Robust sampling of natural resources using a GIS implementation of GRTS

1
Robust sampling of natural resources using a GIS
implementation of GRTS

• David Theobald
• Natural Resource Ecology Lab
• Dept of Recreation Tourism
• Colorado State University
• Fort Collins, CO 80523 USA
• 23 September 2004

2
Funding/Disclaimer

• The work reported here was developed under the
  STAR Research Assistance Agreement CR-829095
  awarded by the U.S. Environmental Protection
  Agency (EPA) to Colorado State University. This
  presentation has not been formally reviewed by
  EPA.  The views expressed here are solely those
  of the presenter and STARMAP, the Program (s)he
  represents. EPA does not endorse any products or
  commercial services mentioned in this
  presentation.

3
Practical sampling needs

• Most information for least cost
• Sample some areas with higher probability than
  others
• Some features are more important than others
• Higher uncertainty of knowing about particular
  situations
• Some locations are more difficult (time, ) to
  access than others
• Flexibility
• In-the-field decisions (e.g., access denied,
  extra time)
• Changes in funding ( or -) for current project
• Subsequent projects (additional funding) augment
  existing dataset (but often different study area)
• GRTS algorithm (Stevens 1997 Stevens and Olsen
  1999 Stevens and Olsen 2004)

4
Why GIS framework?

• Spatial data is needed to as input to describe
  population (frame)
• Spatial data used to describe strata, to describe
  inclusion probabilities, including continuous
  variables (e.g., terrain)
• Ability to sample point, line, and area-based
  ecological resources
• Flexibility in adjusting input to alter sampling
  design
• Visualize sampling design in relation to other
  geographic data (e.g., accessibility, ownership)
• Large, broad user base of GIS technology

5
Existing GIS-based sampling

• Sampling in ArcView v3, ArcGIS v8, v9
• Typically simple random sampling (e.g., random x,
  y constrained to polygon of study area)
• GStat (www.gstat.org) Pebesma and Wesseling.
  1998. Gstat, a program for geostatistical
  modeling, prediction and simulation. Computers
  and Geosciences 24(1)17-31.
• Traditional stratified, simple random sampling
• r.le tools for GRASS
• Stratified sampling

6
Ecological resource types

• Areas (e.g., lakes, land cover patches)
• Discrete represent as point shapefile, GRID
  with single cell
• Convert to centroid or labelpoint then to GRID
• Tesselate surface e.g., watersheds, 8-digit HUCs
• Discontinuous all lakes in Oregon
• Continuous represent as polygon, GRID as zones
• Patches of vegetation types
• Variation of water clarity within selected lakes
• Estuarine resources
• Area bias?
• Lines (e.g., streams, roads)
• Discrete represent as point shapefile, GRID
  with single cell
• Individual stream reaches
• 100 segments
• Continuous
• All possible locations on stream network
• Points (e.g., individual trees, lakes)
• Discrete
• all lakes in Oregon

7
Sample
8
Sample
9
Processing steps

• 1. Input
• raster or GRID of frame, inclusion probabilities
• get spatial extent, grain (resolution), study
  area (inside, outside, holes)
• 2. compute number of quad-levels, L
• 3. generate random permuted 1-4 labels at each L
• 4. add levels together to create reverse-ordered
  address
• 5. compute sequential list order
• 6. threshold against inclusion probabilities
• 7. convert raster to point shapefile

10
Level 1
3
1
2
4
11
Level 2
1
1
3
3
4
2
2
4
1
1
3
3
2
4
4
2
12
Level 3
13
Morton address to sequential list
14
Reverse-Morton address to list
15
Random permutation of quad values
16
Area frame for vegetation survey
17
Continuous listing of sequential points
18
Laramie Foothills vegetation survey
19
(No Transcript)
20
Summary

• Flexibility of input data point, line, area
• Continuous (gradients) and discrete (strata)
  inclusion probabilities
• Visualization of sample design
• Can modify inclusion probability based on
  accessibility constraints
• Develop map of inferred population
• ArcGIS tool

21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
Distribution plans

• Currently alpha test phase
• Beta testing January 2005
• Release Spring/Summer 2005

25

• Thanks!
• Comments? Questions?
• STARMAP www.stat.colostate.edu/nsu/starmap
• GIS-GRTS tools in ArcGIS email
  davet_at_nrel.colostate.edu

Funding/Disclaimer The work reported here was
developed under the STAR Research Assistance
Agreement CR-829095 awarded by the U.S.
Environmental Protection Agency (EPA) to Colorado
State University. This presentation has not been
formally reviewed by EPA.  The views expressed
here are solely those of the presenter and
STARMAP, the Program (s)he represents. EPA does
not endorse any products or commercial services
mentioned in this presentation.
Funding/Disclaimer The work reported here was
developed under the STAR Research Assistance
Agreement CR-829095 awarded by the U.S.
Environmental Protection Agency (EPA) to Colorado
State University. This presentation has not been
formally reviewed by EPA.  The views expressed
here are solely those of the presenter and
STARMAP, the Program (s)he represents. EPA does
not endorse any products or commercial services
mentioned in this presentation.
Funding/Disclaimer The work reported here was
developed under the STAR Research Assistance
Agreement CR-829095 awarded by the U.S.
Environmental Protection Agency (EPA) to Colorado
State University. This presentation has not been
formally reviewed by EPA.  The views expressed
here are solely those of the presenter and
STARMAP, the Program (s)he represents. EPA does
not endorse any products or commercial services
mentioned in this presentation.
CR - 829095
CR - 829095
CR - 829095
26
Stream frame
27
(No Transcript)