ABSTRACT
Previous studies suggest that the expected global warming from the greenhouse
effect could raise sea level 50 to 200 centimeters (2 to 7 feet) in the next century. This
article presents the first nationwide assessment of the primary impacts of such a rise on
the United States: (1) the cost of protecting ocean resort communities by pumping sand
onto beaches and gradually raising barrier islands in place; (2) the cost of protecting
developed areas along sheltered waters through the use of levees (dikes) and bulkheads;
and (3) the loss of coastal wetlands and undeveloped lowlands.
The total cost for a one meter rise would be $270-475 billion, ignoring future
development. We estimate that if no measures are taken to hold back the sea, a one meter
rise in sea level would inundate 14,000 square miles, with wet and dry land each
accounting for about half the loss. The 1500 square kilometers (600-700 square miles) of
densely developed coastal lowlands could be protected for approximately one to two
thousand dollars per year for a typical coastal lot. Given high coastal property values,
holding back the sea would probably be cost-effective.
The environmental consequences of doing so, however, may not be acceptable.
Although the most common engineering solution for protecting the ocean coast--pumping
sand--would allow us to keep our beaches, levees and bulkheads along sheltered waters
would gradually eliminate most of the nation's wetland shorelines. To ensure the long-term
survival of coastal wetlands, federal and state environmental agencies should begin to lay
the groundwork for a gradual abandonment of coastal lowlands as sea level rises.
INTRODUCTION
At the turn of the century, scientific opinion regarding the practical
implications of the greenhouse effect was sharply divided. Since the 1860s, people had
known that by absorbing outgoing infrared radiation, atmospheric CO2 keeps the earth
warmer than it would otherwise be (Tyndall, 1863). Svante Arrhenius (1896), who coined the
term "greenhouse effect," pointed out that the combustion of fossil fuels might
increase the level of CO2 in the atmosphere, and thereby warm the earth several degrees.
Because the 19th century had experienced a cooling trend, however, others speculated that
the oceans and plant life might gradually reduce CO2 levels and cause an ice age (Barrel
et al., 1919).
Throughout the first half of the 20th century, scientists generally recognized
the significance of the greenhouse effect, but most thought that humanity was unlikely to
substantially alter its impact on climate. The oceans contain 50 times as much CO2 as the
atmosphere, and physical laws governing the relationship between the concentrations of CO2
in the oceans and in the atmosphere seemed to suggest that this ratio would remain fixed,
implying that only 2 percent of the CO2 released by human activities would remain in the
atmosphere. This complacency, however, was shattered in 1957 when Revelle and Seuss (1957)
demonstrated that the oceans could not absorb CO2 as rapidly as humanity was releasing it:
"Human beings are now carrying out a large-scale geophysical experiment." Only
then were monitoring stations set up to measure worldwide trends in atmospheric
concentrations. By the mid 1960s, it was clear that Revelle and Seuss had been correct
(President's Science Advisory Committee, 1965).
In the last decade, climatologists have reached a consensus that a doubling of
CO2 would warm the earth 1.5-4.5 o C (3-8 o F), which could leave our planet warmer than
it has ever been during the last two million years (National Academy of Sciences, 1979).
Moreover, humanity is increasing the concentrations of other gases whose combined
greenhouse effect could be as great as that due to CO2 alone, including methane,
chlorofluorocarbons, nitrous oxide, and sulfur dioxide (Ramanathan et al., 1985). Even
with the recent agreement to curtail the use of CFCs, global temperatures could rise as
much as 5 o C (9 o F) in the next century (Smith and Tirpak, 1988).
Global warming would alter precipitation patterns, change the frequency of
droughts and severe storms, and raise the level of the oceans. This article presents the
first attempt to quantify the nationwide impacts of an accelerated rise in sea level. The
study was undertaken in response to a request from the U.S. Congress to the Environmental
Protection Agency, and had to be completed in twelve months with a budget of $300,000.
With these constraints, we were only able to focus on the loss of dry and wet land and the
cost of holding back the sea.
Because our focus was on aspects that can be most readily estimated on a
nationwide basis, we have disregarded impacts who importance is limited to a few areas--in
particular the threat to coastal water supplies from saltwater intrusion and the unique
situation in Louisiana. We hope that this study will motivate others to improve on the
methods presented here and to start quantifying other impacts of global warming.
CAUSES AND EFFECTS OF SEA LEVEL RISE
Climate and Sea Level
The level of the oceans has always fluctuated with changes in global
temperatures. During ice ages when global temperatures were 5 o C (9 o F) lower than
today, much of the ocean's water was tied up in glaciers and sea level was often over one
hundred meters (three hundred feet) lower than today (Donn et al., 1962; Kennett, 1982;
Oldale, 1985). On the other hand, during the last interglacial period (100,000 years ago)
when temperatures were about 1 o C (2 o F) warmer, sea level was approximately 6 meters
(20 feet) higher than today (Mercer, 1970).
When discussing shorter periods of time, one must distinguish worldwide
(eustatic) sea level rise from relative sea level rise, which includes land subsidence.
Although climate affects worldwide sea level, the rate of sea level rise relative to a
particular coast has more practical importance and is all that current monitoring stations
can measure. Because some coastal areas are sinking while others are rising, relative sea
level rise in the United States varies from more than one meter (three feet) per century
in Louisiana and parts of California and Texas, to thirty centimeters (one foot) per
century along most of the Atlantic and Gulf Coasts, to a slight drop in much of the
Pacific Northwest as shown in Figure 1. Global sea level trends have generally been
estimated by combining the trends at tidal stations around the world. These records
suggest that during the last century, worldwide sea level has risen 10 to 25 cm (4 to 10
in) (Barnett, 1984; Peltier and Tushingham, 1989), much of which has been attributed to
the global warming of the last century (Gornitz et al., 1982; Meier, 1984).
(SEE GRAPHICAL CHARTS AND FIGURES IN THE
DOWNLOADABLE PDF ADOBE ACROBAT DOCUMENT (774 KB FILE SIZE))
The projected global warming could raise worldwide sea level by expanding ocean
water, melting mountain glaciers, and causing the ice sheets of Greenland and Antarctica
to melt or slide into the oceans. A report by the Department of Energy estimated that over
a period of two to five hundred years, the West Antarctic Ice Sheet could disintegrate,
raising sea level 6 meters (20 feet) (Bentley, 1983; Hughes, 1983). Studies after 1983,
however, have focused on the next century. Figure 2 illustrates recent estimates of sea
level rise, which generally fall into the range of 50 to 200 cm (2 to 7 feet) by 2100;
Studies since 1990, however, generally suggest that the 50-200 cm rise is more likely to
take 150-200 years. This assessment uses rises of 50, 100, and 200 cm.
Although the most recent reports have been at the lower end of the range, recent
press accounts have exaggerated the extent to which those reports should be viewed as
downward revisions.
Meier's (1990) recent paper (entitled Reduced Rise in Sea Level), led major
newspapers to report that sea level projections are being lowered. However, as Figure 2
shows, the main reason the projections were lower was that Meier changed his target year
from 2100 to 2050. The Intergovernmenal Panel on Climate Change (1990) released
projections that were lower than most previous studies, based on the assumption that
Antarctica will accumulate more ice in response to global warming, an assumption that has
not met universal acceptance from glaciologists. Clearly, the models are unable to
adequately estimate how much sea level will rise; and the experts disagree. Under these
circumstances, revisions of sea level projections are just as likely to result from
changes in who chairs the panel that forecasts sea level rise as from any new
information."
The greenhouse effect would not necessarily raise sea level by the same amount
everywhere.
Removal of water from the world's ice sheets would move the earth's center of
gravity away from Greenland and Antarctica; the oceans' water would thus be redistributed
toward the new center of gravity. Along the coast of the United States, this effect would
generally increase sea level rise by less than 10 percent; sea level could actually drop,
however, at Cape Horn and along the coast of Iceland (Clark and Lingle, 1977). Climate
change could also influences local sea level by changing winds, atmospheric pressure, and
ocean currents, but no one has estimated these impacts.
Effects of Sea Level Rise
A rise in sea level would inundate wetlands and lowlands, accelerate coastal
erosion, exacerbate coastal flooding, threaten coastal structures, raise water tables, and
increase the salinity of rivers, bays, and aquifers (Barth and Titus, 1984). Most of the
wetlands and lowlands of the United States are found along the Gulf Coast and the Atlantic
coast south of the central part of New Jersey, although there is also a large low area
around San Francisco Bay. Similarly, the areas vulnerable to erosion and flooding are also
predominantly in the southeast, while potential salinity problems are spread more evenly
throughout the coast. We briefly describe the impacts that would result if nothing was
done to address sea level rise, and then discuss the responses that have been investigated
in the last several years.
Shoreline Retreat. Coastal marshes and swamps are generally
found between the highest tide of the year and mean sea level. Because they collect
sediment and produce peat upon which they can build, most wetlands have been able to keep
pace with the past rate of sea level rise (e.g. Kaye and Barghoom). Thus, as Figure 3a-3c
illustrate, the area of wetlands today is generally far greater than the area that would
be available for new wetlands if sea level rose too rapidly (Titus, 1986; Titus, 1988).
The potential loss would be the greatest in Louisiana (U.S. Environmental Protection
Agency and Louisiana Geological Survey, 1987). Moreover, in many areas people have built
bulkheads just above the marsh; if sea level rose, the wetlands would be squeezed between
the estuary and the bulkhead (Figure 3d). Such a loss would reduce available habitat for
birds and juvenile fish, and would reduce the production of organic materials on which
estuarine fish rely.
The dry land within two meters (seven feet) of high tide include forests, farms,
low parts of some port cities, communities that sank after they were built and are now
protected with levees, and the bay sides of barrier islands. The low forests and farms are
generally in the mid-Atlantic and southeast, and would provide potential areas for new
wetland formation. Major port cities with low areas include Boston, New York, Charleston,
Miami, and New Orleans; the latter averages about two meters below sea level, and parts of
Texas City, San Jose, and Long Beach California are about one meter below sea level.
Some of the most important vulnerable areas are the recreational barrier islands
and spits of the Atlantic and Gulf Coasts. Coastal barriers are generally long narrow
islands and spits (peninsulas) with the ocean on one side and a bay on the other.
Typically, the ocean-front block of an island ranges from two to four meters above high
tide, while the bay side is less than a meter above high water. Thus, even a one meter
rise in sea level would threaten much of this valuable land with inundation.
Erosion, moreover, threatens the high parts of these islands, and is generally
viewed as a more immediate problem than the inundation of their bay sides. As Figure 4
shows, a rise in sea level can cause an ocean beach to retreat by considerably more than
the retreat due to inundation alone. The shape of a beach profile is determined by the
pattern of waves striking the shore; generally, the visible part of the beach is much
steeper than the underwater portion which comprises most of the active "surf
zone." While inundation alone is determined by the slope of the land just above the
water, Bruun (1962) showed that the total shoreline retreat from a rise in sea level
depends on the average slope of the entire beach profile.
Previous studies suggest that a one meter rise in sea level would generally
cause beaches to erode 50-100 meters from the Northeast to Maryland; 200 meters along the
Carolinas; 100-1000 meters along the Florida coast; and 200-400 meters along the
California coast (Everts, 1985; Kyper and Sorensen, 1985; Kana et al, 1984; Bruun, 1962;
Wilcoxen, 1986). Because most U.S. recreational beaches are less than 30 meters (100 feet)
wide at high tide, even a thirty centimeter (one foot) rise in sea level would require a
response.
Flooding. Coastal areas would become more vulnerable to
flooding for four reasons: (1) A higher sea level provides a higher base for storm surges
to build upon; a one meter rise in sea level would thus enable a 15-year storm 3 to flood
many areas that today are only flooded by a 100-year storm (Kana et al., 1984). (2) Beach
erosion would leave particular properties more vulnerable to storm waves. (3) Higher water
levels would increase flooding due to rainstorms by reducing coastal drainage (Titus et
al., 1987). (4) Finally, a rise in sea level would raise water tables.
Many coastal areas are protected with levees and seawalls, and would thus not
necessarily experience inundation, erosion, or flooding. However, these structures have
been designed for current sea level. Higher water levels would threaten the integrity of
these coastal structures because (1) higher flood levels might overtop them, and (2)
erosion could undermine them from below (National Research Council, 1987). In areas like
New Orleans that are drained artificially, the increased need for pumping could exceed
current capacities (Titus et al., 1987).
Saltwater Intrusion. Finally, a rise in sea level would enable
saltwater to penetrate farther inland and upstream in rivers, bays, wetlands, and
aquifers, which would be harmful to some aquatic plants and animals, and would threaten
human uses of water. Increased salinity has already been cited as a contributing factor to
reduced oyster harvests in the Delaware (Gunter, 1974) and Chesapeake Bays, and for
converting cypress swamps in Louisiana to open lakes (U.S. Environmental Protection Agency
and Louisiana Geological Society, 1987). Moreover, New York, Philadelphia, and much of
California's Central Valley get their water from areas that are just upstream from where
the water is salty during droughts. Farmers in central New Jersey as well as the city of
Camden rely on the Potomac-Raritan-Magothy aquifer, which could become salty if sea level
rises (Hull and Titus, 1986). The South Florida Water Management District already spends
millions of dollars per year to prevent Miami's Biscayne aquifer from becoming salty
(Miller et al., 1989).
Responses
Inundation, Erosion, and Flooding. Possible responses fall
broadly into three categories: erecting walls to hold back the sea; allowing the sea to
advance and adapting to it; and raising the land. The slow rise in sea level over the last
thousand years and the areas where land has been sinking more rapidly offer numerous
historical examples of all these responses. For over five centuries, the Dutch have used
dikes and windmills to prevent inundation from the North Sea. By contrast, many cities
have been rebuilt landward as structures and land were lost to erosion; the town of
Dunwich, England has had to rebuild its church seven times in the last seven centuries.
More recently, rapidly subsiding communities such as Galveston, Texas, have used fill to
raise land elevations; The U.S. Army Corps of Engineers and coastal states regularly pump
sand from offshore to counteract beach erosion. Venice is a hybrid of all three responses,
allowing the sea to advance into the canals, while raising some low lands and erecting
storm protection barriers.
Most assessments in the United States have concluded that low-lying coastal
cities would be protected with bulkheads, levees, and pumping systems, while in sparsely
developed areas 8.shorelines would retreat naturally (National Research Council, 1987).
This conclusion has generally been based on the commonly accepted assumption that the cost
of these structures would be far less than the value of urban areas being protected but
greater than the value of undeveloped land.
Studies on the possible responses of barrier islands (Titus, 1985; Howard et
al., 1985) and moderately-developed mainland communities (Kana et al., 1986; Titus, 1988)
suggest that environmental and aesthetic factors would be as important as economics.
Figure 5 illustrates four possible responses by which barrier islands could respond to sea
level rise: no protection, engineering a landward retreat, raising the island in place,
and building a levee. A case study of Long Beach Island, New Jersey, concluded that all
three protection options are far less costly than the value of the land that would be
threatened. Although levees and retreat are somewhat less expensive than raising islands,
the latter option would probably be preferred because (1) constructing levees and seawalls
would result in the loss of beaches and waterfront views; and (2) retreat would not be
feasible for islands with high-rises and would only be marginally less expensive for
moderately developed islands, while requiring major changes in how people view ownership
of coastal property (Titus, 1990).
Responses to erosion are more likely to have adverse environmental impacts along
sheltered water than on the open coast. Because the beach generally is a barrier island's
most important asset, economics would encourage ocean beach communities to preserve their
natural shorelines; and preventing the island from breaking up would also protect the
adjacent wetlands. But along most mainland shorelines, economic self-interest would
encourage property owners to erect bulkheads, which would prevent new wetland formation
from offsetting the loss of wetlands that were inundated. Figure 6 illustrates possible
policy responses for ensuring that development does not block the inland migration of
coastal wetlands.
Saltwater Intrusion. Most of the measures for counteracting
saltwater intrusion due to sea level rise have also been employed to address current
problems. The Delaware River Basin Commission, for example, protects Philadelphia's
freshwater intake on the river--as well as New Jersey aquifers recharged by the
river--from excessive salinity by storing water in reservoirs during the wet season and
releasing it during droughts, forcing the saltwater back toward the sea (Hull and Titus,
1986). Other communities have protected coastal aquifers by erecting underground barriers
and by maintaining freshwater pressure through the use of impoundments and injection wells
(Sorensen et al., 1984).
PROJECTING THE NATIONWIDE IMPACTS: OBJECTIVES AND STRATEGY
Ideally, we would like to know the economic and environmental impacts of all the
possible scenarios of sea level rise for all possible policy responses for every coastal
community in the nation. Every community would then have sufficient information to
rationally consider how it should respond. Moreover, we could estimate the nationwide
impact by picking the best policy response for each community and adding all the costs.
However, because such a comprehensive analysis is not yet possible, our study had the more
limited objective of developing nationwide estimates that considered as many factors as
possible.
[GRAPHIC GOES HERE]
Our first step was to choose which of the impacts to study. We chose shoreline
retreat (i.e., erosion and inundation) for several reasons: First, we excluded saltwater
intrusion because only two case studies had examined the physical impacts and none had
examined the economic impact of rising sea level; the processes are too complicated to
meaningfully represent without detailed models; and the unavoidable economic and
environmental impact of increased salinity appeared to be an order of magnitude less than
shoreline retreat--and much more sensitive to possible changes in the frequency of
droughts than to sea level rise (Hull and Titus, 1986). We would have liked to include
flooding, which is closely related to shoreline retreat, but the cost of applying flood
models to a large number of sites was prohibitive, and models of the resulting property
damage are inaccurate without detailed surveys of the elevations and types of structures.
By contrast, estimating the impacts of (1) natural shoreline retreat and (2)
holding back the sea seemed feasible. In the former case, estimating inundation of dry
land simply requires one to determine its elevation; wetland loss requires the elevation
and an assumption regarding how rapidly the wetlands might accrete; beach erosion can be
approximated using topographic maps and the Bruun (1962) rule; and the value of lost land
can be estimated using tax maps. The costs of holding back the sea are also fairly
straightforward: Wetland loss is estimated the same way as under the natural retreat
scenario, except that one must specify which areas are likely to be protected from the sea
(and hence, unavailable for creation of new wetlands); the cost of nourishing beaches can
be derived using data collected by the Corps of Engineers; and the cost of elevating land,
houses, and of erecting shore-protection structures can be estimated by engineers based on
experience.
Moreover, the procedures for assessing shoreline retreat tend to implicitly
account for flooding caused by storm surges (at least after the first foot of sea level
rise). Where development is protected from sea level rise, levees and pumping systems used
for preventing inundation would also prevent sea level rise from increasing flood damages;
and raising barrier islands and the structures on them by the amount of sea level rise
would leave flood risks constant. Where development is unprotected, the estimates of
lost land and structures would probably account for the costs of increased flooding;
although flood plains would move inland, the value of structures standing in the new flood
plain would be approximately balanced by the inundated structures that are lost.
Nevertheless, for the first foot of sea level rise, examining shoreline retreat probably
does not account for flooding: if development is protected, major measures would probably
not be taken to counteract the first foot, so the frequency of flooding would increase. If
development is not protected, the first foot would increase flooding but not threaten many
structures with inundation.
At the outset, it was clear that it would not be possible to estimate both the
cost of holding back the sea and the cost of not holding it back with the resources we
were allotted. We chose to focus on the former because it currently seems more likely. We
would learn little, for example, from estimating the value of buildings on Manhattan
Island that would be lost if the sea was not held back; because of its value, we know
Manhattan will be protected, and that assuming otherwise in a nationwide study would
substantially skew the results. By contrast, assuming that shores would be protected in a
lightly-developed area more likely to be abandoned would introduce a more modest error.
Accordingly, we divided the assessment into three tasks:
(1) Estimating the areas of dryland and wetlands that would be lost for various
scenarios.
(2) Estimating the cost of protecting developed areas along sheltered coasts.
(3) Estimating the cost of protecting developed areas along the open coast.
Each task used four scenarios of sea level rise: the historic trend of 12 cm per
century, and accelerated rises of 50, 100, and 200 cm by the year 2100. In all cases,
local subsidence was added to the projections of global sea level rise.
LOSS OF COASTAL WETLANDS AND DRY LAND
Methods
This part of the study was based on a sample of 46 coastal sites selected at
regular intervals along the coast, accounting for 10 percent of the contiguous U.S.
coastal zone. Because of funding constraints, we had to exclude Alaska, Hawaii, and the
U.S. territories.
This task required us to (1) characterize existing elevations and (2) model the
impact of sea level rise. We estimated elevations of dry land by interpolating between
contours on U.S. Geological Survey topographic maps. For wetlands, however, the procedure
was more complicated: First we determined the dominant wetland types (e.g. high marsh, low
marsh, mangrove) for 57- by 79-meter "pixels", based on LANDSAT imagery
(DeSarthy, 1974). (Because of computer limitations, these pixels had to be aggregated to
500-meter squares.) We then calculated the distribution of wetland elevations based on
known tidal ranges,6 the relationship between tides and wetland types (Lefor et al.,
1987), and the assumption that elevations for particular wetland types are distributed
uniformly between their upper and lower bounds.
Given our estimates of current elevations, we estimated loss of wet and dry land
for three scenarios of shoreline protection: no protection, (standard) protection of
currently developed areas, and (total) protection of all shores. Although estimating the
loss of dry land depends solely on current elevations, determining the loss of coastal
wetlands requires an assumption regarding vertical accretion. Because no one had assessed
the possible impact of global warming and sea level rise on accretion, we assumed that
current accretion rates would continue.
An evaluation of the implications of coastal land loss is outside the scope of
this report. Nevertheless, the reader interested in understanding the entire economic
impact of sea level rise requires at least a rough conversion of the wetland loss
estimates into monetary terms. We use high and low assumptions of $6,000 and $30,000 per
acre. The first assumption is based on the low end of previous studies that valued each of
the various services from wetlands that can be quantified, such as fisheries, flood
control, and water purification (e.g. Gosselink et al. 1974; Farber and Costanza 1987).
This assumption is almost certainly too low because it does not consider the various
aesthetic benefits of better environmental quality. Our high assumption is based both on
the upper end of previous studies and on the observation 7 that wetland mitigation
programs generally cost about $30,000 per acre; this assumption is also an understatement,
since the artificially created wetlands generally have a lower quality than the natural
wetlands that are lost.
We also made a rough estimate of the undeveloped land that would be lost if only
currently developed areas are protected. These estimates were based on the subjective
assessments of real estate agents regarding the value of undeveloped land, as reported in
Yohe (1990).
Limitations
The most frequently criticized aspect of our approach is that we have assumed
constant wetland accretion rates. Because sea level itself can limit the rate of vertical
accretion of wetlands, one might hope that accelerated sea level rise would enable wetland
vertical accretion to accelerate as well. Although our results for the 50-cm
scenario are sensitive to this assumption, they are not particularly sensitive for the
one- and two-meter scenarios: No one has suggested that wetlands could generally keep pace
with sea level rising 1 to 4 cm/yr, which these scenarios imply for the second half of the
21st century.
This is not to say that most areas would lose all their wetlands. Even in areas
that are rapidly losing wetlands due to sea level rise, such as Louisiana, one can find
parcels of wetlands that have kept pace with sea level. In areas with substantial sediment
supplies, accelerated sea level rise simply implies that a fixed supply can sustain a
proportionately smaller are of wetlands. Even in areas with relatively little sediment
supply, the loss of wetlands due to erosion, inundation, or saltwater intrusion frees up
sediment that had been previously bound up with the wetlands that are lost; this extra
sediment may help to accelerate the vertical accretion of the remaining wetlands. Finally,
some types of low-salinity marshes and mangroves appear to be able to sustain much larger
rates of sea level rise than are prevalent today (Stevenson et al. 1990).
Although remote sensing provides greater vertical resolution than topographic
maps, it requires interpretation which can cause errors. In California, for example,
redwoods have a spectral signature very similar to some marsh grasses. Although we checked
our interpretation against other sources such as U.S. Fish and Wildlife Service Wetland
Inventory maps, detailed survey information was only available for a few sites.
Nevertheless, as Table 1 shows, our estimate of current nationwide wetland acreage is
fairly close to the estimate developed by the National Oceanic and Atmospheric
Administration.
Our failure to consider the potential implications of alternative ways of
managing river flow is a particularly serious limitation for application to Louisiana,
where a wide variety of measures have been proposed for increasing the amount of water and
sediment delivered to the wetlands.
We made no attempt to predict which undeveloped areas might be developed in the
next century. Currently, only about one- seventh of the coastal lowlands are developed;
even a low growth rate would imply substantially greater protection costs and higher land
values for the currently undeveloped land. Finally, our conversion of the area of lost
wetlands into dollar values is completely dependent on a few case studies that have never
considered the implications of the large nationwide loss that could result from a rise in
sea level. The law of diminishing returns implies that the cost to society of 50 percent
of the coastal wetlands would be much greater than, for example, fifty times the cost of
losing 1 percent.
Results
In presenting results, we group the sites into seven coastal regions, as shown
in Figure 7. New England, Mid-Atlantic, South Atlantic, South Florida/Gulf Coast
Peninsula, Louisiana, Other Gulf (Texas, Mississippi, Alabama, Florida Panhandle), and
Pacific Coast.
[GRAPHIC GOES HERE]
Figure 8 illustrates our point estimates of nationwide wetland loss for the
three policy scenarios -- total, standard, and no protection. If all shorelines were
protected, a one meter rise would result in a loss of 50-82 percent of U.S. coastal
wetlands, while a two meter rise would result in a loss of 66-90 percent. If only
currently developed areas are protected, the losses would be 29-69 percent and 61-80
percent for the one and two meter scenarios. Table 2 provides regional detail for the
one-meter scenario; Table 3 provides nationwide results for all three scenarios. The
greatest losses of wetlands would be in the southeast, which currently has 85 percent of
U.S. coastal wetlands; for a one meter rise 90-95 percent of the loss would take place in
this region, 40-50 percent in Louisiana alone. By contrast, neither the northeast nor the
west 9 would lose more than 10 percent of their wetlands if only currently developed areas
are protected. Except for the northeast, no protection results in only slightly lower
wetland loss than protecting only densely developed areas.
Figure 9 illustrates our projections of the inundation of dry land for the seven
coastal regions. If shorelines retreat naturally, a one meter rise would inundate 20,000
square kilometers (7,700 square miles) of dry land, an area the size of Massachusetts;
rises of 50 and 200 cm would result in losses of 13,000 and 31,000 square kilometers
(5,000 and 12,000 square miles), respectively.
Seventy percent of the losses would occur in the southeast, particularly
Florida, Louisiana, and North Carolina; the eastern shores of Chesapeake and Delaware Bays
would also lose considerable acreage. Table 4 provides estimates of the value of wet and
dry land that might be lost is currently developed areas are protected.
COST OF PROTECTING CURRENTLY DEVELOPED SHORES ALONG
SHELTERED WATERS
Methods
Our approach here was to develop cost estimates 10 for protecting six index
sites and extrapolate 19.those estimates to the other sites in the sample based on the
total amount of developed shoreline to be protected and the portion of the protected area
threatened with inundation.
The six index sites were metropolitan New York; Long Beach Island, New Jersey;
Dividing Creek, New Jersey; metropolitan Miami; Corpus Christi, Texas; and parts of San
Francisco Bay.
Because our resources were not sufficient estimate the costs of shore-protection
strategies for several sea level scenarios, we did so only for the 2-meter scenario, (and
interpolated for the other scenarios, as we describe below).
Our assessments of the index sites assumed that developed areas below the 10-ft
(NGVD) contour would require protection. Because contours are only available in 5-ft
increments, this assumption was a computational necessity. Nevertheless, it is fairly
reasonable: with a 2-meter rise, the contour would only be about 60 cm above mean sea
level and hence very close to spring high tide. Levees were assumed to cost $1.6 million
per kilometer ($500 per linear foot) (Sorensen et al., 1984). As an alternative, we
considered protection with bulkheads, whose cost is only $0.42 million per kilometer ($130
per foot) (Sorensen et al., 1984). In the case of areas that already have bulkheads, we
netted out the costs that would be incurred without sea level rise due to routine
replacement, using the standard engineering assumption that that bulkhead costs increase
with bulkhead height raised to the 1.5 power. In cases where the area requiring protection
is isolated and connected to higher ground by low-lying roads, we estimated the cost of
raising the elevation of the roadway. In areas that are too lightly developed to warrant
protection, we estimated the cost of moving buildings assuming a cost of $10,000 per
structure.
Developing the nationwide estimates involved two steps: extrapolating the index
sites to the entire sample, which gave us an estimate for the 2-meter scenario, and
interpolating the results for each site for the 50- and 100-cm scenarios. The first step
required us to estimate the total length of shorelines that would require bulkheads or
levees, the total length of roads that would have to be rebuilt, and the number of
structures that would be moved. We used the digitized.maps described above to estimate
shoreline length, assuming that the ratio actual shoreline to digitized shoreline would be
the same for the rest of the sample as it had been for the 6 index sites. We calculated
the amount of roads that had to be rebuilt and structures to be moved by fitting
regression equations that expressed those variables as functions of the amount of
low-lying land that is developed and the slope of the land, both of which were available
from the digitized maps.
We then interpolated for the 50 and 100 cm scenarios. The cost estimates assumed
that the fraction of shoreline protected by levees would correspond to the fraction of
lowland in the particular site that was inundated, and that the remainder of protected
shorelines would be bulkheaded. We interpolated unit protection costs on the assumption
that costs rise with the 1.5 power of the height of the structures. Finally, we use Yohe's
(1990) data for the average value of vulnerable land in developed areas ($91,000 to
266,000 per acre) to estimate the value of the land that would be required to build the
necessary dikes, assuming that the dikes and related infrastructure average 9 meters (30
feet) wide.
Limitations
This task is almost certainly the least accurate part of our assessment. Only
one of the index-site studies involved a visit to the site. Unit-cost estimates were based
on the literature, not site-specific designs that take into consideration wave data for
bulkheads and potential savings from tolerating substandard roads. Although digitized
topographic and development data were considered for all 46 sites, the results depend
largely on a very small sample of only six sites.
Results
Table 4 illustrates our estimates of the value of the necessary land, while
Table 5 provides estimates of the cost of the dikes themselves. Unlike wetland loss, the
cost of protecting developed areas from the sea would be concentrated more in the
northeast than the southeast, because a much greater portion of the coast is developed in
the northeast. (The southeast still accounts for a large percentage of the total costs due
to its majority share of U.S. sheltered shorelines.)
COST OF PROTECTING THE OPEN COAST
Methods
Because resources were limited, we could only consider a single technology for
the entire coast. We chose the island raising approach, based on the Long Beach
Island study discussed in the background section. We assumed that all developed barrier
islands would be protected in this fashion, as well as a few undeveloped recreational and
Louisiana's barrier islands, where shore protection is a state policy.
This task consisted of estimating (1) the cost of placing sand on the beach
profile and low parts.of barrier islands and (2) the cost of elevating houses and
infrastructure.
Sand Quantities. This part of the analysis required us to
estimate the amount of sand each state would require and the average unit cost of sand in
that state. For the former task, we employed the "raise the profile" method,
which is consistent with the Bruun (1962) rule but does not require as many assumptions
(Titus and Greene, 1989). This method simply states that the amount of sand required is
equal to the area being raised times the rise in sea level. For small amounts of sea level
rise, only the active beach profile must be raised (to curtail erosion); for larger rises,
dry land must also be raised (to prevent inundation).
The only difficulty in applying this method is deciding the extent of the beach
profile being raised. Hallermeier (1981) recommended that analysts calculating the amount
of sand necessary for beach nourishment use the critical closure depths for significant
offshore transport, which he estimated as generally about 5-7 meters on the Atlantic Cost,
4-5 meters on the Gulf Coast, and 5-10 meters along the Pacific Coast. He also estimated
the closure depths for significant transport of any type, but he recommended these depths
only for use in erosion calculations.
Thus, in the peer review draft, we used bathymetric charts to estimate the
distance out to sea of the closure depths for offshore transport, and topographic maps to
estimate the inland distance of the dune crest, added the two distances, and multiplied by
the shoreline length to arrive at an estimate of the area of the beach profile that needed
to be raised.
In a previous publication (Titus and Greene 1989), we had noted that this
procedure probably provided estimates that were too low. The closure depth and distance
recommended by Hallermeier approximate the profile adjustment in a single year; should a
major hurricane come along, material will be carried much further out to sea. Indeed, most
cost studies of beach nourishment may be acknowledging this implicitly by also including
an allowance for periodic "renourishment" projects. However, it would be
inappropriate to use the normal renourishment formulas, because they generally lump the
additional sand necessitated by long-term profile adjustment in with other factors that
are outside the scope of this analysis (such as alongshore transport) or already included
(such as post-project sea level rise).
Nevertheless, about half the reviewers of the original version of this paper
suggested that its most serious flaw was that the sand quantity estimates were too low.
Accordingly, we decided to call the original estimates our low scenario and develop a high
scenario adapted from the Titus and Greene paper: Low: Sandt = A * SLRt High: Sandt = A *
SLRt - B * SLRt-1 + c * Sandt-1, where the asterisk "*" denotes multiplication,
B = A * (1 - (1-c) * (di/dl)1.5) and A is the ratio of the horizontal and vertical extents
of the beach profile, 1/(1-c) is the "e-folding" time for long-term profile
adjustment in years, and (di/dl) 1.5 is an estimate of the ratio of the profile lengths
from the two closure depths suggested estimated by Hallermeier, using the assumption that
profiles follow the shape y=x 2/3 . We then assumed that c=.96 for the Gulf coast and .98
for the Atlantic and Pacific Coasts.
Simply put, the high scenario assumes that in the first year after a rise in sea
level, the sand required to stabilize the shore is the same as the sand projected by the
low scenario; but in equilibrium, the required sand will be 6.0, 3.5, and 15.7 times as
great for the Atlantic, Gulf, and Pacific Coasts, based on the alternative estimates of
the profile length. The values of c imply that 90 percent of the adjustment will occur in
57 years for the Gulf and 114 years elsewhere.
Sand would also be required for elevating dry land. We assumed that after the
first foot of sea level rise, portions of coastal barrier islands below 152 cm (5 ft NGVD)
would be raised with the sea, and that other parts of coastal barriers would be raised
after the third foot.
We employed this procedure for every developed recreational beach from Delaware
Bay to the mouth of the Rio Grande, as well as the state of California. For the rest of
the coast, we picked one representative site for each state; we then extrapolated the
individual sites to the rest of the state based on the length of the developed shoreline.
Sand Costs. For both of the sand quantity formulas, we
developed high and low cost estimates for each state, based on inventory surveys by the
Corps of Engineers and state agencies contemplating beach nourishment projects. The low
estimates were based on the assumption that unit sand costs remain constant, while the
high estimates assume that unit costs increase as the least expensive nearshore deposits
are exhausted and it becomes necessary to go farther out to sea or mine lower-quality
deposits.
For the most part, existing inventories identify sufficient sand for the
projects that officials are currently contemplating, but not the far greater quantities
that would be necessary to address a substantial rise in sea level. Therefore, we were
unable to specify separate reliable sand cost functions for each state in developing our
high scenario. However, sufficient sand has been identified for such a cost function for
Florida's Atlantic coast, illustrated in Table 6. Therefore, we calculated sand costs for
each state by assuming that costs would escalate by the same pattern (scaling for the
length of the state's shore); i.e. we considered current differences in sand costs for
various states but assumed that those differences would not change as sea level rises.
Cost of Raising Land and Structures. Estimating the cost of
elevating roads and structures required us to obtain estimates of (1) the area of
developed land that would be elevated, (2) the density of structures, and (3) unit cost
factors. Because the only case study to address this issue was the study of Long Beach
Island (Weggel et al., 1989), this part of our analysis consisted primarily of
extrapolating that case study to other barrier islands based on our estimates of bay and
ocean-side areas and Census data on building density (Bureau of the Census, 1980).
Because Census data for barrier islands is limited, this part of the analysis
had to be conducted at a greater level of aggregation than the portion on sand costs; we
divided the coast into Northeast/Mid Atlantic, South Atlantic, and Gulf, ignoring the west
coast where there are no barrier islands. For each group, we randomly sampled coastal
towns for which Census data is available, and calculated the density of structures.
As Table 7 shows, our extrapolation equations assumed that the cost for
elevating structures could be described as Costslr = aslr * Shoreline_length + bslr *
buildings, where the number of buildings is area multiplied by housing density. Our
rationale was that the Long Beach Island study had found that most of the costs were
associated with roads and utilities. We derived parameters for this equation by assuming
that the length of primary roads is equal to the length of the island and that secondary
road mileage is proportional to the number of buildings, with constants of proportionality
and the cost of responding to 50, 100, and 200 cm rises in sea level based on Long Beach
Island. For the 50 cm scenario, we assumed that only the low bay sides would be raised;
for the 200 cm scenario, we assumed that entire islands would be raised. However, for the
100 cm scenario, we developed two equations to acknowledge that we are not certain whether
communities would elect to raise land currently above the 5 ft NGVD contour. On one hand,
much of this land is about 3 meters (10 ft) above NGVD, and hence would still be above sea
level; on the other hand, with a typical spring tide range of 200 cm, historic sea level
rise, and future subsidence, land at the 10 ft NGVD contour would only be about 60 cm
above spring high tide, which would leave the island vulnerable during severe storms.
Limitations
Because state-specific inventories and sand-cost assessments have been
conducted, our estimates of future dredging costs are probably much more accurate than our
estimates of the cost of elevating structures. Nevertheless, both sets of estimates faced
important data and model limitations.
Sand Quantities. Because of its geometric simplicity, the
"raising the profile" method of estimating the necessary sand is conceptually
appealing. But one must decide whether to use a short or long-term profile. Communities
could delay much of the costs by placing most of the sand on the upper part of the beach
(Titus, 1984). By using the annual storm to determine the profile, the low scenario fails
to account for the fact that the success of the project is only temporary; eventually the
entire profile must be raised. The high scenario may be more realisitc, but because there
is no research available to estimate the likely adjustment times, we had to simply take a
guess. Moreover, Hallermeier's estimates do not explicitly consider hurricanes and other
rare but severe storms, which might extend the effective profile even farther out to sea.
This limitation is particularly important given the possibility that hurricanes
may become more severe as global temperatures rise.
Unit Costs. Both our low and high estimates assumed that the
sand placed on the beach has sufficient grain size to remain within the beach system; we
lacked the data to estimate the quantity that would wash away. Moreover, both estimates
assume that developments in technology do not change the costs. Although improvements in
technology and economies of scale have the potential to lower the cost, future increases
in energy prices may offset these economies. Finally, we assumed that the same sources
would be employed for raising the active beach profile as for raising dry land; low-cost
material unsuitable for the beach may prove acceptable for raising building lots.
Our procedure for extrapolating the cost escalation in Florida to the rest of
the nation is even more suspect. In many areas, sand costs are currently higher because
they already have to go farther out to sea, suggesting a greater scarcity, which might
result in the escalation being greater; in other areas, there may be ample sand close to
the shore. We assumed that tidal deltas would not be mined for sand.
Costs of elevating structures. The Bureau of the Census does
not provide data on the number of structures for various barrier islands; one must infer
the density based on (1) data for jurisdictions that are entirely on coastal barriers, and
(2) by assuming that communities that include both mainland and barrier islands have a
uniform density. Moreover, our analysis treated all structures the same, even though
hotels or apartments would be far more costly to elevate than the single family homes that
dominate on Long Beach Island. Finally, the extrapolation of a single community to the
entire nation implies that our estimates are valid only if that community's costs are
"typical", a condition that can only be verified by examining other communities.
Results
Table 7 shows our estimates by state of the dredging costs that would accompany
the various scenarios of future sea level rise. A total of 3,100 kilometers (1,920 miles)
of shoreline would be nourished. An area of 7,233 square kilometers (2753 square miles)
would potentially have to be raised, one quarter of this after the first foot of sea level
rise. One-half to two-thirds of the nationwide cost would be borne by four
southeastern states: Texas, Louisiana, Florida, and South Carolina. However, much of the
costs for Texas result from the width of its barrier islands; if urban centers such as
Galveston are protected with levees, dredging costs for that state could be less than half
as great as shown here.
Table 8 illustrates our estimates of the cost of elevating roads and structures.
For a 50-cm rise, Gulf Coast barrier islands account for over 50 percent of the $32
billion cost, largely due to their lower elevations. By contrast, for a 2-meter rise, the
Mid-Atlantic and Northeast would account for over 50 percent of the $286 billion cost
because they are on average the most densely developed.
SUMMARY AND CONCLUSIONS
We estimate that shoreline retreat from a one meter rise in sea level would cost
the United States 270 to 475 billion dollars. Like all cost estimates involving
unprecedented activities, our estimates ignore the impacts we could not readily quantify
and those we can not foresee, and hence, are almost certainly too low. But policymakers
are accustomed to "soft" estimates, and we see no reason to believe that our
underestimates are any worse than the norm.
Table 9 summarizes our calculations. Thirty six thousand square
kilometers (fourteen thousand square miles) of land could be lost from a one meter rise,
with wet and dry land each accounting for about half the loss. For a few hundred billion
dollars, fifteen hundred square kilometers (six to seven hundred square miles) of
currently developed land could be protected, but the loss of coastal wetlands would be
that much greater.
Our estimates are optimistically low because we assume that it will only be
necessary to protect areas that are developed today, that is, about 15% of U.S. coastal
lowlands. If development continues and (1)we protect those areas as well, the economic
impact could be far greater because more dikes would be necessary and wetland loss would
be greater. If development continues but (2) we eventually abandon those areas, the
wetland loss will be the same as assumed in this article, but there could be a tremendous
loss of homes, offices, and infrastructures as the abandonment takes place. But (3)
prohibiting coastal development would also have costly impacts on the economy, which we
would have to add. Thus, this article is a servere underestimate of the nationwide cost of
sea level rise unless we implement a means of abandoning low-lying areas at little or no
cost.
At the national level, protecting developed coastal areas appears to be
cost-effective. The cumulative figure would be spread over one hundred years; even at the
end of the century, the annual cost of protection on barrier islands would be about $2,000
for a quarter-acre lot--hardly a welcome prospect for coastal property owners but
nevertheless one well worth bearing in order to maintain the property. The cost of
protecting developed mainland areas would be only about one-tenth as great.
The fact that it may be cost effective to protect property does not necessarily
imply that it would be in the interest of society to do so. We must also consider the loss
of natural shorelines and coastal wetlands that would result. Our results suggest that up
to a point, the objectives of protecting wetlands and coastal property may be compatible.
Abandoning developed areas would increase the area of surviving wetlands by only 5 to 10
percent--but at great cost. By contrast, limiting coastal protection to areas that are
already densely developed (and allowing currently-undeveloped areas to flood) would
increase the area of surviving coastal wetlands by 40 to 100 percent, depending on how
much the sea rises.
However, estimates in areal losses understate the differences in environmental
impacts for the various policy options. Although a substantial loss would occur even if
developed areas were abandoned, most of today's wetland shorelines would still have
wetlands; the strip would simply be narrower. By contrast, protecting all mainland
shorelines could result in wetlands being confined to a small number of isolated reserves,
a situation that humanity has already imposed on many terrestrial species.
Our results are consistent with the hypothesis of a 1987 study by the National
Academy of Engineering that shore protection will be cost-effective for most developed
areas (Dean et al., 1987). From the perspective of civil engineers, that study concluded
that little action is necessary today because shore protection structures can be erected
rapidly compared with the rate of sea level rise. However, the speed with which
communities could build these structures is small comfort to the birds and fish whose
habitat would be destroyed by doing so.
Sea level rise is an urgent issue for coastal environmental planners for the
very reason that it lacks urgency for directors of public works. If environmentalists do
not lay the necessary groundwork today to institutionalize a gradual abandonment of the
coastal plain as sea level rises, the public will almost certainly call upon engineers to
protect their homes in the years to come.
NOTES
1. Mean sea level refers to the average water level over the course of a year.
2. A critical question requiring research is: How rapidly is "too
rapidly"? If salinity is low, mangroves may be able to keep up with almost any rate
of sea level rise. In deltas and other areas with large sediment supplies, at least some
of the marshes can keep up with a 1-2 cm/yr rise. In this case, a net loss will still
occur, since there is a fixed sediment supply (e.g. a threefold increase in the rate of
relative sea level rise would eventually cut the area of the delta by a factor of three).
3. A 15-year storm is a storm whose flood levels have a probability of 1/15 of
being exceeded in any given year.
4. In many estuaries, the direct environmental impact of saltwater intrusion
would probably be greater than shoreline retreat, especially in the short run. However,
the upstream penetration of the salt front due to sea level rise can generally be offset
by relatively modest releases of freshwater from dams or river diversion structures, at a
modest cost compared with abandoning developed areas, erecting dikes, and other options
investigated by this report.
5. Assuming that storm frequency does not change, which is appropriate for an
analysis designed only for sea level rise. Note, however, that hurricanes require a water
temperature of 79oF to form; thus, as global temperatures warm, hurricanes may become more
frequent (Emmanuel, 1987).
6. For example, if the tidal range is 4 feet, mean high water is 2 feet above
sea level. If an area appears to be low marsh, we would assume that elevations are
uniformly distributed between 0 and 2 ft. MSL (which would generally be 0.5 to 2.5 ft
NGVD, since sea level has risen 6 inches
(DOWNLOAD THE ORIGINAL PDF FILE SEE
GRAPHICAL CHARTS, NOTES, CREDITS, REFERENCES AND FIGURES (774 KB FILE SIZE))
Top of Page
