APPLICATION OF HYPERSPECTRAL TECHNIQUES

1
APPLICATION OF HYPERSPECTRAL TECHNIQUES FOR
MONITORING AND MANAGEMENT OF INVASIVE WEED
INFESTATION CS2-1143 Susan L. Ustin University
of California Davis In-Progress Review
Meeting SEPTEMBER 18, 2000
2
PERFORMERS

• Dr. Susan Ustin University of California,
  Davis Specialist in Environmental remote sensing
• Dr. John Pierce Annapolis Naval
  Academy Specialist in Applied Mathematics and
  remote sensing
• Dr. Charlene Sailer University of California,
  Davis Specialist in Commercial Applications of
  remote sensing
• Dr. Tom Moser Dynamac Corporation Specialist
  in developing biological structural indicators
  of ecological condition
• Dr. Jerry Barker
• Dynamac Corporation
• Specialist in impact of physical encroachment on
  vegetation dynamics
• Ms. Teresa Magee Dynamac Corporation Specialist
  in plant inventory techniques and vegetation
  disturbance factors

3
TECHNICAL OBJECTIVE

• ? Develop and demonstrate methodology for mapping
  invasive weeds using hyperspectral Imaging (HSI)
• ? Develop new analysis methods for analyzing HSI
  data suitable for detecting weeds.
• ? Develop species and site-specific indicators to
  characterize site conditions and predict invasive
  potential.

4
TECHNICAL APPROACH

• ? Identify environmental weed spectral
  characteristics using field and laboratory
  measurements characterize features
  distinguishing weed species as basis for
  determining density and rate of spread.
• ? Establish candidate indicators of invasive
  potential based on species-specific attributes,
  habitat susceptibility and landscape metrics.

• ? Assess validity of spectral, spatial and field
  indicators predicting weed locations and spread
  potential at new sites.
• ? Evaluate linear and non-linear spectral methods
  to identify weeds in images.
• ? Develop model for predicting weed spread
  potential.

5
Technical ApproachSpectral Analysis Management
System (SAMS)

• ? Develop toolbox for efficient analysis of
  field-measured spectral characteristics of weedy
  species.
• ? Connect Metadata to image analysis.
• ? Integrate field spectral analysis with image
  processing procedures.
• ? Methodology for problem-specific image analyses.

6
Technical ApproachANALYSIS SCHEMA
WEED SPECTRA
HYPERSPECTRAL IMAGES
SPECTRA DATABASE
REFLECTANCE RETRIEVAL
FUSE GIS DATA CANDIDATE INDICATOR DATA
SPECTRAL ANALYSIS/QUERY
SITE CONDITION ANALYSIS
SITE-SPECIFIC SPECTRAL CHARACTERISTICS AND
CANDITATE INDICATOR METRICS FOR INVASIVE WEEDS
GIS ECOL. INDICATORS
7
NASA/DOI Demonstration HSI Project Roosevelt
National Park
False Color IR Image of North Dakota badlands
region Most of the park is vegetated
except steep canyon faces.
8
NASA/DOI Demonstration HSI Project Roosevelt
National Park
Leafy Spurge
9
PROGRAM PLAN FY 00
Task 1. Acquire existing GIS image
databases Task 2. Field sampling, design test
draft results Task 3. Acquire Analyze HSI
data Task 4. Identify candidate indicators of
invasive potential
15.3K 171K 90K 30K
92K 12K
Task 5. Develop test draft map of invasive
weeds and Invasive potential Task 6. Evaluate
nearby bases for FY01, including
Camp Pendleton, Yuma Training Center
10
Accomplishments
1. Coordinated research and data collection
Vandenberg AFB, Camp Pendleton, Yakima Training
Center 2. Identified appropriate weed targets
at each site and control sites of native
vegetation. 3. Obtained GIS databases for
Vandenberg and Pendleton 4. Identified
flightlines at Vandenberg, Pendleton, Yakima.
5. Contracted with the Jet Propulsion Lab to
acquire AVIRIS HSI data ( 4-5m pixels) at each
site . 6. Acquired field ecological spectral
data at Vandenberg and Camp Pendleton.
11
PROGRAM PLAN FY 00
Completed ?
? acquired
in progress
in progress Acquired HSI data at
Pendleton (field) and Yakima To be done
Task 1. Acquire existing GIS image
databases Task 2. Field sampling, design test
draft results Task 3. Acquire Analyze HSI data
Go/No Go Task 4. Identify candidate
indicators of invasive potential
Task 5. Develop Test Draft Map of Invasive
Weeds and Invasive Potential Task 6. Evaluate
nearby bases for FY01, including Camp Pendleton,
Yuma Training Center Task 7. Annual Report
12
Number date of AVIRIS flightlines
Flightlines 2 km wide and 20 km long Data
has been geocorrected

• Camp Pendleton
• 4 flightlines 10/13/99
• 5 flightlines 08/05/00
• 7 flightlines 09/08/00
• Vandenberg
• 8 flightlines 09/06/00
• Yakima TC
• 3 flightlines 07/27/00

13
Vandenberg Weeds

• Iceplant Carpobrotus sp.
• Jubatagrass Cortaderia jubata

14
Vandenberg Air Force BaseCarpobrotus edulis
(Iceplant)
15
Vandenberg AFB iceplant invasion of coastal
dune scrub
16
Vandenberg AFB Corderteria jubata (jubatagrass)
17
Ecological Data collected
4- 100m vegetation transects per plot (422) 40
plots per vegetation type were measured at each
Base studied
Vegetation types studied Coastal dune scrub,
Coastal soft chaparral, Mixed chaparral, Riparian

• Vegetation related
• Height of dominant species
• No. of layers of vegetation
• cover of each species
• Plot topography
• Soil description

• Environmental disturbance
• cover bare ground
• cover litter
• standing dead vegetation
• Level and type of disturbance

18
Vandenberg AFB Distribution of sampling plots
by vegetation type of plots (total 352)
Ecological Analysis

• Coastal scrub intact 23
• Coastal scrub with Iceplant 19
• Coastal scrub with Jubata 1
• Chaparral intact 23
• Chaparral with Iceplant 11
• Chaparral with Jubata 23

19
Camp Pendleton Weeds

• fennel Foeniculum vulgare
• tamarisk Tamarix ramosissima,
• T. parviflora

• giant reed
• Arundo donax

20
Camp PendletonIceplant invasion of maritime
chaparral
21
Camp Pendleton Distribution of Sampling plots by
vegetation type of plots sampled (Total 72)
Ecological Analysis

• Coastal scrub intact 11
• Coastal scrub with fennel 89
• due to density of giant reed and Tamarask along
  the Santa Margarita River, the site was sampled
  differently.

22
Camp Pendleton Vegetation
Predicted by Spectral Angle Mapper
TechniqueData 10/13/99
Preliminary Results of Image Analysis
23
Preliminary Results of Image Analysis
Camp Pendleton Vegetation Predicted by
Maximum Likelihood Classification Data
10/13/99
24
Yakima TC Weeds

• knapweed
• Centaurea maculosa
• cheatgrass
• Bromus tectorum

25
ACTION PLAN

• Tasks FY 01
• Coordination with SERDP projects.
• No specific tasks identified.

26
PROGRAM PLAN FY 01
Budget 94.0 10.0
10.0 124.0
40.0 130.0 0.0
Task 1. Conduct field work at FY01 bases Task 2.
Test Revise field sampling techniques Task 3.
Document methods in procedures manual Task 4.
Acquire HSI data at FY01 bases Go/No Go
Task 5.Develop Invasive Weed Indicators Task 6.
Predict Weeds at FY01 Bases Task 7. Annual Report
27
Program plan
FY/TASK FY00 FY01 FY02 FY03 Acquire
existing GIS/Image Databases ///////// Field
sampling, design, test results ////////////////
// Acquire and Analyze HSI data /////////////
/////////////////// GO/NO GO Decision
X Identify candidate
invasive indicators ////////////////////////////
//// develop test map of invasive weeds
//////////////////////////////// Initiate
study for Pendleton, Yuma, Yakima
/////////// Annual Report

////// Acquire field data FY01
bases ///////// Test Revise sampling
procedures ///////////////////// Document
methods in procedures manual
//////////////////////////////// Acquire and
Analyze HSI data //////////////////////////////
// GO/NO GO Decision
X Develop invasive weed
indicators ////////////////////////////////
Predict weeds at FY01 bases
//////////////////////////////// Annual
Report
////// Extend Analyses
to FY02 bases ///////////////////// Evaluate
Revise analysis of FY01 bases ///////////////
////// Acquire and Analyze HSI
data ///////////////////// GO/NO GO
Decision
X Develop invasive weed indicators
//////////////////////////////// Predict weeds
at FY02 bases ///////////////////// ////// Pr
ototype DBMS Toolbox
//////////////////////////////// Annual
Report
////// Final Tests for
Image Analysis Tools ///////////////////// Co
mplete Site-specific Indicators //////////////
/////// Complete Technology Transfer ////////
///////////// Provide Software Toolbox to
DOD ///////////////////// Demonstration of
Toolbox ///////////////////// Produce and
Deliver Final Report /////////////////////
28
PROGRAM FUNDING
SERDP 407.6 94.0 10.0
10.0 124.0 40.0 130.0 0.0
UCD COST SHARE 64.8K 5.8 13.0 10.0
6.0 10.0 10.0 10.0
SERDP 410.3 15.3 90.0 171.0 30.0
92.0 12.0 0.0
FY 00 Task 1 Task 2 Task 3 Task 4 Task
5 Task 6 Task 7 FY 01
Task 1 Task 2 Task 3 Task 4 Task 5 Task
6 Task 7

UCD COST SHARE 67.2K 18.0 5.0 5.0
10.0 9.2 10.0 10.0
29
PROGRAM FUNDING

UCD COST SHARE 61.6K 6.6 10.0 10.0
10.0 10.0 10.0 10.0
SERDP 402.5 72.0 40.0 155.0
50.0 50.0 35.0 0.0
FY 02 Task 1 Task 2 Task 3 Task 4 Task
5 Task 6 Task 7

30
DELIVERABLES

• Final Report
• Description of methodology
• to isolate spectral features of weedy species
• to assess the feasibility for remote monitoring
  weed distribution
• to assess the effective use of coarse vs fine
  resolution spatial and spectral sensors.
• to establish candidate indicators of invasive
  potential
• to validate indicators and assess prediction of
  invasive potential and distribution
• Weed species and extent maps from remotely sensed
  hyperspectral data
• Peer Review Publications

31
DELIVERABLES 2

• Site condition maps by fusing field data,
  spectral image data and GIS database layers
• Site-specific predictive models for weed species
  distribution
• Field Ecological Indicator Tool
• Training and Support Manuals
• Spectral Database Management System
• Nonlinear Spectral Analysis Tools
• Interactive, Web-Based Technical Support
• Field protocols for establishing indicator metrics

32
TRANSITION PLAN

• DBMS Toolbox at Test Bases
• LMS Compliant Invasive Weed Potential Tool
• Web-based Technology Support

33
Technical ApproachEcological Indicators
CANDIDATE INDICATORS of INVASIVE POTENTIAL Weed
Species Attribute Indicators Prolific Seed
Production High Germination Rate Sexual and
Asexual Reproduction Multiple Dispersal
Mechanisms Wide Tolerance Range Stress
Tolerant Habitat Susceptibility Indicators
Susceptible Habitats and Conditions (e.g.,
Disturbance) Biodiversity of Native Plants
Presence of Noxious Weeds Native Plant Cover
Water and Nutrient Stresses Soil Properties,
Topography Landscape-Level Indicators
Proximity to Established Weeds Mode of
Dispersal Land Use Susceptible Habitat Types
History of Land Use
34
Proposed Ecological Analyses(conducted at
quadrant, plot, and site scales)

• Cluster analysis
• to identify vegetation structure and pattern
• Ordination analysis
• to correlate environmental and disturbance
  variables
• to relate weed distribution to environmental
  variables

35
Vandenberg Air Force Base AVIRIS flightline
locations
36
Camp PendletonSanta Margarita River flightlines

37
Camp Pendleton AVIRIS single band image