-alone
sponse using models of ice sheet surface mass balance
OCR for page 85
PROJECTIONS OF SEA-LEVEL CHANGE 85
FIGURE 5.1 Thermal expansion contribution to global sea-level rise calculated by a range of models for three emission scenarios:
A1B, A2, and B1. SOURCE: Figure 10.31 from Meehl et al. (2007).
ice sheet models with far more sophisticated dynamic and initial conditions or other model parameters are
capabilities have long been in use, they are difficult not available. Consequently, some investigators use
to drive in a realistic fashion with only climatic forc- extrapolation methods to project the cryospheric con-
ing variables, and are still not a feature of integrated tribution to sea-level rise. Extrapolations carry past
atmosphere-ice-ocean models. Consequently, the and present-day observed rates of change forward in
IPCC (2007) treated ice sheets as fixed geographic time at rates that remain constant or vary according to
features that could gain and lose mass through accu- assumed rules.
mulation and ablation, but would not otherwise change A number of recent studies have projected the
size or undergo variations in flow. The IPCC (2007) future contributions of land ice to sea-level change by
attempted to account for rapid transfers of ice from extrapolating observed trends in ice loss rates. Meier
land to ocean by scaling up certain components of the et al. (2007) extrapolated loss rates for the Greenland
modeled results, shown in Table 5.1 under "Scaled-up and Antarctic ice sheets and for aggregate glaciers and
ice sheet discharge." However, the estimates were not ice caps, and estimated that land ice would contribute
based on physical models of ice sheet processes, and ca. 816 cm to sea-level rise by 2050 and 1756 cm
they were not included in the projections of global by 2100 under plausible future conditions. The lower
sea-level rise. The IPCC (2007) projections of the estimate assumed that present-day loss rates continued
cryospheric contribution to sea-level rise are widely unchanged in the future. The higher estimate assumed
regarded as too low (e.g., Kerr, 2007; Pfeffer et al., that the present-day loss rate continued to increase
2008; van der Veen and IMASS, 2010; AMAP, 2011; in the future. Future sea-level rise could be less than
Price et al., 2011). the lower estimate only if global loss rates actually
decreased in the future, an unlikely outcome of most
Extrapolation of Land Ice Contributions climate and mass balance and ice dynamics modeling.
Whether the higher estimate, which was not proposed
As noted above, some aspects of the cryospheric as a firm upper limit, bounds the true upper range of
system are not yet understood well enough to be outcomes is uncertain.
confidently represented in physical models, and many Pfeffer et al. (2008) made extrapolations that
of the observations needed to characterize boundary were intended to constrain the upper limits of glacier
OCR for page 86
86 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
and ice sheet contributions to sea level. Rather than ice sheets may undergo changes in the next century that
project present-day observed rates forward in time, the are quite unlike the changes recorded over the past few
authors calculated what loss rates would be required decades, such as an increase or decrease in the speed
to achieve certain hypothesized future sea-level values. of marine-ending outlet glaciers. Analyses to evaluate
For example, a hypothesized 2 m rise in global sea level the effects of non-stationarity (time-varying processes)
by 2100 from the Greenland Ice Sheet alone would re- and to qualitatively estimate the timescale for which
quire the average velocity of Greenland's outlet glaciers extrapolations are valid are described in the committee's
to immediately rise to 49 km yr-1, which is highly projections of global sea-level rise (see "Cryosphere
unlikely and thus not a plausible future scenario. The Contributions" below).
authors also hypothesized a range of accelerated but
reasonable glacier dynamic behavior for the Greenland Semi-Empirical Models
and Antarctic ice sheets and for glaciers and ice caps,
and they projected land ice contributions ranging from Projections of 21st century sea-level rise are subject
0.8 m to 1.7 m by 2100, with roughly equal contribu- to uncertainties arising from the nonlinear responses
tions from Greenland, Antarctica, and glaciers and of the Greenland and Western Antarctic ice sheets
ice caps. (Pfeffer et al., 2008; Rahmstorf, 2010), steric changes
Using comprehensive data extending back to 1992, (Domingues et al., 2008; Leuliette and Miller, 2009),
Rignot et al. (2011a) constructed detailed mass loss and contributions from mountain glaciers (Meier et al.,
time series for both the Greenland and Antarctic ice 2007). One way to avoid the difficulties of accurately
sheets, and extrapolated linear trends fit to that data estimating these individual contributions is to postulate
to estimate future sea-level contributions from the ice a simple link between observed sea-level rise and ob-
sheets. The observed mass loss trends for 19922009 served global temperature changes in the past (Rahm-
were 21.9 ± 1 GT yr-2 for the Greenland Ice Sheet storf, 2010). Such semi-empirical models are based on
and 14.5 ± 2 GT yr-2 for the Antarctica Ice Sheet. Ex- the simple physical concept that sea level rises faster as
trapolating these loss trends forward to 2100, Rignot the Earth gets warmer. This concept is supported by
et al. (2011a) estimated a sea-level contribution from observations on long timescales.
the ice sheets of 15 ± 2 cm by 2050 and 56 cm (with no Early semi-empirical models assumed a linear
stated uncertainty) by 2100. To arrive at a total land ice relationship between global temperature and sea-
projection, Rignot et al. (2011a) used the glacier and ice level rise (e.g., Gornitz and Lebedeff, 1987), but
cap values calculated by Meier et al. (2007). The uncer- subsequent refinements have included corrections for
tainty attached to the projection reflected the quality of the time-response characteristics of sea level to tem-
fit of the linear regression of the trend to the loss rate perature forcing. A frequently cited semi-empirical
data, rather than uncertainty of the data. The authors model to project future sea-level rise was developed by
suggested that their calculations provide an indication Rahmstorf (2007), who related rising sea level to global
of the potential contributions of ice sheets to sea level in near-surface air temperature as follows:
the next century, but should not be regarded as projec-
tions, given the uncertainty in future acceleration of ice dH/dt = a (T(t) -T0 ),
mass loss and the simplicity of their model.
The extrapolation methods used by Meier et al. where H is the sea level, T is the mean global tempera-
(2007) and Rignot et al. (2011a) assume geostatistical ture, T0 is the baseline temperature at which sea level is
stationarity--that the statistical characteristics dur- stable, and a is the sea-level sensitivity, which measures
ing the period of observation remain valid over the how much the rate of sea-level rise accelerates per unit
period of extrapolation. For unvarying processes or for change in global temperature. The model postulates
short extrapolation periods relative to the observation that if the temperature rises above T0, sea level will
period, this assumption is justifiable. For time-varying rise indefinitely at a rate determined by the magnitude
processes or for long extrapolation periods, this as- of the temperature rise, so a linear rise in temperature
sumption is more questionable. Glaciers, ice caps, and with time leads to a quadratic change in sea level. The
OCR for page 87
PROJECTIONS OF SEA-LEVEL CHANGE 87
unknown parameters a and T0 are determined from Grinsted et al. (2009) used a much longer tem-
global sea-level reconstructions (e.g., Church and perature record and a different semi-empirical model
White, 2006) and global temperature data archived to project sea-level rise:
by the National Aeronautics and Space Administra-
tion Goddard Institute for Space Studies. Rahmstorf dS/dt = (Seq - S)/ ,
(2007) found that the parameter a is 3.4 mm yr-1/°C.
Projecting the equation forward using the IPCC (2000) where is the response time on the order of centuries,
scenarios for temperature change yielded a rise in sea Seq is the equilibrium sea level at a fixed global tem-
level between 0.38 m and 1.2 m by 2100. perature, and S is the global mean sea level relative to
A subsequent revision to the model (Vermeer and the mean over a well-documented time interval. Seq is
Rahmstorf, 2009) included an extra term b to allow sea assumed to change linearly with temperature. As the
level to respond directly to temperature change: atmospheric temperature rises, the sea level rises at a
rate that depends both on the magnitude of the total
dS/dt = a (T -T0 ) + b dT/dt. warming (which determines Seq - S) and the response
time . Grinsted et al. (2009) calibrated their equation
To gain confidence in the model, the authors using several historical global temperature data sets,
calibrated the a and b coefficients with temperature then used the IPCC (2000) scenarios to project into
data from 1880 to 2000, then verified the model over the future. For all IPCC (2000) emission scenarios, they
a 1,000-year time frame using sea-level proxy data for projected that sea level would rise between 0.21 m and
the past millennium. With this model and the IPCC 2.15 m by 2100.
(2000) emission scenarios, Vermeer and Rahmstorf All of the semi-empirical model projections are
(2009) projected that sea level would rise between higher than the IPCC (2007) projections by a factor
0.81 m and 1.79 m by 2100. Their projections for three of two or even three (Rahmstorf, 2010; Figure 5.2).
of the IPCC (2000) emission scenarios are shown in The two projection approaches rest on different
Figure 5.2. foundations--GCMs on the physical processes that
FIGURE 5.2 Projections of sea-level rise from 1990 to 2100, based on the Vermeer and Rahmstorf (2009) semi-empirical model
and three IPCC (2000) emission scenarios (A1FI, A2, and B1). Uncertainty ranges are 1 standard deviation from the model means,
and the gray shading is an added ± 7 percent, representing uncertainty in the fit of the data. The corresponding sea-level projections
by IPCC (2007; labeled AR4) are shown for comparison in the bars on the bottom right. Also shown are the observations of annual
global sea level (red line). SOURCE: Vermeer and Rahmstorf (2009).
OCR for page 88
88 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
cause sea level to rise and semi-empirical models on The projections are for individual years (2030, 2050,
the observed relationship between temperature and and 2100, relative to 2000), and were derived using
sea level--so it is not surprising that they do not agree. single-year values from low-order curves, except for
Moreover, the IPCC (2007) projections are likely the steric values, as explained below. The projections
underestimates because they do not account fully for are given in Table 5.2 and discussed below.
cryospheric processes. The highest projections made by
semi-empirical models (more than 2 m of sea-level rise) Steric Contribution
are likely overestimates because they would require un-
realistically rapid acceleration of glaciological processes The most recent GCM results for the steric contri-
(Pfeffer et al., 2008). bution that were available to the committee were from
An advantage of semi-empirical models is that, the Coupled Model Intercomparison Project Phase
by parameter fitting, they reproduce the observed past 3 (CMIP3), which were used in the IPCC Fourth
sea-level rise. However, the simple empirical connec- Assessment Report. Although outputs from a new
tion found for the past may not hold in the future. In generation of GCMs are beginning to be available, per-
particular, the ice sheets appear to have been negligible forming computations of derived quantities like global
sea-level contributors during the observational periods sea-level changes from these new outputs is beyond the
used by Gornitz and Lebedeff (1987), Rahmstorf charge and capability of the committee. Consequently,
(2007), and Vermeer and Rahmstorf (2009), but ice the committee drew on the work of Pardaens et al.
sheet dynamic response is widely regarded as the most (2010), who analyzed an ensemble of IPCC (2007)
uncertain aspect of sea-level change. Indeed, some model projections using the A1B emission scenario
events, such as ice shelf melting triggering an instability (Figure 5.3). Drs. Pardaens and Gregory1 provided the
of the West Antarctic Ice Sheet, would not be factored gridded annual mean sea-level data used in their paper,
into semi-empirical models. and the committee analyzed the combine steric and
ocean dynamic height data for the globe.
COMMITTEE PROJECTIONS OF GLOBAL The models in Pardaens et al. (2010) yielded time
SEA-LEVEL RISE series of annual mean sea level spanning roughly the
21st century: the first year in the various model simu-
The committee was charged with projecting both lations ranged from 2000 to 2004, and the final year
the individual contributions to global sea-level rise was 2099. The committee performed a quadratic fit on
(e.g., thermal expansion, melting of land ice) and the each model's time series at each grid point and, using
total global sea-level rise for the years 2030, 2050, and the values on the quadratic curves, obtained steric sea-
2100 (Task 1, see Box 1.1). Given the state of knowl- level changes for 2030, 2050, and 2100 relative to year
edge and the limited time and computational capability 2000 for each model. The results are presented in the
available for a National Research Council study, the first row of Table 5.2.
committee chose a combination of approaches for its
projections. The output of GCMs was used to project Uncertainties
the steric contribution (primarily thermal expansion)
to global sea-level rise over the three time frames. For The committee endeavored to incorporate and
the land ice projections, the committee extrapolated describe as accurately as possible the known sources of
mass balance estimates. Like the IPCC (2007), the uncertainty in the steric projections. These uncertain-
committee did not project land hydrology contribu- ties are related to future greenhouse gas and aerosol
tions because uncertainties are too large, and a recent emissions and concentrations (human forcing), the
comprehensive assessment (Milly et al., 2010) found response of global temperatures to human forcing, and
that the primary sources (groundwater depletion) and the response of the ocean to those global temperature
sinks (reservoir storage) appear to effectively cancel out. distributions. The IPCC (2007) treated uncertainty in
The individual components were then summed and
compared with results from semi-empirical methods. 1 See .
OCR for page 89
PROJECTIONS OF SEA-LEVEL CHANGE 89
TABLE 5.2 Committee's Global Sea-Level Rise Projections (in cm) Relative to Year 2000
2030 2050 2100
Term Projection Range Projection Range Projection Range
Sterica 5.4 ± 1.6 1.711.0 9.9 ± 2.4 4.018.9 24.2 ± 5.9 9.646.2
(B1A1FI) (B1A1FI) (B1A1FI)
Glaciers and ice capsb 2.9 ± 0.1 2.73.6 5.5 ± 0.2 5.17.3 14.3 ± 0.7 12.919.4
Greenlandb 2.3 ± 0.2 1.84.0 5.6 ± 0.7 4.310.2 20.1 ± 2.7 14.833.8
Antarcticab 2.9 ± 0.7 1.55.1 7.0 ± 2.1 3.013.3 24.0 ± 8.3 7.746.2
Total Cryosphereb 8.1 ± 0.8 6.612.2 18.0 ± 2.2 13.729.4 58.4 ± 8.8 40.994.1
Sumc 13.5 ± 1.8 8.323.2 28.0 ± 3.2 17.648.2 82.7 ± 10.6 50.4140.2
Semi-empiricald 18 1422 37 2847 121 78175
(B1A1FI) (B1A1FI) (B1A1FI)
a For the steric contribution, the projection is for scenario A1B from Pardaens et al. (2010), ±1 standard deviation computed for 20-year windows across
models, and the range was determined by scaling the A1B projections for 2100 to the low value of B1 and the high value of A1FI for A1B, from Table 5.1.
b The cryospheric projection is an extrapolation from observed changes, ±1 standard deviation. The range column includes an additional dynamic contribution,
described in Appendix E, which is used only for the high-end estimates.
c The low value of the range for each year (2030, 2050, 2100) was computed by subtracting twice the standard deviation from the mean in the projection
column, and adjusting to the difference between A1B and B1. The high value of the range was computed by adding twice the standard deviation to the mean,
adjusting to the difference between A1FI and A1B, and adding the dynamical imbalance contribution.
d Data from Vermeer and Rahmstorf (2009).
A. K. Pardaens et al.: A model study of factors influencing projected changes in regional sea level
GISS-ER GFDL-CM2.0 GFDL-CM2.1 CGCM3.1(T47)
GISS-AOM ECHO-G MRI-CGCM2.3.2 UKMO-HadCM3
GISS-EH MIROC3.2(medres) BCCR-BCM2.0 FGOALS-g1.0
IPSL-CM4
-0.24 -0.12 0 0.12 0.24
FIGURE Fig.5.3
1 SeaCombined steric
level changes and wind-driven
(19801999 to 20802099,sea-level changes (19801999
units of meters) to 20802099,
applying a high-pass unitswith
Chebyshev filter in m) for the indicated
a half-power models,
cut-off of
relativefor
toeach
eachof model's
the AR4_13global mean.
ensemble The overlying
of models, relative tocontour lines are
each model's of the
about 165 sea-level distribution
years). A Student- in the
t test (with baseline
degrees control
of freedom fromsimulations,
lag-
averaged over
global a 120-year
mean. period
The overlying (contours
contour lines are
are every
of the0.2sea m). SOURCE:
level Pardaens et
1 autocorrelation) al.applied
was (2010). to obtain the significance level. The
distribution in the baseline control simulations, averaged over the unforced variability was removed from the control simulation point-
period parallel to the 19802099 projection period (contours are every by-point (using the filter described above) before calculating and
0.2 m, thick countours for zero and positive deviations; thin contours removing model drift, so as to better determine significance of the
for negative deviations). Changes shown above the 95% significance projected changes. All model data on grids as provided to CMIP3
level given by unforced variability, as determined from the control database. A version of this figure including changes below signifi-
simulation (using 20 year averages, model drift removed by first cance level is shown in Online Resource 1
OCR for page 90
90 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
human forcing by calculating results for six emission continuous records. Dyurgerov and Meier (2005) used
scenarios (Box 5.1, Table 5.1). The most common known area and area-altitude distributions by region
approach for treating uncertainty in the global tem- to scale up limited point mass balance observations.
perature response is to use the results of many climate Their analysis considered surface mass balance changes
models, which also provides a range of projected values directly modulated by climate and excluded losses by
for the global ocean response. The IPCC (2007) global calving. Dyurgerov (2010) reevaluated the data used in
model simulations for the thermal expansion and Dyurgerov and Meier (2005) and made significant cor-
dynamical components are available for only the B1, rections to changing glacier areas during the period of
A1B, and A2 emission scenarios, and Pardaens et al. observations. Cogley (2009) presented an independent
(2010) performed their calculations for only the A1B data set, evaluated in 5-year increments from ca. 1850
scenario. To provide a range of projections for all six to 2009, that includes both climatically forced and
scenarios, the committee used the ratios of thermal calving losses. The data for glaciers and ice caps were
expansion projections from Table 5.1. For example, averaged using techniques that weight the data accord-
the global projection for the low value of B1 in 2030 ing to its quality as measured by the magnitude of the
was computed from the digital values in Pardaens et stated uncertainty (see Appendix E for details).
al. (2010) using the quadratic fit for 2030 as described The base-rate extrapolation assumes that present-
above. To determine the range, this value was multi- day observed trends in loss rates continue in the future.
plied by the ratio of the low value of the thermal ex- To investigate the effect of varying rapid dynamic dis-
pansion term in 2100 for B1 to the low value for A1B charge on these projections, the committee performed
(0.10/0.13). This approach slightly underestimates the model experiments to calculate the effects of both
B1 values for 2030 and 2050 because sea-level change acceleration and deceleration in ice discharge relative
under the B1 scenario is fairly linear (see Figure 5.1), to observed present-day rates. Both possibilities have
but the difference is estimated to be within rounding been examined in the literature. For example, Pfeffer
error (a few mm). et al. (2008) discussed the consequences of large-scale
losses from both Greenland and Antarctica in hypo-
Cryosphere Contribution thetical terms, and Rignot et al. (2011a) projected
a large dynamic contribution to sea-level rise from
The committee projected the cryosphere contribu- the ice sheets on the basis of past observations. On the
tion to global sea-level rise using adaptations of the other hand, observations in Greenland (e.g., Moon et
Meier et al. (2007) extrapolation techniques and the al., 2012) show that recently active outlet glaciers are
Pfeffer et al. (2008) methods for evaluating uncertainty slowing down, suggesting that rapid dynamics may
and establishing projection boundaries. The commit- have an episodic or periodic nature and that future
tee's extrapolations were based on selected observa- increases in sea level from rapid dynamics may not be
tional data for glaciers, ice caps, and the Greenland as dramatic as have been postulated elsewhere. Price
and A ntarctic ice sheets. The most comprehensive time et al. (2011) used a high-order numerical model to
series of mass loss of the Greenland and Antarctic ice explore the effect of outlet glacier dynamics, their in-
sheets is the Rignot et al. (2011a) compilation, which fluence on upstream ice dynamics, and time variations
combines modeled surface mass balance and measured in outlet glacier dynamics on future losses from the
and modeled ice discharge to produce net balances Greenland Ice Sheet.
for both ice sheets for 19922010, the earliest date Increased ice discharge beyond presently observed
from which continuous observations are available. rates was simulated by extrapolating a multiple of
For glaciers and ice caps, the committee used data present-day observed discharge forward in time to
from Dyurgerov and Meier (2005), Cogley (2009), and 2100 (see Appendix E). For glaciers and ice caps, an
Dyurgerov (2010). At the time this report was being increment of flux equal to 50 percent of the present-
written, Dyurgerov and Meier's (2005) mass balance day discharge was added, equivalent to 162.4 GT yr-1.
data from glaciers and ice caps for the 19602005 For Greenland, the average speed of all outlet glaciers
period were the most recent global compilation of was increased by 2 km yr-1, equivalent to a net dis-
OCR for page 91
PROJECTIONS OF SEA-LEVEL CHANGE 91
charge of 375.1 GT yr-1. These values are consistent tions. Among the most important reasons for this
with the observed doubling of Greenland's mass increase are the following:
balance deficit between ca. 1996 and 2000 ( Rignot
and Kanagaratnam, 2006). For Antarctica, the net 1.Observed rates of loss from the ice sheets have
outlet flux was doubled from its ca. 2006 value to accelerated significantly since the IPCC Fourth Assess-
264 GT yr-1. All values were increased linearly over ment Report was finalized. Prior to 2004, published
20 years and held constant thereafter. The exact choice mass balances for the ice sheets were near zero or even
of values for the individual components is less impor- negative, but subsequent work indicates that loss rates
tant than the net added flux after the 20-year increase, are rapidly accelerating (see Chapter 3). Thus, the
which is approximately 800 GT yr-1 (2.2 mm yr-1 present-day loss rates from the ice sheets constitute
sea-level equivalent). significantly different initial conditions than were ap-
Decreased ice discharge was simulated by reduc- plied in the IPCC (2007) model calculations.
ing the projected output of the Greenland Ice Sheet 2. The extrapolation method gives more weight to
by 25 percent from its projected base value. Currently, recent observations than to past observations (Appen-
about 50 percent of Greenland's ice loss rate is caused dix E). Thus the high present-day observed loss rates
by iceberg calving; a hypothetical 50 percent reduction have a larger effect on extrapolations than on model
in calving discharge yields a 25 percent reduction in calculations.
the total ice loss rate. Other cryosphere terms were left 3.Rapid dynamic response was hypothesized as
unchanged. For Antarctica, systems likely to experi- significant in the IPCC (2007) analyses, but was incor-
ence rapid change are concentrated on the Amundson porated at only a rudimentary level in the projections. In
Coast, and there are no known geographic features in the committee's analysis, added dynamics can account
the region that would likely serve as points of stabili- for 26 percent to 58 percent of total sea-level rise.
zation. Moreover, there is no reason to think that the
dynamic slowdown seen recently in Greenland is likely Even accounting for the possibility of slowing
to occur soon in Antarctica. Given the larger size of discharge in Greenland, the committee's cryosphere
the Amundson Coast outlet glaciers, it is reasonable extrapolations are substantially higher than the IPCC
to hypothesize that any reversals will occur on longer (2007) cryosphere projections. A 25 percent reduction
time scales than the committee's projections. For gla- in the Greenland dynamic discharge lowers the com-
ciers and ice caps, future discharge was left unchanged mittee's sea-level projections by 6 percent for 2100 (see
from the base-rate projection in this experiment. The Table E.4 in Appendix E). This result is not surprising,
fraction of glacier and ice cap loss from calving dis- given the fraction of Greenland's contribution to global
charge is unknown, but is probably less than 50 percent. sea-level rise. If calving is responsible for 50 percent of
Thus, the committee assumed that direct climatically- Greenland's ice loss rate, or about 10 percent of total
forced surface mass balance is the primary control on global sea-level rise, then halving the amount of calv-
future changes in the loss rate of glaciers and ice caps. ing should affect sea level at about the 5 percent level.
The variations listed above were intended to cap- The committee's extrapolations also are higher
ture the general magnitude of plausible changes in ice than recent numerical model projections. For example,
dynamics. Although these exact events may not occur, Price et al. (2011) simulated the net dynamic losses
the calculations provide a means to develop a quanti- from the Jakobshavn, Helheim, and Kangerdlugssuaq
tative, albeit crude, estimate of the influence of rapid glaciers, including their effects on the interior of the
glacier dynamics on sea-level rise. ice sheet, to 2100, and then scaled up that response
to the entire Greenland Ice Sheet. They projected a
Results dynamic sea-level contribution from Greenland of
5.8 ± 2.1 mm SLE by 2100, regarding it as a lower
The results of the extrapolation are presented in bound. By comparison, the committee's projection for
Table 5.2. All of the cryosphere extrapolations to 2100 Greenland is 20.1 ± 2.7 cm SLE, which includes both
are higher than the IPCC (2007) cryosphere projec- surface mass balance and dynamic contributions. If the
OCR for page 92
92 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
dynamic response to climate change constitutes half of response of outlet glaciers, such as the modeling study
Greenland's recent ice loss rate, then the Greenland of Price et al. (2011), may lead to constraints on the
dynamic contribution to sea-level rise is about 10 cm timescales of rapid dynamic response.
SLE by 2100, more than an order of magnitude higher
than the Price et al. (2011) projection. Even if dynamics Discussion of Global Projections
constitutes only 13 percent of the Greenland's recent
ice loss rate, as estimated by Price et al. (2011), the The committee's projections of global sea-level
Greenland dynamic contribution projected by the com- rise are summarized in Table 5.2 and illustrated in
mittee would be 2.6 cm, greater than the Price et al. Figures 5.4 and 5.5. For the three projections periods
(2011) projection by a factor of 4. The two projections (2030, 2050, and 2100), the committee provides a
differ in part because of simplifications and uncertain- projection and a range, which attempt to incorporate
ties in both approaches, underscoring the need for more the various sources of uncertainty discussed above and
complete knowledge of processes and more complete to provide guidance on possible outcomes. The projec-
information about initial and boundary conditions. tion for the steric component is derived from the A1B
emission scenario, which was used in the Pardaens et
Uncertainty al. (2010) analysis, and the range is the corresponding
value for the lowest emission scenario (B1) and the
The cryosphere projections presented here have highest emission scenario (A1FI). Extrapolations are
two types of uncertainties: quantified uncertainty and based on observations and thus take no account of
unquantified uncertainty. The quantified uncertainty, emission scenarios. For the cryospheric component, the
which is calculated from the 595 percent projec- projection is the extrapolation from observed changes
tion intervals (Appendix E) then converted to 1 and the range includes a possible additional dynamic
uncertainties, is a statistical product representing the contribution. No formal probability analysis of the
uncertainty of the curve fitting process. The unquan-
individual contributors of uncertainty was performed,
tified uncertainty is associated with the assumption so the projections are not necessarily the likeliest out-
that past system behavior is a good predictor of future comes, and the ranges are not the highest or lowest
system behavior. Rapid dynamic response may play a possibilities.
different role in future sea-level rise than it did during The committee's projected contributions of the
the period of observations, making that period poten- steric and cryospheric components of future sea-level
tially a poor predictor of future system behavior. How- rise are illustrated in Figure 5.4. The steric component,
ever, deviations of actual sea-level rise from the simple which the IPCC (2007) projected as substantially larger
extrapolation will take time to emerge. Extrapolation than the cryospheric component (Table 5.1), is roughly
of unstable or unpredictable dynamics will thus be similar in magnitude to the cryospheric component
reliable initially, but the errors may increase dramati- for the first few decades. By mid-century, however,
cally as the timescale of the extrapolation exceeds the the cryospheric component greatly exceeds the steric
characteristic timescale of the dynamics. component for all GCM simulations. The steric pro-
In theory, the uncertainty of the extrapolations jection of the various models ranges from 15 cm to
could be evaluated by determining the characteristic almost 40 cm in 2100, relative to 2000, with a model
timescale of rapid dynamic response of vulnerable land- average of 24 cm. The cryospheric extrapolation ranges
based ice. The timescale for dynamics of individual from about 50 cm to 67 cm in 2100, and ice dynamics
outlet glacier systems is thought to be decades, whereas would add 18 cm.
the timescale of aggregate outlet glacier systems, such Figure 5.5 shows the range of projections of global
as the marine-ending glaciers draining the Greenland sea-level rise. For the projection (middle line), the steric
Ice Sheet, may be a century or longer. This timescale estimate for the all-model average was added to the
has not been established, however, and contributes central value of the cryospheric extrapolation. The low
uncertainties that are both quantifiable and unquantifi- estimate was derived by subtracting twice the standard
able. New work on the time-varying aspects of dynamic deviation of the steric values (shown in the first row of
OCR for page 93
PROJECTIONS OF SEA-LEVEL CHANGE 93
FIGURE 5.4 Committee projections of individual components of global sea-level rise. The colored lines are the steric contributions
from various models with the A1B emission scenario, and the heavy black line (labeled "all model") is the model average. Gray shad-
ing is the cryospheric contribution, and the black line within the gray swath is the cryosphere average. The black line at the bottom
is the added ice dynamics component.
FIGURE 5.5 Range of committee projections for the sum of all individual components of global sea-level rise.
OCR for page 98
98 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
unadjusted extrapolations is given in Table 5.4. The tion models and geophysical models, respectively. The
gravitational and deformational effects reduce the cryo- total vertical land motion, including tectonics, GIA,
spheric contribution to relative sea-level rise projected compaction, and fluid withdrawal and recharge, can be
for 2100 by 721 percent along the north coast and by projected using continuous GPS (CGPS), assuming
112 percent along the central coast. Along the south that the vertical land motion is predominantly secular
coast, these effects can increase the cryospheric contri- within ~100 years. Results from these three projection
bution to sea-level rise by up to 2 percent or decrease methods are summarized in Table 5.5 and discussed
it by up to 6 percent. below.
The projection assumes that the fingerprint scale
factors remain constant for the small reduction in land Tectonic Projections
ice volume expected by 2100, which is likely reasonable
for the next several decades. Assuming that the sea-level Tectonics causes significant vertical land motion
fingerprint is correct, uncertainties in the calculation are along the coast above the Cascadia Subduction Zone.
associated with the ice loss rate. When uncertainties in The tectonic component of vertical land motion in
the ice loss rate are factored in, the fingerprint-adjusted this area can be projected using earthquake cycle de-
contribution of Alaska, Greenland, and Antarctica to formation models. The CAS3D-2 model (He et al.,
relative sea-level rise ranges from 3375 cm along the 2003; Wang et al., 2003; Wang, 2007), which is the
north coast and 3976 cm along the south coast for most sophisticated and complete model of earthquake
2100 (Table 5.3). deformation along the Cascadia Subduction Zone, is
the only model that has been used to make forward
Vertical Land Motion projections. Like other tectonic models, it is limited by
incomplete knowledge of the temporal behavior of the
Major causes of vertical land motion along the earthquake process, such as the degree of periodicity of
west coast of the United States include tectonics, gla- the Cascadia earthquake cycle, which adds uncertain-
cial isostatic adjustment (GIA), and subsidence as a ties that are difficult to quantify. The model excludes
result of sediment compaction and/or fluid withdrawal vertical land motion from glacial isostatic adjustment.
(see "Vertical Land Motion" in Chapter 4). The verti- The projected rates of interseismic deformation
cal land motion signal in Oregon, Washington, and for the Cascadia Subduction Zone from the CAS3D-2
northern California is dominated by regional tectonics model are given in Table 5.5. The projections suggest
associated with the Cascadia Subduction Zone. In that coastal sites, which are closest to the offshore sub-
California south of Cape Mendocino, vertical land duction boundary, will undergo uplift, whereas more
motion depends on varying combinations of GIA, inland locations (Anacortes and Seattle) will undergo
sediment compaction, fluid withdrawal or recharge, subsidence. Projected vertical land motions for the
and local compressional tectonics that may or may not coastal sites range from -1.0 mm yr-1 (subsidence) to
be related to the San Andreas Fault. Projections of the +3 mm yr-1 (uplift), with most rates varying by less
regional tectonic and GIA components of vertical land than 0.2 mm yr-1 over the 21st century. The vertical
motion can be made using earthquake cycle deforma- land motions projected using the CAS3D-2 model
TABLE 5.4 Effect of Sea-Level Fingerprints of Alaska, Greenland, and Antarctica Ice Masses Expressed as Percentage
Differences from Cryosphere Projections with No Fingerprint Effecta
North Coast (Neah Bay) Central Coast (Eureka) South Coast (Santa Barbara)
Year Low Central High Low Central High Low Central High
2030 -19.9% -13.4% -9.6% -10.4% -5.8% -3.1% -4.9% -1.6% 0.4%
2050 -20.3% -12.2% -8.6% -10.6% -4.8% -2.2% -5.1% -0.8% 1.2%
2100 -21.1% -10.8% -7.4% -12.3% -3.7% -1.2% -5.5% 0.2% 2.1%
a Uncertainties, expressed as low-high values, were derived from the spread of projected contributions from Alaska, Greenland, and Antarctica.
OCR for page 99
PROJECTIONS OF SEA-LEVEL CHANGE 99
TABLE 5.5 Vertical Land Motion Projections for the Two Tectonic Regimes
Tectonic Component GIA Component Total Vertical Land Motion
(mm yr-1)a (mm yr-1)b (mm yr-1)c
Location Latitude Longitude 20102030 20302050 20502100 20102100 20102100
Cascadia Subduction Zone
Cherry Point, WA 48.87 -122.75 0.2 ± 0.4 1.0 ± 1.5
Anacortes, WA 48.56 -122.64 -0.9 -0.9 -1.0 -0.1 ± 0.5 1.0 ± 1.5
Seattle, WA 47.85 -122.73 -0.6 -0.6 -0.6 -0.5 ± 0.4 1.0 ± 1.5
Long Beach, WA 46.58 -123.83 1.9 1.8 1.7 -1.1 ± 0.5 1.0 ± 1.5
Pacific City, OR 45.38 -123.94 1.7 1.6 1.5 -1.0 ± 0.4 1.0 ± 1.5
Waldport, OR 44.42 -124.02 1.7 1.6 1.5 -0.9 ± 0.3 1.0 ± 1.5
Coos Bay, OR 43.36 -124.30 2.3 2.2 2.1 -0.9 ± 0.3 1.0 ± 1.5
Eureka, CA 40.87 -124.15 3.0 2.8 2.6 -0.7 ± 0.3 1.0 ± 1.5
San Andreas Fault Zone
Point Reyes, CA 38.00 -122.98 -0.5 ± 0.3 -1.5 ± 1.3
San Francisco, CA 37.80 -122.47 -0.5 ± 0.3 -1.5 ± 1.3
Monterey, CA 36.60 -121.88 -0.5 ± 0.3 -1.5 ± 1.3
Port San Luis, CA 35.17 -120.75 -0.5 ± 0.3 -1.5 ± 1.3
Santa Monica, CA 34.02 -118.50 -0.3 ± 0.3 -1.5 ± 1.3
Los Angeles, CA 33.72 -118.27 -0.4 ± 0.3 -1.5 ± 1.3
San Diego, CA 32.72 -117.17 -0.4 ± 0.3 -1.5 ± 1.3
NOTE: Positive rates denote uplift and negative rates denote subsidence.
a Rates provided by Kelin Wang, Geological Survey of Canada, using the CAS3D-2 model described in Chapter 4.
b Rates were averaged from an ensemble of 16 GIA models (see Table 4.3) and are represented so positive GIA means falling relative sea level.
c Rates (± 1 standard deviation) were determined from CGPS data from the Scripps Orbit and Permanent Array Center taken within 15 km of the coast; see
Table A.1.
generally agree with rates determined from leveling GPS Projections
(Burgette et al., 2009) and from GPS (Mazzotti et al.,
2008; this report). The total vertical land motion, including signals
from tectonics, GIA, sediment compaction, and/or
fluid withdrawal or recharge, is recorded in GPS data.
GIA Projections
Consequently, the committee used CGPS data in its
Projections of the GIA component of vertical land projections of sea-level rise for 2030, 2050, and 2100. It
motion were made using an ensemble of 16 models. would be attractive to use the relatively densely-spaced
The projections show subsidence at all locations except CGPS vertical land motion data to make projections
for northernmost Washington, which shows negli- at high spatial resolution along the coast. However,
gible uplift (Table 5.5). Mean GIA model predicted vertical land motions can vary at length scales that are
rates of vertical land motion range from +0.2 mm yr-1 considerably smaller than the CGPS station spacing, so
in northernmost Washington to -1.0 mm yr-1 in interpolation using the CGPS data carries substantial
southern Washington and northern Oregon. This risk of spatial aliasing. Moreover, the data exhibit sig-
strong latitudinal gradient illustrates the importance nificant scatter because of local sediment compaction
of GIA in regions underneath or at the margins of and/or fluid withdrawal (see Figure 4.14b and associ-
the extinct Laurentide ice sheet. In southern Oregon ated discussion). Consequently, the committee chose
and C alifornia, mean rates are generally between the conservative approach of projecting vertical land
-0.4 mm yr-1 and -1.0 mm yr-1. Given the slow pace motion for the two tectonic regions--Cascadia and the
of glacial isostatic adjustment, these rates are assumed San Andreas region--and characterizing them using
constant for the three projection periods (2030, 2050, simple statistics (mean and 1 standard deviation). With
and 2100). obvious outliers removed, the current rates of vertical
OCR for page 100
100 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
land motion for these regions are 1.0 ± 1.5 mm yr-1 for and Antarctica; the local steric and dynamical ocean
Cascadia and -1.5 ± 1.3 mm yr-1 for the San Andreas component; and the vertical land motion component.
area. The values used for the projections appear in Table 5.3
In using the current rates of vertical land mo- and are illustrated in Figure 5.8. The cryosphere is the
tion in its projections, the committee assumed that only component with pronounced upward curvature
the CGPS spatial pattern and rates in the two tec- (acceleration) over the 21st century. Ice mass loss rates
tonic regions would remain constant for 2030, 2050, for Alaska, Greenland, and Antarctica were adjusted
and 2100. This assumption is supported by leveling for gravitational and deformational effects and then
data in California (Appendix D) and in Washington added to loss rates from other glaciers and ice caps.
and Oregon ( Burgette et al., 2009). In addition, the The sum was then extrapolated forward. The steric
CAS3D-2 tectonic model suggests that, in the absence and dynamical ocean components (blue swath in Fig-
of a great earthquake, the general vertical land motion ure 5.8) were extracted from the ocean data provided
pattern or trend in Cascadia will not change signifi- by Pardaens et al. (2010), averaged for the west coast,
cantly in the coming century. The projected rates of then smoothed for plotting using locally weighted
vertical land motion are given in Table 5.3. regression. The vertical land motion components and
their uncertainties for the northern and southern part
Discussion of Regional Projections of the coast are shown in the shaded areas; the bars on
the right margin indicate the range for 2100. North of
The projections of sea-level rise off California, Cape Mendocino, the coast is experiencing mean uplift,
Oregon, and Washington were made by summing so vertical land motion contributes negatively to rela-
the cryosphere component, adjusted for the effects tive sea-level rise (although uncertainties are large and
of the sea-level fingerprints of Alaska, Greenland, include positive contributions), whereas the coast south
FIGURE 5.8 Committee projections of components of sea-level rise off California, Oregon, and Washington. The blue band rep-
resents the model results for combined global steric and local dynamical sea-level change, averaged between 32° and 49° latitude,
from 13 GCMs. Light gray shading in the middle of the figure shows estimated effects of vertical land movement in the San Andreas
region (VLM S), and dark gray shading at the bottom of the figure shows the vertical land movements for Cascadia (VLM N). Light gray
shading at the top of the figure shows the global cryosphere, including added ice dynamics. The red line is the effect of the sea-level
fingerprint of ice melt from the Alaska, Greenland, and Antarctica sources, shown for the north coast (49°N). The fingerprint effect is
subtracted from the global cryosphere.
OCR for page 101
PROJECTIONS OF SEA-LEVEL CHANGE 101
of Cape Mendocino is experiencing mean subsidence, the models, and the need to make assumptions about
so vertical land motion contributes positively to relative future conditions (e.g., population growth, technologi-
sea-level rise. cal developments, large volcanic eruptions) that drive
Figure 5.9 shows the total regional sea level pro- the climate system. Although a systematic analysis of
jected for the years 2030, 2050, and 2100, relative to these uncertainties was beyond the ability of the com-
year 2000, for a transect along the west coast. The shape mittee, this report attempts to describe and combine
of the curve is dominated by the change in vertical land the most important uncertainties. For the committee's
motion at about 40° latitude from uplift in the north global sea-level rise projections, important uncertain-
to subsidence in the south. The sea-level fingerprint ties are associated with assumptions about the growth
effect reduces the projected sea levels along the entire of concentrations of greenhouse gases and sulfate
coast and is most pronounced in Washington. The aerosol, which affect the steric contribution, and future
fingerprint effect has not been included in previous ice loss rates and the effect of rapid dynamic response,
studies and projections of sea level for the west coast which affect the land ice contribution. Additional,
(e.g., Mote et al., 2008; Cayan et al., 2009; Tebaldi et unquantified uncertainties arise from neglecting the
al., 2012). The ocean components have little effect on terrestrial water component in the projections and from
the north-south gradient in projected sea-level change. combining model-projected steric contributions with
The committee's projections for the west coast of extrapolation-projected land ice contributions (e.g.,
the United States are significantly different from global model projections account for future emissions whereas
projections (Figure 5.10). The difference is largest off extrapolations do not).
the Washington coast, where sea-level fingerprint ef- Regional projections carry additional uncertainties
fects lower the height of the ocean surface and regional because more components are included and some com-
tectonics raises the height of the land surface, resulting ponents are estimated from global scale analyses. The
in rates of relative sea-level rise that are substantially uncertainties are larger for the committee's projections
lower than the global mean. Off the California coast, for California, Oregon, and Washington than they are
where subsidence is lowering the land surface, the for the global projections, primarily because uncer-
projected relative sea-level rise is slightly higher than tainties in the steric component are larger at smaller
the global mean. The committee's projected values for spatial scales and because some of the additional com-
California are somewhat lower than the Vermeer and ponents (e.g., vertical land motion) have relatively large
Rahmstorf (2009) projections, which are being used by uncertainties.
California state agencies on an interim basis for coastal For both global and regional projections of sea-
planning (CO-CAT, 2010). For California and Wash- level rise, uncertainties grow as the projection period
ington, the committee's projections fall within the range increases because the chances of the observations and
presented in Cayan et al. (2009) and Mote et al. (2008), models deviating from actual climate changes increases.
respectively. The committee's projected values for 2030 Currently, all projection methods--including process-
and 2050 also are comparable to those of Tebaldi et al., based numerical models, extrapolations, and semi-
(2012), although the committee found a larger north- empirical methods--have large uncertainties at 2100.
south difference in the magnitude of sea-level rise. Although the actual value of sea-level rise will almost
surely fall somewhere within these wide uncertainty
UNCERTAINTY bounds, confidence in specifying the exact value is rela-
tively low. At short timescales, the models more closely
Projections of future sea-level rise carry numerous represent the future climate system, so uncertainties
sources of uncertainty. This uncertainty arises from an are smaller and confidence is higher. Confidence in the
incomplete understanding of the global climate system, committee's projections is likely to be highest in 2030
the inability of global climate models to accu rately and perhaps 2050, which are likely of greatest inter-
represent all important components of the climate est to coastal planners, engineers, and other decision
system at global or regional scales, a shortage of data makers tasked with planning for sea-level rise along
at the temporal and spatial scales necessary to constrain the west coast of the United States.
OCR for page 102
102 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
-124° -120° -116°
2030 2050 48°
2100
48° Seattle 48°
Cascadia Subduction Zone
Newport
44° 44°
40° 40°
MTJ Sa
San
nA
Francisco
nd
re
as
36°
Fa
36°
ul
t
Los
Angeles
32° 32°
0 50 100 150
cumulative SLR (cm)
FIGURE 5.9 Projected sea-level rise off California, Oregon, and Washington for 2030 (blue), 2050 (green), and 2100 (pink), rela-
tive to 2000, as a function of latitude. Solid lines are the projections and shaded areas are the ranges. Ranges overlap, as indicated
by the brown shading (low end of 2100 range and high end of 2050 range) and blue-green shading (low end of 2050 range and
high end of 2030 range). MTJ = Mendocino Triple Junction, where the San Andreas Fault meets the Cascadia Subduction Zone.
OCR for page 103
PROJECTIONS OF SEA-LEVEL CHANGE 103
SOURCE
2030
This report, Washington
and Oregon
This report, California
This report, global
Vermeer and
Rahmstorf (2009), global
2050
This report, Washington
and Oregon
This report, California
This report, global
Vermeer and
Rahmstorf (2009), global
2100
This report, Washington
and Oregon
This report, California
This report, global
Vermeer and
Rahmstorf (2009), global
-20 0 20 40 60 80 100 120 140 160 180
SEA-LEVEL RISE (cm)
FIGURE 5.10 Committee's projected sea-level rise for California, Oregon, and Washington compared with global projections. The
dots are the projected values and the colored bars are the ranges. Washington and Oregon = coastal areas north of Cape Mendocino;
California = coastal areas south of Cape Mendocino.
RARE EXTREME EVENTS high astronomical tides, and large waves producing
record sea levels along virtually the entire coast (see
Extreme events can raise sea level much faster "Changes in Ocean Circulation" in Chapter 4). Dam-
than projected above. The rapid rise in sea level could age was extensive (e.g., Figure 5.11), with losses total-
be temporary, as in the case of a severe storm, or ing $215 million (in 2010 dollars; Griggs et al., 2005).
permanent, as in the case of a great subduction zone Some models predict that such extreme events will be-
earthquake. The potential contribution of such extreme come more common and that heightened sea level will
events to future sea-level rise is described below. persist longer as sea level rises, increasing the potential
for damage (Cayan et al., 2008; Cloern et al., 2011).
Extreme Sea Level Cloern et al. (2011) used a GCM forced by the
IPCC (2000) B1 emission scenario to assess possible
In the first 3 months of 1983, the west coast of climate change impacts in the San Francisco Bay and
the United States experienced a sequence of strong delta. As part of the analysis, they used a local sea-level
storms, with the coincidence of El Niño conditions, model, introduced by Cayan et al. (2008), to investi-
OCR for page 104
104 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
FIGURE 5.11 Rio Del Mar on northern Monterey Bay was damaged during the El Niño winter of 1983 by large waves arriving
simultaneously with high tides and elevated sea levels. SOURCE: Courtesy of Gary Griggs, University of California, Santa Cruz.
gate sea-level extremes that occur in conjunction with centile level (1.41 m above historical mean sea level)
broad-scale sea-level rise. Historical (19611999) and increases from the historical rate of approximately
projected (20002100) hourly sea level was simulated 9 hours per decade to more than 250 hours per decade
using predicted tides, simulated weather and El Niño- by mid-century, and to more than 12,000 hours per
Southern Oscillation conditions, and long-term rates decade by the end of the century. The model also shows
of sea-level rise from Vermeer and Rahmstorf (2009). that the duration of these extremes would lengthen
Wind, surface atmospheric pressure, and tropical from a maximum of 1 or 2 hours for the recent histori-
Pacific sea surface temperature were obtained from the
cal period to 6 or more hours by 2100, increasing the
National Center for Atmospheric Research PCM1 exposure of the coast to waves.
climate model simulation. The marked rise in the occurrence of extreme sea
The committee reproduced the Cloern et al. (2011) levels is qualitatively similar for different sea-level rise
analysis using its own sea-level projection for the San scenarios, but the duration of extremes can differ sub-
Francisco area and the Geophysical Fluid D ynamics stantially. For example, for the low end of the Vermeer
Laboratory CM2.1 model. This exercise showed and Rahmstorf (2009) sea-level projection (78 cm by
that as mean sea level rises, the incidence of extreme 2100), extreme water heights (exceeding the 99.99th
high-sea-level events becomes increasingly common percentile) are predicted to occur more than 300 hours
(Figure 5.12). According to the model, the incidence per decade by 2050 and more than 7,500 hours per
of extreme water heights that exceed the 99.99th per- decade by 2100.
OCR for page 105
PROJECTIONS OF SEA-LEVEL CHANGE 105
total time of exceedance
annual sea level hrs above historical 99.99th percentile
240 1200
hrs
160
cm 600
80
0 0
1960 2000 2040 2080
yr
FIGURE 5.12 Projected number of hours (blue bars) of extremely high sea level off San Francisco under an assumed sea-level
rise and climate change scenario. In this exercise, a sea-level event registers as an exceedance when San Francisco's projected sea
level exceeds its recent (19702000) 99.99th percentile level, 1.4 m above historical mean sea level. In the recent historical period,
sea level has exceeded this threshold about one time (1 hour) every 14 months. Sea-level rise (black line) during 19601999 was
arbitrarily set to zero, then increased to the committee's projected level for the San Francisco area over the 21st century (92 cm).
SOURCE: Adapted from Cloern et al. (2011).
Great Earthquakes Along the Cascadia Subduction or terrestrial soil (e.g., Nelson et al., 1996; Leonard et
Zone al., 2010).
Cycles of buried peat-mud couplets beneath coastal
Measurements of current deformation and geologic marshes (Figure 5.14) suggest that 6 to 12 great earth-
records (e.g., Savage et al., 1981; Atwater, 1987; Nelson quakes have occurred at irregular intervals ranging from
et al., 1996; Atwater and Hemphill-Haley, 1997) estab- a few hundred years to 1,000 years along the central
lish the potential for great (magnitude greater than 8) Cascadia margin over the past 6,000 years (Long and
megathrust earthquakes and catastrophic tsunamis Shennan, 1998). Geologic evidence also has been found
along the Cascadia Subduction Zone. In Washington for six great earthquakes along the northern Oregon
and Oregon, a great earthquake would cause some areas coast in the past 3,000 years (Darienzo and Peterson,
to immediately subside and sea level to suddenly rise 1995), 11 or 12 great earthquakes in southern Oregon
perhaps by more than 1 m. This earthquake-induced in the past 7,000 years (Kelsey et al., 2002; Witter et
rise in sea level would be in addition to the relative al., 2003), and seven great earthquakes in southwest
sea-level rise projected above. A great earthquake also Washington in the past 3,500 years (Atwater and
would produce large postseismic vertical land motions Hemphill-Haley, 1997). Turbidite deposits identified
in the area for years to decades. in marine cores suggest that 18 great earthquakes rup-
Sudden subsidence during great earthquakes is tured at least the northern two-thirds of the Cascadia
revealed in the geological record as abrupt changes in margin during the Holocene (Goldfinger et al., 2003,
sedimentary sequences (Nelson, 2007). When a great 2008).
earthquake occurs, salt marsh or terrestrial soils are The last great earthquake on the Cascadia mega
lowered into the intertidal zone, killing the vegetation thrust occurred on January 26, 1700 (Satake et al., 1996,
(e.g., Figure 5.13). These peaty soils are quickly covered 2003). The date of the earthquake was determined by
by tsunami-deposited sand or muddy tidal sediments. radiocarbon dating of suddenly buried marsh herbs,
In the decades after an earthquake, the coast slowly tree-ring records of trees stressed by coastal flooding
rises, producing a gradual transition back to a salt marsh
OCR for page 106
106 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
FIGURE 5.13 Ghost forests, such as this grove of weather-beaten cedar trunks near Copalis River, Washington, are evidence of
sudden subsidence. SOURCE: Courtesy of Brian Atwater, U.S. Geological Survey.
FIGURE 5.14 Stratigraphy and abundance of foraminifera in the sediment sequence recording the 1700 earthquake at Siuslaw
River, Oregon. Also shown is a reconstruction of elevation during this interval (WA-PLS column). Sediment likely deposited by tsunamis
is shaded in gray. SOURCE: Modified from Hawkes et al. (2011).
OCR for page 107
PROJECTIONS OF SEA-LEVEL CHANGE 107
during subsidence (e.g., Yamaguchi et al., 1997), and observations into the future, but the results depend on
Japanese historical records of a tsunami from a distant assumptions about the future behavior of the system.
source. Modeling of the tsunami waveform (Satake et Semi-empirical methods avoid these difficulties by
al., 1996) and estimates of coastal subsidence based on projecting global sea-level rise based on the observed
detailed microfossil studies (Hawkes et al., 2011) sug- relationship between sea-level change and global tem-
gest an earthquake magnitude of 8.8 to 9.2. The coastal perature. However, the highest projections made using
subsidence and associated sea-level rise were spatially this method (e.g., Grinsted et al., 2009) require un
variable, with the largest rise in sea level (12 m) oc- realistically rapid acceleration of glaciological processes.
curring in northern Oregon and southern Washington, Given the strengths and weaknesses of the differ-
where the plate boundary forms a wide, shallow arch ent projection approaches and the resource constraints
(Leonard et al., 2004, 2010; Hawkes et al., 2011). of an NRC study, the committee chose to use GCMs
Other sections of the margin subsided <1 m and the developed for the IPCC Fourth Assessment Report to
southernmost part of the subduction zone was uplifted estimate the steric contribution and extrapolation tech-
(Leonard et al., 2004, 2010; Hawkes et al., 2011). niques to estimate the cryospheric contribution. The
contributions were then summed. The land hydrology
Discussion component was assumed to be near zero and was not
factored into the projection. The committee's global
Changes in regional meteorological and climate projections for 2100 are substantially higher than the
patterns, including El Niños, coupled with rising sea IPCC's (2007) projection, mainly because of a faster
level, are predicted to result in increasing extremes in growing cryosphere component, and are somewhat
sea levels. Models suggest that sea-level extremes will lower than the Vermeer and Rahmstorf (2009) projec-
become more common by the end of the 21st century. tions. The committee estimates that global sea level
Waves riding on these higher water levels will cause will rise 823 cm by 2030, 1848 cm by 2050, and
increased coastal damage and erosion--more than that 50140 cm by 2100, relative to 2000 levels. As the
expected by sea-level rise alone. projection horizon lengthens, the uncertainties grow,
The biggest game changer for future sea level along and hence the ranges widen. The major sources of
the west coast of the United States is a great Cascadia uncertainty in the global projection are related to as-
earthquake. The related coastal subsidence of such an sumptions about the increase in rapid ice dynamics and
earthquake would, in a matter of minutes, produce the growth of future greenhouse gas emissions.
significantly higher sea levels off the Cascadia coast Formal projections of future sea-level rise along the
than 100 years of climate-driven sea-level rise. A great west coast of the United States have not been made,
earthquake could cause 12 m of sea-level rise in some although a few studies have presented ranges of possible
areas, which is significantly higher than the committee's outcomes for California and Washington. Methods
projection for Cascadia in 2100 (0.6 m). Further, the vary but usually involve a combination of global m odels
earthquake-induced sea-level rise would be an addition and local information. The committee's projections
to the expected global warming-related sea-level rise. account for factors that affect sea level in the area, in-
cluding local steric variations; wind-driven differences
CONCLUSIONS in ocean heights; the gravitational and deformational
effects associated with melting of Alaska, Greenland,
Global projections are commonly made using and Antarctic glaciers; and vertical land motions along
ocean-atmosphere GCMs, which provide a reasonable the coast. The local steric and wind-driven compo-
representation of the steric contribution to global sea- nents were estimated by extracting northeast Pacific
level rise, but do not yet fully capture the cryospheric data from the same ocean models used for the global
contribution. The IPCC (2007) projections made projections. The cryosphere component was adjusted
using this method are likely too low, even with an for gravitational and deformational effects and then
added ice dynamic component. Some studies project extrapolated forward. Vertical land motion was esti-
the cryospheric contribution by extrapolating current mated using continuous GPS measurements.
OCR for page 108
108 SEA-LEVEL RISE FOR THE COASTS OF CALIFORNIA, OREGON, AND WASHINGTON
The projected values vary by latitude, with the mean sea level continues to rise, the number of extreme
highest sea levels expected off the coast south of high water events and their duration are expected to in-
Cape Mendocino (430 cm for 2030, 1261 cm for crease. A simulation based on predicted tides, projected
2050, and 42167 cm for 2100, relative to 2000) and weather and El Niño conditions under a mid-range
the lowest sea levels expected off the coast north of greenhouse gas emission scenario, and the committee's
Cape Mendocino (-423 cm for 2030, -348 cm for projections of sea-level rise suggests that the incidence
2050, and 10143 cm for 2100). The lower sea levels of extreme water heights in the San Francisco Bay area
projected for Washington, Oregon, and northernmost would increase from about 9 hours per decade for the
California reflect coastal uplift and gravitational and recent historical period (19611999) to hundreds of
deformational effects, which lower the relative sea level. hours per decade by 2050 and several thousand hours
Major sources of uncertainties in the regional projec- per decade by 2100. In addition, the duration of these
tions are related to assumptions about the rate of future extremes would lengthen from 1 or 2 hours in the his-
ice losses and the constant rate of vertical land motion torical period to about 6 hours by 2100.
over the projection period. Uncertainties are larger for The biggest game changer for future sea-level rise
the regional projections than for the global projections along the U.S. west coast would be a great earthquake
because more components are considered and because (magnitude greater than 8) along the Cascadia Sub-
uncertainties in the steric and ocean dynamic compo- duction Zone. Such earthquakes have occurred every
nents are larger at a regional scale than at a global scale. several hundred to 1,000 years, with the most recent
Extreme events can raise sea level much faster than occurring in 1700. During a great earthquake, some
the rates projected by the committee. For example, land areas would immediately subside and relative sea
unusually high sea levels may occur temporarily when level would suddenly rise, perhaps by 1 m or more. This
major storms coincide with high astronomical tides, earthquake-induced rise in sea level would be added
and especially during years when regional sea levels to the projected rise in relative sea level (about 60 cm
are anomalously heightened during El Niño events. As by 2100).
