Geospatial analysis and modeling with open source GIS: education and research

1
Geospatial analysis and modeling with open source
GISeducation and research

• Helena Mitasova
• North Carolina State University

2
Outline

• NCSU GIST program and open source GIS
• GS Modeling and Analysis course
• Research
• NC Floodplain mapping topobathy
• lidar data time series analysis
• Tangible GIS

3
NCSU GIScience technology

• New MS program in GIST
• planned start in Fall 2009
• distance ed sections
• Updated certificate and minor
• Enhanced courses include
• modeling and analysis
• open source GIS
• python programming

4
Teaching with open source GIS

• Geospatial Modeling and Analysis course,
  assignments in GRASS GIS and ArcGIS
• Data integration, display and 3D visualization
• Proximity analysis, cost surfaces
• Terrain modeling, analysis, geomorphometry
• Flow tracing, watershed analysis, landforms
• Modeling geospatial processes, Tangible GIS
• view implementation in source code

5
Cumulative cost surfaces

• Accident is reported on I-440 find the fire
  stations from which the help can come the fastest
  and find the least cost path

Cumulative cost surface
Deriving a cost map from speed limit
6
Cost surface from friction map

• Map showing the time it will take a walking
  person to get to each grid cell. Orange isochrone
  delineates search area for given time. Cost is
  based on topography, land cover and streets

7
Viewshed analysis
From Royal Bank of Canada headquarters
From RedHat headquarters
8
Solar radiation
Analyze spatial and temporal pattern of solar
radiation
Solar incidence angle at 2pm, building with
cast shadow
Cumulative direct beam solar radiation
summer solstice 7800-8200 W/m2
winter solstice 0-5700 W/m2
winter solstice
9
Flow tracing, watershed analysis

• Flow accumulation SFD D8 Dinf
  MFD D8

Watershed and stream networks extraction
10
Terrain modeling lidar

• Multiple return lidar point cloud data 2001 NC
  Flood mapping

first return second or last return
surface with vegetation and structures
bare ground
11
Increasing LIDAR point density
no. of points/2m grid cell 1996 0.2 1997
0.9 1998 0.4 1999 1.4 2001 0.2 NCflood 2003
2.0 2004 15.0 2005 6.0 2007 3.0 2008 3.0
substantially improved representation of
structures but much larger data sets
2004 lidar, 0.5m resolution DEM binned and
computed by RST (smoothes out the noise and
fills in the gaps)?
1m res. DEM, computed by RST, 1998 lidar data
12
USACE SHOALS LIDAR topo mapping
13
NC Floodplain Mapping topobathy
Data SourcesBeach ProfilesWrightsville Beach,
Topsail, Topsail Island, Pea Island, Ocean
Isle-Shallotte-Holden Beach, Oak Island-Bald
Head, FRF Duck, Fort Fisher, Figure 8 Island,
Dare County, Carolina Kure, Bear Island,
Bogue-Shackleford BanksInletsTopsail,
Shallotte, Rich-Nixon Green Channel, Oregon, New
River, Mason, Masonboro, Lockwoods Folly, Cape
Fear, Bogue, Beaufort, BardenNational Ocean
Service soundings
Lisa Stillwell, RENCI
14
NC Floodplain Mapping topobathy
Coastal Relief Model, NC Flood mapping DEM and
channels
Cape Fear
It was necessary to use the original data and
interpolate new topobathy DEM where possible
Topsail
New River
15
Topobathy 2.3 and 2008 lidar
improved interpolation removes the nearshore
bulge
Topobathy 2001 NCFM lidar 2005 USACE lidar NOS
soundings profiles
16
Topobathy 2.3 and 2008 lidar
improved interpolation removes the nearshore
bulge
Topobathy 2001 NCFM lidar 2005 USACE lidar NOS
soundings profiles
beach and vegetation requires higher resolution
and first return data
17
Topobathy 2.3 and 2008 lidar
elevation loss gain
Elevation difference on beach 1-2m scarp is not
captured in 10m topobathy (photo - spring 2005)?

• differences are due to
• lower resolution
• actual change between 2001-2005-2008
• bare grounds versus surface with vegetation
• and structures

18
NC Floodplain Mapping topobathy
DDigital Elevation Modelrelease 2.3provides
input for ADCIRC mesh which is manually modified

• 10 meter resolution
• 54,000 x 54,000 cells
• 2,916,000,000 total cells

• main issues
• bathy in sounds and channels
• updates in most dynamic locations

More info at http//tornado.renci.org/renci_ncfmp
/
Lisa Stillwell, RENCI
19
Measuring terrain dynamics

• Workflow for multitemporal lidar data processing
• Data integration coordinate system
  transformation
• Point density and noise analysis selection of
  common resolution and gridding method using
  binning (per cell statistics) no. of points per
  cell, z-range within cell, mean z
• Simultaneous spatial approximation, smoothing of
  random noise and computation of topographic
  parameters using splines
• Detection of systematic error and its
  elimination for all DEMs using roads and NCDOT
  benchmarks, evaluation of interpolation accuracy
• Result is time series of corrected, smoothed,
  high resolution DEMs
• 1996-2008 over 10 surveys for OBX

20
Measuring terrain dynamics

• Two approaches
• measuring migration and change of features, such
  as shorelines, ridges, crests, peaks (Jockeys
  Ridge)?
• raster-based time series analysis per cell
  statistics where each grid cell in the resulting
  map is function of grid cells in the entire time
  series in the same location
• challenging task masking data to the same
  coverage which varies due to dynamics and
  differences in survey extents

21
Spatial coastal change indicators

• New, spatial indicators for coastal terrain
  evolution based on per grid cell statistics from
  DEM time series tk, k1, , m
• core surface below which elevation never
  decreased
• zcore(i,j) min z(i,j,tk)?
• k
• outer envelope above which elevation never
  increased
• zenv(i,j) max z(i,j,tk)?
• k
• standard deviation map shows areas with most
  elevation change in red
• Mitasova, Overton, Recalde, Bernstein, Freeman,
  2009, JCR 25(2)?
• Wegmann and Clements, 2004, GRASS Newsletter

22
Spatial and temporal indicators
a) time at minimum and b) time at maximum
maps represent timeyear when the grid cell was
at its minimum and its maximum elevation c)
regression slope maps show spatial pattern of
elevation trends, inset transparency added as
function of correlation coefficient, white areas
have r2lt0.3
increase decrease
23
Coastal change indicators structures
Beach and dunes near Oregon Inlet

• In areas with homes core surface and outer
  envelope can be used to map new and old homes and
  their relation to core
• Some new homes built with exception from
  regulations do not have any core surface
• Derived from 2m res. DEMs 1996-2008

Beach and homes in Rodanthe
0 100m
24
Coastal change indicators structures
Beach and homes in Rodanthe

• core surface
• outer envelope
• 0m elevation
• derived from 0.5m resolution DEMs

old home, no core
old home, thin core
0 100m
new or lost home no core
25
Coastal change indicators structures
2008 DEM
26
Coastal change indicators structures
new or lost homes core-envelope gt 10m
2008 DEM
Time period 1997 - 2008
27
Coastal change indicators structures
vulnerable homes core-envelope gt 10m and core lt 1m
new or lost homes core-envelope gt 10m
2008 DEM
Time period 1997 - 2008
lost homes
28
Coastal change indicators structures
less vulnerable homes core-envelope gt 10m and
core gt 2m
vulnerable homes core-envelope gt 10m and core lt 1m
new or lost homes core-envelope gt 10m
2008 DEM
Time period 1997 - 2008
lost homes
29
Surface evolution as volume
New approach Evolution of terrain surface is
represented as a volume with time used for 3rd
dimension z f (x,y,t)? Evolution of a
contour is then represented as an
isosurface. The approach reveals often neglected
high dynamics of foredunes (zgt 4m)? and stability
of backshore beach (z1.5m)?
30
Surface evolution as volume
31
GIS and spatial simulations
Explore answers to questions What happens with
water, ecosystems, if - we change land use? -
sea level is 1m higher? - area is flooded or
trees are cut? Create new landscape
configurations in 2D using map algebra or
digitizing If the change is 3D and we need to
explore many scenarios, digitizing becomes
tedious and collaboration is limited
touch systems work in 2D
can we make changes in 3D ?
32
Building TanGIS at the VISSTA lab
3D scanners projectors 3D display
workstations web cameras flexible models
System is linked to GIS GRASS, ArcGIS - both
can be used simultaneously Multipurpose facility
at VISSTA Lab at ECE NCSU Prof. Hamid Krim
33
3D landscape design with TanGIS
3D laser scanner projector
flexible model with projected orthophoto
Mitasova, H., Mitas, Ratti, Ishii, Alonso,
Harmon, 2006, Real-time Human Interaction With
Landscape Models Using a Tangible Geospatial
Modeling Environment, IEEE CGApp, 26(4).
34
3D landscape design with TanGIS
Modify model and scan it
3D laser scanner projector
flexible model with projected orthophoto
Compute DEM, run flow simulation, project the
results (img or animation)?
Mitasova, H., Mitas, Ratti, Ishii, Alonso,
Harmon, 2006, Real-time Human Interaction With
Landscape Models Using a Tangible Geospatial
Modeling Environment, IEEE CGApp, 26(4).
35
Case study experimental watershed
Problems sediment deposition, road
flooding Design new land management alternatives
in 3D space and evaluate their impacts
N
1993 photogrammetric DEM
Sediment pollution
N
N
Flooding
0 200m
2001 lidar-based DEM
36
Impact of landscape modification
N
initial terrain
flexible clay model
37
Impact of landscape modification
initial terrain road breaks

flexible clay model take out piece of
clay
38
Impact of landscape modification
initial terrain road breaks
checkdam is added
flexible clay model take out piece of
clay add piece of clay
39
Impact of landscape modification
initial terrain road breaks
checkdam is added
flexible clay model take out piece of
clay add piece of clay
40
Impact of landscape modification

• Modify landscape
• add buildings, ponds, dams, roads
• change land cover properties
• Compute and project
• elevation or volume change,
• slope and aspect
• viewshed, line of sight
• flow accumulation and watershed boundaries
• soil erosion and deposition,
• solar energy potential

41
Terrain with buildings in TanGIS
elevation change
Add buildings
42
Terrain with buildings in TanGIS
elevation change
Add buildings
Runoff from buildings, compacted surface
43
Terrain with buildings in TanGIS
elevation change
Add buildings
Runoff from buildings only
Runoff from buildings, compacted surface
44
Getting creative in TanGIS
Exploring various materials, testing runoff
simulations on surfaces with depressions and
various patterns of roughness
lagoon
porous parking lot
Design by USFWSUSGSNCDENR team
45
Getting creative in TanGIS
Exploring various materials, testing runoff
simulations on surfaces with depressions and
various patterns of roughness
lagoon
summer solstice
porous parking lot

• the model does not crash in spite of all the
  pits and flats
• more rainfall is needed to flood the road

Design by USFWSUSGSNCDENR team
46
Teaching with Tangible GIS

• Tangible GIS in class
• experiment with laser scanning of different
  materials
• process point clouds of features with different
  geometries and surface properties
• test algorithms for analysis and simulations
• explore and demonstrate spatial impacts of
  landscape change on runoff, erosion, solar
  irradiation, ...

47
Open source GIS GRASS6.4 http//grass.osgeo.org
General purpose GIS 2D/3D raster and vector
data management, analysis, modeling,
visualization for Linux, Mac and
MSWindows Developed by US Army CERL
1982-1993 GPL since 1999, current development
coordinated from Trento, Italy, official OSGeo
project GRASS64RC3 available new wxPython GUI,
native MS Windows support, new and enhanced
modules
48
Acknowledgment

Funding by the US Army Research Office NC Sea
Grant NC Floodplain mapping / RENCI NCSU DELTA NC
WRRI is gratefully acknowledged