Multiscale characterization of nearshore environment using Open source GIS technology

1
Multi-scale characterization of near-shore
environment using Open source GIS technology
Helena Mitasova and Tom
Drake Department of Marine, Earth and Atmospheric
Sciences North Carolina State University,
Raleigh, NC Advisor R.S. Harmon, ARO
2
GIS and the Environment
Monitoring processing imagery and site
environmental data, spatial access to data
through Internet (USGS, EPA) Analysis and risk
assessment integration of multiple-source data
spatial analysis Prediction, modeling,
simulation numerical modeling for prediction of
impacts Planning and decision support cost
effective prevention and conservation
Prepared by H. Mitasova ImagesGMSLab
University of Illinois T. Frank, W.Brown,
W.Reez, D. Johnston Slide design A. Mitas
3
Objective

• Coastal field measurements and models involve
  processing, analysis and visualization of large
  volumes of spatial data, generated in different
  environments and formats. To support coastal
  applications GIS needs enhancements to improve
  support for large, heterogeneous, spatio-temporal
  data sets at multiple scales.
• Goal
• Enhance, develop and test Open source GIS GRASS
  tools for
• multivariate interpolation with analysis of
  geomorphology
• multidimensional dynamic cartography
• GIS support for coastal processes simulations

4
Open source GIS GRASS
General purpose GIS for raster, vector, site and
image data processing for all flavors of UNIX
LINUX, MacOS X, Solaris, IRIX and WINDOWS/Cygwin.
Originally developed at US Army CERL. Released
under General Public Licence (GPL) as free to
run, study, modify, re-distribute, improve, and
release (cannot be improved and released as
proprietary) Lead development coordinator
Markus Neteler, ITC, Trento, Italy
http//grass.itc.it/
5
Modeling with GRASS
Raster map algebra, topographic analysis, line
of sight, solar radiation, cost
surfaces, covariant analysis, buffers, Vector
digitizing, overlay Imagery processing of
multispectral data, image rectification,
principal component analysis, edge detection,
Sites spatial interpolation, voronoi polygons,
Projections, transformation between data
models, spatial interpolation, import/export Visua
lization 2D display with zoom and pan,
interactive 3D visualization with multiple
surfaces, vector and site data Link to other OS
projects Map Server, OSSIM, R-stats, GSTATS

6
Digital nearshore data

• USGS DEM or NED insufficient resolution
• Traditional surveying and photogrammetry
  points-gt TINs -gt contours (breaklines)
• RTK GPS profiles directional oversampling
  
• LIDAR sweeps (no breaklines, oversampled)
• Nearshore bathymetry LARC profiles (directional
  oversampling), sonar
• Fast, accurate and consistent transformation
  between the measured data and GIS data models are
  necessary grid, contours (vector), sites
  (points), at various levels of detail

USGS DEM RTK GPS LIDAR
SONAR
7
Challenges

• massive data sets (million points data
  sets)
• unlabeled, oversampled data (no defined
  breaklines)
• surfaces are complex and data are often
  noisy
• anisotropy or directional oversampling is
  common
• spatial coverage and accuracy may be
  heterogeneous
• both statistical and geometrical accuracy is
  needed
• almost all data are spatio-temporal

8
Relaxed Splines
Relaxed spline method (regularized spline with
tension - RST) was implemented in GRASS to
support spatial interpolation of multivariate
data.

• Properties
• 2D, 3D and 4D implementation,
• flexible properties through tension and
  smoothing,
• simultaneous computation of slope, aspect,
  curvatures,
• computation of deviations and cross-validation
  error,
• segmented processing for large data sets.

Examples
9
FRF Duck nearshore profiles LARC
Interpolation by RST with simultaneous
computation of topographic parameters slope
and profile curvature
Elevation
Profile curvature In normal plane in gradient
direction
Slope
10
Jockeys Ridge LIDAR methods
1m rasterization,
3m rasterization
RST interpolation 1m resolution grid all points
with distance gt 0.5m preserved
11
Jockeys Ridge curvature
Curvatures computed by RST at 3 levels of detail,
controlled by tension and smoothing, resolution
is 2m, red is convex blue is concave
12
Jockeys Ridge visualization
LIDAR-based DEM (2m resolution) IR DOQQ with
modified color, visualized in GRASS5 NVIZ
13
Bald Head Island Data
Renourished beach in 2001 Data LIDAR 1997-2000,
RTK GPS, NED
NED
RTK GPS
LIDAR98
Dec. 2001 Jan. 2002
LIDAR2000
14
Bald Head Island change 1998-2000
1998 and 2000 LIDAR 5m resolution rasterized
15
Bald Head slope and curvature
1999
Computed by RST at 2m resolution with high
tension parameter (high level of detail).
2000
legend
16
Bald Head slope and curvature
1999
2000
slope
Computed by RST with lower tension
profile curvature
17
Bald Head RTKS Anisotropic data
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 interpolation with default parameters
RST with anisotropic tension, 110 at 160 deg
anisotropy with incorrect angle (135 deg)
18
Bald Head Island change Dec.-Jan.
January 2002
December 2001
19
Bald Head Island moved volumes
Jan. 2002 - Dec. 2001
RTK GPS
total eroded 149,000m3 total gained 43,000m3
Aug. 2000 Fall 1999
LIDAR
total eroded 445,000m3 total gained 225,000m3
20
Simulation of processes
Path sampling method uses duality between
particle and field representation to solve the
governing equations It is mesh free so it is
easier to use with GIS than finite element or
finite difference methods. Process can be modeled
as evolution of particles or fields.
21
Overland flow sediment transport
sandhigh detachment and deposition rates, short
distance transport
clay low detachment and deposition, long
distance transport
22
Multiscale simulations
Density of walkers is adjusted to resolution and
is controlled by an importance sampling
functionW(r)
Entire area 10m res., 400x400 grid, Inset
2m res., 600x400 grid
23
Conclusions
RST method with simultaneous topographic analysis
was applied to several types of data used for
characterization of nearshore environment
LIDAR Jockey's Ridge was interpolated at 1m
resolution with analysis of surface geometry at
various levels of detail. Improvement of
performance for high density data points is being
implemented. The 3D map of Jockey's Ridge with
draped IR-DOQQ was created using GRASS GIS tool
NVIZ. LIDAR and RTK GPS measurements were
used to assess the change of the Bald Head Island
shoreline, including in geomorphologic
properties. Anisotropic tension was necessary for
processing of RTK GPS profile data. The analysis
has shown that after renourishment the pattern of
erosion and deposition remains the same, possibly
at higher rate. Technology Transfer
? Jockeys Ridge LIDAR is included in the book on
Open source GIS the GRASS GIS approach, to be
published in 2002. All improvements are tested
and immediately released with the development
version of GRASS GIS
24
What is next
New generation interpolation faster,
adaptable, with automatic optimization of
parameters Support for spatio-temporal data
beyond timestamp and animations Finish and
enhance the support for volume data General path
sampling simulation tool for GIS
25
GIS and the Future
Key challenge research, infrastructure and tools
for employment of geospatial data for proactive
protection of environment. Free access to
spatial environmental data and tools rapid
development of new technologies for environment
encouragement of environmentally responsible
behavior Integration of monitoring,
simulations and optimization sustainable
land use management, real-time response to
environmental disasters, prevention
unexpected environmental impacts
Images GMSLab University of Illinois at U-C W.
Brown, H. Mitasova Prepared by H.
Mitasova,