geog537_2002_metrics

July 2002
Measuring Landscape Metrics
Using GIS Data
Geog 537

July 2002
Highlights
• Raster vs vector
• Defining your patch types
• Defining your study area
• GIS inventory data versus imagery
• Patch Analyst
• Precautions

July 2002
Raster versus Vector
• What raster cell size would you use?
• How does selection of raster cell size affect
metrics measurements?
• How does this affect edge metrics?
Contagion? Number of patches?
• Does your raster adequately represent the
feature detail contained in the vector data?
• What processing advantages/disadvantages
does your choice give you.

July 2002
AVI on Cover Group Raster (500m)

July 2002
Raster Resolution
500m
200m
100m
20m

July 2002
Comparison: Number of patches
0
5
10
15
20
25
30
35
C CD DC D
raster 500m
raster 200m
raster 100m
raster 20m
raster 10m
raster 5m
vector

July 2002
Comparison: Class Area
0
20
40
60
80
100
120
140
160
CD DC D
Raster 500m
Raster 200m
Raster 100m
Raster 20m
Raster 10m
Raster 5m
Vector

July 2002
Comparison: Mean Patch Size
0
5
10
15
20
25
30
35
40
45
CD DC D
Raster 500m
Raster 200m
Raster 100m
Raster 20m
Raster 10m
Raster 5m
Vector

July 2002
Vector Resolution: defining patches
AVI polygons AVI dissolved into cover group

July 2002
Processing Power
• Patch Analyst grid uses Fragstats.
• Fragstats has a maximum number of raster
cells it can process.
• High raster resolution can result in a smaller
area that can be run.
• Vector data sets may be more complicated to
create, but may enable a run on a larger area.
• Generally vector operations are more CPU
intensive

July 2002
Defining your patch types
• First, define your problem.
• Then define what you need to measure.
• One patch type or multiple patch types?
• How to generate your required patch type
from the data you have...

July 2002
Forested vs Non-Forested
Stand Origin
Patch Type
Examples

July 2002
Dominant Forest Cover Type
Stand Density
Patch Type
Examples

July 2002
Anthropogenic Disturbance
Forest Type Age Class
Patch Type
Examples

July 2002
Now you know what you want...
• Examine your GIS data for:
– currency
– completeness (both spatial and attribute)
– grain size
– accuracy (both spatial and attribute)
– relevance
…how do you get there?

July 2002
AVI on 1m Orthophoto

July 2002
Defining Patches in Your Data
• Are you using raw GIS polygons?
• Are you using readily available attributes?
• Do you need to “roll up” your data?
[Documenting of the rules is criticial].
• Do you need to incorporate additional
information to create your defined patches?

July 2002
Creating Your Patch Layer: Steps
• Start with your original data set.
• Add missing features and/or updates.
• Create a field identifying patches.
• Dissolve out on patch-type field.
• Rasterize if desired.
• Calculate metrics.

July 2002
Defining Your Study Area
• Use natural boundaries where possible.
• Choose an appropriate size for what you are
studying.
• Choose an appropriate size for your data
resolution.

July 2002
Stratifying Your Study Area
• Don’t
OR
• Stratify one or more ways in order to
understand differences across a landscape.
• Understand how your stratification can
affect your metrics.

July 2002
Landscape Management Units
Study Area Stratification Examples
Landscape
Compartments
Watershed Sub-Basins

July 2002
Example: Anthropogenic Disturbance
4.2 1
8.1 4
1.1 6
12. 09
0.7 0
29. 10
1.3 8
3.0 3
49. 97
0.8 0
13. 36
1.3 3
1.2 6
8.3 6
0.5 5
29. 10
6.2 3
0.5 5
2.1 0
0.5 5
0.5 5
23. 90
13. 03
13. 03
23. 90
0.6 6
Landscape Management
Unit
5.5 9
3.8 3
2.0 9
2.6 7
1.4 8
10. 19
9.1 0
49. 97
0.8 0
0.4 1
0.8 2
11. 98
2.5 4
0.4 2
0.6 7
1.3 3
16. 38
9.5 6
8.2 6
1.6 7
53. 60
13. 48
0.0 4
1.4 7
8.2 8
15. 61
45. 90
0.7 8
0.0 0
15. 44
8.6 8
0.9 2
0.4 9
0.0 1
4.9 0
20. 87
0.1 3
0.7 6
20. 87
6.4 6
21. 10
9.2 8
2.1 4
0.5 7
1.2 0
2.4 2
0.2 5
23. 90
32. 09
0.2 5
24. 84
15. 44
2.4 4
13. 03
0.8 8
13. 03
2.8 3
23. 90
9.2 8
Landscape Compartment
5.7 1
2.1 9
5.9 5
18. 20
30. 10
5.4 0
2.1 6
7.1 9
1.5 7
10. 19
23. 11
4.6 7
0.5 5
15. 59
17. 41
0.8 5
0.5 7
0.2 5
0.0 0
5.4 0
Watershed Sub-Basin

July 2002
Example: Linear Disturbance Density
1.33
1.5 6
1.4 6
1.7 0
0.9 9
1.1 4
2.5 6
1.6 6
0.7 4
1.8 2
1.6 0
0.3 7
1.7 2
2.0 7
3.4 0
1.7 1
2.1 6
0.6 7
4.8 5
0.2 5
2.0 6
0.2 7
2.5 61.8 2
0.2 7
0.2 7
0.2 7
0.2 7
Landscape Management Unit
1.33
1.5 5
2.1 6
1.1 0
1.7 1
1.4 2
1.8 2
2.0 7
0.7 6
1.5 6
0.6 5
1.7 3
1.8 3
1.3 1
2.0 2
1.1 3
2.1 2
1.6 0
2.1 6
0.9 4
0.9 2
1.6 0
0.9 6
0.9 7
0.7 4
0.9 3
0.5 2
3.0 5
3.6 5
2.0 9
0.3 0
2.5 7
0.6 7
4.8 5
1.9 6
2.0 0
0.8 0
2.2 1
2.3 7
3.5 0
1.9 3
0.7 3
1.3 9
0.3 1
1.5 1
1.2 9
2.0 7
0.0 6
2.0 7
0.1 2
2.7 6
2.5 5
1.3 2
3.1 3
0.5 1
3.7 0
2.7 6
0.4 1
0.1 1
1.1 8
0.0 3
0.2 9
2.9 1
0.1 3
2.9 9
0.1 3
2.1 7
0.1 3
0.5 2
Compartment
1.12
1.61
1.72
1.06
0.89
1.35
1.33
0.90
1.74
1.77
1.64
1.04
1.54
1.95
1.62
2.10
2.01
2.10
1.77
1.90
0.80
1.91
Watershed Sub-Basin

July 2002
Example: Linear Disturbance Density
by Watershed Sub-Basin
1.12
1.61
1.72
1.06
0.89
1.35
1.33
0.90
1.74
1.77
1.64
1.04
1.54
1.95
1.62
2.10
2.01
2.10
1.77
1.90
0.80
1.91
Entire data set
1.33
0.7 9
0.9 0
1.0 4
1.6 4
1.9 8
1.6 3
1.5 1
1.7 7
1.6 3
1.1 8
1.7 4
1.0 0
0.9 9
1.7 7
1.0 8
2.0 4
1.9 6
1.5 1
1.9 6
0.9 3
1.0 0
1.9 6
1.0 5
Clipped data set

July 2002
GIS Data vs Imagery
• What is controlling your grain size?
• What do you know about your data?
• What is the accuracy of your data?
• Can you minimize the effects of bias?
• Can you do change-over-time analysis?

July 2002
Example: Fire Dominated Landscape

July 2002
Fire History Sample
L3J

July 2002
Patch Analyst
• ArcView 3.x freeware extension
• Has both raster and vector modules
• Requires Spatial Analyst extension for the
raster module.
• The raster module actually runs Fragstats
• There is a maximum number of raster cells
that a data set can contain.

July 2002
Patch Analyst (cont.)
• Contains pre-processing tools
• Allows setting of threshold values
• Creates a virtual table.
• Results table must be exported in order to
save externally.
• Allows table export directly to Excel.

July 2002
Know your ecology
• Have you defined relevant patch types?
• Are you looking at an ecologically relevant
scale?
• Is your data of appropriate resolution (both
spatial and attribute-wise)?
• What are the natural processes at work?
• Is your study area disturbed or undisturbed?

July 2002
Precautions:
• Use the appropriate metrics for what you
are measuring or concerned about.
• Is the change in the metric(s) greater than
the ‘noise’ of the data?
• Are you using the appropriate study area
size? (ie. No less than 300-500 sq.km for
Landsat, but not so large that noise obscures
important changes)

July 2002
Precautions (cont.)
• Is your study area defined by natural breaks
in the landscape or is it an arbitrary straight
line (ie. ATS). Are you skewing results by
including ‘border’ polygons?
• If you are doing time change analysis, are
your data sets completely consistent in grain
size, capture criteria, etc.? Are they apples
and apples or apples and oranges?

July 2002
Interpreting the metrics...
• Running Fragstats or Patch Analyst is the
easy part.
• Interpreting the calculated metrics is the
hard part. What does it all mean?

July 2002
Murphy’s Law
• Your project area will always fall outside
the bounds of a single data set extent.
• Your data will never be “ready-to-go”.
• You will have areas where no data is
available.
• You will have to harmonize different data
sets/formats/types into a single layer.

July 2002
House River Fire (2002)
LWF-031-2002
L3J
L3J
L4J
L1J
L5J
L5
L6JA14J
A14
A14J

July 2002
Always Remember:
Your project will be 80% data
preparation, 20 % analysis!

geog537_2002_metrics

More Related Content

Viewers also liked

Similar to geog537_2002_metrics

More from Tammy Kobliuk

geog537_2002_metrics