Development of an Integrated Ecological Database: The mapping process used for the LANDFIRE project

1
Development of an Integrated Ecological
DatabaseThe mapping process used for the
LANDFIRE project
GAP 2007Asheville, NC
Donald LongMissoula Fire Sciences Lab
2
LANDFIRE is a nation-wide multi-partner project
designed to map and model vegetation, fire, and
fuel characteristics using a consistent,
peer-reviewed, scientifically based methodology
(www.landfire.gov).
3
LANDFIRE Mapping Process
LANDFIRE
Fire Regime
FRCC
Mapping
Mapping
SCLASS
Reference
Mapping
Conditions
Fire Regime
LANDSUM
Mapping
Simulations
4
LANDFIRE Mapping Units
Training Sites
5
LANDFIRE Mapping Process
Requirements Identifiable Mapable Modelable Sc
aleable Consistent
6
LANDFIRE Mapping Process
LANDFIRE Reference Database 500,000 records, 24
mapzones
7
Zone 55
Zone 56
8
LANDFIRE Mapping Process
9
Average Annual Precipitation
10
Average Annual Actual Evapotranspiration
11
LANDFIRE Mapping Process
Potential Vegetation the potential
distribution of an Ecological system
Plots
Plots
Existing Vegetation the current distribution of
an Ecological system
12
LANDFIRE Potential Vegetation

• What are we mapping?
• Environmental Site Potential (ESP) - native
  vegetation that could be supported at a given
  site in the absence of disturbance, based on the
  biophysical environment (climax)
• Biophysical Settings (BpS) - native vegetation
  that may have been dominant on the landscape
  during a presettlement reference period, based on
  the current biophysical environment and an
  approximation of the historical disturbance regime

13
Assigning plots to an ESP Example
14
ESP Mapping Objective Methods

• Predictive Modeling with Classification Trees
• Advantages
• Supervised classification approach (plot data
  driven)
• Can take advantage of extensive field plot data
• Creates predictive models from inherent patterns
  in data
• Disadvantanges
• Software makes all the modeling decisions (black
  box)
• Predictive models are weak where plot data are
  sparse or poorly distributed
• Software See5 data mining software

15
ESP Map Output Example
Rocky Mountain Subalpine Dry-Mesic Spruce-Fir
Forest and Woodland
16
BpS Mapping Methods

• Spatial recoding use spatial GIS layers to
  directly assign the pixels in certain settings to
  certain BpS units.
• Common GIS layers used elevation,
  section/subsection

17
BpS Gambel Oak Example
Map Zone 16
18
BpS Mapping Methods

• Plots use plots in the LFRDB to build See5
  models that take pixels in a certain ESP and
  assign them to different BpS units.
• Common plot attributes used EVT, dominant
  and/or codominant species, individual species and
  their abundance (relative cover)

19
BpS Mixed Conifer / Ponderosa
Map Zone 16
20
LANDFIRE Existing Vegetation
EVT
EVH
EVC
21
Generating Lifeform Mask
? Use only DEM derivative and imagery as
predictors ? Exclude tree plots with lt 20
canopy cover ? Pseudoplots ?
10 NLCD mask
22
Generating Lifezone Mask
? Use only DEM derivative and imagery as
predictors ? Pseudoplots ?
23
Generating LF-LZ Mask
Combined Lifeform/ Lifezone Mask
Lifeform Mask
Lifezone Mask


24
Current Methodology EVT
Partial table of modeling groups in Zone 6
(Sierra Nevada)





See5 predictor variables include imagery, DEM
derivative, gradient, and probability data
25
Current Methodology EVT
ERDAS Model
EVT MAP
Group 11 Spatial Output
Group 12 Spatial Output
Group 13 Spatial Output
Add riparian and sparse masks and spatial
outputs
Group 21 Spatial Output
Group Spatial Output
Group 22 Spatial Output
26
Existing Vegetation Cover
27
Existing Vegetation Height

28
LANDFIRE Vegetation Rectification
29
LANDFIRE Mapping Process
30
LANDFIRE Fire Behavior Fuel Models
Spatial Intersection
EVT
CC
CH
ESP
31
Methods Fire Behavior Fuel Models

• Spatial Intersection of EVT/CC/CH/ESP
• Develop fuelbed assignment rulesets using expert
  review
• Linked in a GIS and mapped
• Calibration workshops to refine

32
Methods FBFM Assignments
Great Basin Pinyon-Juniper Existing Vegetation
Type
Expert Opinion
33
LANDFIRE Mapping Process
34
Methods Canopy Fuels
35
LANDFIRE Mapping Process
36
Vegetation Modeling
VDDT Vegetation dynamics models

• Synthesize the best available knowledge of
  vegetation dynamics
• Quantify the natural range of variability in
  vegetation composition and structure
• Developed through expert workshops
• Refined through peer-review process

37
VDDT DB NE Lowland Spruce-Fir Forest
Vegetation Modeling
Class B (Mid Closed)
Class C (Mid Closed)
Class A (Early All)
Class E (Late Closed)
Class D (Late Closed)
Succession
Disturbance
38
Vegetation Modeling
Vegetation Models are used to

• Help map biophysical settings

B
D

• Provide rules for succession class mapping

• Help calculate FRCC

39
LANDFIRE Fire Regimes Departure Mapping

• Simulate reference fire regimes and vegetation
  dynamics
• Map current succession classes
• Map vegetation departure (FRCC)

40
Biophysical Settings Topography
Fire Regime Maps
Vegetation Models
Reference Conditions
LANDSUM
Fire Climate Parameters
41
Succession Class Mapping

• Classify current vegetation into successional
  components of BpS models
• SClasses are defined in vegetation models

SClass Mapping Rules
42
FRCC Calculations
Smaller of Reference or Current
Similarity 10 5 35 25 5
80 Departure 100 80 20 FRCC I
43
LANDFIRE Fire Regimes Departure Mapping
44
Fuels
LANDFIRE FinalDeliverables
Vegetation
Fire Regime
45
Data Distribution

• Ways to Obtain LF Data
• USGS LF National Map
• Web-based view and query
• LANDFIRE Data Access Tool (LFDAT)
• An ArcGis Toolbar data download w/o National
  Map
• On DVD from RSAC
• Creates DVD set with metadata, data attributes,
  and geospatial layers - ready for use in Arc

46
?