1.1. Maps of Lands Close to Sea Level along the
Middle Atlantic Coast of the United States: An
Elevation Data Set to Use While Waiting for LIDAR

James G. Titus

Environmental Protection Agency
and

Jue Wang

Pyramid Systems, Incorporated

This section should be cited as:

Titus J.G., and J. Wang. 2008. Maps of Lands Close to Sea Level along the Middle Atlantic
Coast of the United States: An Elevation Data Set to Use While Waiting for LIDAR. Section
1.1 in: Background Documents Supporting Climate Change Science Program Synthesis and
Assessment Product 4.1, J.G. Titus and E.M. Strange (eds.). EPA 430R07004. U.S. EPA,
Washington, DC.

-------
Abstract

This report provides a coastal elevation data set for
the mid-Atlantic for purposes of assessing the
potential for coastal lands to be inundated by rising
sea level. Depending on what we were able to
obtain, our elevation estimates are based on
LIDAR, federal or state spot elevation data, local
government topographic information, and USGS
1:24,000 scale topographic maps. We use wetlands
and tide data to define a supplemental elevation
contour at the upper boundary of tidal wetlands.
Unlike most coastal mapping studies, we express
elevations relative to spring high water rather than
a fixed reference plane, so that the data set
measures the magnitude of sea level rise required
to tidally flood lands that are currently above the
tides. Our study area includes the seven coastal
states from New York to North Carolina, plus the
District of Columbia.

We assess the accuracy of our approach by
comparing our elevation estimates with LIDAR
from Maryland and North Carolina. The root mean
square error at individual locations appears to be
approximately one-half the contour interval of the
input data. We also compared the cumulative
amount of land below particular elevations
according to our estimates and the LIDAR; in that
context, our error was generally less than one-
quarter the contour interval of the input data.

We estimate that the dry land in the region has a
relatively uniform elevation distribution within
the first 5 meters above the tides, with about
1,200-1,500 km2 for each 50 cm of elevation.
With the exception of North Carolina, the area of
nontidal wetlands declines gradually from about
250 km2 within 50 cm above the tides to about
150 km between 450 and 500 cm above the
tides. North Carolina has approximately 3,000
km of nontidal wetlands within 1 meter above
spring high water; above that elevation, the
amount of nontidal wetlands declines gradually
as with the other states. North Carolina accounts
for more than two-thirds of the dry land and
nontidal wetlands within 1 meter above the tides.

We also compare our results to previous studies
estimating the region's vulnerability to sea level
rise. Our results are broadly consistent with an
EPA mapping study published in 2001, which
estimated the total amount of land below the 1.5-
and 3.5-m contours (relative to the National
Geodetic Vertical Datum of 1929). This study
appears to be a significant downward revision,
however, of EPA's 1989 Report to Congress. Our
estimates of the dry land vulnerable to a 50- or
100-cm global rise in sea level are less than one-
half the estimates of the Report to Congress. The
regional estimates of that nationwide study,
however, were based on a small sample.
Therefore, one should not extrapolate our mid-
Atlantic result to conclude that EPA's previously
reported nationwide estimate overstates reality
by a similar magnitude.

-------
1.1.1 Introduction

During the last two decades, the issue of rising sea
level has spread from being primarily a concern of
coastal geologists (e.g., Pilkey et al., 1982) and
those who measure the tides (e.g. Hicks et al.,
1983; Zervas, 2001) to an issue that concerns
planners, policymakers, and the public at large
(e.g., KRISTOFF, 2005; Dean 2006). One reason is
that the sea is rising 3 mm/yr or more along many
low-lying areas (Figure 1.1.1), enough for some
areas that were developed 50-100 years ago to be
flooded by high tides during new or full moons
(Figure 1.1.2). Another reason is that increasing
concentrations of carbon dioxide and other
greenhouse gases appear to be contributing to a
global warming responsible for at least part of the
current rate of sea level rise (e.g., U.S. EPA, 1996;
IPCC, 2007). Most scientists expect greenhouse
gases to accelerate the rise in sea level (IPCC,
2001a), and some have suggested that it may
already be doing so (Church and White, 2006).

Rising sea level inundates low-lying lands, erodes
wetlands and beaches, exacerbates flooding, and
increases the salinity of estuaries and aquifers
(e.g., IPCC, 2001b). Studies over the last two
decades have identified numerous decisions that
may be sensitive to sea level rise (e.g., NRC,
1987; Williams et al., 1995; Titus and
Narayanan1, 1996). During the Administration
of President George W. Bush, the U.S. Climate

Change Research Program (2003) has actively
promoted decision support research to assist with
adaptation to consequences of climate changes
such as rising sea level. Studies sponsored by the
U.S. EPA have suggested that local governments
may be making the most important decisions
regarding the eventual impact of rising sea level on
the United States. Local governments create the
land use plans and issue the construction permits
that determine whether the areas at risk will be
developed enough to require shore protection as
the sea rises or will remain vacant enough for
wetlands to migrate inland (Titus, 1990, 1998).

Over the last several years, EPA staff and
contractors have met with local governments
concerning possible responses to sea level rise
(Titus, 2005). When we have asked what
information might help them to better prepare, the
most common answer has been better elevation
maps. When senior government officials or
newspaper reporters have asked us about
vulnerability to sea level rise, the most common
request has been for a map showing the lands that
might be flooded. Yet maps depicting lands close
to sea level using the best available data are
unavailable for most areas.2

1 Section 3.1 of that paper is an overview of decisions that
depend on the probability of the sea rising a particular
magnitude.

But see Weiss and Overpeck (2006), which provides a
map server using the USGS national elevation data series.

-------
[ 4 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

V

3

*4

»• i".





Rate of Sea Level Rise

•	< -2.00 mrm/yr
•-2.00 to 0.00

•	0.01 to 1.25

•	1.26 to 2.50
2.51 to 3.75

•	3.76 to 6.00

•	> 6.00 mm/yr

Figure 1.1.1. Relative sea level rise at locations with at least 50 years of tide station data. (Data Source:
PERMANENT SERVICE FOR MEAN SEA LEVEL, 2003).

For many years, the EPA sea level rise project
avoided this obvious endeavor because we
expected LIDAR3 data (hereafter LIDAR) to be
available soon, which would make moot the entire
exercise. But the LIDAR was slow in coming.
Finally, we decided that a better way to manage
the risk of creating unnecessary information would
be to create an elevation data set anyway; if the
LIDAR does not arrive, then we will have
provided a useful elevation data set; if LIDAR
does become available, then the public will have
even better maps.

This report presents the methods we used to create
maps and a dataset for analyzing the impacts of sea
level rise. We provide elevation data for the
coastal zone from New York to North Carolina.
The purpose of this data set is to identify and
quantify the land that could potentially be

3Light Detection And Ranging (LIDAR) is similar to
RADAR, except it relies on light instead of radio waves.
The LIDAR instrument transmits light out to a target; the
time it takes for the light to return is used to determine
distance. Land elevations are estimated with low-flying
aircrafts with LIDAR instruments.

inundated as sea level rises, so that EPA and other
researchers can (1) evaluate the land potentially
available for wetland migration, (2) identify the
areas that might require shore protection and
quantify shore protection costs, and (3) estimate
the population and assets within the area
potentially at risk to sea level rise. Although this
report has only a few example maps, map
templates accompany the data set that we are
distributing so that those with GIS software can
easily create elevation maps to suit their needs.

Although our focus is on coastal elevations rather
than scenarios of inundation, the elevation of our
study area was broadly guided by the available
literature estimating future sea level rise. IPCC
(2007) estimated that global sea level is likely to
rise 18-59 cm over the next century, but also
indicated that the sea could rise 9-17 cm more if
polar ice sheets begin to disintegrate.4 Along the
mid-Atlantic coast, sea level rise is generally
expected to be 10-20 cm more than the global
average rise.5 Thus, as you examine our maps, you

4IPCC (2007) at table 10.7.

'See, e.g., Titus and Narayanan (1996) at Chapter 9.

-------
[ SECTION 1.1 5 ]

Photo 1.1.1. Tidal flooding at Ship Bottom, New Jersey (Labor Day 2002).

might reasonably assume that sea level is likely to
rise 30-100 cm in the next century. But IPCC also
estimates that by the 21st century, global sea level
could be rising 4-14 mm/yr,0 which would imply a
rise of 5-16 mm/yr along the mid-Atlantic coast.
Thus a rise of several meters over the next few
centuries is possible. Our maps all provide
elevations up to either 3 or 6 meters above the ebb
and flow of the tides. Our primary motivation for
extending the maps this far inland was to convey at
least a rough sense of the topography of the coastal
zone, and doing so requires one to look above the
elevation that is most immediately at risk.
Nevertheless, sea level could rise 3-6 meters over
the next few centuries.

The next section (1.1.2) discusses our general
approach, which was to obtain the best available
elevation data from the U.S. Geological Survey
(USGS) and other federal, state, and local
government agencies; create an extra contour at
the inland boundary of tidal wetlands; and express
elevations relative to the ebb and flow of the tides.
After that, we describe (1.1.3) how we applied that

approach to the data we were able to obtain, and
explain (1.1.4) our accuracy assessment. The final
section (1.1.5) presents the maps, estimates the
area of land close to sea level, and compares our
results to previous assessments.

Before proceeding with the analysis, a word of
caution on the use of units in this report. We
generally use metric units, with English units in
parentheses where we cite a report whose results
originally used English units. However, when
discussing contour intervals of specific maps used
in the analysis, we refer to the units that the maps
actually used, which are often English units. We
believe that it is more accurate to say (for
example) that the USGS maps of Maryland
include 1-m and 5-ft contour intervals, than to say
that they include 1-m and 1.524-m contours.
Although most writers would normally prefer to
avoid mixing units of measurement, the underlying
reality is that there is currently a relatively
confusing patchwork of available elevation data
sets, and different units of measurement is part of
that reality.

bSee IPCC (2007) at Table 10.7 (high estimate includes
lines called "sea level rise" and "scaled-up ice sheet
discharge").

-------
1.1.2 General Approach

This study is based on the relationship between the
tidal elevations, tidal wetlands, and the reference
elevations used by available elevation data. Figure
1.1.2 illustrates the relationship between these
three factors along a typical shore profile, using
the tidal elevations for Hampton Roads (VA). In
this particular case, mean sea level is 17.2 cm
above NGVD29, which is the reference elevation
used by the USGS topographic maps. Spring high
water is 43 cm above mean sea level, and thus 60
cm above NGVD29. Thus, the 5-ft contour is only
90 cm (3 feet) above spring high water. Because
tidal marshes are found between mean sea level
and spring high water, the 5-ft contour is also 90
cm (3 feet) above the tidal wetlands.

Our general approach has five main steps:

1.	Obtain the best elevation data from usual
sources of topographic map data, such as the
USGS, as well as state and local governments
and other federal agencies.

2.	Use wetlands data to determine the location of
the upper boundary of tidal wetlands, which
we treat as the land flooded by spring high tide

7	Older maps generally measure elevations relative to the
National Geodetic Vertical Datum of 1929, which was
originally meant to be a fixed reference plane. NGVD was
set equal to the sea level of 1929 at specific reference
stations along the U.S. coast. The reference "plane"
(actually a spheroid) in all other locations was based on
leveling techniques. As a result, even in 1929, NGVD was
not sea level in areas where average water levels diverge
from the ideal "plane" because of winds, freshwater
inflow, and other factors. Since 1929, rising sea level and
subsidence have caused sea level and the NGVD to
diverge 10-20 cm in most areas. Recognizing the
problems with the deteriorating benchmarks, the USGS
and the National Geodetic Survey converted to the North
American Vertical Datum (NAVD) of 1988. The reference
plane associated with this benchmark is based on a single
fixed site. New data generally are relative to NAVD-1988.
See, e.g., NATIONAL GEODETIC SURVEY et al.

(1998).

8	Spring tides refer to the extreme tides that occur during
new and full moons, when the tidal forces of the moon and
sun are aligned. Spring high water is the average height of
high water during spring tides. See, e.g., NOS (2000).

and, hence, the horizontal position of our
wetland supplemental contour.

3.	Use tidal data to estimate the elevation
(relative to NGVD29), of spring high water,
which we use as the vertical position of our
wetland supplemental contour.

4.	Interpolate elevations relative to the vertical
datum for all land above spring high water
using elevations obtained from the previous
three steps.

5.	Use the information from step 3 to calculate
elevations relative to spring high water.

Figures 1.1.3 and 1.1.4 illustrate the results of
these steps for a portion of Long Beach Island
(New Jersey) and the adjacent mainland, including
the portion of Ship Bottom (Figure 1.1.2) that is
often flooded by spring tides. The USGS maps
have a 10-ft contour interval (Figure 1.1.3a), but
the U.S.

Army Corps of Engineers provided spot elevation
data for the islands and some of the mainland
(Figure 1.1.3b). For the mainland areas without
spot elevation data, we created a supplemental
contour representing spring high water and the
upper edge of tidal wetlands. The wetlands data
define the horizontal position of this contour
(Figure 1.1.3a). We used tidal data to define the
vertical position of the contour relative to
NGVD29 (Figure 1.1.4). With that supplemental
contour defined, we interpolated elevations in
between the contours (Figure 1.1.3c), which yields
elevations relative to NGVD29. Finally we
subtract the tidal elevations from Figure 1.1.4 to
express land elevations relative to spring high
water (Figure 1.1.3d).

Difference from Other Elevation Mapping
Assessments

Our approach differs from other elevation mapping
studies in two fundamental ways. First, our final
product represents elevations above the tides rather

-------
[ SECTION 1.1 7 ]

Transition

Upland

Open Wfeter
(subtidal)

Figure 1.1.2. Relationship between tides, wetlands, and reference elevations for an example estuarine
shore profile. The example elevations are based on the Hampton Roads (Virginia) Tide Station. See Gill

and Schultz 2001. The wetland characterizations are based on Kana et al. 1988.

11

than above a fixed reference elevation such as
NGVD29. Second, we use tidal wetlands data to
produce a single (but important) supplemental
elevation contour, in addition to the conventional
topographic information.

We estimate elevations relative to the sea
because the intended use of these maps is to
analyze the implications of sea level rise. Early
assessments often ignored the difference
between the NGVD29 and mean sea level.
Because (local) mean sea level tends to be 10-20
cm above NGVD29, equating these two
reference elevations was harmless when
analyzing the impact of a 4.5-7.5-m (15-25-ft)
sea level rise by 2030 (SCHNEIDER and Chen,
1980), or even a 50-300-cm rise by 2100
(Barth and Titus, 1984). A more recent
analysis provided maps relative to NGVD29 for
the U.S. Atlantic and Gulf coasts, with an
explicit warning about the difference between
that benchmark and sea level (Titus and
RICH MAN, 2001). The print media generally
ignored the caveat and rewrote the map key from
"1.5 meters above NGVD" to "future shoreline
resulting from 1.5 meter rise in sea level," not
only confusing NGVD29 with sea level but also
equating elevation with shoreline change.'

ySee, e.g., "Coasts in Peril: Exhibit E" in "Life in the
Greenhouse", Time. 157:14:24,29 (April 9, 2001) "These
maps show how much of the shoreline we know today will

Unfortunately, the lack of interest in tidal datums
is not limited to sea level rise assessments: In New
Orleans, flood control engineers used NGVD29
and mean sea level interchangeably for decades,
even though mean sea level was 50-60 cm higher
than NGVD29. As a result, the levee along the
Inner Harbor Navigation Canal (which failed
during Hurricane Katrina) was about 60 cm lower
than intended (INTERAGENCY PERFORMANCE
EVALUATION TASKFORCE, 2006).

It is axiomatic that maps ought to depict the
information they seek to convey rather than leave
it up to the reader's imagination or ability to obtain
additional data. Unfortunately, the absence of a
data layer relating sea level to the fixed vertical
benchmarks has made it impractical for coarse-
scale national mapping studies to provide
elevations relative to the sea.10 As a byproduct of

vanish if sea levels rise by the indicated amount." But see
the -Yew York Times, January 1, 2000 (closely
paraphrasing the caveat that the journal article had
recommended).

' 'Consider for example, the state-scale maps showing land
below the 1.5- and 3.5-m (NGVD) contours in Titus and
Richman (2001). Expressing elevations relative to the
tides would have more than doubled the $75,000 cost of
that study.

Gill. S.K. and J.R. Schultz. 2001. Tidal Datums and
Their Applications, NOAA Special Publication NOS CO-
OPS 1. February 2001.

-------
[ 8 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Figure 1.1.3. Estimated elevations around Long Beach Isiand, New Jersey. The first three maps
show elevations relative to NGVD29 according to (a) the USGS 1:24,000 scale map, (b) spot
elevations provided by the Corps of Engineers where available and USGS data elsewhere, and (c)
our interpolations using wetlands data as a supplemental contour. The final map (d) shows the
same elevations as (c), relative to spring high water. The first map also shows the location of
spring tide flooding depicted in Photo 1.1.1.

I 1200 - 250
r~~l 150-200
¦1100-150
H50-100
M <50
Tidal
™ Wetlands

-------
[ SECTION 1.1 9 ]

Centimeters above
NGVD 1929

¦ 60 - 80
80-100
I I 100-120
Ei > 120

# Tidal Range
Measurement

Measurement

Figure 1.1.4. Elevation of spring high water
relative to NGVD29. The red and black dots
depict the locations of reported observations for
mean tide level and spring tide range,
respectively. The various colors of the land
represent the interpolated values of spring high
water. For the actual values of mean tide level
and tide range observations, see Figures 1.1.6
and 1.1.7, respectively.

this effort, we create such a layer that others may

12

find useful even as LIDAR becomes available.

In ordinary conversations, people refer to
elevations "above sea level." This report provides
elevations relative to spring high water instead, for
two reasons: First, one can map the existing high
water mark more accurately than mean sea level
Wetland maps generally show the upper boundary
of tidal wetlands, but maps illustrating the location
of the shore at mean tide level are rarely available.

Second, showing elevations relative to high
water is more useful than mean sea level,
because the character of land changes
fundamentally once it is subject to the ebb and
flow of the tides: Marsh grasses replace trees,
lawns, or crops that cannot tolerate the saltwater

"NO A A is developing a software tool that converts data
between fixed benchmarks and elevations relative to the
tides, making this aspect of our analysis obsolete as well in
a few years. See, e.g., Parker et al. (2003) and Myers
(2005). Results are available for Pamlico Sound, which we
used.

13

and frequent flooding ' (see generally Teal and
Teal, 1969). Moreover, the land becomes
subject to tidal wetlands regulations,14 and
ownership shifts from the upland owner to the
public in most states (SLADE et al. 1990). By
contrast, elevation above mean sea level implies
little about the impact of sea level rise. Because
tide ranges vary, knowing that a parcel is 50 cm
above mean sea level does not tell one whether it
is even wet or dry land, let alone the rise in sea
level necessary to convert the area to open water.

Difference from Other Sea Level Rise
Impact Mapping Studies

This effort also differs from most sea level rise
impact mapping studies because we report
elevations rather than projected future shorelines.
Maps of future shorelines are important, but
elevations alone say something about vulnerability
to sea level rise.

Converting our results into maps of future
shorelines would represent, in effect, a separate
study—and the final results would be more
speculative. Elevation is a necessary precursor for
estimating shoreline change due to sea level rise.
But projecting future shorelines requires more
questionable assumptions than one must make
when estimating elevations. The Bruun (1962)
Rule produces an approximation of sandy beach
erosion that is useful for some purposes, but many
geologists decline to project erosion of beaches
without applying a more site-specific model

1 'This generalization does not always apply in areas with
low salinity. In nanotidal areas (i.e., areas where the
astronomic tide range is only a few centimeters) the "tidal
wetland" vegetation may be irregularly flooded because of
winds rather than regularly flooded from astronomic tides.
There may be a gradual transition between the irregularly
flooded wetlands and adjacent nontidal wetlands, and even
if the line is well-defined, the elevation is not a function of
the tides. The Pamlico and Albemarle sounds are the most
important example in our study area. Other exceptions
include tidal freshwater forests and areas where extensive
tidal freshwater wetlands are adjacent to nontidal wetlands
with similar vegetation types.

"Clean Water Act § 404, 33 U.S.C. § 1344(a) (1994) and
Rivers and Harbors Act of 1899, 33 U.S.C. §§ 403, 409
(1994).

-------
[ 10 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

requiring data collected over several years (e.g.,
Dean and Maurmeyer, 1983; Cowell and
Thom, 1994; Cowell et al., 1995; Young and
Pilkey, 1995). And sandy beaches are the best
known shores! Wetland accretion is too poorly
understood for coastal scientists to quantify how
rapidly the sea could rise before it began to drown
the wetlands (Kana et al., 1988; Park et al.,
1989; Cahoon et al., 1995).15 The ability to
predict erosion of muddy shores is so poor that for
many geologists, the term "coastal erosion" refers
only to sandy beaches.

Even in areas where we have a good model of
shore erosion, projecting future shoreline can
require a rather cumbersome set of analytical steps.
Future sea level rise is uncertain; so one must
evaluate the implications of several scenarios,
taking care to ensure that one has encompassed the
range of uncertainty. Different readers have
different time horizons, so one typically must
prepare maps for a few different projection years.
Finally, actual shoreline migration will also
depend on the type and extent of human activities
to hold back the sea, so one must consider
alternative shore-protection scenarios. Handling all
these issues well on a regional scale requires too
much effort to be undertaken as a final step of this
study.

Fortunately, coastal elevations do tell us something
about the impacts and responses to sea level rise.
Coastal wildlife managers who want to ensure that
new wetlands are created as sea level rises must
identify the dry land that might be tidally flooded
in the future. The importance of reserving a given
parcel depends on how much the sea has to rise
before the tides inundate the land. The need to
elevate streets and back yards depends on the
land's elevation. Moreover, an elevation map
makes a suitable graphic for those attempting to
convey the broad ramifications of sea level rise to
the general public because it shows both existing
wetlands and the dry land that will be inundated as
the sea rises.

15Elsewhere in this report however, a panel of wetland
accretion specialists provide a consensus subjective
assessment about whether mid-Atlantic wetlands could keep
pace with three sea level rise scenarios. (See Reed et al..
Section 2.1 of this report.)

-------
1.1.3 Application of Our Approach

Step 1: Obtain Best Elevation Data

Table 1.1.1 summarizes the elevation data we
used. For a given state, the order in which the table
lists the data represents the quality of the data and,
hence, the order in which we selected data layers.
For example, in Maryland, we used the 10-ft
contour from the Department of Natural Resources
spot elevation data set in areas where the USGS
maps had a contour interval of 20 feet.16 If USGS
7.5-minute maps had a contour interval of 10 feet
(or better), however, we preferred the USGS data.
For four counties, we had county data with 2- or 5-
ft contours, which was even better.

Our approach was more systematic than one might
initially assume, given the variation in data quality
indicated by Table 1.1.1. Our goal was to estimate
the land potentially inundated by a 1-m rise in sea
level where possible and, where we could not, at
least map the area vulnerable to a 2- or 3-m rise.
Although we would have preferred to rely solely
on a nationwide data set, the USGS 7.5-minute
maps do not have a consistent contour interval—
and for much of the coast their contour intervals
are too great, especially in New Jersey and
Maryland (see Figure 1.1.5). Therefore, we
attempted to supplement the USGS data where
feasible.

Outside North Carolina, the USGS 5-ft contour is
generally within 1 meter above the ebb and flow of
the tides. Therefore, outside North Carolina,
wherever the USGS maps had a contour interval of
5 feet or better, we did not actively seek better
data.

16We obtained LIDAR for the lower Eastern Shore of
Maryland after the analysis was complete. We use that
data to assess the accuracy of our DEM in the section on
Quality Control and Review. The primary data set we
make available to the public will include these LIDAR
data; we will also make the original data set available.

A 20-ft contour interval, by contrast, provides no
information about lands vulnerable to sea level rise
in any meaningful time horizon (although it does
identify areas that are not vulnerable). Therefore,
we made a relatively exhaustive effort to obtain
alternative data in the portions of New Jersey and
Maryland where the USGS maps had a 20-ft
contour interval. Fortunately, Maryland had spot
elevations on a 90-m grid with a vertical precision
(90 percent interval) of 5 feet. Thus, according to
national map accuracy standards (Bureau of the
Budget, 1947; Federal Geodetic Control
Subcommittee, 1998), the Maryland data provide
a 10-ft contour interval at a 1:180,000 scale. The
horizontal scale is considerably poorer than the
USGS 1:24,000 maps; but unlike a map with a 10-
ft contour interval, the spot elevations provide
estimates for points with intermediate elevations,
allowing us to derive, for example, a 5-ft contour
(albeit with twice the vertical error of national
mapping standards). Unfortunately, New Jersey
had no similar statewide data set.

Many counties have elevation data for coastal
floodplain management, pollution runoff
modeling, and identification of areas where slopes
make land undevelopable. Unlike the federal
government, however, the counties usually charge
for the data—sometimes tens of thousands of
dollars per quad. In some cases, the bonds used to
raise the money to collect the data contain

17

restrictions against giving the data away. The
restrictive county policies generally allow the GIS
department to provide the data to a genuine partner
doing work primarily to benefit the county. This
study probably would not—by itself—qualify
because we are analyzing the vulnerability of a
multistate region to rising sea level and creating a
product for researchers who will not, in general,
collaborate with county staff to attain county
objectives. Nevertheless, our collaboration with

17The planning director of Monmouth County, New Jersey,
expressed this concern.

-------
[ 12 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

four counties led four county GIS departments to
see this effort within the context of a joint federal-
local partnership to understand the implications of
rising sea level:

•	Monmouth County (New Jersey): The only
county along New Jersey's Atlantic Coast
where USGS maps have a 20-ft contour
interval.

•	Anne Arundel County (Maryland): This
county includes both Annapolis and the largest
low-lying area on Maryland's Western Shore
of Chesapeake Bay.

•	Harford County (Maryland): This county
includes the second largest area of very low-
lying lands along the Western Shore.

•	Baltimore County (Maryland). Unlike the
other counties, Baltimore insisted that we
provide elevation maps using their superior (2-
ft contour) elevation data before they would
even consider responses to sea level rise.

We also examined some maps that had been stored
in the warehouse where FEMA keeps the
documentation for the flood insurance rate maps.
In general, whenever the USGS maps had contour
intervals greater than 5 feet, FEMA obtained
topographic maps. As a test, FEMA searched their
archives for specific communities in Monmouth
County, New Jersey, and found that for about half
the townships and boroughs, the archives
contained numerous maps with 2-fft contours at a
scale better than 1:10,000. We were tempted to
have those maps all digitized. FEMA, however,
was reluctant to allow the entire collection to leave
their premises—and we were not sure that the
effort was worthwhile for areas with only partial
coverage. We did persuade FEMA to lend us their
map of Kent Island, Maryland, the eastern landing
of the Chesapeake Bay Bridge, where our own

eyes told us that the land is very low but the USGS
maps have a 20-ft contour interval.

The supplemental data sources left us with only 24
quads where we have nothing better than a 20-ft

18

contour interval. All of those quads are along
tidal rivers well inland or upstream from a major
estuary, except for six quads along the north shore
of Long Island, which is dominated by substantial
bluffs, and three quads in northern New Jersey (see
Figure 1.1.5).

Maps with 10-fft contour intervals give some
insight on vulnerability to sea level rise, but not
enough to justify their use if one can find a
practical alternative. Unfortunately for us, most
USGS maps have 10-fft contours in coastal New
Jersey, New York, Pennsylvania, Virginia, the
District of Columbia, and Virginia west of
Chesapeake Bay. For most low land along the
Atlantic Ocean and back barrier bays in New
Jersey (except for Monmouth County, where we
had 2-ft contours), the Corps of Engineers
provided spot elevations with sufficient precision
and density to identify 4-ft contour intervals at a
1:100,000 scale. The District of Columbia
provided 1-m contour data. The City of
Philadelphia provided 2-ft contours. Most of
Pennsylvania's remaining low land is in Delaware
County; because of the high tide range, the 10-fft
contours in that region are only about 150 cm (5
feet) above spring high water.

North Carolina is a special case. Currituck,
Pamlico, and Albemarle sounds have almost no
tides because the areas of these bodies of water are
large compared to their inlets to the ocean. With
the high water mark barely above sea level, the sea
would have to rise more than 1 meter to inundate
the 5-ft contour during a high tide, unlike areas
with larger tidal ranges. In the wake of Hurricane
Floyd, however, the state collaborated with FEMA
to substantially improve the already-good
elevation data with LIDAR. Early on in the study,

18Those 24 quads also included all or part of the upper
tidal portions of the Delaware River (Bucks County,
Pennsylvania, and Burlington County, New Jersey),
Choptank River (Caroline County, Maryland) Wicomico
River (Worcester County, Maryland), and several small
rivers or creeks in New Jersey.

-------
[ SECTION 1.1 13 ]

we obtained LIDAR for most of the low-lying
counties in the state19; and by the end of the study
we had data for the entire state. As we discuss
below, however, the absence of tides and tide data
for this area diminishes the usefulness of our
analysis for evaluating the possible impacts of sea
level rise in North Carolina.

Step 2: Use Wetlands Data to Obtain the
Location of the Upper Boundary of Tidal
Wetlands

We used tidal wetlands to define a supplemental
topographic contour, approximately equal to
spring high water. The precise elevation of that
contour varies, but it is almost always between
zero and the lowest contour above zero. This
supplemental contour is useful and important for
two reasons: First, for many purposes we are
interested in knowing elevations above the tides;
so a contour that defines the upper boundary of the
tides is essential. Second, where elevation
information is poor, a supplemental contour is
likely to be more accurate than elevations
estimated by interpolating with a model.

Table 1.1.2 lists our wetlands data sources. Just as
the USGS provides 7.5-minute quadrangles at a
1:24,000 scale for topography, so too the US Fish
and Wildlife Service's National Wetlands
Inventory (NWI) provides 1:24,000 maps with
broad wetland categories. Several states, however,
have developed their own wetlands maps;
representatives from New Jersey, Maryland, and
North Carolina asked us to use their data instead of
the NWI maps.20

The key limitation of the NWI data is its age: the
aerial photographs for New Jersey were from the
1970s, and the Maryland and North Carolina

19As we discuss, we obtained LIDAR for the rest of the
state after we developed our elevation data. As with the
Eastern Shore of Maryland, we use these data to assess the
accuracy of our procedure and will make the better
LIDAR data available to the public.

2uNew York also provided wetland data for a portion of its
coastal zone. Delaware also has its own wetland data, but
state officials did not specifically ask us to use their data,
and given the cost of interpreting each new data set, we
did not.

21

photographs were from the 1980s. Since then
wetland shores have eroded, low dry land areas
have converted to wetlands, human activities have

22

converted wetlands to dry land, and some
previously drained areas have converted back to

23

wetlands. A second limitation of NWI is scale.
Small fringing wetlands along tidal creeks
sometimes do not show up in the NWI data set,
even though they are large enough to be seen on a
1:24,000 scale map. Given these limitations, and
the availability of data that state agencies trust
more for their uses, we took the three states' advice
and used their data.

We use wetlands only to define the inland limit of
tidal wetlands. Kana et al. (1988) originally
proposed the approach that we apply here. While
surveying marsh transects around Charleston
(South Carolina) and Long Beach Island (New
Jersey), they recalled that low marsh is generally
flooded twice daily and high marsh is flooded at
spring tides but not every day. With an estimate of
mean high water and spring high water, they
reasoned, the wetland zonation can give
supplemental elevation contours at both mean high
water24 and spring high water. PARK et al. (1989)
first applied that approach. Although their
LANDSAT imagery did not distinguish between
low and high marsh vegetation, Park et al.
attempted to do so by obtaining imagery at high
tide during "half moons" (i.e., at mean high water)
and delineating the flooded areas. The NWI and
state wetlands data we used, however, made no
such distinctions.

01See NWI Status Photo Page, accessed April 1, 2005, at
http ://www. nwi.fws. gov/statusphotoage. htm.

"Tidal wetlands are rarely converted to dry land for
development, but occasionally some loss will be permitted
for water-dependent uses such as marinas and ports. See
generally U.S. EPA and U.S. Army Corps of Engineers
(1990) (explaining the federal policy on wetland
mitigation under section 404(b)(1) of the Clean Water
Act).

23Along Delaware Bay, for example, diked wetlands had
been converted to agriculture for more than a century. As
part of an enviromnental mitigation program for a PSE&G
nuclear power plant, most of the coastal zones of
Cumberland County, New Jersey, and areas across the Bay
in Delaware are being returned to nature.

24Mean high water is the average water level at high tide.

-------
[ 14 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Table 1.1.1. Elevation data sources used in original analysis.

Area Included

Data Source

Scale

Contour Interval or
Eauivalent Precision

Benchmark

New York

Entire state

USGSa

1:24.000

5 or 10 ft

NGVD29

New Jersey

Monmouth County

County datab

1:1,200

2ft

NAVD88

Atlantic coast east of US-9

Corps of Engineers, spot elevations0



2ft

NGVD29

Atlantic, Delaware Estuary

USGS

1:24,000

5 or 10 ft

NGVD29

North Jersey

USGS

1:24.000

10 or 20 ft

NGVD29

Pennsylvania

Philadelphia

City datad

1:2,400

2ft

PVDd

Delaware and Bucks counties

USGS

1:24.000

10 and 20 ft

NGVD29

Delaware

Entire state

USGS

1:24.000

Mostly 5 ft

NGVD29

Maryland

Baltimore County

County data®

1:100

2ft

NAVD88

Anne Arundel County

County data'

1:2,400

5ft

NAVD88

Harford County

County data9, excludes Aberdeen
Proving Grounds

1:2,400

5ft

NGVD29

Kent Island

Hard-copy map FEMA used for flood
insurance rate maph

1:7,200

2ft

NGVD29

Dorchester County and
nearby

Southern part of state, both E

USGS

1:24,000

1 m

NGVD29

USGS

1:24,000

5ft



and W shores



Potomac and Western Shore

USGS

1:24,000

Mostly 10 ft,
some 20 ft

NGVD29

Statewide except for a few
small areas

Maryland DNR spot elevations'

90-m grid

10ft

NGVD29

Northern Eastern Shore

USGS 1:24.000

1:24.000

Mostly 20 ft

NGVD29

District of Columbia

Entire district

City dataj

1:1.000

1 m

NAVD88

Virqinia

Entire state

USGS

1:24,000

5 and 10ft

NGVD29

State LIDAR project with FEMA*

40 cm
Mostly 5 ft, some 2 m

NAVD88
NAVD88

North Carolina

Most of Pamlico and
Albermarle sounds to ocean

Elsewhere	USGS	1:24,000

a USGS. Large Scale Digital Line Graphs, http://edc.usgs.gov/products/map.html accessed May 1, 2006.
b U.S. Army Corps of Engineers, St. Louis District. 1999. Intercoastal Waterway, NJ: Spot Elevations (LFHYPELS)._Prepared by ADR,
Inc., Pensauken, NJ. Vertical position accuracy: 1 ft. Horizontal position accuracy: 5.0 ft.

Monmouth County Office of Geographic Information Systems. 1997. Contours. Contour interval: 2 ft. Scale: 1:1200. Complies with
National Map Accuracy Standards.
d City of Philadelphia Water Department, Information Systems and Technology. 1996. Philadelphia Vertical Datum. "The Philadelphia
datum was first established in 1682 by Wlliam Penn with a metal spike in the Delaware River pier at the foot of Chestnut St based on
the mean height." Metadata file accompanying Philadelphia 2-ft contours. NAVD (1988) is 4.63 ft lower than the PVD.

Baltimore County, Maryland. 1997. Baltimore County Topo Data. Towson, Maryland: Baltimore County OIT/GIS Services Unit. Complies
with standards of the American Society Photogrammetry and Remote Sensing as well as with National Map Accuracy Standards.
f Anne Arundel County, Maryland. 1995. Anne Arundel County 1995 Topographic Mapping. Prepared by Photo Science, Inc. (now
EarthData International, Inc.) for Anne Arundel County, Department of Public Works. Scale: Annapolis: Anne Arundel County Office of
Information Technology. Complies with National Map Accuracy Standards.

9 Harford County, Maryland, undated. Harford County 5-ft contour elevation maps. Contour Aberdeen: Harford County GIS Department.
h GEOD Surveying and Aerial Mapping Corporation. Kent Island, Maryland. Map prepared for the Flood Insurance Rate Maps of Kent
Island, Project No. 1381-107. Archived by Dewberry and Davis, Annapolis, Maryland. Provided by the FEMA Flood Insurance
Administration. Scale: Contour. Complies with National Map Accuracy Standards.

' Maryland Department of Natural Resources. 1992. Digital Elevation Models. Vertical position accuracy: 5 ft. Horizontal Accuracy, 33 ft.
' National Capital Planning Commission and District of Columbia Department of Public Works. 2001. Rooftop Elevation and Ground

Elevation. Washington, D.C.: Office of Chief Technology Officer. Complies with the National Map Accuracy Standards.
k Floodplain Mapping Program, North Carolina Division of Emergency Management. May 2002. NC Floodplain Mapping: 50 ft

Hydrologically Corrected Digital Elevation Modelv.1. White Oak, Tar-Pamlico, Neuse, and Pasquotank basins. Vertical accuracy: 20 cm
for coastal counties.

-------
[ SECTION 1.1 15 ]

Figure 1.1.5. The elevation data used in this study. Rectangles generally signify USGS 1:24,000 data.
The USGS maps had a 20-ft contour interval for the (pink) quads in Maryland where we used state
data, and most or all of the four counties where we obtained 2- or 5-ft contour data (but 10-ft contours
in the City of Philadelphia). We obtained the Maryland LIDAR and some of the North Carolina LIDAR
after interpolating the elevations, and hence use that data to assess the accuracy of our approach.
The final data set we provide has LIDAR for all of North Carolina.

ATLANTIC
OCEAN

Contour Intervals

| Spot Elevation
Lidar

H 2 Feet

| 1 Meter
5 Feet
10 Feet

10 Feet, State Data
20 Feet

-------
[ 16 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Table 1.1.2. Wetlands data used in this study

State

Countv

Data Source

Year for
Imaaerv

Scale



Suffolk, Nassau, Rockland,

NWIa

1980,1990

1:24,000



Hudson above Tappan Zee

New York

New York City, Westchester:
Long Island Sound and
Hudson below Tappan Zee

New Yorkb

1974

1:12,000



All but part of Delaware River

Rutgers Land Cover0

1995

30-m grid

New Jersey

Delaware River upstream of
Commodore Barry Bridge

NJ Upper Wetland
Boundary01

1970

1:12,000

Pennsylvania

Entire state

NWIa

1980

1:24,000

Delaware

Entire state

NWIa

1982

1:24,000

Maryland

All but three areas where
data laggedf

Three areas where DNR data
laggedf

MD-DNR8
NWIa

1988-1995
1980s

1:12,000
1:24,000

District of
Columbia

Entire district

NWIa

1983

1:24,000

Virginia

Entire state

NWIa

1990,2000

1:24,000

North Carolina Entire state	NC DENRS	1981-1983 and 1994 1:24,000

a U.S. Fish and Wildlife Service. National Wetlands Inventory, http://www.nwi.fws.gov/ [accessed January 2006.] Scale: 1:24,000. The following
types of polygons were treated as tidal wetlands: M2RS (Rocky Shore), M2US (Unconsolidated Shore), E2RS (Rocky Shore) E2US
(Unconsolidated Shore), E2EM (Characterized by erect, rooted, herbaceous hydrophytes), E2SS (Scrub-shrub), E2FO (Forested). In
addition, for areas with a water-regime characterized as N (Regularly Flooded-exposed daily), P (Irregularly Flooded, less than daily), S
(Temporary-Tidal), or R (Seasonal-Tidal), we also used L2EM and PEM (both characterized by erect, rooted, herbaceous hydrophytes),
L2RS (Rocky Shore), L2US and PUS (Unconsolidated Shore), PML (Moss-lichen), PSS (Scrub-shrub), and PFO (Forested). In cases
where the water regime was unknown, we included a polygon-specific inspection.

b New York State Department of Environmental Conservation. 2000. Tidal Wetlands Map: 1974.

0 Rutgers Land Cover. Richard Lathrop. 2000. New Jersey 1995 Level III Land Cover Classification. Grant F. Walton Center for Remote Sensing
and Spatial Analysis, Rutgers University. Reported in: Richard G. Lathrop. 2001. Final Report. Land Use/Land Cover Update To Year
2000/2001. NJ DEP. http://www.nj.gov/dep/dsr/landuse/landuse00-01.pdfaccessed January 2006.]

d New Jersey Department of Environmental Protection. NJDEP Upper Wetlands Boundary/Upper Wetlands Limit for New Jersey.

6 Maryland Department of Natural Resources Wetlands Data. 2001. Chesapeake and Coastal Watershed Services, Geographic Information
Services Division. Minimum mapping unit: % acre. Scale: 1:12,000. Codes same as for NW.

f The three areas where DNR data lagged: (a) Caroline, Talbot counties south of Easton-St Michaels, and parts of Dorchester along the

Choptank River; (b) Cecil and Kent except for Chester River and Chesapeake Bay south of Rock Point; (c) Baltimore County west of Glenn L.
Martin airport and Baltimore City north side of Baltimore Harbor.

9 North Carolina Department of Environment and Natural Resources. 1999. DCM Wetland Mapping in Coastal North Carolina. Scale: 1:24,000
for portions relying on NWI and soils data. (Note: Scale is 1:58,000 for nontidal wetland boundaries areas relying on 30-m grid data [1994
update]. Those boundaries do not inform our elevation data, but do affect calculations of the area of nontidal wetlands vulnerable to sea level
rise.) Polygons with a code of 1, 3, or 15 were treated as tidal wetlands, as well as any polygons identified with the code "e" (estuarine wetland).

-------
[ SECTION 1.1 17 ]

New Jersey and North Carolina were special cases.
For most of New Jersey, we used the 30-m grid
data developed by Richard Lathrop for the State of
New Jersey. These data provide a more detailed
vegetation classification system, and it would have
been possible to differentiate low from high
marsh.25

For much of North Carolina, by contrast, we
lacked the wetlands data necessary to completely
apply our approach. Pamlico and Albemarle
sounds, as well as their tributaries, are nanotidal
estuaries: their astronomic tide ranges are so small
that for most practical purposes there are no tides.
As a result, wetlands data sets misleadingly
classify the wetlands along the shore as "nontidal
wetlands." Unlike true nontidal wetlands, these
wetlands are at sea level and experience the full
force of the tides in the bodies of water to which
they are attached. Thus, unlike nontidal wetlands,
which would eventually be inundated by a rising
sea level, the nanotidal wetlands are already
inundated. But the wetlands data do not distinguish
the nanotidal wetlands from the nontidal wetlands
(other than areas where salinities are high enough
in the estuary to support brackish marsh). Thus, the
wetlands data do not provide the location for the
supplemental contour we would have hoped to
create. North Carolina's LIDAR provided us with
better elevation data than we would have been able
to derive using the wetlands data; but the failure of
the data to distinguish nontidal wetlands from
nanotidal wetlands changes the meaning of any
estimates of the area of tidal wetlands in North
Carolina.

Step 3: Use Information on Tide Ranges
and Benchmark Elevations to Estimate the
Absolute Elevation of the Upper Boundary
of Tidal Wetlands

Creating a supplemental contour using wetlands
data requires us to have an estimate of the
elevation of the upper tidal wetland boundary.

That elevation depends on the elevation of mean

25This study did not make such a differentiation. However,
we provided the data for Ocean County for the study by
Jones and Strange (Section 3.20 in Section 3 of this
report), which did make such a distinction.

tide level26 (MTL) (relative to the benchmark) and
the tidal range.

Relate benchmark elevations to mean tide level.

NOAA's Published Benchmark Sheets (NOS,
2005) and the corresponding National Geodetic

27

Survey (NGS) Data Sheets' provide estimates of
the difference between mean tide level and the
benchmark elevations at 125 locations throughout
the study area. As Figure 1.1.6 shows, the majority
of those locations are in or adjacent to New Jersey.
Observations are especially sparse, by contrast, in
the sounds of Long Island and North Carolina.
Typically, mean tide level in the ocean is approx-
imately 20-30 cm above NGVD29 in the mid-
Atlantic, reflecting the rise in relative sea level
since the benchmark was established. The average
water level in a back bay, however, is often several
centimeters higher than the mean tide level on the
ocean side of the barrier island:28 The cross
sections of inlets and channels are greater at high
tide than low tide. As a result, a flood tide brings
more water into the bay when the inlet is 1 meter
above the bay than the ebb tide carries away when
the inlet is 1 meter lower. Therefore, ignoring
rainfall, the flows during the ebb and flood tides
are in balance only if the average bay level is
somewhat higher than the ocean.

Rainfall and runoff are additional sources of water
in estuaries, further increasing water levels relative
to the nearby ocean. During wet periods, water
levels in back bays behind barrier islands may be
10-30 cm higher than normal. This effect may be
greatest in freshwater tidal rivers, given their
distance from the ocean and prevailing seaward
flow of water. Hence, as Figure 1.1.6 shows, mean
tide level is higher at Philadelphia and
Washington, D.C., than along the shores of

26Mean tide level is the average of mean high water and
mean low water. It is generally very close to mean sea
level (the average water level) but requires fewer data
points to calculate.

27The Published Benchmark Sheets include links to the
NGS information relating mean tide level to the vertical
benclunark elevations.

28For example, the stations inside Little Egg Harbor Bay
near Long Beach Island (New Jersey) show MTL to be
about 1.25 feet above NGVD29.

-------
[ 18 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Delaware and Chesapeake bays, respectively.
Along major bodies of water, the coverage is
sufficient to estimate the elevation of mean tide

29

level through interpolation. For back bays
lacking such data, we assumed that the elevation of
mean tide level was similar to that of a nearby bay
where data are available. The complete lack of
data for Albemarle and Pamlico sounds was the
most problematic. Fortunately, we had the best
land elevation data—LIDAR—for that area; thus
our need for a supplemental contour based on
wetlands and tidal data was least. NOAA has
developed a hydraulic model to estimate water
levels in Pamlico Sound, but not Albemarle
Sound; we used the NOAA results wherever they
were available (Parker et al., 2003; Myers,
2005).

Use tidal range data to estimate the elevation
of spring high water. Estimates of tide ranges are
more prevalent than the absolute elevation of mean

30

sea level. NOAA's tide tables provide

31

estimates of the mean and spring-tide range at
768 discrete locations in the study area (see Figure

32

1.1.7). As with the elevation of mean tide level,
coverage is poor in Albemarle and Pamlico
sounds, where astronomic tides are small
compared to wind-generated tides; and again we
used NOAA's model for Pamlico Sound. The
NOAA estimates consider only astronomic tides,
whereas tidal wetlands are also found in areas that
are flooded irregularly by the winds. The
distinction is minor in areas with a large tidal
range, but where astronomic tide ranges are small,

29We interpolated elevations using the TopoGrid function
inESRI's Arclnfo Grid module (ESRI, 1998). See the
section on Step 3 for additional details on interpolation
algorithms. The algorithm allowed us to treat intervening
land as a "barrier" in the interpolation. In a back bay, for
example, we use measurements from the bay—but not
nearby ocean locations, because the impoundment effect
of an inlet can elevate mean tide level within the bay.
" See, e.g., NOS (2004). The hard-copy report "Tide
Tables" is now provided online. In 2004 it was still called
"Tide Tables" but more recent versions of the web site
have dropped the traditional title.

The estimates in the NOAA tide tables are long-term

averages.

32EachUSGS 1:24,000 scale map includes an estimate of the
mean tide range, which Park et al. (1989) used in their
assessment.

the wind-generated tides tend to enable wetland
vegetation to form tens of centimeters above mean
tide level, even if the spring tide range is
negligible.

As with mean tide level, we used the available data
to estimate the spring tide range through
interpolation. We then calculated the elevation of
spring high water relative to NGVD29 for the tidal
epoch 1983-200133 as one-half the spring tide
range plus the elevation of mean tide level
calculated in the previous subsection. Based on
various wetland transect studies relating wetland
elevations to the tides (e.g., Kana et al., 1988), we
assume that this elevation also represents the
elevation of the upper boundary of tidal wetlands.

This assumption is only an approximation:
wetlands may extend above spring high water, for
example, in areas with small tide ranges where
winds frequently cause areas above spring high
water to flood. This discrepancy will not affect our
estimate of the amount of dry land within (for
example) 50 cm above spring high water; but it
does lead us to overlook that some of the land (for
example) 50-75 cm above spring high water
would be flooded enough to support tidal wetlands
if sea level rises 50 cm. This error is small
compared to the accuracy of most USGS
topographic maps—but it would be very
significant in areas where LIDAR is available.34

33NOAA's Published Benchmark Sheets adjust estimates
of mean tide level so that they refer to the mean tide level
averaged over a 18.6-yr lunar cycle. See, e.g.. Gill and
Schultz (2001).

4 Commenting on this report, Christopher Spaur of the
Corps of Engineers provided the following: Regularly
flooded tidal marshes have a predictable—and easily
ascertainable—flooding regime controlled by astronomical
tides, and along the Atlantic Coast possess broad areas
dominated by tall-form Spartina alterniflora. Irregularly
flooded marshes are found in areas where the pattern of
flooding is at most partly related to astronomical tidal
regime instead of wind and seasonal tides (e.g., wet and
dry periods causing water levels to vary). These marshes
lack the pronounced break between tall-form Spartina
alterniflora and other marsh plants that occurs in regularly
flooded marshes (Frey and Basan, 1985). Surfaces are
subject to long periods of exposure and inundation (Stout,
1988). Because duration of inundation determines the
lower limits of marshes, the longer duration of inundation
causes the lower limit of marshes to be higher than in an
area where tides dominate. The surface of irregularly

-------
[ SECTION 1.1 19 ]

How accurate is our surface estimating spring high
water? The NOAA data on spring tide range and
mean tide level are based on substantial data and
thus are precise for our purposes. Interpolation
model error, however, can be significant. In large
estuaries with substantial data, the variations of
spring tide range from location to location are on
the order of 5 cm; hence our interpolation error is
likely to be small. In back barrier bays, however,
tide ranges can vary by tens of centimeters. In
many cases, the tide range simply dampens away
from the inlets, and interpolation between stations
can largely account for this dampening. In some
cases, however, there are tidal creeks with no tide
stations. In these locations, our error in calculating
spring tide range—and hence spring high water—
is likely to be on the order of tens of centimeters.

Adjusting tidal elevations to account for sea
level rise. Only by sheer coincidence would the
wetland maps be based on imagery taken during
the midpoint of the 19.6-year tidal epoch that
NOAA used to define local mean sea level. Given
our assumptions, the wetlands maps provide the
location for spring high water the year the photos
were taken. In parts of New Jersey, sea level has
risen 10 cm since the photos were taken (e.g.,
Permanent Service for Mean Sea Level,
2003). Even though this discrepancy is less than 5
cm in most areas, we corrected for it because it is a
systematic error that can be corrected, unlike the
substantial random error resulting from large
contour intervals of most elevation data.

We used the regression coefficients published by

35

the Permanent Service for Mean Sea Level for all
locations with more than 40 years of data (see
Figure 1.1.1). We then estimated the current rate of
sea level rise at intermediate locations through
interpolation. Multiplying that rate by the number

flooded marsh occurs at about mean high water (Reimold,
1977). In the coastal bays, where tidal range is generally
30-50 cm, the elevation range across the marsh surface is
much less and the marshes tend to lack much habitat
below MHW or so. The short form of Spar tin a alterniflora
is dominant on the seaward edge, and the tall form is either
lacking or very local in occurrence along tidal creeks.

The Service obtains the data for the United States from
NOAA's National Ocean Service.

of years between the map date and the NOAA base
year provided us with site-specific adjustments to
our surface estimating the elevation of spring high
water.

Step 4: Interpolate Elevations Relative to
the Vertical Datum for All Land above the
Tidal Wetlands Using Elevations Obtained
from the Previous Three Steps

From the aforementioned steps, we had the
standard elevation contours, plus a supplemental
contour along the upper boundary of tidal
wetlands. We now examine how we used those
contours to characterize elevations of locations
between the contours. Doing so required us to
decide on a rule for addressing data conflicts and
pick an interpolation algorithm.

The primary potential for data conflicts concerned
discrepancies between topographic contours and
the tidal wetland boundary. Along the Delaware
River and parts of Delaware Bay, the upper edge
of the tidal wetlands is about 4 to 5 feet
(NGVD29), so we expected the wetland boundary
to occasionally be landward of the 5-ft contour. In
areas with 2-ft and 1-m contours, we expected to
see a similar overlap even in areas with low tidal
ranges. We limited ourselves to two possible
solutions: Either the wetlands data or the contour
always takes precedence over the other.

We decided that the wetlands data should take
precedence over the contour information, for three
reasons. First, accepting both data sets at face
value, the elevation and wetlands data typically
had a scale of 1:24,000 and hence an allowable
horizontal error of 12 m. But the topographic maps
also have a vertical error of one-half contour
interval. Therefore, by its very terms, the typical
topographic map allows for the possibility that the
5-ft contour may be as low as 75 cm (2.5 ft)
(NGVD29), which would be tidal wetlands in most

37

areas. Second, the wetlands data are newer.

36With the exception of Maryland, we used only one
source of standard topographic data in a given location.
For Maryland, where we used USGS and MD-DNR
information, the USGS contours took precedence.

37USGS has made planimetric updates to most of the maps
since 1970. However, the contour dates are generally from

-------
[ 20 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Because of shore erosion, wetlands may now exist
in areas that had previously been above the 10- or
20-ft contour. Finally, this approach leaves us with
a reasonable landscape. If we gave precedence to
the 5-ft contour, we would be left with a bluff over
the wetlands along the 5-ft contour; removal of the
5-ft contour, by contrast, means that we interpolate
between the wetlands and the 10-ft contour.

In selecting an interpolation algorithm, we had to
consider our three objectives:

•	Estimate the amount of land that could be
inundated by rising sea level to the level of
precision allowable by existing data.

•	Produce maps depicting the elevations of land
close to sea level.

•	Provide an elevation data set for other
researchers.

Estimate the amount of land that could be
inundated. The contours provide polygons of
various elevation classifications. In a location with
a 5-ft contour interval, for example, we have
polygons that represent the land between spring
high water and the 5-ft (152-cm) (NGVD29)
contour, as well as 152 to 304 cm, etc. If spring
high water happens to be 60 cm above NGVD29
in a given area, then we have polygons that tell us
how much land is 0 to 92 cm above spring high
water, as well as 92 to 244 cm, etc. However,
contour intervals and the elevation of the tides
vary, so the polygons in different locations
represent different elevation ranges relative to the
tides. This situation prevents us from simply
adding the calculated area across all localities.38
Estimating the amount of land at particular
elevations requires an assumption about how
elevations are distributed in the land between the
contours.

We considered two approaches for estimating
elevations between contours: using linear

the 1945-1970 period. See, e.g., the order forms provided
by the New York State Center for Geographic Information
(showing the planimetric and contour dates for every
USGS 7.5-minute quad). Accessed on August 12, 2006, at
http://www.nysgis.state.ny.us/mapssales/orderfnn/brantlk.
htm.

38In areas where we have accurate spot elevations (e.g.,
LIDAR), we do not face this problem.

interpolation and using a digital elevation model
(DEM) to fit an estimated land surface through the
contour data we had. We tried the DEM approach
first, because it was going to be necessary for
creating the maps. We quickly concluded,
however, that readily available algorithms would
unreasonably skew our results. When the shore
and the contours are all fairly straight or well-
behaved, results seem reasonable; but when the
contours have sharp turns, the algorithms assume
that a disproportionate amount of land has an
elevation close to that of the contour. In effect, the
algorithms tend to create plateaus on either side of
the contours.39

Therefore, our estimates are based on linear
interpolations; i.e., we assume that elevation is
uniformly distributed between contours. To keep
the calculations manageable, we interpolated
elevations at the quad level.40

Produce maps and elevation datasets. We

interpolated between the contours and spot
elevations using the TopoGrid algorithm provided
by ESRI (1998) software. This procedure was
developed based on Hutchinson's (1988, 1989)
approach to estimating DEMs. The fundamental
insight embodied in that algorithm is that ground
surfaces have many local peaks, but few local
minimums, because water generally flows toward
the sea rather than being impounded. For our
purposes, that aspect was not important because
we are not concerned about slopes; instead we are
concerned with improving the accuracy of

"9In Figure 1.1.8, TopoGrid correctly creates a stream valley
in an area with a fairly simple topography. But when we
applied that algorithm over our entire study area, we found
numerous plateaus along the contours. Someone more skilled
with the algorithm may have been able to set parameters to
better replicate normal topography; but this algorithm was
designed for correct drainage, not for correctly duplicating
the distribution of elevations.

4uFor each quad, we estimated the average elevation of
spring high water (SHW). We then calculated the amount
of land between SHW and the 5-ft contour, and allocated it
proportionally between 0 and 5-SHW. We then calculated
the land between 5 and 10 feet and allocated it between 5-
SHW and 10-SHW. We stored the results in bins of 0.1
feet. We followed this approach twice for each quad, so that
we could distinguish nontidal wetlands from dry land (using
nontidal wetlands polygons from the data sources displayed
in Table 1.1.2).

-------
[ SECTION 1.1 21 ]

elevations at particular locations and correctly
describing the overall distribution of elevations.
Nevertheless, the algorithm's use of stream data to
characterize slopes should tend to ensure that
stream valleys are captured below the lowest
topographic contour, even in areas where the
stream is above the tides (and hence would not
provide a contour)41 We used a cell size of 30
meters because when we began the study, a 10-m
cell size slowed processing time too much.

Before settling on TopoGrid, we tested three other
readily available algorithms: inverse distance
weighting42 (IDW), spline,43 and triangulated
irregular networks (TIN),44 using the (5-ft interval)
contours in the general vicinity of Ocean City,
Maryland. All four algorithms created plateaus
near contours with sharp curves, with
approximately the same amount of land having an
elevation within 15 cm (0.5 feet) above or below
the contour as the amount of land with an elevation
15 to 75 cm above or below the nearest contour.

Figure 1.1.8 compares the four algorithms. Each
started with the same set of set of contours, with a
circular hill to the right, a U-shaped bluff to the
left, a stream valley in between, and a shore that is
otherwise fairly straight. The various colors
represent the elevations that the four algorithms
estimated. Between the 20- and 40- ft contours, the
yellow-brown and pink-red shades in the
TopoGrid and TIN maps suggest that these
algorithms create intermediate elevation contours
that are evenly spaced between the input contours.
IDW, by contrast, assigns virtually all of this land
to elevations of 20, 30, or 40 feet, as if the land

41If there is a 10-fit contour on either side of a creek,
without additional information, an interpolation algorithm
is likely to assume that the land between the contours (the
stream valley) is also at 10 feet.

42IDW interpolates by defining the elevation of a point X
as the weighted average of points A within a given
neighborhood, with the weights being the inverse of the
distance between X and the various points X, possibly
raised to a power. See, e.g., Shepard (1968), Fisher et al.
(1987), and Childs (2004).

43See, e.g., Childs (2004).

44A TIN is a digital data structure that represents terrain
with a series of triangles. We used the ESRI command
"CreateTin". See, e.g.. Price (1999).

were a series of steps. Spline creates 50-ft and 200-
ft hills between the 20- and 30-ft contours for no
obvious reasons. This example generally
confirmed the literature: IDW is more appropriate
when one has many points that already outline the
shape of the surface (e.g., Childs, 2004). Spline
tends to produce spurious hills, especially with
unevenly spaced input data (e.g., Rogers and
Satterfield, 1980; Olsen and Bliss, 1997).
Given the relatively large study area, we needed an
algorithm that required less supervision than
spline.

Our choice between TIN and TopoGrid was a
close call. In areas where the contours are one or
two cells apart, TIN faithfully interpolates between
the contours, whereas TopoGrid seems prone to
horizontal errors of one or two cells. TIN
completely misses the stream valley, however,
treating both the valley and the U-shaped hill as a
single flat area. TopoGrid, by contrast, creates a
stream valley with a reasonably constant slope
between the 10- and 20-ft contours. Similarly, TIN
assumes that all the land within the 40-fft contour is
at precisely 40 feet, whereas TopoGrid creates a
peak in the center just above 45 feet. We decided
to use TopoGrid because we were more willing to
tolerate its one- or two-cell errors than maps that
missed hills and streams.

Step 5: Use the Information from Step 3 to
Calculate Elevations Relative to Spring
High Water

We conducted both sets of interpolation relative to
the fixed benchmark elevation. We created maps
and a data set of elevations relative to spring high
water by subtracting our estimate of the elevation
of spring high water from every data point. We
derived our estimates of the area of land within a
given elevation above spring high water by
subtracting the average elevation of spring high
water within a given USGS quad from the
elevation of the contours between which we were
interpolating. The effect of this conversion is that
our maps show the land below a given contour
(e.g., USGS 5-ft contour) to be lower in areas with
large tide ranges than in areas with small tide
ranges.

-------
[ 22 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

PA

NY

NJ

>

<

v



DC

MD

x
* ¦/
(

f

{^

. \ 4
J \
"** \

DE •

is* i

C J

v.-

ATLANTIC

VA



OCEAN

NC

Elevation of Mean Tide Level
(cm above NGVD29)

5 15 25 35 45 55
Each dot represents one measurement

Figure 1.1.6. Observations of mean tide level used in this study. This map depicts the 125
observations from NOAA's Published Benchmark Sheets and the National Geodetic Survey's data
sheets used in this study to create a surface depicting mean tide level relative to NGVD29.

-------
[ SECTION 1.1 23 ]

NY

PA

¦> NJ J «

\ S*'
&

\



? -OP



NC

/

• /

J

X

ATLANTIC

OCEAN

Spring Tide Range (centimeters)

0 30 60 90 120 150 180 270
Each dot represents one measurement

J?

K:

Figure 1.1.7. Observations of tide ranges used in this study. This figure depicts the 768 observations from
NOAA's tide tables used in this study to create a surface depicting spring tide range.

-------
[ 24 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Elevation (ft)

o

Spline		o.s_

Comparison of
Interpolation Methods

11
11
11
i i
i i

i »

Transect

Figure 1.1.8. Test of four interpolation models. The colors represent the elevations calculated by each
of the four algorithms for the same set of contours. The box to the left shows a cross section of the
elevations as one travels from point A to point B for the surfaces created bv each alqorithms.

-------
1.1.4 Quality Control and Review: Error Estimation

We corrected all the errors and questionable
aspects that we noticed—but a test of what we did
not notice was also necessary. We enlisted Russ
Jones of Stratus Consulting to test the algorithm
and validate the results against an independent data
set. Let us briefly examine the results of the tests
we asked him to perform.

Testing the interpolation algorithm

The intent of the algorithm is to interpolate
between contours. Values that are outside the
contours represent a failure, regardless of whether
the problem is caused by TopoGrid or some other
error. We asked Jones to pick 12 representative
quads, including at least 1 quad for each state,
encompassing the different contour intervals and
data sources. After he picked the quads, we sent
him the DEM (grid) results and the input data (see
Table 1.1.1) representing polygons of land
between the wetlands and the first contour (e.g., 5
ft), the first and second contour (e.g., 5-10 ft), etc.
For each quad, we asked him to compare the areas
of the source contour polygons with those of the
interpolated DEM for the same elevation range
and to produce a histogram of the DEM elevations
by contour polygon.

Table 1.1.3 shows the results of this comparison.
Topogrid did not duplicate the area of dry land
below the lowest contour as well as we had hoped,
with percentage errors of 20 percent or more in 5
of the 12 quads. For the second and third contours,
the percentage error was less than half as great, but
hardly inspiring.

The errors do not seem as large, however, when
viewed as vertical error. The fourth column in
Table 1.1.3 provides the "effective" elevation of
the polygon contour as estimated by the DEM.45

45That is, the elevation below which the DEM estimates an
area equal to the polygon area below the first contour.

For example, the polygon area below the 2-m
contour of the Merry Hill quad is 241 ha. Although
the DEM found only 152 ha below the 2-m (6.56-
ft) contour, it also finds 241 below 6.67 ft.46 In 8 of
the 12 quads, this effective elevation is less than
0.11 feet above the corresponding USGS contour.
The area error is large and the vertical error is
small, because the algorithm created a plateau
along the contour; more land was slightly above
the contour than slightly below it.

Moreover, in most quads, most of the land below
the first contour is tidal wetland. Hence, an error of
20 percent of the dry land is typically about 5
percent of the total land. Thus, if we included all
low land in the denominator, our percentage error
estimates for the lowest contour would have the
same magnitude as for the other contours. This is
particularly true for the Middle River quad in
Baltimore County, Maryland, where the dry land
below the 2-ft (NAVD88) contour is a very narrow
strip adjacent to the tidal wetlands, whose inland
boundary is often 1.5 feet above NAVD88. Thus,
any such dry land in our data set may largely
represent errors in the input data. We are unable to
explain why our algorithm underestimated the low
land below the first contour for the other 10

47

quads, but it may be an artifact related to the
relative complexity of wetland shores.

Comparing results with an independent
dataset

As this study proceeded, LIDAR data became
available for the entire state of North Carolina as
well as Maryland's Eastern Shore south of Rock
Hall (MD DNR, 2004). JONES (2007) converted

46For some of the quads, plateaus emerged at the contours
in spite of our efforts to avoid them.

47We used Corps of Engineers spot elevation data for most
of the Atlantic City quad, so the comparison with the
USGS polygon area represents a comparison of Corps data
to USGS maps more than a test of our algorithm.

-------
[ 26 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

the LIDAR from NAVD88 to NGVD, and then
compared our results to the LIDAR. Table 1.1.4
shows (a) the mean error and (b) the root mean
square error of the DEM, by LIDAR elevation
increment and source of input data. Figure 1.1.9
depicts the difference in elevation estimates for
Maryland, which has the wider variety of data
quality.

Overall, wherever we relied on USGS maps, our
RMS error was approximately one-half the contour
interval. In those areas where the USGS maps had
a contour interval of 2 meters or better, the mean
error was usually less than 1 foot (30 cm). Note,
however, the tendency for the DEM to
overestimate the elevations of the lowest land
while underestimating the elevations of higher
ground. For example, in Maryland where we had
USGS 5-ft contours, the DEM overestimated
elevations by an average of 48 cm in the area
below 50 cm while underestimating elevations an
average of 10 cm in the area between 450 and 500
cm. This pattern occurs for two reasons. First, the
LIDAR has a random error on the order of 20 cm,
and the set of dry land locations where LIDAR
suggests an elevation of 40 cm will include cases
where the true elevation is 60 cm, but few if any
cases where the true elevation is 20 cm because
such land would be tidal wetland. Second, if land
has an elevation between 0 and 50 cm, the error of
our interpolation algorithm will assign some of this
land as between 50 and 150 cm; but by design
none of it will be assigned values below spring
high water (typically about 30 cm NGVD for these
areas).

It would be wrong to conclude, however, that our
analysis is systematically understating
vulnerability to sea level rise. We also created a
similar table (not shown) with the 50-cm
increments based on DEM elevation. That table
shows that most of our lowest DEM elevations are
less than the LIDAR elevation at that location,
while higher DEM elevations overstate the LIDAR
elevation. That pattern resulted largely from the
plateau problem (see previous discussion of Step
4). If the DEM assigns an elevation barely above
the contour, it is often underestimating an
elevation; but if it assigns an elevation barely
below the next contour, it is probably overstating
the elevation. Thus, if the DEM finds a very low

elevation, quite often the land is truly higher; but
when the land is truly very low, often the DEM
assigns a higher value. Does either tendency
dominate?

Table 1.1,4c suggests that the DEM is about as
likely to overstate as understate the amount of land
below a particular elevation. For each data source,
we calculated the cumulative elevation

48

distribution. We then took the area of land below
a particular elevation (e.g., 1 meter) as estimated
by our DEM interpolation, and then looked up the
elevation below which the LIDAR estimated the
same elevation. For example, our interpolated
DEM estimates 24.75 km (excluding tidal
wetlands) below 1 meter SHW in the part of
Maryland where USGS maps have a 1-m contour
interval, and the LIDAR shows the same amount
of land below 72 cm. Thus, the land vulnerable to
a 1-m rise according to the DEM would be
inundated by a 72-cm rise according to LIDAR.
Hence the table shows a vertical error of 28 cm.

Our analysis of the cumulative error shows that
errors offset to a large extent in Maryland, with the
vertical error generally less than Vi contour
interval, and generally less than the mean error
(except for the undocumented Kent Island map
provided by FEMA). In North Carolina, however,
the error appears to be more systematic: the
cumulative error is not substantially less than the
mean error (and in some cases is greater).
Fortunately, we now have LIDAR for all of North
Carolina, so our problems there may have no
practical importance, provided they are confined to
that state. Is Maryland alone a good test of our
method, or must one give weight to North Carolina
as well?

Considering the probable causes of the systematic
error in North Carolina, the accuracy assessment
of Maryland alone is probably more representative
of the error in the rest of the study area. Our
approach of defining a supplemental contour along
the upper boundary of tidal wetlands breaks down
in North Carolina, for three reasons. First, as we
have mentioned, the failure of available wetlands

48The cumulative vertical error in Table 1.1.4c is similar to
the difference between the two contour elevations in Table
1.1.3 and discussed in the last section.

-------
[ SECTION 1.1 27 ]

data sets to distinguish nanotidal wetlands from
nontidal wetlands in Albemarle Sound and its
tributaries led our interpolation to treat them as
ranging in elevation from just above the tides to
50-100 cm above spring high water, even though
some are at sea level. Second, several nontidal
rivers have wide floodplains consisting of nontidal
wetlands, with a bank at approximately the 2-m (or
5-ft) contour. Lacking a supplemental contour,
TopoGrid has no basis for estimating

how much below the 2-m contour those lands
might be, and hence tends to assign elevations
close to, albeit below, the 2-m contour. Finally, as
Figures 1.1.4 and 1.1.6 show, the stations for
estimating mean tide level and spring high water
are sparse.

Table 1.1.3. How well the DEM duplicated the polygons in our input data.

USGS Quad
Name

Central Park

Atlantic

City6

Port Norris

Marcus

Hook

Bethany

Middle River

South River

Ocean City

Broomes

Accomack

Irvinaton

Merry Hill

DEM p. First ^ prror Second	Third

. . Elevation ^ . Contour °, . Contour % Error Contour % Error
Lowest Qf Contour po|vgo of Area DEM ofAr0a DEM ofArea

°?«\U Contour . b Area s 'l'a e Area Estimate Area Estimate
<«'	(ha)	(ha)	(ha)

New York

New Jersey

New Jersey

Pennsylvania

Delaware
Maryland
Maryland
Maryland
Maryland
Virginia
Virginia
North
Carolina

10
5
5

10

5
2
5
5
5
5

10

10.1

4.6

5

10.1

5.1

1.25

7.3

5.1

5.8

5.1

10

399.3 428.1
164.2 132.2
202.0 209.9

-6.7 1472.2 -1.5
24.2 1011.4 -5.6
-3.8 522.2 2.6

1350.6 -1.3
146.7 11.0
107.9 0.4

69.2

885.2
54.8
162.4

416.3
84.2
123.7
471.2

87.8

1002.6
28.3

344.5

426.6

108.2
151.6

471.3

-21.2 308.3

-11.7
93.4
-52.9
-2.4
-22.2
-18.5
0.0

2211.8
96.3
444.4
505.3
228.2

1193.9
1309.9

0.3

7.9
-11.0
-5.2
-20.9
-38.2
0.5
2.1

6.56 6.666 151.7 240.8 -37.0 247.8 35.4

257.9

1412.2

117.5
606.4
391.9

607.6
1382.0
1654.6

333.2

-1.7

10.7

-0.9

0.8

16.7

17.4

0.5

13.2

8.4

Total

3184.2 3632.0 -12.3 9551.6 -1.2 8368.4

6.4

For example, the Central Park quad has 428.1 ha below the USGS 10-ft contour according to the input polygon
data, and 428.1 ha below an elevation of 10.1 feet according to the DEM. Therefore, we say that the DEM's
estimated elevation of the USGS 10-ft contour is 10.1 feet, and the vertical error of DEM's estimate of the contour
elevation is thus +0.1 feet.

b For example, area of land (other than tidal wetlands) between 0 and 10 feet in the Central Park quadrangle,
according to our DEM.

c For example, the area of land below the 10-ft contour in the Central Park quadrangle (other than tidal wetlands).
Comparing the difference between the areas of the DEM and polygons is a measure of our procedure.

d The DEM's estimate of 399.3 ha is 6.7% less than the area of the input polygons. Therefore, the error of our
area estimate is -6.7%.

8 The Atlantic City Quad is not a test of the algorithm because we had Corps of Engineers spot elevation data for
most of the quad. However, it does provide an indication of the difference between the Corps data and the USGS
maps.

-------
[ 28 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Table 1.1.4. Accuracy of DEM Results: Comparison with LIDAR

Source:



Maryland Eastern Shore





North Carolina I

Kent Island

1-m

5-ft

MD-DNR

20-ft

5 ft

2m I

Contour (cm)

60

100

152

305

610

152

200

Elevation3



A. Mean Error (Difference between DEM and LIDAR)b





50

60

26

48

102

17

58

56

100

12

18

9

74

-25

54

67

150

0

56

27

71

-67

38

53

200

4

23

54

72

-109

23

27

250

-1

-7

43

59

-155

13

19

300

-7

-9

21

37

-193

2

15

350

-8

12

9

2

-240

-2

7

400

-9

2

3

-35

-276

-4

1

450

-13

-2

-3

-59

-304

-3

-5

500

-42

-9

-10

-80

-341

2

-11

B. RMS Error (Root Mean Square Difference between LIDAR and DEM)°

50

102

38

107

160

18

113

116

100

74

59

70

151

28

92

92

150

72

83

95

146

74

99

87

200

80

41

100

135

124

100

84

250

80

37

71

110

170

100

94

300

76

52

56

91

218

91

96

350

78

57

61

82

263

84

90

400

94

49

63

84

308

77

81

450

121

59

65

98

345

76

73

500

135

71

66

119

387

86

73

C. Vertical Error (Difference in Cumulative Elevation Distributionf

50

-48

3

-21

-36

-54

36

52

100

-65

28

-1

-19

-75

38

80

150

-35

41

3

-18

-93

49

105

200

-30

35

9

-7

-106

52

109

250

-32

-3

23

-12

-110

46

130

300

-30

1

10

-11

-113

52

115

350

-29

-5

-2

-46

-111

73

67

400

-21

-5

-6

-101

-108

94

68

450

7

-14

-7

-104

-101

104

82

500

16

-16

-8

*

*

113

98

a In parts A and B, results are presented for 50-cm increments relative to NGVD29 as measured by LIDAR.

For example, the second row in each case provides results averaged over all lands with elevations
between 50 and 100 cm according to LIDAR. In part C, results are cumulative, and relative to spring high
water as estimated by the DEM interpolations. For example, the second row is based on the area of land
whose DEM interpolated elevation is less than 100 cm above SHW.
b The mean of LIDAR-DEM. For example, in parts of Maryland where USGS maps had a 5-ft contour and
LIDAR showed elevations between 100 and 150 cm, the DEM estimate was 27 cm higher than the LIDAR
value, on average. If LIDAR represents the true elevation, the mean difference represents mean error.
c Root mean square difference is calculated by taking the difference between the LIDAR and DEM
elevations at each point, squaring that value, adding all the squares and dividing by the number of data
points, then taking the square root. If the mean difference is zero, it is the same as the standard deviation.
If LIDAR represents the true elevation, this value is the root mean square error.
d A measure of the sensitivity to sea level rise of an estimate of the amount of land vulnerable to inundation.
For example, in parts of Maryland where USGS maps have a 1-m contour interval (excluding tidal
wetlands), our interpolated DEM estimates 24.75 km2 below 1 meter, while the LIDAR shows the same
amount of land below 72 cm. Assuming LIDAR to be accurate, the land vulnerable to a 1-m rise according
to the DEM would actually be inundated by a 72-cm rise. Hence the table shows a vertical error of 28 cm.

-------
[ SECTION 1.1 29 ]

Difference (meters)

Tidal Wetland



Figure 1.1.9. Elevation as estimated by LIDAR minus elevation estimated by our DEM: Eastern Shore of
Maryland.

-------
1.1.5 Maps and Results

Maps

Figures 1.1.10 to 1.1.14 show our maps using
various scales and formats. Figure 1.1.10 compares
our new maps of Maryland with the coarser-scale
maps published by Titus and Richman (2001).49
At that scale, our new maps do not appear to be a
major improvement over the previous maps—
except that that we have a smaller contour interval.
Unlike the previous effort, however, the current
study provides elevations relative to the tides. In
Figure 1.1.10, the 1-m (spring high water) contour
looks like the 1.5-m NGVD29 contour, but that
varies from place to place. Perhaps more
important, the current data provide maps at a much
larger scale. Figure 1.1.11 shows the area around
Washington, D.C.

Depending on the needs of a particular audience, it
may be useful to distinguish nontidal wetlands
from dry land rather than simply presenting
elevations. Figure 1.1.12 shows the lands along the
Delaware River between the
Delaware/Pennsylvania border and Northeast
Philadelphia. The maps show open water and tidal
wetlands as light and dark blue, respectively. For
other lands, Figure 1.1.12a depicts elevations
relative to the upper tidal wetland boundary using
a 50-cm contour interval with colors following the
spectrum from green to yellow to red. The
contours look relatively smooth outside of
Philadelphia, because we had to interpolate
between the upper tidal wetland boundary and the
USGS 10-fft contour, which is about 2 meters
above the tides. For the city itself we had 2-ft
contours. One limitation of our approach is that we
do not make use of contours below the tidal
wetlands. Both Philadelphia and Gloucester
County, New Jersey, have land below sea level
protected by dikes; it simply shows up as land less
than 50 cm above the tides in our maps.

Figure 1.1.12b is similar, except that the green-to-
red spectrum applies only to dry land; we show
nontidal wetlands using two shades of purple. The
rationale for this format is that elevation alone is
not always the best guide to risk of inundation as
sea level rises. From the perspective of many
property owners and planners, the tidal inundation
of previously dry land represents a significant loss
of property, whereas inundation of nontidal
wetlands may be viewed as less problematic.
Nontidal wetlands tend to be found well inland
from tidal waters. Because a dike runs along the
Delaware River in Gloucester County, New Jersey,
nontidal wetlands are found very close to the river,
albeit on the other side of a dike.

Figures 1.1.13 and 1.1.14 show the entire study
area using the same two formats. At this scale, one
notices that the lowest lands are mostly dry land in
Maryland and Delaware, but nontidal wetland in
North Carolina, and split evenly between the two
in New Jersey.

We are making both our maps and the underlying
data available to the public. We will provide the
maps with and without nontidal wetlands, at the
1:100,000 scale, county by county, state by state,
and a few multicounty and multistate regions. The
maps will generally show elevations above the
tidal wetland boundary, with the 50-cm contour
interval generally used in this report. However, the
county- and 1:100,000-scale maps will use a 1-m
contour interval wherever the underlying
topographic data had a 10-ft contour interval50;
where we relied on maps with a 20-ft contour
interval, we will not include those maps in
materials oriented for the general public. Both the
digital elevations and the coastal wetland maps
will be available from the authors as well.

49We added the tidal wetlands to the Titus and Richman
map to make them more comparable.

-------
[ SECTION 1.1 31 ]

V

Figure 1.1.10. Maryland: Comparison of Titus and Richman (2001) with this study. Map (a) shows
elevations relative to NGVD29 from Titus and Richman (2001). Map (b) shows elevations relative to
sprinq hiqh water according to this studv-

-------
[ 32 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Elevation (meters)

Fiqure 1.1.11. Elevations relative to sprinq hiqh water: Washinqton, D.C., and vicinitv

-------
[ SECTION 1.1 33 ]











%

ElevaUori (meters)





Upland



¦

25



	

15















m

OS





Tidal Wetland

0 10



Figure 1.1.12. Lands close to sea level in Pennsylvania and nearby New Jersey. Map (a) shows elevations
relative to spring high water. Map (b) distinguishes dry land from nontidal wetlands, depicted in purple.

Elevation (meters)

Dry Land





Upland

3.0





2.5





2.0



m

1.5



m

1.0
0.5
0.0

Wetland



1

3.0
1.0
0.0
Tidal

0

7.5

=aHkm

-------
[ 34 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

.V.

f

t£, J*

ujJ,
3'L.^

J



& ? *1

/ — fv	>

\ % „ W!

s \ !* %,Je *//

f \ %

\

4

\ f ) \ r 7

1 /	f N S

t /	}	'w.

H . \ 1.

ri * *

'V.

H

v ' $is $

/ ¦. —I

> -'a1 /

\ 4.

;s

/

\-l' h

*K dk V
* F*s ?



+¦

150
^3 km

Elevation

(Meters)
Dry Land

H Upland

3.0
2.5

2 0

	 1.0

0.5
0.0
JTidal
Wetland

Figure 1.1.13. Elevations relative to spring high water: New York to North Carolina.

-------
[ SECTION 1.1 35 ]

A a

§L~'< J*

r" n«,fC'

"\

jr

~~- \

-
-------
[ 36 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Results

Tables 1.1.5 and 1.1.6 provide our estimates of the
land within 6 meters above spring high water, in
50-cm elevation increments. Although the maps
show that the distribution of elevations varies from
place to place, Table 1.1.5 shows that at the
statewide level, the amount of dry land at various
elevations is fairly uniform. Previous studies that
were forced to rely on 5-m contour intervals and
assume that the dry land below 1 meter is one-fifth
that amount, for example, appear to have made a
reasonable assumption. For the most part, the
amount of dry land within 1 meter of high water is
within 25 percent of the amount of land between 4
and 5 meters. Given the various geological
processes that cause land to form just above sea
level, it is not surprising that that area of land
within 1 meter would be slightly greater than the
area between 4 and 5 meters.

At first glance, North Carolina appears to be an
important exception to this tendency, with less
land below 50 cm than at other elevation
increments. At most of the elevations depicted,
North Carolina has 500-700 km , almost as much
as the 600-800 km for the other seven states
combined. Below 50 cm, however, North Carolina
has only approximately 100 km of dry land. Table
1.1.6 (and Figure 1.1.14) shows why: close to
2,000 km2 of nontidal wetlands. Looking at all
lands above the tides, North Carolina has about
2,000 km2 between 0 and 50 cm, 1,400 km2
between 50 and 100 cm, and 800-900 km2 for each
of the other 50-cm increments below 5 meters.
Thus, considering Tables 1.1.5 and 1.1.6 together
gives us the opposite picture as Table 1.1.5 alone:
more land between 0 and 1 meters than between 1
and 2 meters and other elevation increments.

That result, however, is probably an artifact of the
definition of tidal wetlands. As we have discussed,
the nanotidal wetlands of Albemarle and Pamlico
sounds and their tributaries are generally classified
as nontidal wetlands. These wetlands depend on
sea level, however, as much as most tidal
wetlands: their vertical accretion is in part a
function of sea level rise. The nontidal wetlands
may be vulnerable to sea level rise as well: their
irregular flooding tends to occur either from high
water levels in the sounds or because of a
combination of rainfall and the very slow drainage

that results from being barely above sea level.
Agricultural and other dry lands just above these
nontidal wetlands could become wet if the sea
rose, just as lands above tidal wetlands can be
inundated as sea level rises. Thus, it is somewhat
misleading to classify those wetlands with other
nontidal wetlands in an analysis of sea level rise.
As Table 1.1.6 shows, North Carolina is unique in
that its area of nontidal wetlands below 50 cm is
greater than the area of tidal wetlands; the
remaining states, by contrast, have about 20 times
as much tidal wetlands as nontidal wetlands below
50 cm.

Tables 1.1.5 and 1.1.6 also support previous
assessments suggesting a potential for a significant
net loss of wetlands if sea level rise accelerates.
This report focuses solely on the topographic
vulnerability of wetlands, that is, the ratio of
current tidal wetlands to the area of low land that
could potentially become inundated. Companion
studies are examining the potential for vertical
accretion and the extent to which shore protection
might thwart landward migration. From New York
to Virginia, the area of dry land within 1 meter
above the tides is only about one-fourth the current
area of tidal wetlands. North Carolina has
approximately 3,000 km of wetlands less than 50
cm above the tides, but only 700 km of dry land
within 1 meter above the tides. Figure 1.1.15
shows county-by-county variability of the ratio of
tidal wetlands to dry land within 1 meter above the
tides.51 Because 1 meter is somewhat arbitrary,
Figure 1.1.15b shows a similar ratio, but with the
area of land within one-half the tide range (instead
of 1 meter) above spring high water in the
denominator. This ratio indicates the net loss of
tidal wetlands that would occur if sea level were to
rise one-half the tide range instantaneously. (We
exclude North Carolina because the small tide
range would give us a meaninglessly large ratio.)
Equivalently, this figure shows the ratio of the
average slope immediately above spring high
water to the average slope between spring high
water and the open water. Across the region
depicted, the average ratio is about eight. That is, if
wetlands were able to migrate inland unfettered by

51 Counties that are partly along the ocean and partly along
Chesapeake Bay, Delaware Bay, or Long Island Sound are
split.

-------
[ SECTION 1.1 37 ]

shore protection, but were not able to vertically
accrete, sea level rise would eventually cause the
area of tidal wetlands to decline by 7/8.

Thus, the fate of tidal wetlands in the mid-Atlantic
is likely to depend more on their ability to accrete
vertically than to migrate inland. The potential for
wetlands to keep pace with an accelerated rise is
sea level uncertain (see Reed et al., Section 2.1 of
this report). A priority for additional research
would thus be to determine whether human
activities are impairing—and how they might be
able to enhance—the ability of wetlands to keep
pace with rising sea level.

Comparison with Comparable Studies

Two previous mapping studies funded by the EPA
assessed the amount of mid-Atlantic land
vulnerable to sea level rise. Titus and Richman
(2001) reported results only for the 1.5 and 3.5
contours, relative to NGVD29, without
distinguishing wet from dry land, for each state in
the Atlantic and Gulf coasts. Table 1.1.7 compares
our results to their results for the eight Mid-
Atlantic states. For both elevation increments, this
analysis finds more low land than the previous
effort for every state except North Carolina. In
fact, for the seven states from New York to
Virginia, our estimate of tidal wetlands alone is
greater than the previous estimate of land below
1.5 m. Because the 2001 study was largely based
on the USGS 1° (1:250,000 scale) maps, our
results are almost certainly more accurate. Our
primary reliance on 5- and 10-ft contour
intervals—as well as the coarse (30-m) cell size—
suggests that comparable (or greater)
improvements are likely whenever this assessment
can be revised using LIDAR.

Finally, Table 1.1.8 compares our results to that of
EPA's 1989 Report to Congress, which remains
the sole nationwide estimate of the land vulnerable
to a 50- or 100-cm rise in sea level (Park et al.
1989; Titus and Greene 1989). That study was
primarily designed to estimate the vulnerability to
a 2-m global rise in sea level, for which the
available elevation maps seemed adequate. The
need for an assessment of the more likely and
near-term scenarios led the authors to interpolate

elevations below the contours, primarily using
triangular irregular networks. PARK et al. included
a dynamic model of how wetlands respond to sea
level and provided land loss in 5-year increments,
based on 48 sites equally dispersed around the
nation representing 10 percent of the coastal 7.5-
minute quads. Like this effort, that study used
wetlands data to distinguish dry land and defined
elevations using the USGS 7.5 minute quads—but
the cell size was 500 meters. The Report to
Congress (Titus and Greene 1989) grouped the
sites into seven regions, so that confidence
intervals could (barely) be developed to capture
uncertainty regarding the extent to which the
sample sites were representative of the coastal
zone; the mid-Atlantic region was defined as New
York to Virginia. The authors of the Report to
Congress no longer have the intermediate results,
so our only available comparison is the aggregate
land loss for New York to Virginia. As Table 1.1.8
shows, our estimate of the land vulnerable to a 2-m
rise is about 30 percent less than the estimate52
from the Report to Congress. Our estimates of the
land vulnerable to a 50- or 100-cm rise, however,
are 50-60 percent less than those of the 1989 study.
The key difference is that our newer data suggest
that that dry land is close to uniformly distributed

53

by elevation below 5 meters, although Park et
al. found the dry land to be disproportionately
close to sea level. As the final column shows, the
Report to Congress, in effect, estimated land to be
30—40 cm lower on average than this study.

Does our downward revision for the mid-Atlantic
imply that the Report to Congress also
overestimated the nationwide loss of land
vulnerable to a 50-cm rise by a factor of three?
Probably not: only three of the regions showed
such a disproportionate amount of low land in the
1989 study. Moreover, the Report to Congress was
based on a nationwide sample of 48 sites, only 8 of

52The Report to Congress did not report a confidence
range for the regional estimates of dry land loss because
the central estimates were less than two times the standard
deviation.

530f course, our results assume linearity between
contours—but Tables 1.1.6 and 1.1.7 show elevations to
be fairly uniform from contour interval to contour interval
as well. Our areas with spot elevation data and LIDAR
also show a fairly constant pattern.

-------
[ 38 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Table 1.1.5: Area of Dry Land Close to Sea Level (km2, 50-cm elevation increments)

Elevation
above

Tidal	mm	r.- t t *	m x/ ¦ m -iu New York

	New New	^ ,	.. . . District of ... . . New York North . .. ..

Wetlands	Pennsylvania Delaware Maryland _ . .. Virainia	,. to North

. .	York Jersev	Columbia	to Virainia Carolina _

(m)	Carolina

< 0





2.4















0.5

82

127

10.2

72

184

2.4

172

651

741

1391

1.0

81

148

11.1

54

265

1.2

177

737

626

1364

1.5

86

150

15.0

52

240

1.4

223

768

582

1350

2.0

86

125

13.4

56

265

1.4

237

785

637

1422

2.5

78

111

11.3

66

226

1.8

253

748

633

1381

3.0

71

108

11.3

69

244

1.8

332

837

572

1409

3.5

67

104

9.8

71

246

1.8

346

846

618

1464

4.0

61

100

9.2

74

231

1.8

338

816

715

1531

4.5

58

99

9.3

75

203

1.7

275

721

567

1288

5.0

52

95

9.1

73

195

1.6

253

679

412

1090

5.5

35

80

8.3

70

164

1.4

254

613

294

907

6.0

20

80

8.2

71

108

1.3

230

519

156

676

Table 1.1.6: Area of Wetlands Close to Sea Level (km

2, in 50-cm increments)









Elevation





















above
tidal
Wetlands
(m)

New
York

New
Jersev

Pennsylvania

Delaware

Marvland

District of
Columbia

Virainia

New York
to Virainia

North
Carolina

New York
to North
Carolina

Tidal
Wetlands

149

980

6

357

1116

0.8

1619

4228

1272

5500

< 0





0.4















0.5

5.0

99

1.5

22.2

64

0.04

73

266

2372

2637

1.0

4.8

73

1.5

9.8

57

0.02

75

221

719

940

1.5

3.4

71

1.7

9.2

54

0.03

70

209

394

604

2.0

3.2

64

1.6

8.9

58

0.02

69

204

321

525

2.5

2.8

43

1.1

7.9

41

0.02

73

168

296

464

3.0

2.0

41

1.0

7.8

47

0.02

74

173

259

432

3.5

1.9

40

1.0

7.9

54

0.03

74

178

233

411

4.0

1.9

36

1.0

7.6

47

0.03

74

168

238

405

4.5

1.9

36

0.8

7.5

41

0.05

67

154

219

373

5.0

1.8

35

0.3

7.4

40

0.05

64

148

234

372

5.5

1.7

30

0.4

7.3

42

0.02

81

162

166

328

6.0

1.3

30

0.4

7.5

38

0.02

84

161

79

240

which were in the mid-Atlantic. Nevertheless, until
a nationwide revision of the Report to Congress is
undertaken, those needing a nationwide estimate of
land loss for a 50-cm rise would probably be better
advised to linearly interpolate the 2-m estimate
from that study rather than rely on the reported
results. Doing so yields an estimate at the lower
end of the 8,500 to 19,000 km2 range from the

Report to Congress. Alternatively, viewing our
newer results as a vertical revision, instead of
saying that a 50-cm rise could inundate (or require
shore protection for) 8,500-19,000 km2 of land, it
would seem more reasonable to suggest that this
area of land is vulnerable to a 50-100 cm rise in
sea level.

-------
[ SECTION 1.1 39 ]

~

Map A

Ratio of Wetland to Dryland (1 meter)

¦ ooo.oso
Ho SO- 1 00

< 04-1 50
9 155 - 2 50
251-4
¦I >A
¦¦>9-12

>50

Figures 1.1.15 Topographic Vulnerability of Tidal Wetlands in the Mid-Atlantic, (a) County-by-county
ratios of the area of tidal wetlands to the area of dry land within 1 meter above spring high water. The
figure shades polygons from our tidal wetlands data set. Small polygons are exaggerated to ensure
visibility, (b) Ratio of tidal wetlands to the area of dry land within one-half the tide range above spring
high water. Calculation of the denominator was undertaken quad by quad, using procedures similar
to the approach for calculating land within 1 meter above spring high water. The map shows county-
by-county ratios

-------
[ 40 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Table 1.1.7: Comparison of this Study with Previous Studies (km ).

Elevation
(m)

New
Jersey

Pennsvl

District

Delaware Maryland of Virainia
Columbia

This Study (7.5-minute maps)
<1.5 277 1552 10
1.5-3.5 341 798 53
Titus and Richman (1-degree maps)
<1.5 240 1083 2.5
1.5-3.5 266 638 2.5

452
262

388
172

1737
1158

1547
806

3.2
5.9

1.5
4.0

2061
1322

969
1041

New York

To
Virainia

6092
3941

4230
2930

New York
North To
Carolina North
Carolina

5716
3559

5836
3865

11808
7500

10066
6794

Table 1.1.8. Mid-Atlantic Dry Land Potentially Inundated by Sea Level Rise: Comparison of this study with
EPA's 1989 Report to Congress

Global
Sea Level Rise
(cm)

Relative
Sea Level Rise3 EPA 1989b
(cm)	(km2)

This Study
(km2)

This Study's Estimate of the
relative rise reguired to
inundate Corresponding
estimate
from EPA (1989)

50	70	2341	948	112	42

100	120	3121	1697	162	42

200	220	4587	3242	253	33

a The EPA Report to Congress assumed that relative sea level rise in the mid-Atlantic would be 20 cm more
than the global sea level rise over the period being analyzed.
b From Titus and Greene (1989) Table 5 (p. 5-26).

-------
Literature Cited

Barth, M.C., and J.G. Titus. 1984. Greenhouse Effect and Sea Level Rise: A Challenge for this
Generation. Van Nostrand Reinhold, New York.

BRUUN, P. 1962. Sea level rise as a cause of shore erosion. Journal of Waterways and Harbors
Division (ASCE) 88: 116-130.

BUREAU OF THE Budget. 1947. National Map Accuracy Standards. Government Printing Office,
Washington DC. http://rockyweb.cr.usgs.gov/nmpstds/nmas.html. (cited October 1 2006).

Cahoon, D.R., D.J. Reed, and J.W. Day. 1995. Estimating shallow subsidence in microtidal salt
marshes of the southeastern United States: Kaye and Barghoorn revisited. Marine Geology 128: 1-9.

Childs, C. 2004. Interpolating surfaces in ArcGIS Spatial Analyst. ArcUser July-September: 32-35.

Church, J.A., and N.J. White. 2006. A 20th century acceleration in global sea-level rise.
Geophysical Research Letters 33, L01602, doi:10.1029/2005GL024826.

COWELL, P. J., andB.G. Thom. 1994. Morphodynamics of coastal evolution. In Coastal Evolution.
R.W.G. Carter and C.D. Woodroffe (eds.). Cambridge University Press, Cambridge, U.K. and
New York, pp. 33-86.

Cowell, P.J., PS. Roy, and R.A. Jones. 1995. Simulation of large-scale coastal change using a
morphological behavior model. Marine Geology 126: 45-61.

Dean, C. 2006. Next victim of warming: The beaches. New York Times (Science Times). _pp. Dl, D4.
January 20.

Dean, R.G., and E.M. Maurmeyer. 1983. Models of beach profile response. In CRC Handbook of
Coastal Processes and Erosion, P. KOMAR and J. MOORE (eds.). CRC Press, Boca Raton, FL: pp.
151-165.

ESRI. 1998. ARC/INFO, 7.2.1 edition. Environmental Systems Research Institute, Redlands, CA.

Federal Geodetic Control Subcommittee, federal Geographic data Committee. 1998.
Geospatial Positioning Accuracy Standards. Report FGDC-STD-007.1-1998. Washington, DC.

Fisher, N.I., T. Lewis, and B.J.J. Embleton. 1987. Statistical Analysis of Spherical Data.
Cambridge University Press, Cambridge, U.K.

FREY, R.W., andP.B. BASAN. 1985. Coastal salt marshes. In: Coastal Sedimentary Environments,
R.S. Davis (ed.). Springer-Verlag, New York, p. 225-301.

Gill, S.K., and J.R. Schultz (eds.). 2001. Tidal Datums and Their Applications. National Ocean
Service, Silver Spring, MD.

Hicks, S.D., H.A. DeBaugh, and L.H. Hickman, 1983. Sea Level Variations for the United States
1855-1980. U.S. Department of Commerce, National Ocean Service, Rockville, MD.

-------
[ 42 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

HUTCHINSON, M.F. 1988. Calculation of hydrologically sound digital elevation models. In Third
International Symposium on Spatial Data Handling, Sydney. International Geographical Union,
Columbus, OH, pp. 117-133.

Hutchinson, M.F. 1989. A new procedure for gridding elevation and stream line data with
automatic removal of spurious pits. Journal of Hydrology 106: 211-232.

Interagency Performance Evaluation Taskforce. 2006. Performance Evaluation of the New
Orleans and Southeast Louisiana Hurricane Protection System. U.S. Army Corps of Engineers,
Washington, DC.

IPCC [Intergovernmental Panel on Climate Change], 1996. Climate Change 1995: The
Science of Climate Change. Cambridge University Press, New York.

IPCC [Intergovernmental Panel on Climate Change], 2001a. Climate Change 2001: The
Scientific Basis. Cambridge University Press, New York.

IPCC [Intergovernmental Panel on Climate Change], 2001b. Climate Change 2001: Impacts,
Adaptation, and Vulnerability. Cambridge University Press, New York.

IPCC [Intergovernmental Panel on Climate Change], 2007. Climate Change 2007: The
Physical Science Basis. Cambridge University Press, New York.

JONES, R. 2007. Accuracy Assessment of EPA Digital Elevation Model Results. Memorandum and
attached spreadsheets prepared for the U.S. EPA under Work Assignment 409 of EPA Contract #68-
W-02-027.

Kana, T.W., B.J. Baca, andM.L. Williams. 1988. Charleston case study. In Greenhouse Effect,
Sea Level Rise, and Coastal Wetlands, J.G. Titus (ed.). U.S. EPA, Washington, DC.

KRISTOF, N.D. 2005. The storm next time. New York Times Section 4, Page 15, Column 6. (Sept.

11).

MD DNR [Maryland Department of Natural Resources]. 2004. LIDAR 2004 Gridded DEM.
Department of Natural Resources, Annapolis.

Myers, E.P. 2005. Review of progress on VDatum, a vertical datum transformation tool. Marine
Technology Society/IEEE OCEANS Conference, Washington, DC., September 19-23

National Geodetic Survey, Dewberry and Davis, and Psomas & Associates. 1998. Report to
Congress: National Height Modernization Study. National Ocean Service, Silver Spring, MD.

NOS [National Ocean Service], 2000. Tides and Current Glossary. National Oceanic and
Atmospheric Administration, National Ocean Service, Silver Spring, Maryland.

NOS [National Ocean Service! 2004. Tide Tables 2004. National Oceanic and Atmospheric
Administration, National Ocean Service http-//co-ops nos noa.a. gov/tides04/tpred2 html accessed
October 1, 2007.

NOS [National Ocean Service! 2005. Published Benchmark Sheets. National Oceanic and
Atmospheric Administration, National Ocean Service. httrv/Zco-ons nos noaa gov/hench html
accessed May 1, 2005.

NRC [National Research Council, Marine Board], 1987. Responding to Changes in Sea Level.
National Academy Press, Washington, DC.

-------
[ SECTION 1.1 43 ]

Olsen, L.M., and N.B. Bliss. 1997. Development of a 30-arc-second digital elevation model of
South America. User Conference Proceedings: 1997 ESRI International User Conference (Julv 8-
11). ESRI, San Diego, CA.

Park, R.A., M.S. Treehan, P.W. Mausel, and R.C. Howe. 1989. The effects of sea level rise on
U.S. coastal wetlands. In Potential Effects of Global Climate Change on the United States. U.S.
EPA, Washington, DC.

Parker, B., K.W. Hess, D.G. Milbert, and S. Gill. 2003. National V-Datum: The implementation
of a national vertical datum transformation database. U.S. Hydrographic Conference, Biloxi, MS,
March 24-27.

PERMANENT SERVICE FOR MEAN SEA LEVEL. 2003. Some Example Applications of the Use of
PSMSL Data: A Table of Long-Term Trends. Proudman Oceanographic Laboratory, Liverpool,
United Kingdom.

Pilkey, O., J. Howard, B. Brenninecmeyer, R. Frey, A. Hine, J. Kraft, R. Morton, D.
Nummedal, and H. Wanless. 1982. Saving the American beach: A position paper by concerned
coastal geologists. Results of the Skidaway Institute of Oceanography conference on America's
eroding shoreline. Shore and Beach J. (April): 3-8.

PRICE, M. 1999. Terrain modeling with Arc View 3D analyst. ArcUser January-March.
http://www.esri.com/news/arcuser/0199/web4.html (accessed July 1, 2006).

REIMOLD, R.J. 1977. Mangals and salt marshes of eastern United States. In Ecosystems of the World.
1. Wet Coastal Ecosystems, V.J. Chapman (ed.). Elsevier Scientific Publishing Company, New
York, pp. 157-166.

ROGERS, D.F., and S.G. SATTERFIELD. 1980. B-Spline curves and surfaces for ship hull design.
Computer Graphics Quarterly 14(3): 211-217.

Schneider, S.H., and R.S. Chen. 1980. Carbon dioxide flooding: Physical factors and climatic
impact. Annual Review of Energy 5: 107-140.

Shepard, D. 1968. A two-dimensional interpolation function for irregularly-spaced data.
Proceedings of the 23rd National Conference (ACM), pp. 517-524.

See httpV/portal a cm org/citation cfm?id=810616 or

httpV/portal a cm org/ft_j?a.teway cfm9iH=81 0616&tvpe=pdf or google report title.

Slade, D.C., R.K. Kehoe, and M.M. Fleming. 1990. Lands, waters and living resources subject to
the public trust doctrine. In Putting the Public Trust Doctrine to Work, D.C. Slade, (ed.). Coastal
States Organization, Washington DC, pp. 13-128

Stout, J.P. 1988. Irregularly flooded salt marshes of the Gulf and Atlantic Coasts of the United
States. In The Ecology and Management of Wetlands Volume 1: Ecology of Wetlands, D.D. Hook,
W.H. McKee, Jr., H.K. Smith, J. Gregory, V.G. Burrell, Jr., M.R. DeVoe, R.E. Sojka, S. Gilbert, R.
Banks, L.H. Stolzy, C. Brooks, T.D. Matthews, and T.H. Shear (eds.). Timber Press, Portland, OR,
pp. 511-525.

TEAL, J., and M. TEAL. 1969. Life and Death of the Salt Marsh. Ballantine Books, New York.

-------
[ 44 MAPS OF LANDS CLOSE TO SEA LEVEL ALONG THE U.S. MIDDLE ATLANTIC COAST ]

Titus, J.G. 1990. Greenhouse effect sea level rise, and barrier islands: Case study of Long Beach
Island, New Jersey. Coastal Management 18:65-90.

Titus, J.G. 1998. Rising seas, coastal erosion, and the takings clause: How to save wetlands and
beaches without hurting property owners. Maryland Law Review 57: 1279-1399.

Titus, J.G. 2004. Maps that depict the business-as-usual response to sea level rise in the
decentralized United States of America. In Global Forum on Sustainable Development:
Development and Climate Change. Paris, 11-12 November 2004. Organization of Economic
Cooperation and Development. As of January 1, 2007, available at
http://www.oecd. org/document/60/0,2340, en_2649_34361_33868732_l_l_l_l, 00. html.

Titus, J.G., and M. Greene. 1989. An overview of the nationwide impacts of sea level rise. In
Report to Congress: Potential Impacts of Global Climate Change on the United States. U.S. EPA,
Washington, DC.

Titus, J.G., and V. Narayanan. 1996. The risk of sea level rise: A Delphic Monte Carlo analysis in
which twenty researchers specify subjective probability distributions for model coefficients within
their respective areas of expertise. Climatic Change 33: 151-212.

Titus, J.G., and C. Richman. 2001. Maps of lands vulnerable to sea level rise: Modeled elevations
along the U.S. Atlantic and Gulf coasts. Climate Research 18:205-228.

U.S. EPA [U.S. Environmental Protection Agency], 1996. The Probability of Sea Level Rise.
Climate Change Division, Washington, DC.

U.S. EPA [U.S. Environmental Protection Agency] and U.S. Army Corps of Engineers.
1990. Mitigation Memorandum of Agreement (Feb. 6). U.S. EPA and U.S. Army Corps of
Engineers, Washington, DC.

US CLIMATE Change Research Program. 2003. Strategic Plan for the U.S. Climate Change Science
Program. National Oceanic and Atmospheric Administration, Washington, DC.

Young, R.S., and O.H. Pilkey. 1995. A discussion of the generalized model for simulating
shoreline changes (GENESIS). Journal of Coastal Research 11(3): 875-886.

Weiss, J.L., and J.T. Overpeck. 2006. Climate Change and Sea Level: Maps of Susceptible Areas.
Department of Geosciences, University of Arizona. Available at
httpV/www gen arizona edn/dgesl/index html" as of January 12, 2007.

Williams, S.J., K. Dodd, and K.K. Gohn. 1995. Coasts in Crisis. U.S. Geological Survey Circular
1075. U.S. Geological Survey, Washington, DC.

Zervas, C.E. 2001. Sea Level Variations of the United States 1854-1999. NOAA Technical Report
NOS CO-OPS 36, National Oceanic and Atmospheric Administration, Silver Spring, MD.

-------