LIDAR and the North Carolina Floodplain Mapping Program NCFMP

1
LIDAR and the North Carolina Floodplain Mapping
Program (NCFMP)

• Symposium on Terrain Analysis
• for Water Resources Applications
• December 16, 2002
• David F. Maune, Ph.D., CP, CFM
• Dewberry Davis LLC
• Fairfax, Virginia 22031-4666
• Tel [phone]
• E-mail dmaune_at_dewberry.com

2
Outline

• National Flood Insurance Program (NFIP)
  requirements for Digital Elevation Data
• Overview of LIDAR technology
• The North Carolina Floodplain Mapping Program
  (NCFMP)
• LIDAR standards, QA/QC issues

3
National Flood Insurance Program (NFIP)

• Created by Congress in 1968 to
• Transfer costs of private flood losses from
  taxpayers to floodplain property owners through
  insurance premiums
• Guide development away from flood hazard areas
• Require new buildings be constructed to
  minimize/prevent flood damages

4
How the NFIP Works

• The NFIP is based on a mutual agreement between
  the Federal Government and participating
  communities.
• Federally guaranteed flood insurance is made
  available in those communities that agree to
  regulate development.
• Community Rating System (CRS) credits may further
  reduce insurance premiums for participating
  communities.

5
Flood Insurance Rate Maps (FIRMs)

• Official maps of a community on which FEMA has
  delineated both the Special Flood Hazard Area and
  the risk premium zones applicable to the
  community.
• Digital FIRMs (DFIRMs) are more easily
  revised/updated than paper FIRMs and can be
  incorporated in a communitys Geographic
  Information System (GIS).

6
FIRM/DFIRM Uses

• The FIRM or DFIRM is used to determine
• Whether a property is in the floodplain
• The flood insurance zone that applies to the
  property
• The approximate base flood elevation (BFE) at the
  site
• Used to mitigate future flood losses in
  floodprone communities through management of
  development

7
Ingredients for a DFIRM

DFIRM
8
Hydrologic Modeling

• Models entire watershed (river basin) to compute
  peak discharges at key locations.
• Discharges are predicted from rainfall,
  watershed characteristics (e.g., land cover,
  soils, and terrain slope), and flood routing.
• For flood routing and slope, topographic data
  does not need to be highly accurate hydrologic
  models are not very sensitive to moderate
  elevation errors.

9
Hydraulic Modeling

• Models floodplain to compute surface water
  velocities and elevations predicted from
  hydrologic model discharges and channel and
  floodplain cross section characteristics (e.g.,
  area, slope, vegetation roughness) and
  information on bridges, culverts, dams.
• Computes flood profiles, flood boundaries.
• Cross sections, water-surface elevations, flood
  profiles and floodplain boundaries all require
  accurate topographic data.

10
Most Accurate Elevation Data are Needed in Low
Elevations of Floodplains
11
Types of Digital Elevation Data

• Mass points
• Breaklines
• Triangulated Irregular Networks (TINs)
• Digital Elevation Models (DEMs)
• Digital Terrain Models (DTMs)
• Digital Surface Models (DSMs)
• Cross sections

12
Triangulated Irregular Network (TIN)

• TIN data structure is based on mass points and
  breaklines
• Superior for hydraulic modeling
• LIDAR is poor for generation of breaklines.
  Imagery is normally required
• Photogrammetry and LIDAR may be combined for best
  results, especially if controlled imagery is
  already available

13
Digital Elevation Models (DEMs)

• DEMs have uniform post spacing where ?x and ?y
  5m or 20 ft, for example.
• DEMs are interpolated from mass points and
  breaklines, or TIN data.
• Neither TIN nor DEM points are clearly defined on
  the ground.

14
Digital Terrain Model (DTM) (Example Breakline
in Red)

• DEMs can jump over key breakline features,
  e.g., tops/bottoms of stream banks
• DTMs add significant topographic features and
  change points (breaklines) that are irregularly
  spaced to better characterize the shape of the
  terrain

15
Hydraulic Models Require Representative Cross
Sections

• Traditionally, cross sections are selected to
  represent reaches that are as long as possible,
  without permitting excessive conveyance change
  between sections.
• Either photogrammetry or high-density LIDAR data
  may be suitable for cutting cross sections above
  the water surface, but not subsurface.

16
Field-Surveyed Cross Sections

• Surveyed cross sections immediately upstream or
  downstream of bridges
• Intermediate cross sections from LIDAR or
  photogrammetry
• Supplemental breaklines at tops/bottoms of stream
  banks and/or hydro-enforced breaklines

17
LIDAR Data Acquisition
Courtesy Dodson Associates, Inc.
18
LIDAR Laser Sensor

• Airborne GPS provides x/y/z at pulse origin
• Inertial Measurement Unit (IMU) provides roll,
  pitch and heading of sensor for each pulse
• LIDAR system measures laser scan angles and time
  (distance) to hard and soft points on the ground
  for up to 5 returns for each pulse

Cost effective for digital elevation data
equivalent to 2 contours
19
Multiple Returns
20
LIDAR Data Collection(Fixed Wing or Helicopter)
21
LIDAR System Calibration
22
Lidar Last-Return Data of Baltimore
23
Lidar Intensity Returns of Baltimore
24
FEMA LIDAR Applications

• Most cost-effective digital topographic data for
  cross sections and hydraulic modeling, but value
  for breaklines is still to be determined (depends
  on post spacing)
• LIDAR may also prove to be ideal for determining
  flood risks to individual buildings, comparing
  lowest adjacent grade (LAG) with BFE

25
Bare-Earth Post-Processing
26
Color-Coded Last Return LIDAR Elevations and
Gray-Scale Void Map
Images courtesy of U.S. Army Topographic
Engineering Center
27
After Post-Processing forVegetation/Building
Removal
Images courtesy of U.S. Army Topographic
Engineering Center
28
This Slide Shows at Least Three Problems from
these LIDAR Points
29
LIDAR Comparisons
Last return LIDAR
Bare Earth LIDAR
First return LIDAR
Images courtesy of U.S. Army Topographic
Engineering Center
30
Are the Purple Lines on the Ground?
31
Artifacts

• Artifacts result when LIDAR fails to penetrate
  vegetation and/or when post-processing procedures
  incorrectly model the bare-earth terrain
• May not impact hydraulic modeling because cross
  sections can be cut to avoid such artifacts

32
LIDAR TINs, DEMs and Contours are Unforgiving
33
LIDAR irregular contours and smoothed with
cartographic license
34
FEMA Lessons Learned(prior to NC LIDAR project)

• Under ideal conditions, LIDAR bare earth DEMs can
  achieve RMSEz of 15 cm for hydraulic modeling in
  open terrain only. If possible, acquire LIDAR
  during leaf-off and low-water conditions for
  accurate hydraulic modeling.
• Data voids from water and deliberate
  post-processing to remove vegetation and manmade
  features may cause problems in generating DEMs
  through interpolation.

35
FEMA Lessons Learned(continued)

• For higher accuracies, LIDAR is now cost
  competitive with photogrammetry
• All mapping technologies, including
  photogrammetry, yield obscured areas where bare
  earth elevations are uncertain
• Photogrammetry needs to see the ground from two
  (angular) perspectives whereas LIDAR only needs
  to see the ground from one (near vertical)
  perspective

36
Cost of Photogrammetry vs. Lidar
37
LIDAR Cost Comparisons
38
Hurricane FloydSeptember, 1999

• Revealed limitations in NCs FIRMs
• Majority over 10 years old, compiled in 1970s by
  approximate methods
• FEMAs funds average one updated countywide study
  per state per year
• Most of NC needed to be immediately re-mapped
  digitally consistent with FEMAs Map
  Modernization Plan
• State decided to take decisive action as FEMAs
  first Cooperating Technical State (CTS)

39
North CarolinaCooperating Technical State

• North Carolina entered into a CTS agreement with
  FEMA whereby the state assumed primary ownership
  of, and responsibility for, the NFIP maps for all
  North Carolina communities, to include
  resurveying the state, conducting flood hazard
  analyses, and producing updated, digital FIRMs
  (DFIRMs).
• DFIRMs allow the latest GIS technologies to be
  used, facilitate updating in the future, and
  allow on-line distribution.

40
North Carolina Solicitation

• Phase I advertised for TINs and DEMs with
  specified accuracy for six Eastern watersheds in
  2001 ( 50 counties)
• All firms proposed use of LIDAR as opposed to
  photogrammetry
• Phase II advertise for TINs and DEMs with
  specified accuracy for six central watersheds in
  2003 ( 38 counties)
• North Carolina Geodetic Survey (NCGS) and
  Dewberry Davis are actively involved in QC

41
Phases of the North Carolina Floodplain Mapping
Program
Phase IPhase II Phase III
42
Objective of NC QC Surveys

• Establish quality control (QC) checkpoints to
  evaluate the vertical accuracy of the TIN data
• 20-cm vertical RMSE in coastal counties (39.2 cm
  vertical accuracy at 95 confidence level)
• 25-cm vertical RMSE in inland counties (49.0 cm
  vertical accuracy at 95 confidence level)
• 5 of checkpoints allowed to be discarded from
  RMSE computations to account for uncleaned
  artifacts this controversial issue will be
  discussed in detail later in this workshop.

43
Checkpoint Land Cover Classes

• FEMA requires TINs to be tested separately for
  major land cover classes within the floodplain
  being studied, 20 checkpoints per category. NC
  selected 120 checkpoints per county for the
  following 5 categories
• 20 in open terrain (bare-earth and grass)
• 20 in weeds and crops
• 20 in scrub
• 40 in forests (higher weight than other areas)
• 20 in built-up or urban areas

44
Open Terrain nominally 20
checkpoints/county
45
Weeds and Crops nominally 20 checkpoints/county
46
Scrub nominally 20 checkpoints/county
47
Forested Areas nominally 40 checkpoints/county
48
Built-Up or Urban Areas nominally 20
checkpoints/county
49
(No Transcript)
50
(No Transcript)
51
LIDAR/Field Cross-SectionsWhite Oak River Basin
LIDAR Elev.s from TIN vs. Field Survey Elev.s
Section Upstream of Rhodestown Rd in Onslow
County
52
LIDAR/Field Cross-SectionsWhite Oak River Basin
LIDAR Elev.s from TIN vs. Field Survey Elev.s
Section Upstream of Northwest Bridge Rd in
Onslow County
53
Outreach Tools

• Website
• Newsletters
• Direct correspondence
• Toolkits (fact sheets, pamphlets, references,
  etc) for the following
• Final meetings
• Media
• Congressional briefings
• General public, educators, and interest groups

54
Web site

• www.ncfloodmaps.com

55
National Standard for Spatial Data Accuracy
(NSSDA) 1998

• Vertical accuracy at 95 confidence level
  Accuracyz 1.9600 x RMSEz, when errors follow a
  normal distribution
• No clear guidance on what to do when errors dont
  follow normal distribution
• No provision to offset vertical errors by
  allowable horizontal errors
• A minimum of 20 check points tested.
• Users generally do not understand meanings of
  RMSEz and Accuracyz

56
NSSDA (continued)

• When 20 points are tested, the 95 confidence
  level allows one point to fail the threshold
  given in product specifications.
• Accuracy reported at the 95 confidence level
  means that 95 of the positions in the dataset
  will have an error with respect to true ground
  position that is equal to or smaller than the
  reported accuracy value.
• This allows 95th percentile method

57
Equivalent Contour Interval compared with NMAS
and NSSDA
58
Major NC Requirements(Phase I)

• Vertical RMSE 20 cm in coastal counties
• Vertical RMSE 25 cm in inland counties
• Worst 5 of 120 checkpoints per county discarded
  to accommodate outliers in vegetated areas.
  RMSE is based on the best 95 of the checkpoints
  thus, all NC RMSE statistics are biased. Cannot
  say 95 of checkpoints are accurate within 1.9600
  x RMSE

59
Lidar Accuracy Assessment Craven County Example
60
Craven County Histogramwith Best 114 Checkpoints
RMSE 14.8 cm Easily Passes States 20 cm
criteria
61
Craven County Histogramwith Total 120 Checkpoints
RMSE 21.3 cm Would have failed 20 cm criteria
62
Statistical Indicators(Craven County)
63
(No Transcript)
64
100 Checkpoint Example(95th percentile for 105
pts 28.2 cm)
Outliers
381.6 51.7 49.6 42.0 32.4
65
Accuracy Statistics for Original 105 checkpoints,
then 104, then 103
No Outlier Deleted Delete 381.6
cm Delete 51.7 cm
66
Accuracy Statistics for 102 checkpoints, then
101, then 100
Delete 49.6 cm Delete 42.0 cm Delete 32.4 cm
67
www.fema.gov/mit/tsd/dl_cgs.htm

• Replaces Appendix A to FEMA 37
• Replaces Appendix 4B (LIDAR)
• Updates to FGDC standards
• Will update to guidelines of the National Digital
  Elevation Program (NDEP)

68
National Digital Elevation Program
A consortium of agencies coordinating the
collection and application of high-resolution,
high-accuracy elevation data
69
NDEP Draft Guidelines (as of September 2002)

• Fundamental Vertical Accuracy. For open terrain
  only, compute RMSEz. Report Accuracyz 1.9600 x
  RMSEz as Tested __ (meters, feet) Fundamental
  Vertical Accuracy at 95 confidence level in open
  terrain.
• Supplemental Vertical Accuracy. For all other
  land cover categories, determine 95th percentile
  error(s). Report Accuracyz as Tested __
  (meters, feet) Supplemental Vertical Accuracy at
  95th percentile in weeds, crops, scrub, forests,
  urban areas, etc. and document outliers. AND/OR
• Consolidated Vertical Accuracy. Report Accuracyz
  as Tested __ (meters, feet) Consolidated
  Vertical Accuracy at 95th percentile in open
  terrain, weeds, crops, scrub, forests, urban
  areas, etc. and document outliers.

70
Hydro-Enforcement of Bridges

• Rivers flow smoothly downstream?
• Bridges, box culverts cut?
• Corrugated pipe culverts cut?
• Ditches?
• Puddles drained?
• Puddles filled?

71
LIDAR bare-earth mass points - river not yet
hydro-enforced
72
TIN without breaklines -river not yet
hydro-enforced
73
TIN with breaklines river now hydro-enforced
74
The North Carolina Flood Insurance Rate Map (FIRM)
75
Flood Insurance Rate Map

• 1 annual chance floodplain-cyan dots (FEMA
  Standard)
• Floodway White Hatch (FEMA Standard)
• 0.2 annual chance floodplain-black dots
• Numbered cross sections

76
Questions?