A Successful Pictometry Implementation Story - 2016Oct
interior_forest_analysis
1. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Interior Forest Analysis
Presenter: Tammy Kobliuk
Resource Analysis Section
Forest Management Branch
Public Lands and Forest Division
Alberta SRD
2. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Brief Overview
• Some background…
• Interior Forest Criteria
• Methodology: Vector or Raster?
• Input Data
• Case Study: Tolko High Level
• Seismic Lines – Do they have an impact?
• Technical Challenges and Things To Think About
3. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Interior Forest Analysis:
Background
4. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Initial Questions
• Can Interior Forest patches be identified?
• What criteria should be used to identify
Interior Forest patches?
• Can we develop a procedure to identify
Interior Forest patches?
• Should seismic lines be accounted for?
5. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Who?
• All FMA-holders may, in the future, be
required to run an Interior Forest Analysis as
part of their DFMP.
• The Crown will run the analysis on Crown
units for which it is undertaking a DFMP.
• The Interior Forest Analysis procedure was
developed by the Resource Analysis Section of
FMB, in consultation with the FMB ecologist.
6. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
What?
• INTERIOR forest refers to those areas of
forest which meet the habitat criteria for
interior forest species.
• Non-interior forest would be EDGE forest
and is generally undesirable for non-edge
species.
7. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
When?
• The Resource Analysis Section was
approached, in the summer of 2003, to
scope out the feasibility of identifying
“Interior” forest patches.
• Methodology was worked out in September
2003.
• Analysis for Tolko High Level was carried
out from September 2003 to January 2004.
8. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Why?
• To identify biodiversity issues.
• To identify conservation issues.
• To identify access issues and the general
impact of access and disturbance on the
landscape.
• The Planning Manual may require the
identification of Interior Forest patches in
the Biodiversity Annex.
9. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Interior Forest Criteria
10. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Attribute Criteria
• B, C, or D density forest. (> 30% crown
closure)
• Stand height > 2m.
• Not “Regen” or “Young” seral stage.
(Mature or Old Growth)
11. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Spatial Criteria
• 60m or more away from a non-forest edge.
• 30m or more away from a non-interior
forest edge (soft edge).
• At least 100 hectares in size.
12. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
General Methodology:
14. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Execution:
Raster versus Vector
• A fundamental choice that affects the
implementation of the analysis and the
interpretation of final landscape metrics.
• A choice that affects the creation of the
input data.
• A choice that will determine how much disk
space and processing time will be required.
15. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Why choose Vector?
• Vector analysis is technically more
accurate, but may be unwieldy on overly
large or complex areas.
• Vector analysis will likely take up less disk
space.
• There are fewer steps.
16. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Vector
Disadvantages
• Vector commands may take an inordinate amount
of time to run on a complex dataset.
• Vector outputs are more difficult to QC and audit
than raster outputs.
• ArcInfo is unreliable when datasets reach in the
area of 1 million polygons. Commands may run
incomplete without giving an error message.
• Software and/or hardware limits may be reached.
17. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Why Choose Raster?
• Raster analysis is more efficient on large
and/or complex areas.
• An appropriate cell size will enable efficient
and accurate analysis.
• Intermediate raster outputs are easy to QC.
18. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Raster Disadvantages
• Datasets are potentially very large – data
storage may become an issue.
• More steps than vector.
• May take more time than vector to run on a
smaller area.
• Raster will affect Edge-type landscape
metrics.
19. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Detailed
Raster
Process
Flow
20. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Input Data
• Vegetation data (ie. AVI) required
attributes:
– Density (crown closure)
– Height
– Seral Stage – a derived attribute
• Linear access polygons
• *** Essentially a simple net landbase ***
21. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Spatial Net Landbase
27. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Non-Forest Input Grid
28. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Soft Edge Input Layer
29. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Distance From Hard Edges
30. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Distance From Soft Edge
31. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Remove Areas Failing the Minimum
Distance Criteria From a Hard Edge
32. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Remove Areas Failing the Minimum
Distance Criteria From a Soft Edge
33. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Remove Soft Edge Effects to
Create a Candidate Patch Layer
34. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Calculate the Size of Each Patch
35. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Final Interior Forest Patches:
Remove Patches Failing the Minimum Size Criteria
36. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Case Study:
Interior Forest Analysis
• Study Area: single FMA 3.5 million Ha
• Source data: ArcInfo Library (~2.25
million polygons in 406 tiles)
• Attributes in separate .dbf file.
• Problem undefined.
• Solution undefined.
37. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Year 0 Timeline
• 2 weeks to nail down the exact problem and
formulate the solution(s). Solution(s) were
programmed for automation.
• 2 weeks to complete data prep and spatial
analysis.
• 1 day to compile final datasets.
• 5 minutes to run Patch Analyst for final
landscape metrics.
38. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Why a Raster/Vector hybrid?
• A single vector buffer command was taking
in excess of 15 hours to run with no
guarantee of completion.
• Raster seemed to be the most efficient
method.
• The intent was 100% raster – only one
section was run as vector because it
exceeded the raster limitations.
39. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Obstacles
• Had to create input datasets tile-by-tile since
the entire spatial net landbase could not be
extracted.
• Could not grid off the entire FMA at 5 m.
• Difficult to partition the landscape into
smaller pieces.
• Vector processes took an inordinate amount
of time.
• We seemed to hit every software and
hardware limit possible.
40. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Raster vs Vector
• Raster data sets were very large.
• Raster processes were comparable to vector
processes on simpler landscapes.
• Raster processes were faster than vector
processes on complex landscapes.
• Raster outputs were easier to check than
vector.
41. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Analysis Details
• Raster cell size: 5m
• Subdivided into 8 portions.
• 7 portions run as raster, 1 as vector.
• Source input – net landbase library
• Significant massaging required to create the
input datasets.
42. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
From Plan A to Plan D
• Plan A: Grid off the entire FMA and run
analysis.
• Plan B: Run entire FMA as vector
• Plan C: Partition the FMA and run pieces
as raster.
• Plan D: Run remaining partition as vector.
43. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Year 10 Analysis
• Input data must be “aged” by 10 years:
– Recalculate seral stage for Year 10.
– Incorporate planned harvest blocks.
– IF good height-age relationships are available, the forest
may also be grown to recruit stands that may move into
3m or greater height. In most areas, this is likely not
available.
– It is not recommended that crown closures be changed
(aged) – not enough information is available to properly
model crown closure changes over time.
44. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Final Disk Space Requirements
• 75GB:
– source data
– spatial net landbase library
– Year 0 analysis – includes many interim files
and both Seismic and No Seismic analyses.
– Year 10 analysis – includes Seismic analysis
only.
45. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Seismic Lines:
Do They Really Matter?
46. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Seismic
Thoughts
• If they are big enough to be mapped, they
are big enough to have an impact.
• Some areas are heavily impacted by linear
disturbance. Some are not.
• Failure to account for cutlines may result in
unrealistic assumptions and interpretations.
47. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
No accounting for seismic
48. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Accounting for seismic
49. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Final Results: No Seismic
50. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Final Results: Seismic
There are no interior forest patches in this
area once seismic lines have been accounted
for.
51. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Technical
Challenges
52. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Hybrid Raster/Vector Analysis
• The Raster and Vector analysis outputs
contain fields with differing definitions.
• Additional massaging of the outputs must
be done before a seamless final layer can be
created.
53. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Size of study areas…
• Extremely large areas of contiguous
industrial forest land.
• Massive size of individual tenures and
management units.
• Undergoing amalgamation of units.
• Multi-jurisdiction study areas.
54. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Boundary Effects
• Arbitrary administrative boundaries or
dataset edges may influence the final
results.
• In order to derive an accurate picture, a
buffer or data around the study area is
desireable to mitigate boundary effects.
55. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Extremely Large Geographic Areas
• Large areas may require subdivision for
processing.
• Subdivision must be based on real non-
forest (hard edge) features. (Example:
large rivers, highway rights-of-way).
Arbitrary lines are not recommended.
• Subdividing the study area may add
significantly to the effort required.
56. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
The Multi-Use Landbase
• There is a phenomenal amount of activity
modifying the landscape in some areas of
the province.
• Oil & gas, grazing, forestry and recreation
are all on some of the same areas.
• Datasets are out-of-date the moment they
are completed.
57. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Lots of Data Means…
• Disk space issues:
– Either way you slice it, this analysis will
require a lot of disk space. Backup issues.
• Archiving issues – what to keep and what to
blow away. Choose your intermediate data
sets carefully.
• File management issues.
58. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Large Datasets Mean…
• Streaming live data requires lots of
network bandwidth.
• Hardware, operating systems, and software
need to be optimally tuned.
• Unix systems may still out-perform PC’s.
• Data display needs to be managed.
• Data format needs to be carefully chosen.
59. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Size is Relative
• What’s large for you?
• What’s large for the
software developer?
• Are software limits
documented? Can you
find limits prior to running
a process?
60. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
So Your Command Ran…
So What?
• ArcInfo commands may run to completion
on large datasets, but:
– Did it really complete?
– Did it run correctly?
– How do you check it?
– Did it clean up after itself?
61. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Check Your Outputs
• Spatially view your data.
• Query out results.
• Do spot checks.
• Do topology checks.
62. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Software Limits
• ArcInfo 10,000 temporary file limit
• Grid byte limit
• “Too many arcs in a scan line”
63. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Hardware Limits
• Insufficient disk space for outputs.
• Insufficient disk space for temporary files.
• Byte limits.
• Possible insufficient swap space.
64. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Raster Cell Size
• Choose your cell size wisely.
• The cell size should be able to appropriately
model the smallest linear feature.
• The cell size must work with the minimum
distances. (ie. must divide evenly)
• The cell size should be optimized for
performance and accuracy.
65. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Choosing Your Cell Size
• Pick an area to test (ie. a mapsheet or
township) that contains features
representative of the greater area.
• Sample various cell sizes.
• Compare samples against vector data:
– Number of Patches
– Mean Patch Size
71. FMA GIS Meeting
April 15, 2004
A
Sustainable Resource Development
Conclusions
• This analysis is feasible and reasonable.
• Cutlines must be accounted for.
• The optimal cell size is likely to be 5 or 10m.
• Small areas should be straightforward. Large areas
make the analysis much more complicated.
• Both raster and vector methodologies are workable.
• Software and hardware limits can cause a lot of grief.