spatial analyst - identifying the best paths with cost ... · two step process for performing cost...
TRANSCRIPT
Spatial Analyst – Identifying the Best
Paths with Cost Distance AnalysisKevin M. Johnston
Elizabeth Graham
Cost distance analysis - Outline
• What is cost distance analysis
• Cost Analysis 1 - Cost Connectivity (and creating a cost surface) - Demo
• Cost Analysis 2 - Two step process – Cost Distance and Cost Path - Demo
• Source characteristics - Demo
• Directionality of the movement - Demo
• Incorporating barriers - Demo
What is cost distance analysis
• One of the most common applications in Spatial Analyst
• Euclidean distance – as the crow flies
• Cost distance – as the phenomenon moves across the landscape
• Euclidean is how far while cost analysis is a total amount
• Cost can be:
- Preference
- Energy expended
- Time
- Dollars
- Risk
Problems addressed by cost distance analysis
• Constructing a road to a proposed shopping center
• Conserving wildlife corridors between habitat patches
• Supplying and reinforcing military troops in a deployment
• Providing movement paths for fire fighters between posts
• Locating a pipeline to connect energy fields to a refinery
• Siting electrical lines
Why do patches need to be connected?
• Fragmentation Metapopulation
- Logging Roads
- Supply routes for military locations
- Fire fighting routes
Cost distance analysis - Outline
• What is cost distance analysis
• Cost Analysis 1 - Cost Connectivity (and creating a cost surface) - Demo
• Cost Analysis 2 - Two step process – Cost Distance and Cost Path - Demo
• Source characteristics - Demo
• Directionality of the movement - Demo
• Incorporating barriers - Demo
The input to cost distance analysis
• Define the problem (which includes creating submodels)
• Identify and derive the criteria
• Transform values to a common scale
• Weight the criteria relative to one another and combine.
• Analyze the results
How to create cost surface
• Similar to creating a suitability
model
• Cost per map unit to move through
the cell
• The lower the cost the better
• Diagonal accounted for
Cost analysis 1 – Cost Connectivity
• Input regions
• Input cost
• Create a network of paths
- Output optimum paths
- Optional neighboring paths
Cost Connectivity – Optimum network
C
A
B
Based on cost not Euclidean distance
Creating a cost surface
Cost Connectivity
Creating the
optimum network
Cost distance analysis - Outline
• What is cost distance analysis
• Cost Analysis 1 - Cost Connectivity (and creating a cost surface) - Demo
• Cost Analysis 2 - Two step process – Cost Distance and Cost Path - Demo
• Source characteristics - Demo
• Directionality of the movement - Demo
• Incorporating barriers - Demo
Two step process for performing cost distance analysis
• Cost Distance tool
- Input
- Sources – starting point
- Cost surface – cost per map unit for travel
- Output
- Cost distance – total accumulative least-cost for
each cell to reach a source
- Back link – direction to move from each cell to
reach a source
- Cost allocation – for each cell, which is the least-
cost source
• Cost Path tool
- Input
- Destination – ending point
- Cost distance and Back link output rasters from
Cost Distance tool
- Output
- Least-cost paths – the least-cost paths
Step 1: How to perform the cost distance analysisThe Cost Distance tool
• Sources
• Cost surface
• Output distance raster – for each cell, the
lowest total accumulative cost to reach a source
Step 1: How to perform the cost distance analysisThe Cost Distance tool (continued)
• Back link
• Cost allocation
Step 2: How to create the least-cost pathThe Cost Path as Polyline and Cost Path tools
• Destination and the cost distance and back link output from
the Cost Distance tool
• Creates the least cost path from the destination to the sources
- Best single path
- Each zone
- Each cell
Cost Distance
Cost Path as Polyline
Creating the least-
cost path
Cost distance analysis - Outline
• What is cost distance analysis
• Cost Analysis 1 - Cost Connectivity (and creating a cost surface) - Demo
• Cost Analysis 2 - Two step process – Cost Distance and Cost Path - Demo
• Source characteristics - Demo
• Directionality of the movement - Demo
• Incorporating barriers - Demo
Source characteristics
• Multiplier
- Different modes of travel from each source - an ATV versus walking
- Different magnitudes at each source – number of firefighters at each headquarter
• Start cost
- Time it takes to prepare before leaving the source
a1 = starting_cost + (((cost1 + cost2) / 2) * cost_multiplier)
• Resistance rate
- A hiker getting tired
• Capacity
- Identify potential locations for refueling stations for military tanks
accum_cost = (a1 * (1 + resistance_rate)) + ((((cost2 * HF(2)) + (cost3 * HF(3)))/2) *
Surface_distance(23) * VF(23) * cost_multiplier)
Source Characteristics
Controlling the
mover
Cost Connectivity – Neighbor paths and Network Analyst
Cost distance analysis - Outline
• What is cost distance analysis
• Cost Analysis 1 - Cost Connectivity (and creating a cost surface) - Demo
• Cost Analysis 2 - Two step process – Cost Distance and Cost Path - Demo
• Source characteristics - Demo
• Directionality of the movement - Demo
• Incorporating barriers - Demo
Adding complexity Adding surface distance with the Path Distance tool
• Actual distance traveled
• Endure the cost longer because going uphill
or downhill
• Surface raster
• The Path Distance tool
Adding complexityAdding directionality with the Path Distance tool
• Adjustment to overcome going uphill and
downhill
• Vertical factor
- Surface raster endure cost longer
- Vertical factor the additional cost to over come the
slope
• Horizontal factor
- Additional cost to overcome a horizontal factor such
as wind
Accum_cost_distance = a1 + (((Cost_Surface(b) * Horizontal_factor(b)) +
(Cost_surface(c) * Horizontal_factor(c)))/2) * Surface_distance(bc) * Vertical_factor(bc)
Directionality of the movement
Controlling the
mover
Cost distance analysis - Outline
• What is cost distance analysis
• Cost Analysis 1 - Cost Connectivity (and creating a cost surface) - Demo
• Cost Analysis 2 - Two step process – Cost Distance and Cost Path - Demo
• Source characteristics - Demo
• Directionality of the movement - Demo
• Incorporating barriers - Demo
Incorporating barriers in Cost Distance
• Locations that are barriers are set to
NoData in the cost surface through the
Environment mask
• Cost Distance and Backlink are
calculated around the barrier locations
• Cost Path and Cost Path As Polyline
are determined from the Backlink
therefore the paths do not pass
through the barriers
Incorporating barriers in Euclidean Distance (new capability)
• Barriers are specified by an
input dataset
• Diagonal cracks are resolved
• Distance and Direction rasters
created (0 – 360 not 1 – 8)
• Paths from back direction raster
• Differs from Cost Distance with
constant cost surface
Euclidean Distance
Incorporating
barriers
Additional capability
• Distance tools are parallelized and
distributed
• Raster Analytic tools have been created for
the distance tools to work on large data in
the server environment
• Euclidean distance supports both planar and
geodesic calculations
The road ahead – the near future
• Break from the constraints of processing cell
centers in a lattice
- “Raster is faster but vector more accurate” no
longer true – implement level set methods for
evolving boundaries (costs, fire, water)
• Resulting cost rasters will not have octagon
artifact and back direction 0 – 360 (not 1 – 8)
• Add barriers, address cracks, and implement
geodesic to cost analysis
• Simplify distance analysis
- Distance can now be a matter of whether you
have cost surface or barriers
Conclusion
• Defining the cost units can be difficult
• Can create the optimum network of paths – Cost Connectivity
- Compatible with Network Analyst
• Two step process – Cost Distance and Cost Path
- When source and destination known
• Source Characteristics
- Difference modes of travel from a source
- Starting costs
- Dynamic adjustment – getting tired
• Directionality – Path Distance
• Barriers can be incorporated in Cost and Euclidean Distance
• Creates the optimum least-cost solution
- Assumes memory
- Has visited all locations
Acknowledgements: The Vermont Center for Geographic Information and British Columbia for the use of their data in this presentation
Additional resource
• Cost distance analysis case study
http://desktop.arcgis.com/en/analytics/case-studies/understanding-cost-distance-
analysis.htm
• Suitability modeling case study
http://desktop.arcgis.com/en/analytics/case-studies/understanding-the-suitability-
modeling-workflow.htm
• Spatial Analyst Resources
http://esriurl.com/spatialAnalystResources
• We need your feedback, please take the Raster survey
https://www.esri.com/arcgis-blog/products/arcgis-pro/analytics/arcgis-spatial-analyst-
survey/
Please Share Your Feedback in the App
Download the Esri
Events app and find
your event
Select the session
you attended
Scroll down to
“Survey”
Log in to access the
survey
Complete the survey
and select “Submit”