Airborne laser altimetry for predictive modeling of coastal storm-surge flooding

Full text

Chapter 7
AIRBORNE LASER ALTIMETRY FOR PREDICTIVE MODELING OF
COASTAL STORM-SURGE FLOODING
TIM L. WEBSTER1 AND DONALD L. FORBES2
1Applied Geomatics Research Group, Center of Geographic Sciences, Nova Scotia
Community College, 50 Elliot Road, RR# 1 Lawrencetown, Nova Scotia, Canada B0S
1M0
2 Geological Survey of Canada, Bedford Institute of Oceanography, P.O. Box 1006,
Dartmouth, Nova Scotia, Canada B2Y 4A2
1. Introduction
1.1 THE CHALLENGE
Next to fire, floods are the most common and widespread natural disasters. For
coastal communities, the risk of flooding associated with sea-level rise and storm
surges is of particular concern. Global mean sea-level has been rising at a rate between
0.1 and 0.2 m per century over the past 100 years (McLean et al., 2001). The
Intergovernmental Panel on Climate Change (IPCC) has predicted a further increase of
0.09 to 0.88 m, with a central value of 0.48 m, from 1990 to 2100 (Church et al., 2001).
Adjusted to a 100 year base, the central value (0.44 m) is between 2.2 and 4.4 times the
global mean rate of sea-level rise during the 20th century, implying an acceleration in
response to global warming. This will result in more frequent extreme high water levels
(storm-surge flood levels) and the return interval for a given flood level may be further
reduced by potential increases in storm intensity (e.g. Walsh and Katzfey, 2000).
Relative sea-level rise at any one place is a combination of changes in regional sea-
level and vertical ground motion. Subsidence in many coastal areas will result in more
rapid relative sea-level rise. Projections of the number of people potentially subject to
storm-surge flooding on an annual basis increase from about 10 million today to 50-80
million in the 2080s, depending on adaptive response and population growth, and
assuming a sea-level rise comparable to the central value of the IPCC projections
(Nicholls et al., 1999). This points to the need for mapping of flood risk hazard zones in
areas of low relief where existing maps are rarely adequate. A high-resolution digital
elevation model (DEM) produced using airborne laser altimetry (LIDAR) provides a
cost-effective and efficient method to achieve this end. These types of maps and
information products provide the background required for coastal zone managers to
make policy decisions related to future developments, as well as adaptation possibilities
to mitigate damage and loss of existing coastal resources.
Superimposed on sea-level rise, coastal erosion and flooding associated with large
waves and runup riding on high storm-surge water levels pose additional hazards. Other
factors being equal, coastal erosion will increase with a rise in mean sea-level (Bruun,
1962; Cowell and Thom, 1994; Leatherman et al., 2003). On dune-backed coastal
beaches and barriers, erosion often takes the form of dune-front scarping, while
backshore flooding may occur if runup and overwash exploit low points in the dune
crest to create breaches and washover channels. The potential for breaching is difficult
to assess, but availability of a high-resolution DEM from LIDAR surveys would enable
157
L.L. Richardson and E.F. LeDrew, (eds.), Remote Sensing of Aquatic Coastal Ecosystem Processes: Science
and Management Applications, 157-182.
© 2006 Springer. Printed in the Netherlands.

rapid pinpointing of risk sites, while repetitive LIDAR can be used to assess storm
impacts on coastal geomorphology and sand storage (Stockdon et al., 2002).
Surges along the east coast of Canada can add up to 2 m or more of water to the
predicted water level along the coast (Parkes et al., 1997). A storm surge is defined by
the difference between observed water level and the level predicted for the astronomical
tide. Surges are caused by high winds and atmospheric low pressure systems associated
with storms. Coastal communities are most vulnerable if a storm surge makes landfall
during a high-tide event, especially during large spring high tides. With accelerated sea-
level rise as described above, the extent and frequency of flooding will increase in the
future, as will the impact of related factors such as shoreline erosion. Thus, there will be
a growing demand for information products to predict areas at risk from such events, at
present and into the future, as a basis for sustainable coastal zone management,
effective planning, and the development of appropriate adaptation strategies. In this
chapter, following the introductory discussion of coastal flooding associated with sea
level rise and storm surges, we consider remote-sensing technologies available to
produce information products suitable for defining susceptible coastal areas. We then
describe a case study from Atlantic Canada, where a multi-disciplinary scientific team
has produced information products to aid local resource managers in identifying areas
prone to coastal flooding and erosion. The Planning Department for the City of
Charlottetown was involved in the project and has incorporated the flood risk and flood
depth maps into their GIS system to allow them to develop a policy to minimize
damage from future flooding events (Webster et al., 2003).
1.2 REMOTE SENSING TECHNOLOGIES FOR FLOOD RISK MAPPING
Remotely sensed data can be used in flood risk applications in two main ways: 1)
using remote sensing to map the extent of flooding (active, cloud-penetrating, sensors
such as synthetic aperture radar (SAR) on Radarsat are ideal for this application); and
2) using remote sensing to obtain high-resolution elevation information that can be used
in predicting flood-risk areas. Because storm surges are typically 0.6 to 2 m in height,
technologies with vertical precision significantly finer than these values must be
employed to generate flood risk maps of sufficient resolution. Airborne LIDAR (light
detection and ranging) is an emerging technology that offers the vertical accuracy and
high spatial sampling density required for this purpose. Many LIDAR systems have
vertical accuracies on the order of 15 cm or better. LIDAR technology has been
employed for a number of years in atmospheric studies (e.g. Post et al., 1996; Mayor
and Eloranta, 2001) and as an airborne technique for shallow bathymetric charting (e.g.
Guenther et al., 2000), although cost remains an impediment to widespread acceptance
for the latter purpose. The technology can also be used to image the land and water
surface (Hwang et al., 2000), as in the case study presented here. A general overview of
airborne laser scanning technology and principles is provided by Wehr and Lohr
(1999). Applications have been demonstrated in forestry (Maclean and Krabill, 1986),
sea-ice studies (Wadhams et al., 1992), and glacier mass balance investigations (Krabill
et al., 1995, 2000; Abdalati and Krabill, 1999). Use of LIDAR in coastal process
studies in the USA have been reported by Sallenger et al. (1999), Krabill et al. (1999),
and Stockdon et al. (2002), among others. Preliminary trials in Atlantic Canada were
reported by O’Reilly (2000) and subsequent efforts described by Webster et al. (2002).
Most of the coast of the conterminous USA has now been mapped using this
technology (Brock et al., 2002).
158 Webster and Forbes

Another applicable remote-sensing technique, with resolution on the order of 1 m,
is interferometric SAR (inSAR). This technology has been used for flood risk mapping
in the United Kingdom (Galy and Sanders, 2002). A complete review of LIDAR,
photogrammetry, inSAR and other technologies used in the production of digital
elevation models (DEMs) is provided by Maune (2001).
1.3 CASE STUDY: FLOOD RISK MAPPING IN PRINCE EDWARD ISLAND,
CANADA
Storm-surge flood risk mapping was one of the major objectives of a recent project
(McCulloch et al., 2002) to evaluate coastal impacts of climate change and sea-level
rise on Prince Edward Island in southeastern Canada (Figure 1). Airborne LIDAR
surveys were employed in this project and the resulting data sets provided the essential
foundation for flood risk mapping in the urban centre of Charlottetown and in a
representative rural area in the vicinity of North Rustico (Webster et al., 2002). It was
recognized at the outset that a high-resolution representation of the coastal topography
would be essential for predicting areas at risk of storm-surge flooding (Webster et al.,
2001, 2003). A multi-disciplinary scientific team was involved in this project and
contributed related analyses of sea-level change, storm-surge climatology, wave and
sea-ice climatology, statistics of flood probability, coastal erosion, socio-economic
impacts, and adaptation options (Chagnon, 2002; Forbes and Manson, 2002; Forbes
et al., 2002; Manson et al., 2002; Milloy and MacDonald, 2002; Parkes and Ketch, 2002;
Parkes et al., 2002; Thompson et al., 2002). The City of Charlottetown Planning
Department also participated in the project and incorporated the results into their
information system. The data have wide applicability, beyond climate-change impacts
assessment, among other fields, including geological and ecological research, urban and
regional planning, coastal management, and agriculture. The remainder of this chapter
describes the flood risk mapping component of this project utilizing LIDAR for coastal
areas on Prince Edward Island.
1.3.1 LIDAR mapping
LIDAR mapping involves an aircraft emitting laser pulses toward the ground and
measuring the return time of the pulse (see Webster et al., 2004). The laser scan is
acquired by rapid repetition of the laser pulse transmitter and cross-track deflection of
the beam using an oscillating mirror to produce a zigzag pattern of laser hits on exposed
surfaces below the aircraft (Figure 2). A Time Interval Meter (TIM) records the mirror
scan angle, the time when the pulse is transmitted from the sensor, the time of the
returning reflected pulse, and in some cases the intensity of the return. The
configuration of the TIM is what determines if the sensor captures the first or last
reflected returns. New generation LIDARs are capable of capturing first, last, and
intermediate returns, along with multiple intensities. The data volume with such sensors
is a potential problem and the information content of the intermediate returns is an area
of active research. Using precise differential Global Positioning System (GPS)
technology to determine the location of the aircraft (Krabill and Martin, 1987) and an
Inertial Measurement Unit (IMU) to measure the aircraft attitude (pitch, yaw, and roll),
the location of individual laser returns measured by the TIM can be determined (Figure 2).
159
Airborne Laser Altimetry

Figure 1. Study area locations for LIDAR acquisition, a) Charlottetown, b) North Rustico (from
Webster et al., 2004).
Figure 2. LIDAR helicopter configuration. GPS control for sensor location, Inertial Measurement
Unit (IMU) for attitude angles to position each laser shot. Note the zig-zag pattern of LIDAR
ground shots as a result of the forward motion of the aircraft and the across-track beam displacement
by the oscillating mirror (from Webster et al., 2004).
160 Webster and Forbes

The nominal accuracy of the system used in this study is ±30 cm both horizontally
and in the vertical. The preliminary data output includes geographic coordinates
(longitude and latitude) and elevation (in meters) for each laser point reflection, all
referenced to the World Geodetic System ellipsoid of 1984 (WGS 84), the ellipsoid
employed in the GPS system. If the data are to be used for GIS applications such as
flood risk mapping, the horizontal coordinates are converted to an appropriate map
projection, in this case using the Universal Transverse Mercator (UTM) grid.
Elevations on most land-based topographic maps are measured relative to a geodetic
vertical datum. For Canada this is known as the Canadian Geodetic Vertical Datum of
1928 (CGVD28). Thus, for many applications, including flood risk mapping, the
LIDAR elevations are transformed from ellipsoid heights to orthometric heights.
Orthometric heights are based on the geoid, an equipotential surface defined by the
earth’s gravity field, approximately equal to mean sea-level (Figure 3). To obtain
orthometric heights, an adjustment must be made for the local vertical separation
between the ellipsoid and the geoid. The difference between the WGS84 ellipsoid and
the CGVD28 geoid is obtained by using the HT1_01E model, since replaced by
HTv2.0, with an accuracy of ±5 cm with 95% confidence in southern Canada (Geodetic
Survey Division, Natural Resources Canada at the following website:
www.geod.nrcan.ca/index_e/products_e/software_e/gpsht_e.html).
Figure 3. Relationship between ellipsoidal height and orthometric height. The ellipsoid is a
smooth mathematical surface. The geoid is an equipotential surface defined by the earth’s gravity
field. Orthometric heights are measured from the earth’s surface normal to the geoid. The HT1_01
model is used to determine the separation between the ellipsoid and geoid for this study.
When the LIDAR system scans the ground, several targets are potentially
illuminated on each laser pulse, including the ground, tree canopy, and building tops.
After initial processing, the suite of reflections forms the LIDAR point cloud and must
be further classified. The most common scheme is to classify the LIDAR points into
two categories: ground points and non-ground points. This is required to enable
creation of a bald earth representation, without vegetation or buildings, when
generating a DEM for flooding analysis.
The standard delivery product for many LIDAR vendors at the time of this survey
was an ASCII file of (x,y,z) data. Currently, there is a proposal from the American
161
Airborne Laser Altimetry

Society of Photogrammetric Engineering and Remote Sensing for a new binary data
format standard for LIDAR data (Schuckman, 2003). Both ellipsoidal and orthometric
heights were requested for the final delivery of the LIDAR data in this study. The study
area was split into 1 km by 1 km tiles and two ASCII files were delivered for each tile,
one for ground points and one for non-ground points. Each ASCII file consisted of the
following fields:
xUTM easting (m)
xUTM northing (m)
xheight above WGS84 ellipsoid (m)
xorthometric height above CGVD28 (m)
xGPS time (s) from the start of each GPS week.
The addition of the GPS time stamp allowed us to analyze the data based on flight
lines, made it possible to extract data for times when the GPS constellation was poor,
and provided a time stamp for water level (from the tide gauge). Without times, it is
difficult to separate LIDAR returns between flight lines.
Terra Remote Sensing Inc. of Sidney, British Columbia, Canada, was contracted to
acquire the LIDAR survey data for two study areas: low-lying parts of the City of
Charlottetown and a coastal strip extending about 50 km along the central North Shore
of Prince Edward Island (Forbes and Manson, 2002). The data were acquired on 1-2
August 2000. The aircraft was positioned using phase kinematic GPS, referenced to a
geodetic ground monument north of the Charlottetown Airport. The LIDAR system was
a diode-pumped I/R YAG laser operating at a pulse repetition rate of 10 kHz (10,000
laser pulses per second) with a scanning mirror oscillation rate of 15 Hz and a scan
angle of 50o. The LIDAR was a first-return system, so no subsequent returns were
measured in this case. Down-looking video was acquired simultaneously to assist in
interpreting the LIDAR data. With the aircraft at a flying height of 600 m, the LIDAR
ground swath was approximately 600 m wide and the ground spacing between LIDAR
points was less than 2 m. The technical specifications required the horizontal and
vertical accuracy to be 95% within 30 cm of measured GPS points.
2. Validation of LIDAR Elevation Models
Highly accurate DEMs of the coastal zone are required to predict flooding extent
from storm-surges of the order of 1 m. The accuracy of a DEM derived from LIDAR
data depends on the successful removal of systematic errors associated with the
acquisition system, and on a validation process to confirm that the specifications are
met. Filin (2003) provides an overview of systematic error types and treatment of these
errors in LIDAR systems. In order to ensure that the LIDAR data meet high vertical
accuracy specifications, independent ground validation data are required. However, in
the Charlottetown study area, the available topographic control at the time of the survey
was inadequate and a variety of approaches were used to test the LIDAR data.
During the initial quality assurance of the LIDAR data, water levels from the
Charlottetown tide gauge (Parkes et al., 2002) were used to assess the accuracy of water
surface hits in the LIDAR data set (Webster et al., 2002). During the LIDAR survey,
the winds were light, the harbour surface had no significant waves (although it may
have been rippled), and it was assumed that the water level at the tide gauge was a good
measure of water level throughout the harbour. A total of 16,515 LIDAR point hits on
162 Webster and Forbes

the harbour surface in eight areas showed a mean offset of -0.85 m. A bimodal
distribution with five areas revealed a mean offset of -0.93 m, while three others gave a
mean of -0.66 m (Webster et al., 2002). The standard deviation of the LIDAR
elevations on the water surface in the eight sample areas ranged from 0.14 to 0.31 m for
sample sizes between 1,013 and 3,634. The disagreement in the LIDAR elevations and
water level data was later interpreted as possibly relating to tidal hydraulics in the
Yorke and Hillsborough River arms of the harbour (Webster et al., 2004).
A separate comparison was made between a Canadian Hydrographic Service
(CHS) benchmark CHTN 1-1963 on the Coast Guard wharf near the tide gauge at the
foot of Queen Street in Charlottetown and LIDAR hits on the wharf surface within a
radius of 3 m. The ellipsoidal and geoidal (CGVD28) elevations of the benchmark were
determined as part of the vertical datum control survey reported in King et al. (2002).
Initial comparison with presumed ground (bald-earth) points in the LIDAR data gave
an offset of -2.2 m, but it turned out that the classification algorithm had erroneously
identified water surface points as ground points and wharf deck points as non-ground.
Subsequent identification of points on the wharf deck in the vicinity of the benchmark
showed an offset of -0.92 m ellipsoidal. On the basis on these results and consultation
with the data acquisition contractor in which no systematic errors were identified, an
adjustment of +0.9 m was applied to all data points prior to incorporation in the DEM.
This produced realistic flooding levels for a simulation of the 21 January 2000 storm
surge (Webster et al., 2002, 2003), whereas the original DEM without adjustment had
suggested much more extensive flooding.
There was some question whether the 0.9 m adjustment would be equally
applicable in the North Shore study area some distance from Charlottetown. Therefore,
validation work was undertaken in that area as well, consisting of surveying cross-shore
profiles using real-time kinematic (RTK) differential GPS techniques with horizontal
and vertical resolution better than 5 cm. These data were available at eight monitoring
sites maintained by the Geological Survey of Canada in the study area (Forbes and
Manson, 2002). Comparison of gridded LIDAR points (processed using GRASS as
described below) with RTK survey points along one such transect at Brackley Beach
(Figure 4) showed good correspondence. This profile crossed two dune ridges and
revealed that the crest elevation of the high narrow dune crest was somewhat
underestimated in gridded LIDAR topography. Reflection from the tops of rose and
bayberry plants was also evident in an overestimation of ground elevations in the
depression between the two dune crests. Otherwise, this example showed successful
validation and indicated that the 0.9 m offset adjustment was appropriate in the North
Shore study area.
To further test the validity of the 0.9 m adjustment, a high-precision GPS campaign
was carried out within the Charlottetown survey area in the summer of 2001 (Fraser,
2001). Carrier phase static GPS measurements were collected and processed at 15 sites
throughout the city so that a more detailed analysis of the height differences could be
done (Figure 5). The GPS sites were selected based on a variety of factors including:
spatial distribution throughout the LIDAR study area; GPS satellite geometry to
minimize obstructions (e.g. away from large buildings where multipath could be a
problem); flat smooth areas of dense LIDAR point coverage such as grass fields in city
parks; and critical waterfront features such as wharf decks. Local GPS base stations
were established using the provincial geodetic control network. Baselines between the
163
Airborne Laser Altimetry

Figure 4. Comparison of raw gridded LIDAR points with RTK survey profile on cross-shore
transect over two dune crests and beach at right on central survey line at Brackley Point
monitoring site, North Shore of Prince Edward Island. Apparent step functions on steep slopes
are artifacts of gridded elevation extraction corresponding to closely spaced RTK survey points.
When the 0.9 m adjustment is added to the LIDAR elevations, the LIDAR profile shows good
correspondence with the ground survey. Lower elevation at narrow dune crest is related to the
grid size. Higher elevations in trough between dune crests are related to the height of rose and
bayberry plants growing in that area. Offset at base of beach profile is a function of topographic
change between the ground and airborne surveys.
GPS base and rover units were kept to less than 5 km. Combinations of Trimble dual
and single frequency P-code receivers were used to collect the GPS observations and
post processed in order to maintain centimeter level accuracy of the points.
The processed GPS data were brought into an Arc/Info GIS to be integrated and
compared with the LIDAR survey points. The validation procedure consisted of two
approaches:
xcomparing the GPS points to the interpolated LIDAR DEM surface, and
xcomparing the GPS points to LIDAR points within a fixed radius around each
GPS point (as in the comparison with the CHS benchmark on the wharf).
second method, although more time consuming and complicated in both computation
method was computationally simple and gave details on the accuracy of the DEM
In both cases the orthometric heights measured by GPS were compared to those
obtained from the LIDAR data set. The HT1-01 model was used to transform the GPS
were used because they provided different and complimentary information. The first
ellipsoidal heights into orthometric heights. As noted above, two validation methods
produced from the LIDAR data. This method involved comparing each GPS point with
the interpolated DEM surface, thus ensuring availability of a LIDAR surface elevation
regardless of the original LIDAR point distribution. The drawback of this method was
that the details of the original LIDAR points compared to the GPS points were lost. The
164 Webster and Forbes

Figure 5. Digital Elevation Model (DEM) derived from LIDAR ground points for Charlottetown
with GPS locations (yellow triangles), with the red box denoting the inset map location. Inset
map shows LIDAR ground points (red and grey points) and GPS location (yellow triangle) for a
city park lawn near the waterfront. The grey and red LIDAR points are colour coded based on the
GPS time stamp of the aircraft.
and interpretation, addressed the specific details of individual LIDAR point hits.
Because the LIDAR points rarely if ever coincided exactly with the GPS ground
validation points, in the second method the cluster of LIDAR points within a specified
radius around the GPS point was used to ensure an adequate sample. It was important to
consider the radius of the search area to ensure that the LIDAR points selected were
representative of the ground feature surveyed by GPS (e.g. if GPS points were collected
on a road, a large search radius could include LIDAR points in the ditch and thus
indicate an erroneous vertical offset). With those conditions in mind, the second method
provided more details on the raw LIDAR data and systematic errors could be more
readily detected.
The first validation procedure involved overlaying the GPS points on the LIDAR
DEM in order to obtain the cell elevation value (Figure 5). Fifteen GPS points were
compared to the adjusted DEM surface (Table 1). The average difference between GPS
measurements and the surface was -4.1 cm, thus confirming that the +0.9 m offset was
appropriate throughout the study area (Figure 5). However, the standard deviation of
the differences between the GPS and LIDAR surface values was 0.54 m and the
average magnitude of the height difference was 0.45 m indicating a high degree of
variance in the data (Table 1). This simple validation approach gave a sense of the
potential resolution and validity of the DEM, but observed offsets could be attributed to
the influence of height differences in adjoining cells.
165
Airborne Laser Altimetry

Table 1. Comparison of 15 GPS carrier phase measurements to adjusted LIDAR DEM surface.
GPS (m) LIDAR (m) '= (m)
4.46 5.36 -0.90
4.60 4.02 0.57
4.22 4.10 0.13
10.42 9.77 0.65
5.26 4.27 0.98
9.07 8.84 0.22
2.34 2.92 -0.59
14.99 15.54 -0.55
19.30 19.81 -0.52
6.20 6.19 0.00
18.64 18.93 -0.29
19.48 19.86 -0.37
6.90 7.03 -0.13
2.09 2.44 -0.35
2.78 2.28 0.50
Mean offset -0.04
Standard
deviation 0.54
mean
absolute
difference 0.45
standard
deviation 0.28
The second validation approach involved an analysis of all LIDAR ground points
within 5 m of each of the GPS ground validation points. Many of the latter were located
in city parks and sports fields, with short grass (typically less than 5 cm), and level
terrain. Two outputs were generated to assess the height differences. The first
considered all LIDAR ground points in the vicinity of each GPS point and computed
the following values:
xthe number of LIDAR points within 5 m,
xthe minimum height difference (GPS-LIDAR),
xthe maximum height difference (GPS-LIDAR),
xthe mean height difference (GPS-LIDAR), and
xthe standard deviation of the height difference (GPS-LIDAR).
The second output represented the LIDAR points within 5 m of a GPS point and
computed the following values:
166 Webster and Forbes

xthe distance to the nearest GPS point,
xthe GPS orthometric height, and
xthe height difference.
For all 15 GPS control points, a total of 342 LIDAR points fell within the 5 m
radius limits. The average difference in orthometric heights between the LIDAR and
GPS points was -0.37 m, with a standard deviation of 0.25 m. Because the majority of
the GPS validation points were on extensive flat surfaces within city parks and sports
fields, one would expect the height differences between the locations of individual
LIDAR hits and the nearest GPS points to be nearly constant within the 5 m radius.
Figure 6 shows a typical site in a city park near the waterfront with an extensive, flat,
grass field. In this case, the observed height difference based on the LIDAR elevations
was not constant, and the spatial distribution of differences showed a pattern. Because
we recorded the GPS time tag, we were able to classify the LIDAR points based on the
GPS time to distinguish points from different flight lines.
Taking the site shown in Figure 6 as an example, Figure 7 shows the LIDAR
ground points coded by flight line. The LIDAR points within 5 m of the GPS point are
represented by larger symbols and consist of two flight lines. Figure 8 shows the same
LIDAR points within 5 m of the GPS location, in this case coded by height difference.
From these figures, it appears that the height difference is related to the LIDAR flight
line. When the height differences are plotted against GPS time, the systematic height
difference is more apparent (Figure 9). One set of points flown at GPS time 47,600 s, is
designated line 1; another at time 62,010 s is termed line 2. The height differences for
line 1 range from -0.26 m to +0.03 m, with a mean difference of -0.12 m and a standard
deviation of 0.08 m. The height differences for line 2 range from -0.46 m to -0.20 m,
with a mean difference of -0.33 m and a standard deviation of 0.07 m. Thus the height
differences are distinct for each flight line although they have a similar variance range
and overlap slightly. It is clear that if adjustments in the form of a vertical offset are
applied to the data, these should preferably be specific to flight lines. Examining the
rest of the LIDAR points within 5 m of each of the GPS survey validation points
reveals a similar pattern, with distinctive height differences for different flight lines
(Figure 10). Another way to look at this relationship is to plot the GPS-LIDAR height
differences against GPS time (Figure 11). It is clear from this plot that, while the 0.9 m
adjustment to the LIDAR elevations was appropriate in an aggregate sense, the
variation in offsets between flight lines add an extra error term to the DEM elevations.
In many cases this difference is reduced because the mean elevation for a DEM grid
cell is computed from all points lying within the cell.
Another problem with the LIDAR data collected in 2000 involved large
variations in the density of returns. A reduction in the laser power resulted in flying the
Charlottetown LIDAR survey at lower altitude than planned (Webster et al., 2004).
This also resulted in a lack of LIDAR returns from near-infrared targets such as black
asphalt pavement and rooftops (Figure 5). This resulted in the clear delineation of the
167
Airborne Laser Altimetry

centimeter
level accuracy.
interpolated
from the LIDAR ground points as shown in figure 5.
Figure 6. Example of a typical GPS site. City park near the waterfront with a flat level grass
field. This location corresponds to the inset map on figure 5. The GPS unit is a Trimble model
4600 and was used for post processing to determine the height of the ground surface to
Figure 7. Larger grey and red points are LIDAR ground points within 5 m of the GPS site at the
city park in figures 5 and 6. LIDAR points (large and small) are colour coded by GPS time and
denoted as flight lines 1 and 2. The raster grid in the background represents the DEM
168 Webster and Forbes

flight lines of the LIDAR data.
denoted by GPS time 47,600 seconds and flight line 2 is denoted by GPS time 62,010 seconds.
Figure 8. Larger points are LIDAR ground points within 5 m of the GPS point shown in figure
5, 6, and 7. These points are coded into three classes based on the height difference compared to
the GPS point (GPS-LIDAR). The smaller grey and red points are colour coded based on GPS
time as in figure 7. Since the field is flat, the height difference should be constant. However, as
can be seen from the spatial pattern of the height differences there is a range between –0.45 to –
0.029 m difference. The pattern of the height differences corresponds to the differences in the
Figure 9. Plot of GPS time (flight line) and height difference (GPS-LIDAR) for the LIDAR
points within 5 m of the GPS point within the city park shown in figures 5-8. Note the distinct
difference in height differences between flight lines covering the city park site. Flight line 1 is
169
Airborne Laser Altimetry

Figure 10. Plot of GPS time (flight line) and height difference (GPS-LIDAR) for all LIDAR
ground points within 5 m of all 15 GPS locations. Note the distinct difference in height
differences between flight lines for the entire GPS survey area.
Figure 11. Plot of height difference (GPS-LIDAR red-left y-axis) and GPS time (flight line)
(seconds blue-right y-axis) for all the LIDAR points within 5 m of all 15 GPS points. Note the
relationship between the height variations and the flight lines.
road pattern and individual vehicles on the road in plots of the LIDAR point cloud
(Webster et al., 2002, 2004). After the 2001 GPS campaign in Charlottetown and
another more extensive campaign in the Annapolis Valley of Nova Scotia (where
another LIDAR survey had been flown during the summer of 2000), the vertical offset
problems in the LIDAR system were found to be related to calibration procedures
170 Webster and Forbes

(Webster et al., 2004). Specifically, they were related to differences in flying altitude
between the calibration flights and the survey flights. The laser range calibrations were
carried out at the planned survey altitude of 900 m. However, because of the power loss
problems, the actual survey was flown at an altitude nearer 600 m, introducing a range
bias. The problem was initially difficult to assess, because the LIDAR data showed a
good match in the calibration areas used by the data acquisition team (usually near the
airport where the GPS base station was established). To confirm this interpretation,
several lines of the LIDAR data have been reprocessed by applying a range scale factor
and offset, resulting in agreement of the orthometric heights in overlapping lines and a
better match to the GPS data.
The above analysis demonstrates the necessity for independent validation of
LIDAR data. The expected level of accuracy of this technology is such that most
existing data (e.g. 1:10,000 scale topographic maps) are inadequate for this purpose.
Therefore carrier-phase GPS data must be collected specifically for validation purposes.
3. DEM Construction from LIDAR
3.1 INTERPOLATION METHODS AND CLASSIFICATION OF THE LIDAR
POINT CLOUD
As mentioned above, it is standard practice to classify the LIDAR returns into
ground and non-ground points to enable the production of a bald earth DEM. An
accurate representation of the ground surface is critical for coastal zone flood risk
mapping from storm-surge events. As outlined in Maune (2001), several methods have
been developed for constructing DEMs from point data. Most involve interpolation
between the LIDAR points to generate a continuous surface. For this project two
approaches were used: 1) for both the Charlottetown and North Shore survey areas, a
DEM was constructed by direct gridding of the LIDAR points; and 2) for the
Charlottetown area, a DEM was constructed through interpolation.
In the first approach, public domain GRASS software was used to build a grid from
the LIDAR ground points. This is the method we have been using for a number of years
to process multibeam bathymetric data to generate digital seabed models and shaded-
relief imagery (cf. Courtney and Shaw, 2000). A 2 m grid was overlaid on the LIDAR
points and each grid cell was assigned the mean orthometric height for the point(s)
lying in the cell. This produces a DEM without interpolation, but areas of sparse or
missing LIDAR points will not be assigned a value in the DEM. Limited interpolation
can be used to fill small gaps in the resulting grid. Shaded-relief images derived from
the resulting DEMs were reproduced in Forbes and Manson (2002) and Forbes et al.
(2004).
In the second approach, an Arc/Info geographic information system (GIS) was
used to construct a triangular irregular network (TIN) from the LIDAR ground points.
A 2 m grid was then built from the TIN using the quintic interpolation method (5th order
polynomial). Although it is computationally more intensive than linear interpolation,
this method ensured a smooth surface that honored all data points. The DEM grid was
then transferred to the PCI Geomatica suite of image processing tools for visualization
and modelling. A colour shaded-relief model was constructed from the DEM and used
for qualitative assessment and flood modelling. At this stage, two problems were
identified:
171
Airborne Laser Altimetry

xthe TIN interpolation crossed areas of no data, such as the inner gap in the ‘V’
shape of the study area (Figure 1) and sparse data points in the water, thus
producing a surface in these areas that was not reliable; and
xwharves and sea walls with vertical faces along the waterfront appeared in the
shaded-relief image to have slanted sides (Dickie, 2001).
3.2 GROUND SURFACE REFINEMENT
The problem of inaccurate surface interpolation across areas of no valid data can be
resolved by clipping the grid using a mask covering only the areas of interest. This is
also required along the shoreline to exclude water surface returns. Complications can
develop where the area of interest is large, as in the North Shore survey area, and there
are significant variations in tide level during the survey or hydraulic effects causing
different water levels inside and outside estuaries (Figure 12).
Figure 12. Filtering the LIDAR data at the water line. Upper panel: Initial topographic model
(darker blues are higher). Lower panel: Colour shaded-relief image after trimming at an
appropriate water line (yellows to reds are higher). Note that minor nadir reflections from the
water surface remain inside the bay at bottom centre right.
The problem of the wharves and waterfront structures not being accurately
modeled was serious for our intended use of the DEM for flood visualization.
Overlaying the ground points on the colour shaded-relief DEM, it was immediately
clear that there were very few around the edges of the wharves. Numerous presumed
ground points were located on the water surface and in the central areas of the wharf
decks away from the edges (Figure 5 inset). As noted earlier, we discovered that the
classification algorithm had coded the wharf edges as non-ground. This problem has
been encountered elsewhere with LIDAR datasets – for example, large flat roofed
buildings are often misclassified. The rooftops near the edge will be correctly classified
as non-ground, but in the center of the roof the points will often be coded as ground.
Similarly, in our Prince Edward Island data set (e.g. in the area shown in Figure 12),
dune crests adjacent to dune-face scarps were miscoded as non-ground points. In the
case of the Charlottetown waterfront, this issue was resolved by manually extracting the
correct ground points from the non-ground files. A set of software tools developed by
Helical Systems was used to examine the non-ground points in 3-D and select those that
represented the wharf-edge and sea-wall ground features. The original ground points
172 Webster and Forbes

were combined with the new points extracted from the non-ground file to generate a
new TIN along the waterfront (Dickie, 2001). In this case, a linear interpolation method
was used to construct a 2 m DEM. A linear interpolation was chosen to better represent
the abrupt vertical changes associated with the waterfront. The final DEM consisted of
a mosaic combining the quintic 2 m DEM for the area landward of the waterfront with
the linear 2 m DEM for the waterfront itself (Figure 13).
Figure 13. Revised DEM of waterfront, ground and some non-ground points used to construct
the surface.
A GIS database of the street network, building footprints, and other infrastructure
was overlaid on the DEM to assess the horizontal accuracy of the model. The DEM fit
the vector data sufficiently. As mentioned above, the GPS survey data were examined
in relation to the DEM to assess the LIDAR adjustment and the final DEM product.
A digital surface model (DSM) was also constructed that incorporated all of the
LIDAR points, both ground and non-ground. Because it represents the top of the
canopy in wooded areas (Figure 14), this model is not appropriate for flood risk
modeling. The DSM was generated for visualization purposes only, but it does provide
useful information on land cover, buildings and other structures, and more realistic
visual clues.
4. Flood-risk Mapping Using a LIDAR DEM
4.1 WATER LEVELS FOR THE FLOOD MODELING
Once a reliable DEM is constructed that accurately represents the natural and man-
made coastal morphology, flood-risk modeling can begin. The coastal area of Prince
Edwards Island was selected for this study due to its vulnerability to flooding and other
impacts during storm-surge events with rising relative sea-level (Shaw et al., 1998).
Shortly after the study began, the City of Charlottetown was severely impacted
during a storm on 21-22 January 2000 (Forbes et al., 2000, 2001). This event, which
caused extraordinary damage, was recorded by the Charlottetown tide gauge (Figure
15). The surge occurred during a run of perigean spring tides (Parkes and Ketch, 2002).
The downtown waterfront was flooded by a record high water level of 4.229 m CD
(Chart Datum) resulting from a surge of almost 1.5 m superimposed on a large high
tide.
173
Airborne Laser Altimetry

Figure 14. Digital Surface Model (DSM) all LIDAR points used to construct this surface.
The heights are in millimeters above Chart Datum for Charlottetown.
The Charlottetown gauge provides one of the longest continuous records of sea-
level in the region (1911 to present) as documented by Parkes et al. (2002). That study
showed a rising trend of relative sea-level amounting to 32 cm per century over the
length of the tide-gauge record. This rise in sea-level relative to a fixed reference point
on land results from a combination of climate-induced sea-level rise and regional
crustal subsidence. As relative sea-level rises, the probability of flooding to a given
level increases and the maximum potential flood level rises. Thus, the combined effect
of subsidence and sea-level rise results in an increased vulnerability to coastal damage
by storm-surges.
The 21-22 January 2000 storm provided an opportunity to test our flood-risk
modeling efforts and to validate the results by comparing the areas flooded in the model
to those observed during the event. After the flood risk maps were constructed by
Figure 15. Tide-gauge water level records for Jan. 21, 2000 storm-surge event. The predicted
tide is in green and the observed water level is in blue. The difference between the observed and
predicted water level represents the storm-surge and is shown by the red line. The largest storm-
surge event of 1.5 m occurred at the highest tide level, resulting in significant coastal flooding.
174 Webster and Forbes

modeling the flood level of the January 2000 storm, the results were shared with city
engineers, planners and other municipal officials for a qualitative validation. The maps
of flooding extent were in agreement with flood limits observed by officials during the
event. Three water levels were used to generate flood risk maps from the LIDAR DEM:
1) the peak water level for the 21-22 January 2000 storm event (4.23 m CD); 2) the
coming 100 years (4.93 m CD).
In order to map flood limits in this project, it was also necessary to relate the
LIDAR elevation data to hydrographic Chart Datum (CD), the vertical datum used in
the tide-gauge records from which storm-surge flood levels are obtained. This datum is
approximately equivalent to the lowest astronomical tide and varies from place to place
around the coast. The local Chart Datum at each secondary port has typically been
related to a local benchmark near the tide gauge, often a permanent marker on the wharf
or nearby structures. A separate component of the Prince Edward Island project focused
on determining the vertical differences between Chart Datum, the ellipsoid, and
CGVD28 throughout the study area (King et al., 2002). In the case of Charlottetown,
Chart Datum was determined to be 1.685 m below CGVD28, the vertical reference for
the LIDAR DEM. Thus, a simple translation of the 21-22 January 2000 water level and
future water levels to the DEM was made by subtracting 1.685 m from the water levels
referred to Chart Datum.
4.2 GIS FLOOD MODELING OF STORM-SURGE WATER LEVELS
Many sophisticated numerical models have been developed for simulating tidal
hydraulics and these can be particularly useful for flooding of tidal reaches in rivers. In
the present study of coastal flooding, it was decided to use existing GIS and image
processing capabilities combined with the LIDAR DEM to visualize the potential areas
of flooding. Galy and Sanders (2002) recently used a similar approach where they used
a DEM derived from a SAR data to map flood risk along the River Thames in the
United Kingdom.
In our study, the initial DEM was built using tools within Arc/Info and the data
were transferred to PCI Geomatics image processing software for visualization and the
generation of raster flood risk maps. The maps were then transferred back to Arc/Info
for final vector processing and overlay analysis. It was assumed that a given water level
from the storm-surge event would form a horizontal flood plane extending landward
from the open harbour. Thus hydraulic effects, associated time lags, or flood expansion
or dampening were not considered in the modeling effort.
With the water levels now referenced to CGVD28, a model was written to
threshold the DEM into two classes for a given water level, one wet and one dry. This
initial threshold procedure did not include any connectivity checks with the source
harbour area. The resultant raster image was converted to vector polygons. With the
flood extent data in this form, it was quite simple to select only those polygons that
were connected to the open harbour, thereby excluding low-lying areas landward of
barriers that would check the spread of the flood. Specific conditions such as bridges
and causeways with culverts had to be dealt with individually. Areas separated from the
harbour in the LIDAR DEM by an apparent barrier such as a bridge were included in
the flooded areas. Other less obvious situations, such as causeways with culverts, were
more problematic and local municipal officials were consulted in these cases. Many of
the culverts are equipped with one-way valves to allow water to flow toward the
same storm event superimposed on a moderate sea-level rise (4.73 m CD); and 3)
the same storm event superimposed on a realistic estimate of relative sea-level rise over the
175
Airborne Laser Altimetry

harbour but stop it from flowing upstream. In these cases the low-lying areas beyond
the barrier were not included. Presently there are no built in tools within the
commercial GIS and image processing systems we used in this project to automate this
type of connectivity while mapping flood risk. Therefore, a software program was
developed by M. Gould at the Applied Geomatics Research Group, Centre of
Geographic Sciences to automate the procedure. The program allows the user to enter a
water level and a starting location, and the application will use a DEM to determine all
of the areas below that water level that are connected to the starting point (a location in
the harbour was used in this case).
As mentioned earlier, the 21-22 January 2000 storm-surge flood provided an ideal
validation test. Areas flooded during that storm were compared to those generated from
the DEM model and a very good match was observed. One circumstance that the model
did not account for was the backup of seawater through the storm drain system with
water flowing up and out of street drains near the waterfront. Many of these areas were
in low-lying areas that were predicted to be flooded in the model.
Having obtained good agreement between the observed and predicted water levels
of the 21-22 January 2000 storm, we proceeded to model two additional future water
levels using a similar approach (Figure 16). The flood risk vectors show the variation of
slope within the coastal zone. Many areas have steep slopes and are not vulnerable to
flooding, although erosion may be a problem. The waterfront of Charlottetown is
vulnerable to flooding, as is the residential area to the west of the downtown core
(Figure 17).
4.3 FLOOD-DEPTH MAPS
Another set of layers from our analysis of the flood risk was the depth of the water
within the flooded areas. This was calculated by subtracting a constant value of the
flood water level from the DEM only for the areas at flood risk. This produced a grid of
flood depth values and provides more information than just the flood risk vector that
shows the area of inundation from flooding. The amount of damage will be related to
the depth of the floodwater, where more damage is expected with deeper flood depths.
This is especially true for residential areas where the floodwater must reach a high
enough point to enter the basement through windows that are low to the ground or
doorway entrances. An example of the flood depth map is shown on Figure 18 for the
21-22 January 2000 flood level. A revised economic impact assessment has not yet
been implemented using these maps. However, such maps can add significantly to the
information regarding flood risk, impact mitigation, and adaptation planning required to
minimize the effects of flooding.
4.4 FLOOD-RISK IMPACT ANALYSIS AND ADAPTATION
An additional component of the project focused on the potential economic impacts
of flooding events (Milloy and MacDonald, 2002). Much of the analysis involved the
use of GIS to summarize the areas potentially affected by flooding. This involved
compilation of the municipal GIS database, including property boundaries and building
footprints, linked to other databases such as property and building assessment
information, in relation to flood extent at various water levels (Dickie, 2001). Initially a
simple overlay of the flood-risk areas and the property boundaries was used to
determine that the flood affected more than 460 properties. By summarizing the tax
176 Webster and Forbes

ground” points to ensure an accurate representation.
Figure 17. Close up of flood risk areas. The downtown waterfront is on the right and a dense
residential area is at the center of the map. The large area in the center of the map that would be
flooded at the 4.93 m level corresponds to an ancient stream channel.
Figure 16. Flood risk areas, flood water levels are referenced to chart datum (CD). The 4.23 m
water level corresponds to the Jan. 21, 2000 storm-surge event. The DEM in the background was
constructed from the “ground” LIDAR points and refined along the waterfront using some “non-
177
Airborne Laser Altimetry

Figure 18. Floodwater depth associated with the Jan. 21, 2000 storm-surge event overlaid on a
shaded relief of the Digital Surface Model (DSM). The DSM was constructed from the combined
“ground” and “non-ground” LIDAR points. Water depth can have a significant impact on the
amount of damage an area sustains from flooding.
assessment values of these properties, an economic impact was calculated (Milloy and
MacDonald, 2002). It was also shown that public lands, critical infrastructure, and
transportation links (e.g. access to the hospital) were affected during the surge event.
The flood-risk vectors were passed to the City of Charlottetown Planning
Department to assist in defining future development plans for the waterfront. On a
municipal basis, some planning strategies can be pursued to limit vulnerability. These
include measures such as appropriate zoning, acquisition of flood-prone properties,
flood-proofing, or taking advantage of replacement schedules to move key
infrastructure to less vulnerable locations (Forbes et al., 2002).
Retreat from flooding may not be a broadly viable adaptation option in urban
settings such as Charlottetown, so alternative accommodation and protection strategies
may need to be considered. Furthermore, horizontal setbacks may not be appropriate in
areas where the risk is primarily related to flooding. In such cases, vertical thresholds or
setbacks in which the location is defined by elevation and flood probability may be
more effective in reducing exposure. Accommodation options are already being
considered in Charlottetown as a result of this study. These include flood-proofing of
basements and other measures to reduce damage in the event of flooding, as well as
more stringent assessment of building proposals in potentially flood-prone areas. These
are progressive moves. Other straightforward accommodation measures may include
raising foundation heights or the heights of protection structures, wharves, and other
coastal infrastructure (O’Reilly et al., 2003).
178 Webster and Forbes

5. Conclusions
This work demonstrates the application of LIDAR technology to the mapping of
storm-surge flood risk for coastal areas. The high-resolution of the LIDAR data allowed
a DEM to be constructed that can now be used to model the inundation effect of water
levels of 2 m or more higher than usual. However, the results of this study also
demonstrate the need for independent validation data to ensure the reliability of such
high-resolution topographic mapping.
Careful analysis of validation data using several different approaches revealed the
presence of an altitude calibration bias in the LIDAR elevations for the present study. A
bulk adjustment of the elevations by 0.9 m provided a reasonable representation of
flood levels, but variation in the flying altitude between flight lines resulted in failure to
meet the intended 0.3 m vertical error specification. This was partially mitigated by the
density of survey points, so that the mean elevation determined for each small grid cell
was more often within specification. New LIDAR surveys recently undertaken in
another coastal area of eastern Canada are building on the lessons learned in the Prince
Edward Island study to provide more accurate and precise DEM data. Nevertheless, the
LIDAR data obtained in Prince Edward Island provided unprecedented topographic
detail and enabled highly detailed delineation of flood hazard zones. The flood risk
maps and information products have made it to the hands of the coastal resource
managers, who have to deal with these risks on an annual basis, and have been
incorporated into their GIS system. The maps have provided a tool to allow the local
planning officials to begin a long-term adaptation process, and to initiate community
discussions. For the short term, planning officials now use the maps to inform local
developers of the future predictions of flooding events and possible water depths
associated with different areas of the waterfront. The planning department has used the
information to identify areas where development may be restricted along the waterfront
because of the risk of flooding.
6. Acknowledgements
We are happy to acknowledge the efforts of the data acquisition contractors,
including Rick Quinn, Roger Shreenan and others (Terra Remote Sensing Inc.), Herb
Ripley, Andrew Cameron, and Laura Roy (Hyperspectral Data International), the pilots
and other support staff. We thank Paul Fraser and Dan Deneau (Applied Geomatics
Research Group, Centre of Geographic Sciences, NSCC) for the Charlottetown GPS
campaign and preliminary analysis. We would like to acknowledge George Dias and
the other members of the AGRG class (2002) for coding the ARC AML that does the
comparison between GPS points and LIDAR points within a fixed radius. We are also
grateful to Steve Dickie for his work on the LIDAR processing and report for the PEI
project. We acknowledge the contributions of Mike Butler and Brent Rowley for
helping in the coordination of data acquisition during the summer of 2000. Gavin
Manson (Geological Survey of Canada) provided critical field and office support. Glen
King (CHS) assisted with vertical control data, as did Charles O’Reilly (CHS), who
was the prime inspiration behind the initiative to acquire laser altimetry for this project.
Don Poole (Planning and Development Officer, City of Charlottetown) provided
invaluable assistance. This study was funded in large part by the Climate Change
Action Fund (CCAF) of the Government of Canada and we are grateful to the entire
CCAF project team for their ideas and support on the project. Additional funding to the
179
Airborne Laser Altimetry

AGRG was through the Canada Foundation for Innovation, Industry Canada. The
manuscript benefited from internal reviews by Bob Maher of the AGRG and Gavin
Manson (GSC). This paper is a contribution to the Natural Resources Canada Program
on Reducing Canada’s Vulnerability to Climate Change and is Geological Survey of
Canada contribution 2004000.
7. References
Abdalati, W. and W.B. Krabill. 1999. Calculation of ice velocities in the Jakobshavn Isbrae area using
airborne laser altimetry. Remote Sensing of the Environment, 67:194-204.
Brock, J.C., C.W. Wright, A.H. Sallenger, W.B. Krabill, and R.N. Swift. 2002. Basis and methods of NASA
airborne topographic mapper LIDAR surveys for coastal studies. Journal of Coastal Research, 18:1-13.
Bruun, P. 1962. Sea level rise as a cause of shore erosion. Journal of Waterways and Harbors Division,
American Society of Civil Engineers, 88:117-130.
Chagnon, R. 2002. Sea-ice climatology. In Coastal Impacts of Climate Change and Sea-Level Rise on Prince
Edward Island. Edited by Forbes, D.L. and Shaw, R.W. Geological Survey of Canada, Open File 4261,
Supporting Document 5, 24 pp. (on CD-ROM).
Church, J.A., J.M. Gregory, P. Huybrechts, M. Kuhn, K. Lambeck, M.T. Nhuan, D. Qin, and P.L.
Woodworth. 2001. Changes in sea level. Chapter 11 in Climate change 2001: the scientific basis.
Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on
Climate Change. Cambridge University Press, Cambridge and New York, 639-693 pp.
Courtney, R.C. and J. Shaw. 2000. Multibeam bathymetry and backscatter imaging of the Canadian
continental shelf. Geoscience Canada, 27:31-42.
Cowell, P.J. and B.G. Thom. 1994. Morphodynamics of coastal evolution. In Coastal Evolution: Late
Quaternary Shoreline Morphodynamics. Edited by Carter, R.W.G. and Woodroffe, C.D. Cambridge
University Press, Cambridge and New York, pp. 33-86.
Dickie, S. 2001. Storm surge innundation mapping for Charlottetown, Prince Edward Island: Analysing
flood risk using high-resolution imagery. Unpublished Applied Geomatics Research Group report,
NSCC, Middleton NS.
FGDC ([US] Federal Geographic Data Committee). 1998. Geospatial accuracy standards, Parts 1 to 3
(www.fgdc.gov/standards/status/textstatus.html, documents FGDC-STD-007.1, FGDC-STD-007.2,
FGDC-STD-007.3).
Filin, S. 2003. Recovery of Systematic Biases in Laser Altimetry Data Using Natural Surfaces.
Photogrammetric Engineering and Remote Sensing, 69(11):1235-1242.
Forbes, D.L. and G.K. Manson. 2002. Coastal geology and shore-zone processes. In Coastal Impacts of
Climate Change and Sea-Level Rise on Prince Edward Island. Edited by Forbes, D.L. and Shaw, R.W.
Geological Survey of Canada, Open File 4261, Supporting Document 9, 84 p. (on CD-ROM).
Forbes, D.L., G. Parkes, C. O’Reilly, R. Daigle, R. Taylor, and N. Catto. 2000. Storm-surge, sea-ice, and
wave impacts of the 21-22 January 2000 storm in coastal communities of Atlantic Canada. Abstracts,
Canadian Meteorological and Oceanographic Society, 34th Congress, Victoria BC.
Forbes, D.L., R.W. Shaw, and G.K. Manson. 2002. Adaptation. In Coastal Impacts of Climate Change and
Sea-Level Rise on Prince Edward Island. Edited by Forbes, D.L. and Shaw, R.W. Geological Survey of
Canada, Open File 4261, Supporting Document 11, 18 p. (on CD-ROM).
Forbes, D.L., G.S. Parkes, G.K. Manson, and L.A. Ketch. 2004. Storms and shoreline retreat in the southern
Gulf of St. Lawrence. Marine Geology, 210:169-204.
Fraser, P. 2001. LIDAR validation using GPS, Charlottetown, Prince Edward Island. Unpublished Applied
Geomatics Research Group report, NSCC, Middleton NS.
Galy, H.M. and Sanders, R.A., 2002. Using synthetic aperture radar imagery for flood modelling.
Transactions in GIS, 6(1):31-42.
Gould, M., 2001. Flood risk modeling program. Unpublished Applied Geomatics Research Group report,
NSCC, Middleton NS.
Guenther, G.C., M.W. Brooks, and P.E. LaRocque. 2000. New capabilities of the SHOALS airborne LIDAR
bathymeter. Remote sensing of the Environment, 73:247-255.
Hwang, P.A., W.B. Krabill, W. Wright, R.N. Swift, and E.J. Walsh. 2000. Airborne scanning LIDAR
measurement of ocean waves. Remote sensing of the Environment, 73:236-246.
King, G., C. O’Reilly, and H. Varma, H. 2002. High-precision three-dimensional mapping of tidal datums in
the southwest Gulf of St. Lawrence. In Coastal Impacts of Climate Change and Sea-Level Rise on
Prince Edward Island. Edited by Forbes, D.L. and Shaw, R.W. Geological Survey of Canada, Open File
4261, Supporting Document 7, 16 p. (on CD-ROM).
Krabill, W.B. and C.F. Martin. 1987. Aircraft positioning using global positioning system carrier phase data.
Navigation, 34:1-21.
180 Webster and Forbes

Krabill, W., W.Abdalati, E. Frederick, S. Manizade, C. Martin, J. Sonntag, R. Swift, R. Thomas, W.
Wright, and J. Yungel. 2000. Greenland Ice Sheet: high-elevation balance and peripheral thinning.
Science, 289: 428-430.
Krabill, W.B., R.H. Thomas, C.F. Martin, R.N. Swift, and E.B. Frederick, E.B. 1995. Accuracy of airborne
laser altimetry over the Greenland ice sheet. International Journal of Remote Sensing, 16:1211-1222.
Krabill, W.B., C. Wright, R. Swift, E. Frederick, S. Manizade, J. Yungel, C. Martin, J. Sonntag, M. Duffy,
and J. Brock, J. 1999. Airborne laser mapping of Assateague National Seashore Beach.
Photogrammetric Engineering and Remote Sensing, 66:65-71.
Leatherman, S.P., B.C. Douglas, and J.L. LaBrecque. 2003. Sea level and coastal erosion require large-scale
monitoring. Eos, Transactions American Geophysical Union, 83 (2):13 and 16.
Maclean, G.A. and W.B. Krabill. 1986. Gross-merchantable timber volume estimation using an airborne
LIDAR system. Canadian Journal of Remote Sensing, 12:7-18.
Manson, G.K., D.L. Forbes, and G.S. Parkes. 2002. Wave climatology. In Coastal Impacts of Climate
Change and Sea-Level Rise on Prince Edward Island. Edited by Forbes, D.L. and Shaw, R.W.
Geological Survey of Canada, Open File 4261, Supporting Document 4, 31 p. and 1 attachment (on CD-
ROM).
Maune, D. F. 2001. Digital elevation model technologies and applications: The DEM Users Manual. Edited
by Maune, D.F. American Society of Photogrammetry and Remote Sensing.
Mayor, S.D. and E.W. Eloranta. 2001. Two-dimensional vector wind fields from volume imaging LIDAR
data. Journal of Applied Meteorology, 40:1331-1346.
McCulloch, M.M., D.L. Forbes, R.W. Shaw, and the CCAF A041 Scientific Team. 2002. Coastal Impacts of
Climate Change and Sea-Level Rise on Prince Edward Island.Edited by Forbes, D.L. and Shaw, R.W.
Geological Survey of Canada, Open File 4261, xxxiv + 62 p. and 11 supporting documents (on CD-
ROM).
McLean, R.F., A. Tsyban, V. Burkett, J.O. Codignotto, D.L. Forbes, N. Mimura, R.J. Beamish, V. and Ittekkot.
2001. Coastal zones and marine ecosystems. In Climate Change 2001: Impacts, Adaptations and
Vulnerability. Contribution of Working Group II to the Third Assessment Report of the
Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge and New York,
pp. 343-379.
Milloy, M. and K. MacDonald, K. 2002. Evaluating the socio-economic impacts of climate change and sea-
level rise. In Coastal Impacts of Climate Change and Sea-Level Rise on Prince Edward Island.Edited by
Forbes, D.L. and Shaw, R.W. Geological Survey of Canada, Open File 4261, Supporting Document 10,
90 p. (on CD-ROM).
Nicholls, R.J., F.M.J. Hoozemans, M. and Marchand. 1999. Increasing flood risk and wetland losses due to
global sea-level rise: regional and global analyses. Global Environmental Change, 9:S69-S87.
O’Reilly, C. 2000. Defining the coastal zone from a hydrographic perspective. Proceedings, Workshop on
risk assessment and disaster mitigation: Enhanced use of risk management in integrated coastal
management. International Ocean Institute, Bermuda. Backscatter, April 2000, pp. 20-24
O’Reilly, C.T., D.L. Forbes, and G.S. Parkes. 2003. Mitigation of coastal hazards: adaptation to rising sea
levels, storm surges, and shoreline erosion.Proceedings, 1st Coastal, Estuary and Offshore Engineering
Specialty Conference, Canadian Society of Civil Engineering, Moncton, NB, CSN-410-(1-10).
Parkes, G.S. and L.A. Ketch. 2002. Storm-surge climatology. In Coastal Impacts of Climate Change and
Sea-Level Rise on Prince Edward Island.Edited by Forbes, D.L. and Shaw, R.W. Geological Survey of
Canada, Open File 4261, Supporting Document 2, 87 p. (on CD-ROM).
Parkes, G.S., D.L. Forbes, and L.A. Ketch, L.A. 2002. Sea-level rise. In Coastal Impacts of Climate Change
and Sea-Level Rise on Prince Edward Island.Edited by Forbes, D.L. and Shaw, R.W. Geological Survey
of Canada, Open File 4261, Supporting Document 1, 33 p. and 5 attachments (on CD-ROM).
Parkes, G.S., L.A. Ketch, and C.T. O’Reilly. 1997. Storm surge events in the Maritimes. Proceedings,
Canadian Coastal Conference 1997. Canadian Coastal Science and Engineering Association, Guelph,
Ontario, pp.115-129.
Post, M.J., C.J. Grund, A.M. Weickmann, K.R. Healy, and R.J. Willis. 1996. A comparison of the Mt.
Pinatubo and El Chichon volcanic events: LIDAR observations at 10.6 and 0.69 Pm. Journal of
Geophysical Research, 101(D2):3929-3940.
Sallenger, A.B., Jr., W. Krabill, J. Brock, R. Swift, M. Jansen, S. Manizade, B. Richmond, M. Hampton,
and D. Eslinger. 1999. Airborne laser study quantifies El Niño- induced coastal change. Eos,
Transactions, American Geophysical Union, 80:89, 92.
Schuckman, K. 2003. Announcement of the Proposed ASPRS Binary LIDAR Data File Format Standard.
Photogrammetric Engineering and Remote Sensing, 69(1):13-19.
Shaw, J., R.B. Taylor, D.L. Forbes, M.H. Ruz, and S. Solomon. 1998. Sensitivity of the Coasts of Canada to
Sea-Level Rise.Geological Survey of Canada Bulletin 505.
Stockdon, H.F., A.H. Sallenger, J.H. List, and R.A. Holman. 2002. Estimation of shoreline position and
change using airborne topographic LIDAR data. Journal of Coastal Research, 18(3):502-513.
181
Airborne Laser Altimetry

Thompson, K., H. Ritchie, N.B. Bernier, J. Bobanovic, S. Desjardins, P. Pellerin, W. Blanchard, B. Smith,
and G. Parkes, G. 2002. Modelling storm surges and flooding risk at Charlottetown. In Coastal Impacts
of Climate Change and Sea-Level Rise on Prince Edward Island.Edited by Forbes, D.L. and Shaw,
R.W. Geological Survey of Canada, Open File 4261, Supporting Document 6, 48 p. (on CD-ROM).
Wadhams, P., W.B. Tucker III, W.B. Krabill, R.N. Swift, J.C. Comiso, and N.R. Davis. 1992. Relationship
between sea ice freeboard and draft in the Arctic Basin and implications for ice thickness monitoring.
Journal of Geophysical Research, Vol. 97 (C12):20,325-20,334.
Walsh, K.J.E. and J.J. Katzfey. 2000. The impact of climate change on the poleward movement of tropical
cyclone-like vortices in a regional climate model. Journal of Climate, 13:1116-1132.
Webster T.L., S. Dickie, C. O’Reilly, D.L. Forbes, K. Thompson, and G. Parkes, G. 2001. Integration of
diverse datasets and knowledge to produce high resolution elevation flood risk maps for Charlottetown,
Prince Edward Island, Canada. In CoastGIS2001, Halifax, Nova Scotia, Canada.
Webster, T.L., D.L. Forbes, S. Dickie, R. Covill, and G.S. Parkes. 2002. Airborne imaging, digital elevation
models and flood maps. In Coastal Impacts of Climate Change and Sea-Level Rise on Prince Edward
Island. Edited by Forbes, D.L. and Shaw, R.W. Geological Survey of Canada Open File 4261,
Supporting Document 8, 36 p. (on CD-ROM).
Webster T.L., S. Dickie, C. O’Reilly, D.L. Forbes, G. Parkes, D. Poole, and R. Quinn. 2003. Mapping storm
surge flood risk using a LIDAR-derived DEM. In Elevation, A Supplement to Geospatial Solutions,
May, pp 4-9.
Webster, T.L., D.L. Forbes, S. Dickie, and R. Shreenan. 2004. Using topographic LIDAR to map flood risk
from storm-surge events for Charlottetown, Prince Edward Island, Canada. Canadian Journal of Remote
Sensing, 30(1):1-13.
Wehr, A. and U. Lohr, U. 1999. Airborne laser scanning – An introduction and overview. Journal of
Photogrammetry and Remote Sensing, 54:68-82.
182 Webster and Forbes