Nearest Neighbor Conference - PowerPoint PPT Presentation

1 / 55
About This Presentation
Title:

Nearest Neighbor Conference

Description:

Most Similar Neighbor (MSN) ... The MSN application is used to impute available ground-based inventory data to ... the user whether the MSN run should be used ... – PowerPoint PPT presentation

Number of Views:94
Avg rating:3.0/5.0
Slides: 56
Provided by: erictwombl
Category:

less

Transcript and Presenter's Notes

Title: Nearest Neighbor Conference


1
Nearest Neighbor Conference
  • INFORMS Decision Support
  • Using Most Similar Neighbor Analysis
  • To Create Landscape Data
  • For NEPA Analysis

2
Overview
  • Create a current wall to wall vegetation layer.
  • Grow stand data to current and future year
    conditions.
  • Create alternatives and model vegetation
    treatments (i.e. thinning) for NEPA analysis and
    impacts evaluation.
  • Run models to evaluate and compare alternatives.

3
Included Applications
  • NRIS INFORMS
  • A project-level landscape analysis framework.
  • NRIS FSVeg
  • Forest Service Field-Sampled Vegetation
    database.
  • Forest Vegetation Simulator (FVS)
  • An individual-tree, distance-independent growth
    and yield model.
  • Most Similar Neighbor (MSN)
  • A powerful application used to impute available
    ground-based inventory data to non-inventoried
    units.
  • Stand Visualization System (SVS)
  • Visualize time-step data on a stand by stand
    basis.
  • Fire Area Simulator (FARSITE)
  • A fire spread simulator.

4
Data Requirements
  • FSVeg stand exam data
  • Local stand coverage (which relates to the data
    in the FSVeg database)
  • Non-forested survey data (i.e. rangeland data).
  • Forest FVS Variant Files
  • Most Similar Neighbor analysis
  • DEM derived grids for Slope (in radians), Slope
    Catchment Area, Insulation and Duration.
  • Landsat grids for reflectance values for bands 1,
    2, 3, 4, 5 and 7.

5
NEPA Analysis Process
FSVeg Data
Sampled Vegetation and Fuels Data
FVS/FFE
Most Similar Neighbor (MSN)
DEM Landsat
Wall-to-Wall Current Vegetation and Fuels Data
Legend Yellow Inputs Blue Outputs Green
Models
Treatment Alternative
FVS/FFE
Wall-to-Wall Future Years Vegetation and Fuels
Data (No Action and Alternative)
Fire Area Simulator (FARSITE)
Stand Visualization System (SVS)
6
Process Overview
  • Running MSN
  • Creating the No Action Alternative
  • Alternative Building
  • Evaluating Alternative Impacts
  • INFORMS Prescription Definition
  • Summary

7
  • 1. Running MSN

8
How is Most Similar Neighbor Used?
  • The MSN application is used to impute available
    ground-based inventory data to non-inventoried
    units.
  • The MSN method uses available data from the
    ground-based sample units and globally available
    data measured on all sample units to guide the
    imputation.
  • Examples of global information for all sample
    units include topographic data and satellite
    imagery.
  • Landscape of vegetation data is available for
    analysis based on imputations from the MSN
    process.

9
MSN Calibration
  • MSN analysis command files are prepared and
    tested for each FVS variant. Calibration
    (selection of variables) is the most critical
    part of Most Similar Neighbor analysis.
  • Variables contained in the command files are
    carefully selected in cooperation with the
    researchers who developed the Most Similar
    Neighbor application and methods.
  • Once calibrated, there is a standardized
    methodology for each FVS variant.

10
MSN Calibration Variables
Calibration is the process of finding the correct
combination of global and sampled data for each
FVS variant. Global Data Data that is available
for all polygons (i.e. slope, aspect, Landsat,
etc.). Sampled Data Data that is available for
sampled polygons (i.e. stand exams) or other
vegetation sampling (i.e. range data, fuels
plots). Examples include Basal Area, Trees per
Acre, QMD, Volumes, etc.
11
Where MSN is calibrated by FVS Variant
Completed
Not Done Yet
12
Forests Mapped with MSN
13
MSN MappingStatus by National Forest
Forests Completed
Forests In Progress
  • Region 1
  • Idaho Panhandle
  • Region 3
  • Lincoln
  • Carson
  • Gila
  • Coconino
  • Kaibab
  • Apache-Sitgreaves2 districts
  • Region 1
  • Nez-Perce
  • Region 3
  • Santa-Fe
  • Cibola
  • Corrinodo
  • Tonto
  • Region 6
  • Malheur
  • Umatilla2 districts
  • Wallowa-Whitman2 districts
  • Deschutes2 districts
  • Siuslaw1 district
  • Region 9
  • Chippewa
  • Region 8
  • Ouachita2 districts
  • Daniel-Boone

Near completion 2/2006
14
MSN Vegetation Grouping
  • Most Similar Neighbor analysis is run separately
    on Forested and Non-Forested polygons.
  • Vegetation polygons must be divided into three
    groups by the local GIS shop by adding an
    attribute into the local stands layer
  • Forested Vegetation (FV)
  • Non-Forested Vegetation (NF)
  • Non-Vegetated (NV)

15
Preparing Global and Sampled Data in INFORMS
  • Global Data Preparation Tool Summarizes data
    from the DEM and Landsat Scene into an input
    format for Most Similar Neighbor analysis.
  • Sampled Data Preparation Tool for Forested
    Polygons Grows all stand data forward to the
    year of the Landsat scene to calibrate stand data
    to the current condition using FVS.
  • Sampled Data Preparation Tool for Non-Forested
    Polygons Data is currently used to impute Fuel
    Models only. Work is being done in conjunction
    with USDA Risk Management agency to develop
    methods to calibrate MSN to produce detailed data
    for non-forested vegetation.

16
Running MSN
  • MSN is run as a tool in INFORMS once all of the
    data is prepared.
  • Go/no go statistics are presented when MSN is
    run. This advises the user whether the MSN run
    should be used for further analysis.
  • Statistics are also produced for specific
    vegetation attributes resulting from the MSN run.
    (e.g. basal area, stand height, etc.)

17
MSN Forest Vegetation Quality
Gray Reference (sampled) stands Green OK
Quality (Imputed) Red Poor Quality
(Imputed) Yellow Non-Veg (rocks, lakes,
etc.) Brown Non-Forest (grass, shrubs, etc.)
18
MSN Report for Forested Vegetation
Produced with each MSN run is the MSN Report. It
is a text file summarizing the key elements of
the MSN run. These are the attributes in the
vegetation map.
19
MSN Results
  • The MSN results are stored in a copy of the NRIS
    FSVeg database table NRV_MSN_FOR_USE.
  • This table contains a list of links of un-sampled
    stand polygons pointing them to their most
    similar neighbor with sampled data.
  • This process allows INFORMS tools to use imputed
    data without loading hypothetical data into the
    corporate FSVeg stand and tree tables.

20
Reference Stands
  • NRV_MSN_FOR_USE Reference Stands
  • Stands with sampled data.
  • Note FOR_GIS_LINK and USE1_GIS_LINK are the
    same.

21
Imputed Stands
  • NRV_MSN_FOR_USE Imputed Stands
  • Stands that have not been sampled.
  • Note FOR_GIS_LINK and USE1_GIS_LINK are
    different.
  • USE1_GIS_LINK is the best match.

22
Imputed MSN Stands
Red MSN Imputed Stands Blue FSVeg Stand
Exams Yellow No Data
23
How does MSN help the Forest Service?
  • Provides a method to easily maintain and annually
    update a current vegetation layer.
  • Provides a current wall-to-wall vegetation layer
    containing base-scale attributes.
  • Allows a site to grow the current vegetation
    layer forward into the future using FVS for
    analysis of future conditions. Some examples
    are
  • Current and Future Fire Regime Condition Class
    (FRCC)
  • Current and Future Vegetative Structural Stage
    (VSS)
  • Current and Future individual stand burning
    conditions
  • Allows for modeling of treatments to the
    vegetation layer for NEPA analysis and impacts
    evaluation.

24
Vegetation Layer
  • The results from FVS and MSN are used to
    generate current and future vegetation layers for
    each decade in the simulation.

Fuel Model Same Year
With MSN
Without MSN
25
Base FVS Vegetation Layer
  • A wall-to-wall base vegetation layer is built for
    each decade in the simulation. This layer
    contains information such as basal area, stand
    height, qmd, canopy cover, overstory dbh and more.

26
Fuels Analysis Data
  • The results from FVS and MSN are used to generate
    current and future fuels data layers for each
    decade in the simulation.
  • Attributes in the layers include fuel model,
    crown bulk density, torching index, etc.
  • These data layers can be used to compare fire
    behavior changes over time. This is done in two
    ways
  • Data required for FARSITE is prepared for each
    decade.
  • A Fire Intensity Map is created for each decade.

27
MSN Accuracy Assessment
  • Work is being done to produce a standard accuracy
    assessment methodology for MSN as used in
    INFORMS. Cooperators are
  • Natural Resource Information System (NRIS)
  • Rocky Mountain Research Station (RMRS)
  • Remote Sensing Applications Center (RSAC)
  • An accuracy assessment was performed on the
    Lincoln National Forest (Region 3) in 2005. The
    above-mentioned parties will produce a report in
    2006.
  • Once this approach is finalized, it should
    provide a standard by which to evaluate base
    vegetation layer maps.

28
  • 2. Creating the
  • No Action Alternative

29
Creating No Action Alternatives
Basal Area
No Action Plus Disturbance
No Action
30
Crown Fire Initiation Maps
  • 2005
    2045

31
(No Transcript)
32
Fire Regime Condition Class (FRCC)
  • Output 1 Classification of existing vegetation
    into potential natural vegetation groups (PNVG)
    if the local site does not have an existing PNVG
    map.
  • Output 2 Seral stage classifications.
  • Report
  • Calculation of percent of each seral stage by
    PNVG within the analysis area.
  • Calculation of FRCC for each PNVG.
  • Overall FRCC

33
FRCC - PNVG
Potential Natural Vegetation Groups (PNVG)
34
FRCC - Seral Class
2005
2045
35
  • 3. Alternative Building

36
Alternative Building in INFORMS
  • Vegetative future conditions are created by
    defining alternatives and applying vegetative
    treatment prescriptions.
  • There are three methods for applying
    prescriptions to a stand or a portion of a stand.
    A tool is available to split a stand.
  • If MSN analysis is used, prescriptions can be
    applied to imputed stands (stands which do not
    have a stand exam in FSVeg).
  • Treatments are applied through FVS keyword files.
    This changes future condition vegetative values.

37
Prescription Assignment
38
Prescription AssignmentView Results
39
  • 4. Evaluating Alternative Impacts

40
Evaluating Impacts for Alternatives
  • The results of the prescription treatments from
    FVS are used to generate current and future
    vegetation data for each decade in the
    simulation.
  • The treated alternative layers contain treated
    and untreated stands for the complete analysis
    area.
  • This allows for comparison of effects between the
    no action and treatment alternatives at different
    time steps.

41
Examples Alternative Evaluation
  • Current/Future year comparisons
  • Crown Fire Initiation
  • FARSITE
  • Fire Regime Condition Class (FRCC)
  • Vegetation Structural Stage (VSS) (Region
    Specific)

42
Basal Area Before and After Treatment Same Year
Burn Simulation Treatment
No Treatment
43
Risk Map Before and After Treatment Same Year
No Action
Alternative A
44
FARSITE Before and After Treatment Same Year
  • No Action
    Alternative A

45
Crown Fire Initiation
No Action
Treatment
46
  • 5. INFORMS Prescriptions

47
Vegetative Treatments
  • Vegetation treatments can be created using the
    Forest Vegetation Simulator front-end Suppose
    (GUI).
  • In INFORMS, treatments can be applied to stands
    by selecting them on an ArcView map, from a list
    of stands or with a GIS overlay.
  • Treatments are applied through FVS keyword files.
    This changes future condition vegetative values.

48
INFORMS Prescriptions
  • Prescriptions, as used in INFORMS, are a
    collection of FVS keyword files which can be
    assigned to stands.
  • INFORMS uses FVS to modify the vegetative
    landscape to represent alternative future
    conditions.
  • Sources of FVS keyword files
  • Many regions are creating standard keyword files
  • Forest Health is creating keyword files for
    modeling pest impacts
  • Local FVS users can create their own keyword files

49
Stand Visualization System (SVS)
  • SVS visually demonstrates the effects of
    treatments.
  • Better ID team and public communication of
    prescription treatment results.

50
Stand Visualization SystemWildfire Example
Wildfire in 2035
51
Stand Visualization System
Underburn Now, Wildfire in 2035
52
Stand Visualization System
Thin Now, Wildfire in 2035
53
  • 5. Summary

54
Fuels Analysis Process
FSVeg Data
Sampled Vegetation and Fuels Data
FVS/FFE
Most Similar Neighbor (MSN)
DEM Landsat
Wall-to-Wall Current Vegetation and Fuels Data
Legend Yellow Inputs Blue Outputs Green
Models
Treatment Alternative
FVS/FFE
Wall-to-Wall Future Years Vegetation and Fuels
Data (No Action and Alternative)
Fire Area Simulator (FARSITE)
Stand Visualization System (SVS)
55
Summary
  • INFORMS currently provides a methodology for
    vegetative treatments and fuels analysis.
  • INFORMS provides the ability to produce multiple
    alternatives for various treatment scenarios.
  • MSN provides the ability to do landscape-level
    NEPA analysis (i.e. fire spread).

56
Cooperators
  • Rocky Mountain Research Station (Nick Crookston
    Al Stage)
  • Forest Pest Management (Carol Randall Jim
    Stone)
  • Region 3 Fire Fuels (David Martinez)
  • Los Alamos National Laboratory
  • Texas AM (USDA Risk Management Agency)
  • Mescalero Apache Reservation, Taos Pueblo

57
For More Information
  • Eric Twombly Project Leader
  • Lynne Bridgford GIS Developer
  • Jonathan Marston Developer
  • Tom Warmath - Trainer

FS Web fsweb.nris.fs.fed.us/products/INFORMS WWW
www.fs.fed.us/informs
Write a Comment
User Comments (0)
About PowerShow.com