By John K. Dorman, planning administrator for the
North Carolina Department of Crime Control and Public Safety within the
Division of Emergency Management (www.ncem.org),
and Harold Rempel, director of Program Management for EarthData
International (www.earthdata.com).
Under the watchful eye of the Federal Emergency Management Agency (FEMA),
North Carolina has completed the initial phases of a statewide
floodplain mapping project that may serve as the blueprint for other
states to follow in updating digital flood insurance rate maps (DFIRMs).
The first-of-its-kind project leveraged federal and state resources to
map terrain elevations with a higher level of accuracy and within a
tighter schedule than either North Carolina or the U.S. government could
have accomplished on its own.
Impetus for the seven-year, $110 million endeavor came from Hurricane
Floyd in 1999. The storm swept across
North Carolina with torrential rains that caused widespread flooding of
rivers and streams. Along with more than 50 deaths, Floyd destroyed tens
of thousands of homes and businesses. An estimated 80 percent of the
property owners who lost their structures didn’t realize they needed
flood insurance because the official floodplain maps for their areas
were either outdated or inaccurate.
Ultimately, the U.S. government, the state of North Carolina and the
countless individuals who lacked proper insurance paid Floyd’s $6
billion price tag. The hurricane served as a stark reminder of the value
of quality maps. Although the flooding couldn’t have been avoided, the
devastation could have been minimized if more accurate floodplain maps
had been available to help the state’s communities determine who should
have insurance and where homes and buildings should—or should not—have
been built.
The North Carolina Department of Crime Control and Public Safety,
Geodetic Survey, and Floodplain Mapping Program Office conducted an
immediate study of the causes and possible solutions to the flooding
problem. The effort revealed most of the FEMA DFIRMs for the state were
20 to 30 years old and didn’t include many of the manmade topographic
changes resulting from intense development in recent years. In addition,
most of the existing FEMA flood hazard maps had been created using
outdated U.S. Geological Survey data, with contour intervals of
approximately 10-20 feet. These were deemed inadequate for mapping
floodplains in the North Carolina terrain.
With FEMA input, the study group concluded that it cost federal and
state governments $56 million for each year North Carolina lacked
accurate floodplain maps. Equipped with this information, the group
presented three separate, but related, proposals to the governor and
general assembly:
1. Update the DFIRMs and create an ongoing floodplain mapping program.
2. Solicit FEMA to allow the state to assume primary responsibility for
the DFIRMs.
3. Develop a Web-based flood inundation mapping and alert system.
The study group concluded the success of all three initiatives required
collecting elevation points across the entire 48,000-square-mile state,
something never before attempted. The data would have to be new,
accurate and detailed, according to the group. Furthermore, the
participants identified airborne light detection and ranging (LiDAR)
technology as the only method capable of achieving this objective in a
reasonable period of time.
With memories of Hurricane Floyd still vivid, the North Carolina
governor and general assembly approved a
$56 million state investment in record time. FEMA fully supported the
project, eager to find an efficient process that could be duplicated in
other states to keep its maps updated. By early 2001, North Carolina was
ready to kick off the most ambitious elevation mapping project ever
conceived.
Collecting Statewide LiDAR
The first challenge was finding a team of mapping contractors with the
right combination of technical capabilities to handle such a large
project with such demanding specifications. Meetings between FEMA and
the state had determined the vertical accuracy of the LiDAR would be 20
centimeters along the coastal areas and 25 centimeters inland, with
five-meter post spacing.
The state sought proposals from teams that would collect the LiDAR data,
process the data and deliver a variety of end products, including
county-by-county DFIRMs delineating floodplains at 50-, 100-, and
500-year flood levels. North Carolina also asked to receive several
other mapping products that would be developed to generate the DFIRMs.
These included all-return LiDAR data sets, bare-earth digital elevation
models (DEMs) and hydrologically corrected DEMs.
North Carolina initially launched the project using two teams of
mapping contractors, but it quickly became evident the desired goal
of a seamless statewide dataset could best be achieved by one team
applying a consistent set of processes, techniques and procedures
across the state. A single primary contract ultimately was awarded
to the team comprising Cary, N.C.-based Watershed Concepts (http://www.watershedconcepts.com)
as prime contractor and EarthData International (www.earthdata.com),
Frederick, Md. Four other contractors also worked on the statewide
program. Watershed Concepts provided program management, quality
control and hydrologic correction of the DEMs and deliverables. The
company also provided other services, including surveying,
hydrologic and hydraulic modeling, floodplain mapping, flood
insurance study reporting, and DFIRM preparation. EarthData provided
LiDAR data acquisition, processing and production.
From that point, the project proceeded in three main phases. The
first phase covered the coastal region, and the second phase mapped
the piedmont areas. The third phase focused on the mountainous
western part of the state. Looking back on the project, this
division of work contributed significantly to the effort’s overall
success.
Dividing the state into regions based on dominant topography allowed
EarthData to customize and fine tune its LiDAR acquisition
parameters to each type of terrain as the project progressed. Each
topography presented its own particular challenges to acquisition or
processing, and it was beneficial to focus on one at a time,
although there were many variations in land-cover and ground
conditions within a phase.
For the airborne segments, EarthData mobilized two aircraft equipped
with identical Leica ALS series LiDAR sensors. Because LiDAR can
operate in the dark, data collection continued around the clock
during the leaf-off flying season. The operator made constant
adjustments to the laser sensor to achieve the desired data accuracy
and point spacing. This typically involved varying the altitude of
the aircraft and/or the pulse rate of the LiDAR depending on the
terrain and ground cover in the target area. In general, a lower
flying height or a faster pulse rate was required over densely
vegetated or extremely mountainous terrain.
For an acquisition flight, or “lift,” ground crews established a
Global Positioning System (GPS) base station at the nearest airport.
The data were used to differentially correct the elevation data
being collected in the air. At the end of each lift, the LiDAR
operator made an initial quality check of the data on a personal
computer to ensure there were no gaps in the flight lines. That
lift’s data then were shipped to EarthData’s processing facility in
Maryland.
Following calibration, technicians in Maryland applied a series of
proprietary processing techniques to the raw LiDAR data to produce
the bare-earth DEM and the all-return data set. The technicians used
recent aerial photography of the target area and extracted
breaklines relating to each watershed feature in the DEM. A computer
algorithm assigned elevation values from the LiDAR data to each of
the breaklines, which then were delivered to Watershed Concepts with
the DEMs for further processing and modeling.
While the aircraft were in the air, survey
crews contracted by the state were on the ground collecting points
for quality-control purposes. As agreed by FEMA, the crews captured
120 points in each county with about 20-25 points taken from five
different dominant land-cover types. When these third-party teams
checked their ground collection data against the bare-earth LiDAR
DEMs, they confirmed vertical accuracy ranging eight to 18
centimeters. The post spacing also exceeded expectations, averaging
four meters across the state.
“We realized early in the project that vegetation was the biggest
challenge for accurate LiDAR collection, so we focused our quality
control work on the forested areas,” explains Gary Thompson, section
chief for the North Carolina Geodetic Survey. “LiDAR collection has
been completed for all 100 counties in the state, and every dataset
has passed our quality-control procedures.”
Creating New Flood Maps
To create the updated DFIRMs, Watershed Concepts developed
computerized hydrologic and hydraulic models for all of the state’s
streams with drainage areas greater than one square mile. The models
used the bare-earth DEMs supplemented with ground survey data to
determine where flood water will flow when it overflows its banks.
The models also calculated the volume and depth of water in the
floodplain under the various flood scenarios.
“Ground surveys of river and stream cross sections, using both GPS
and traditional techniques, provide the critical data inputs for the
modeling from tree canopy,” says David Key, the company’s vice
president.
Usually within six months of LiDAR collection, Watershed Concepts
put crews on the ground to survey the cross sections, which extended
across the channel of the river or stream perpendicular to the water
flow. In a typical scenario, the crew established a GPS point about
200 feet from the edge of the channel escarpment. From that point,
the crews used traditional equipment to survey points every 10-20
feet down the bank, through the water and up the other bank. Row
boats or hip waders were used to survey the bottom of the stream
bed.
“We used traditional surveying techniques in the channels because
there was usually so much vegetation in the river beds that we found
GPS incurred too much interference,” relates Key.
For major rivers and streams in each county or
large watershed, the crews took cross-section surveys every 2,000 to
3,000 feet down the entire length of the channel. Additional surveys
were taken in the vicinity of bridges and overpasses because these
objects directly influence the flow of flood waters, making accurate
bank slope and depth data within their vicinity crucial to the
modeling work.
“The accuracy of the floodplain maps depends on the quality of the
hydrologic and hydraulic engineering models, which in turn rely on
good LiDAR and survey data,” adds Key. “The LiDAR data supplied in
this project has been better than any we’ve had in the past, and
that means we’re creating excellent floodplain maps.”
Once the engineering modeling is completed, the project will include
detailed hydrologic and hydraulic studies of more than 24,000 linear
miles of river and stream channels throughout the state.
A Flood of Good Results
Watershed Concepts delivers the DFIRMs on a per-county basis to
North Carolina, which performs additional quality control before
submitting the preliminary maps to counties and municipalities for
preliminary review. The flood maps go through a 90-day appeal period
during which citizens and community leaders may comment on the maps
and submit protests and appeals. After the appeal period, the maps
go through a compliance period before they become official or
“effective” maps that guide developers as to where and how new
structures can be built.
“The response from local zoning and floodplain management officials
has been overwhelmingly positive,” says Thompson. “We get calls all
the time from local officials telling us how helpful it is to have
the new maps.”
Getting the updated maps into the hands of local government offices
as quickly as possible has been a priority of this project from the
start. Although much of the post-Floyd reconstruction has occurred,
North Carolina remains one of the fastest growing states in the
nation. With the new maps in place in 68 counties as of December
2006, this extensive new construction is being guided by more
precise floodplain management principles than were previously
available.
As promised to the state legislature in 2001,
the program doesn’t end with delivery of the last updated DFIRM next
year. North Carolina plans to perform spot updating of about 8
percent of the maps and models every three years. The all-digital
nature of the project already has paid dividends by enabling the
team to make quick updates of maps completed in the early phases.
These updates were based on actual flood levels recorded in
hurricanes Ivan and Francis in 2004. FEMA has been extremely
impressed with the state’s ability to incorporate new information
into the project so rapidly.
Elsewhere in North Carolina, response to the project has been just
as positive. The state has disseminated the DEMs and LiDAR data to
interested parties for use in a variety of projects outside of
floodplain mapping. The state Department of Transportation is using
the bald-earth DEM for the preliminary design of new roadways, and
the Forestry Department has found the all-return LiDAR data useful
in evaluating tree maturity. At the federal level, the Federal
Aviation Administration has asked to use the data for flight-line
obstruction modeling around airports. And several state universities
are applying various datasets for environmental analysis.
Unprecedented Value
Summarizing a project that has returned $7 in value for every $1
invested by the state is simple: A statewide floodplain mapping
project, although daunting in scope, can deliver a significant
return on investment for every level of government involved. Flood
waters don’t stop at the city or county border, making it imperative
for the state to take the lead in ensuring its citizens are properly
protected from the inevitability of rivers and streams overflowing
their banks.
