Invasive Species Forecasting System Tamarisk Detection and Modeling

1
Invasive Species Forecasting System Tamarisk
Detection and Modeling

• Jeff Morisette, Jerry Griffith, Asad Ullah,
  Mohammed Kalkhan, Paul Evangelista, Bill Cheatum,
  Tracy Davern, Geneva Chong, Jim Graham, John
  Schnase, Tom Stohlgren
• Lead ISFS Science Team
• Goddard Space Flight Center
• Presented atMonitoring Science and Technology
  Symposium
• 22 September 2004

2
Collaborators
USGS Stohlgren, lead ISFS collaborator Univers
ities Colorado State University Kalkhan, Reich,
Evangelista/Graham/Newman/Crossier/Davern Universi
ty of Southern Mississippi Jerry
Griffith Umaine spatio-temporal helix
modeling George Mason University statistical
analysis U.Alabama, Huntsville Graves, expertise
of Earth Science data/metadata Agencies/NGO USDA,
ARS Hyperspectral airborne imagery over
Tamarisk, Curruthers, Delfouse USDA Forest
Service, RSAC and Colorado mapping pilot project,
Czaplewski Colorado DNR Colorado mapping pilot
project Tamarisk Coalition Bill Cheatum,
Tamarisk modeling NASA GSFCs Invasive Species
Forecasting System Schnase, Pedelty,
Smith Application Division Scheffner and
Turner Stennis Space Center, McKellip
applications program, EO-1 team, Steve Ungar,
Tom Brakke, Ken McGwire ASTER team, Leon
Maldanado, Tim Gubbles
3
Outline

• Background on the Invasive species forecasting
  system
• Current tamarisk study within ISFS
• Spectral issues
• Temporal issues
• Coverage area
• Conclusions

4
Project Overview

• NASA / USGS Invasive Species Partnership
• Launched approximately two years ago
• One of NASA ESEs Twelve National Application
  areas
• Activities involve
• USGS National Institute of Invasive Species
  Science
• NASA Goddard Space Flight Center
• Many others

5
Project Organization
Engineering Team
CommunicationsTeam
Science Team
Jeff Morisette Stohlgren, Smith, Pedelty, Reich,
Kalkhan, Crossier, Chong, Garnet, Griffith,
Madsen, Pinzon, and others
David Kendig Pollack, Ilagan, Graham, Newman,
Bruce, Clark, Graves, Memarsadeghi, Sachs, Ullah,
Tilmes, and others
Jim Closs Gutro, OCarroll, Banowetz, Wilson,
Graham, Newman, Ritrivi, and others
Neal Most Tom Stohlgren, Jim Smith, John Schnase
6
USGS Science / Client Needs

• On-demand, predictive landscape- and
  regional-scale models and maps for biological
  invasions
• Integrated access to tailored NBII / ESE data

7
Invasive Species Forecasting System
8
Invasive Species Forecasting System

• OLS Regession Kriging
• Regression Tree Analysis
• GARP
• EcoNiche

• NIISS Tamarisk
• Phase 1 EOS
• NBII
• ESE

• T-Map

9
Brief BackgroundTamarisk within the Invasive
Species Forecasting System
Cheatgrass/Thistle
Tamarisk
Native/Non-Nativegrass
WoodyEncroachment
Aquatics
SpeciesRichness
ESA presentation Manuscript
Contribute to Grand Challenge
10
Objectives of Work
careful observation of current patterns of the
invasion of nonnative plant species in natural
landscapes might provide a reasonable staring
point for the design of more realistic
experiments and mathematical models, and in the
search of a generalized theory of
invasion. Stohlgren- director of the National
Invasive Species Science Center Beyond Theories
of Plant Invasions Lesson From Natural
Landscapes, Comments on Tehoretical Biology, 7
355-379, 2002. Our current need is to figure out
where Tamarisk is and how it got thereso we can
figure out where it will go. Success
finding/controlling Tamarisk before it is
established
11
Plan for work (in progress)

• Build on existing remote sensing Tamarisk
  monitoring
• Collect hand held spectra, weekly digital photos,
  and multi-resolution, multi-temporal observation
  for large Tamarisk stands through the 2004
  growing season
• Determine most efficient operational monitoring
  of Tamarisk infestation which may be a specific
  airborne sensor flown during one or more specific
  phenological conditions

12
Current StudyFour Tamarisk study sites
Water 0 - .1 .1 - .2 .2 - .5 .5 - .7 .7 1.0

• Background image June 26, 2003 MODIS Enhanced
  Vegetation Index taken from 4 year contiguous US
  coverage created by MODAPS and stored in the ISFS

13
Debeque area

• Site for Fieldspectrum
• (image showingthe importanceof
  includingtopographicinformation)

May 18th, 2004 Debeque ASTER scenefalse color
NIR, 3x DEM vertical exaggeration
14
Fig. 1. Reflectance from branches laid on dark
gravel road shoulder
4 July 2004
Fig. 2. Reflectance from tamarisk branches with
flowers, or small piles of flowers only
15
Fig. 3. Reflectance from trees
5 July 2004
Fig. 4. Reflectance from trees or other
vegetation.
16
Reflectance difference from Tamarisk
17
Temporal Issues

• Importance
• Plant phenology an important driver for animal
  species
• Many change habitats to track available resources
• Current Sources
• Multispectral imagery
• 30m resolution several times per year
• 250m resolution daily
• Future Sources
• Higher temporal resolution multi-spectral
• Satellite-borne hyperspectral
• Meterological data

18
Airborne Multi-spectral Results
Indication of temporal signature
Reflectance of Riparian Vegetation at 550nm
12
10
8
6
Reflectance
4
2
0
June
May
Sept
August
Jan-Feb
November
From Everitt, J. H., C. J. Deloach. 1990.
Remote Sensing of Chinese Tamarisk (Tamarix
chinensis) and Associated Vegetation. Weed
Science. 38273-278.
19
Seasonal photos through 2004
Water 0 - .1 .1 - .2 .2 - .5 .5 - .7 .7 1.0
20
Vegetation Index at 250m resolution
Four years MODIS vegetation index time series
21
Temporal signal at 250m spatial resolution
22
Hyperion 19 July 2004
23
Hyperion 20 August 2004
24
Satellite Observations
25
Satellite Coverage issue
Green 13 Landsat images covering the Colorado
River Blue 250m buffer on either side of the
Colorado River AREA Green 354K sq km Blue
3K sq km
26
Close-up
27
Preliminary findings

• Multi-spectral imagery should suffice, but with
  specific bands, and needs ortho-correction
• Temporal and Spatial resolution is critical
• Need to catch infestation early (small stands)
  implies high resolution imagery
• Detectability influenced by phenology and stand
  size
• Synoptic, watershed-wide, image acquisition is
  difficult with satellite imagery

28
Preliminary conclusions

• MODIS data can relay temporal information but are
  too course for monitoring tamarisk
• Acquisition strategy and area of interest imply
  aircraft as optimal platform
• Possible monitoring system
• airborne system triggered by MODIS phenology
  and/or field observations
• temporal and spectral requirements for airborne
  system determined by result of current study