Combining data from different lidar surveys and photogrammetry to quantify shortterm topographic cha

1
Combining data from different lidar surveys and
photogrammetry to quantify short-term topographic
change on North Carolina coast
H. Mitasova, Department of Marine, Earth
and Atmospheric Sciences, M. Overton,
Department of Civil Engineering, North Carolina
State University, R. S. Harmon, Army Research
Office, D. Bernstein, Geodynamics llt.

2
Monitoring short term topographic change
North Carolina coast Barrier islands with
dynamic terrain Surveyed with increasing
frequency over the past 10 years using
photogrammetry, lidar, and RTK-GPS. 1996-2000
annual lidar NASA/NOAA/USGS ATM II 2001 lidar
survey part of a NC state flood mapping data
from both surveys are available on-line 2003
pre- and post- hurricane Isabel NASA/USGS EAARL
H. Mitasova
3
Case studies Jockey's Ridge, Bald Head Island
Jockey's Ridge
North Carolina
Bald Head Island
H. Mitasova
4
Jockey's Ridge State Park
Established in 1975 to save the dunes, it was
believed to be stable. It is now completely
surrounded by development and it migrates faster
than expected, threatening nearby homes and
roads Challenge keep the naturally migrating
dune within the park
2003 view from NE
H. Mitasova
5
Jockey's Ridge elevation data
1974
1995
Photogrammetry 0.76m vertical accuracy (5ft
contours) Lidar 0.15m vertical accuracy
altitude 700m and 2300m
2001
1999
H. Mitasova
6
Spatial interpolation
Regularized Spline with Tension (RST Mitasova and
Mitas 1993) implemented in GRASS, ArcGIS,
Intergraph

• - sum of radial basis functions trend,
• obtained by minimization of smoothness
  functional
• tension, anisotropy, smoothing
• error analysis CVE and deviations
• computation of slope, aspect, curvatures,
• segmented processing for large data sets based
  on quad and oct-trees
• - works in 2D, 3D and 4D

http//skagit.meas.ncsu.edu/helena/gmslab/viz/
sinter.html
7
Jockey's Ridge 1999 lidar data
what can be gained by high quality interpolation
a
b
A
points assigned directly to a grid by s.to.rast
a) 1m and b) 3m resolution
c
c) RST interpolation 1m resolution grid all
points with distance gt 0.5m are preserved
H. Mitasova
8
1998 DOQQ, 1999 LIDAR, 2002 RTK-GPS
Accuracy of interpolated DEMs
Differences between interpolated DEMs 50pts on
pavements 1995-1999 RMS 0.16m Range
lt-0.71,0.38gt 1999-2001 RMS 0.03m Range
lt-0.09,0.27gt
sand pavement vegetation
H. Mitasova
9
Jockey's Ridge evolution
1974 and 1995 photogrammetry
1999 ATM lidar USGS/NASA/NOAA
N
D
A
1974 (brown) 1995 (yellow)
migration rate 3m/y
1995 (yellow) 1999 (red)
migration rate 7m/y
H. Mitasova
10
Elevation change
1974 1995 21 years
feet
1999- 2001 2 years
C
1995 1999 4 years
Year zm dz/yearm 1950 43 1974
34 0.37 1995 27 0.33 1999 25.7
0.32 2001 24.8 0.45 2002 23.6 1.20
N
H. Mitasova
11
Jockeys Ridge evolution natural man-made
Dune rolled over minigolf
N
C
2002
2000
2003
winter 2003
Nature tries to shift OB but man keeps shoveling
it back Cornelia Dean, NYT Sept. 22
H. Mitasova
12
Jockey's Ridge after Isabel
Jockeys Ridge evolution and management
1. Evolution of the dune field follows a long
term pattern of faster SW and slower SE
migration 2. Migration is accompanied by
flattening and latest surveys indicate
acceleration of both trends 3. The fences in
Central section work as elevation there has been
increasing 4. Does the major relocation of sand
work?
N
13
Bald Head Island
Human and natural impacts on evolution of
topography and bathymetry 1996
nourishment 1998, 1999 hurricanes Bonnie,
Floyd 2001 - channel deepening and re-alignment,
beach nourishment 2003 - Isabel
ATM LIDAR 1997-2000, EAARL LIDAR 2003
USGS/NOAA/NASA RTKGPS 2001-03 supported by BHIC
old channel
elevation m
Single and multi beam sonar 2000, 2001, 2002
USACE FRF
10m resolution bathy-topo model from multiple
sources
14
South Beach evolution 1997-2000
Overlayed 1997 and 2000 LIDAR surfaces central
section is relatively stable, rest erodes while
changing its shape and moving landwards
Annual sand loss rate 4000 m3/ha Shoreline
erosion rate up to 10m/year 15 of sand was
deposited behind the foredune landward movement
West Center East
convex -gt concave
2000 zgt0m zlt0m
stable pivot area
stable
concave -gt convex
15
Slope and curvature change
Severely eroding area approximated and analyzed
by RST
1998
Slope
Profile curvature
concave convex
2000
16
Bald Head Island 2001
Human impact channel deepening and
re-alignment, beach nourishment in 2001.
Elevation m
re-aligned channel
Integrated 10m resolution model from multiple
sources multibeam and single beam sonar
17
RTK GPS 2001-2003
RTK GPS data are oversampled in the direction of
the vehicle movement. Anisotropic interpolation
is needed when distance between profiles is
significantly greater than resolution.
RST with anisotropy
RST default parameters
Impact on volume change estimate
120000m3/160000m3
H. Mitasova
18
RTK GPS surveying pattern
Binned LIDAR data were sampled by RTK-GPS survey
points. DEM was then interpolated and compared
with the LIDAR data 4643 grid points for 5m,
13108 grid points for 3m resolution.
survey pattern / RST gridding no. of
points MAE m RMSE csh profiles / isotropic
179 / 55 0.73
0.78 csh profiles / anis. optim.
179 / 55 0.43
0.36 lsh profiles / anis., optim.
990 / 757 0.27 0.12 cshlsh
profiles / anis., optim. 1169 / 789
0.21 0.08 same at 3m resolution
1169 / 988 0.19
0.07 cshlsh profiles / anis., opt.
subset 0.16 0.05
DEM approximated from cross-shore profiles is the
least accurate. Long-shore profiles, anisotropy
and optimized approximation parameters can
significantly improve the accuracy of DEM.
H. Mitasova
19
Change after nourishment
Dec. 2001 1 million m3 of sand added
Sep 2002
Dec 2002 sand loss over 300,000 m3
August 2003 erosion spreads eastward
H. Mitasova
20
Sand volume change
time period
loss gain loss rate

m3 m3
m3/ha.year 1997 - 2000 Bonnie (98)
Floyd(99) 376,000 65,000 4,000
2001 nourishment
- 1,000,000 -
Dec01 Dec02 361,000
7,400 12,000 Dec02 - Dec03

West section is retreating at the rate
10m/year, compared to 3m/year official long term
rate. Cape Fear is growing but the gain is less
than 30 of the lost sand volume.
H. Mitasova
21
December 2003
Summer 2001
H. Mitasova
22
Shoreline change where we are now
Spatially variable erosion rates accompanied
with shoreline rotation and beach shape change
along the stable pivot point. After
renourishment erosion rates dramatically
increased in the west section and the beach
became more stable in the east. Cape Fear has
grown over 200m. Current shoreline is still
within its historical range.
2000
2001
1914
2003
1962
1914-1962 350m 1962-2003 - 240m
23
Conclusions
Combination of modern mapping techniques with
Open source GIS GRASS (http//grass.itc.it)
provides unique insight into 3D coastal
topography evolution at high spatial and temporal
resolution. GIS based analysis and visualization
allows us to quantify the observed changes
(elevation, shoreline, volume, slope and shape)
and evaluate effectiveness of stabilization
measures. The developed methodology is being
further enhanced and applied to other areas.
H. Mitasova
24
Acknowledgment Research was funded by NRC/ARO
fellowship http//skagit.meas.ncsu.edu/helena/