Spatial Information in Natural Resources
FANR 3800 Raster Analysis Raster Analysis
Objectives
• Review the raster data model
• Understand how raster analysis
f d t ll diff f t l i
fundamentally differs from vector analysis
• Become familiar with basic tools of raster analysis
Raster Data – Discrete Classes
Each cell has one and only one value. The values are coded to land cover, road types, etc.
Example:
Name Color Value Name Color Value slash pine 10 loblolly pine 20
water 70
hardwood 30
pasture 40
Vector vs.
Raster Revisited
• Vector format is good for storing datasets with many attributes
– For example – stand map with species composition, age class, stand density, etc.
Vector vs.
Raster Revisited
Value = the attribute (yr estab)
– Multiple attributes require multiple grids
(y _ )
Count = the number of pixels with that value (yr_estab)
Vector to Raster
• What if we wanted to do analysis using:
– Stand age (“yr_estab”) – Stand type
– Stand volume per acre (“cuft”)
Vector vs. Raster Revisited
• So, if raster format is not so good for datasets with multiple attributes, Why use rasters?
1. Raster format facilitates many spatial analyses 2 Raster format is much better at representing 2. Raster format is much better at representing
attributes that vary continuously across the landscape
• elevation
• distance to streams / roads
• precipitation
Raster Data Analysis
• Single-Layer Analysis
– Distance, Direction, and Allocation – Density
– Neighborhood Analysis Reclassification – Reclassification
• Multiple-Layer Analysis – Map Calculator (Map algebra) – Zonal Summaries
– Masking (raster “Clipping”)
Raster Data Analysis
• Setting up your analysis
– Working directory– Analysis extent and “snap grid”
– Masking (raster “Clipping”)
• Single-Layer Analysis Single Layer Analysis
– Reclassification– Distance, Direction, and Allocation – Density
– Neighborhood Analysis
• Multiple-Layer Analysis
– Map Calculator (Map algebra) – Zonal SummariesSetting grid analysis properties
• Analysis properties
– Working directory – Masking – Spatial extent ofoutput – Cell size
• Once set, analysis property values stay set until changed
Analysis properties determine spatial properties for all newly created output grid layers
Raster Data Analysis:
“Clipping”: Raster Extent and Mask
1. Setting the extent of analysis 2. Masking (raster “Clipping”)
Raster Data Analysis:
“Clipping”: Raster Extent and Mask
1) Extent set to Bfg_stands
2) Mask set to 100m buffer of roads
The results:
3) Raster Calculator: Output Grid = Input Grid
Single Layer analysis: Distance
• From any point on the landscape, how far is it to the nearest…
– Stream?
R d?
– Road?
– Landing?
– Town?
– Sample Plot?
Distance Analysis Example
Spatial Analyst:
Distance: Straight Line Vector or Raster Streams Layer
Distance from source
Single-Layer Analysis: Density
• Ex: Density of Roads
– What if you wanted to know what the density of roads are “across the landscape” (for every location in the landscape) p )
– Why might we want to know this?
• Wildlife applications?
• Forestry applications?
• Fisheries / Water quality applications?
Single Layer Analysis: Density
Ex: Density of Roads
Roads
Road Density
Single-Layer Analysis:
Neighborhoods
• Neighborhoods or “Moving Windows”
– What if you need to know something about the area around a point (raster cell), not just what’s in that cell?
– Example: mean elevation vs. point elevation
How Neighborhood Analysis Works
• “Neighborhood” is defined by a shape (circle, rectangle, etc.) and a size (3x3, 5x5, etc.)
• Functions computed for cell at the center of the neighborhood include sum, mean, standard deviation, minimum, maximum, majority, etc.
St 2 R lt SUM
1 2 4 2 6 9 3 1 2 4 7 8 4 2 7 1 1 2 4 2 6 9 3 1
Step 1
1 2 4 2 6 9 3 1 2 4 7 8 4 2 7 1 1 2 4 2 6 9 3 1 Step 2
1 2 4 2 6 9 3 1 2 4 7 8 4 2 7 1 1 2 4 2 6 9 3 1
Step 3
- 43 34 - - 38 40 - - 44 42 - - - - - - - - - - 32 29 - Results: SUM
=
Neighborhood example 1
• CUFT – continuous attribute, mean in two different moving windows
3 x 3 cells
15 x 15 cells
Single-Layer Analysis:
Reclassification
• Convert One Classification Scheme…
– 1 = Hardwood – 2 = Natural Pine – 3 = Planted Pine – 4 = Clearcut – 5 = Other
• To Another – 1 = Mature Pine – 2 = Other Forest – 0 = Other
Raster Data Analysis
• Single-Layer Analysis
– Distance, Direction, and Allocation – Density
– Neighborhood Analysis g y – Reclassification
• Multiple-Layer Analysis – Masking (raster “Clipping”)
– Raster Calculator (Raster or “Map” algebra)
– Zonal Summaries
Multi-layer Raster Analysis
• Raster calculator:
– Can be used to combine many grids using mathematical functions and Boolean logic
Multi-layer Raster Analysis
• Zonal Analysis:
– Similar to Vector Overlay analysis, except zones are defined by all pixels with the same value
Z R t V l R t
Zone Raster: Value Raster:
2 2 5 2 3 3 1 1 5 5 2 2 5 5 2 1 5 5 3 2 2 2 5 1
1 2 4 2 6 9 3 1 2 4 7 8 4 2 7 1 1 2 4 2 6 9 3 1
Multi-layer Raster Analysis
• Zonal Analysis:
– Similar to Vector Overlay analysis, except zones are defined by all pixels with the same value
Z R t V l R t O t t (Z l S )
Zone Raster: Value Raster: Output (Zonal Sum):
2 2 5 2 3 3 1 1 5 5 2 2 5 5 2 1 5 5 3 2 2 2 5 1
1 2 4 2 6 9 3 1 2 4 7 8 4 2 7 1 1 2 4 2 6 9 3 1
46 46 22 46 19 19 6 6 22 22 46 46 22 22 46 6 22 22 19 46 46 46 22 6
Multi-layer Raster Analysis
• Zonal Analysis Outputs :
Multi-layer Raster Analysis
• Applications of Zonal Analysis :
– Zone = Stands, Values = Soil Productivity – Zone = Stands, Values = Habitat Suitability – Zone = Ownership, Values = Stand type
– Zone = Ownership, Values = Species Richness
Raster Analysis
• Most wildlife habitat assessment is done using raster data and raster analysis functions
• Why?
– Data formats – land cover, elevation, climate variables are often in raster format
– Efficiency – faster to run using raster – Spatial analysis – lots of techniques and
options
Raster Analysis: Example 1
• Wildlife Habitat Analysis:
– Assemble the variables that are correlated with species presence or abundance
– Use field data (known) locations to “sample”
each layer to develop statistical models
Variable of interest measured for a limited number of isolated plots or individuals
Other, possibly correlated variables are available as spatial datasets
Develop a rule-based or mathematical model
Sample Plot Variable = f(Spatial Variables) Use the model to make predictions for plots not sampled
Example: Tree species richness predicted as a function of Landsat TM vegetation indices, climate variables, and land indices, climate variables, and land ownership
Sites used to sample environmental variables, exported to statistical software package
Once the statistical models have been developed, the equations can be applied back in GIS using Map Algebra and the Raster Calculator
Raster Analysis Example 2:
GAP Analysis
• Objective – Identify conservation “Gaps” at a national level
• Projects in all 50 states Projects in all 50 states
• Cooperative effort b/w USGS, other federal agencies, state agencies, universities, non-profits
GAP Analysis
Land Cover Species Records
Habitat
Land Ownership Management
Conservation Status Status
GAP Analysis
- How much habitat is there?
- Where is the habitat?
- How much of the habitat is protected from development, logging, etc?
Habitat Model:
-Created new grid from suitable h bit t i l di
habitats including:
-open water (fresh) -clearcuts -hardwood forests -loblolly/slash pine -cypress-gum swamp -freshwater marsh -evergreen forested wetland -Applied mask of suitable habitat
>530 ha
-Clipped by digitized range