Biodiversity Studies in the NASA Remote Sensing Programs

1
Biodiversity Studies in the NASA Remote Sensing
Programs

• Greg Asner, Scott Goetz, Kathleen Bergen,
  Nicholas Coops, Weihong Fan, Mick Follows, Joanne
  Nightingale, Matt Oliver, Volker Radeloff, Tom
  Smith, Richard Waring and colleagues
• NASA CCE Plenary 2008

2
The Importance of Biological Diversity
Biodiversity underpins nearly all of the services
provided by ecosystems to humans

• Biosphere-atmosphere interactions
• Climate system
• Secondary production/Fisheries
• Carbon storage and loss
• Water quantity and quality
• Cultural, recreational, aesthetic value

3
Traditional Biodiversity Mapping Capabilities

• Global-scale models without remote sensing
• Resolution low spatially, low taxonomically and
  functionally
• Field-based assessments
• Extent local
• Resolution taxonomically high, spatially
  medium-to-high

4
Current Limitations

• We lack regional and global knowledge of
  biodiversity
• Extent millions of sq. km
• Resolution high spatially and taxonomically
• So what?
• We dont know whats out there
• We can't truly understand biogeochemical cycling,
  including water cycling
• We cant determine if diversity is changing with
  climate
• We cant track insipient effects of land and
  ocean use, and invasive species

5
Remote Sensing and Biodiversity Research
Observed Species Richness and Abundance
Correlated Species Richness and Abundance
Organismal Association/Alliance
Functional Types
Land cover, Structure, Chemistry and Physiology
Satellite and Airborne Measurements, Models, and
Field Observations
Remote sensing provides access to biodiversity
information at scales that cant be reached using
ground-based observations alone.
6
Mapping Ocean Biomes from Color and Temperature
Data (Aqua)
Matt Oliver (U. Delaware) and colleagues
Total Area of Oligotrophic Biomes
Low biological productivity zones have increased
in the last 5 years
7
MODIS Land Cover and Bear DensityVolker Radeloff
and colleagues
In 1990, the Soviet Union broke down, as did its
control over eastern Europe. Since then, land
use intensity has decreased, and parts of Eastern
Europe are re-wilding.
Bear density in 2000
MODIS Landcover
8
Spectral Discrimination of Plant Species
(Hyperion)Phil Townsend and colleagues
Townsend and Foster 2003
9
Plant Functional Types from AVIRISGreg Asner and
colleagues
Montane Rainforest in Hawaii Volcanoes National
Park
Canopy Nitrogen Concentration
Canopy Water Content
Canopy Water
Leaf Nitrogen
Kilauea Iki
Kilauea Iki
Kilauea Volcano
Kilauea Volcano
Canopy Water
2500 mm
0 mm
2.5
Leaf Nitrogen
0
Asner and Vitousek 2005
10
Canopy Chemistry ? Invasive Species
Myrica invasion front (high leaf nitrogen)
Myrica infestations (high leaf nitrogen and high
canopy water)
Kilauea Caldera
Hedychium in forest understory(high canopy
water)
Asner and Vitousek 2005
11
Bird and Plant Species Interactions and
InvasionCombining Plant Species Identification
from AVIRIS with Field Bioacoustics for Birds
AVIRIS Data
Native Ecosystems
Invaded Ecosystems
Forest
100
Forest
80
AVIRIS
Plant Species Cover ()
60
40
Invaded Gradient
20
Native Gradient
Savanna
0
savanna
forest
savanna
forest
shrub
shrub
woodland
woodland
Savanna
Shrubland
Shrubland
Asner et al. 2005
12
Bird and Plant Species Interactions and
InvasionAVIRIS and Bioacoustics
Bioacoustic Spectra
Processed bioacoustic Spectra
Area under acoustic frequency curve ? avian
abundance
Field-based bioacoustics station
Bird diversity and the ratio of native to
invasive birds is highly correlated with
vegetation composition.
AVIRIS
Boelman et al. 2007
13
Continental-scale Bird Diversity from Terra-MODIS
Data
Nicholas Coops, Richard Waring and colleagues
Seasonality of fPAR from MODIS, 2000-2006
Data from North American Breeding Bird Survey
Coops et al. 2008
14
Dynamic Habitat Index
Seasonality
Productivity
Minimum Cover
Relationship between DHI and Total Bird Species
Richness r2 0.88, p lt 0.001, n420 species
15
Forest Structure and Bird Habitat from Airborne
LiDAR (LVIS)Scott Goetz and colleagues
From top left, clockwise Tufted Titmouse, Brown
Thrasher, Kentucky Warbler, and Carolina Wren.
Photographs by Scott Somershoe, USGS.
Goetz et al. 2006
16
Forest Structure and Bird Habitat from Airborne
LiDARScott Goetz and colleagues
Breeding bird survey (BBS) grid

• 5681 Individuals
• 90 species
• 6 guilds
• Forest
• Scrub/2nd Growth
• Suburban
• Pond/Wetland
• Open-forest
• Semi-open Forest

Goetz et al. 2006
17
Bird Habitat and Diversity from Multi-sensor
FusionKathleen Bergen and colleagues
MODEL
Landsat land-cover composition

• SAR
• volumetric structure
• -biomass

Species Occurrence point samples from field
Modeling GARP (or GLM, GAM, MaxEnt, etc)
Modeled Habitat
Landsat horizontal structure -majority -variety

• Simultaneous characterization of
  multi-dimensional structure both horizontal
  (landscape structure) and volumetric (biomass)
• Landscape structure from optical sensors (e.g.
  Landsat)
• Volumetric structure (i.e. biomass, height) from
  SAR, InSAR, and/or Lidar

Bergen, Gilboy Brown, 2007
18
Bird Habitat and Diversity from Multi-sensor
FusionKathleen Bergen and colleagues
Known Primary habitat Mature conifers
Secondary habitat Younger conifers
Pine Warbler

• Best model included vegetation type, biomass, and
  patch size (gt 20 improvement in accuracy over
  vegetation type alone)

Bergen, Gilboy Brown, 2007
19
Diversity Mapping for Conservation
PrioritizationTom Smith and colleagues
1.
2.
3.
20
Diversity Mapping for Conservation
PrioritizationTom Smith and colleagues
1.
2.
3.
Areas of particularly high genetic and phenotypic
turnover
Genetic
Wedge-billed woodcreeper
Phenotypic
Currently protected areas
21
Biodiversity of Ocean Phytoplankton from Remote
Sensing and Modeling
Mick Follows (MIT) and colleagues

• 78 initialized phytoplankton types
• Random assignment of physiological traits
• Simple allometric trade-offs

• MIT ocean circulation model
• N, P, Fe and Si cycles
• 2 grazers

Physical and Chemical Environment
Genetics and Physiology
Competition Interaction Predation Selection
Self-Organizing Ecosystem Model
Ecosystem Structure and Function

• 99 of biomass in 16 types

22
Biodiversity of Ocean Phytoplankton from Remote
Sensing and Modeling
Emergent biogeography organized into functional
classes

Prochlorococcus analogs
Synechococcus small eukaryotes
Diatoms
.
Other large eukaryotes
Follows et al. 2007 Science
23
If you could build a biodiversity sensor, what
would it be?
Chemistry and physiology 3-D Structure
Formation Flying and Integrated Data Processing
High
Spaceborne SAR/wLiDAR and HiFIS
Temporal Resolution
Integrated Airborne Imaging Spectroscopy
and wLiDAR or SAR
Low
Geographic Scale and Resolution
Small/Fine
Large/Coarse
24
A Few Take-home Messages

• There are an increasing number of approaches to
  address fundamental biodiversity questions with
  NASA imagery.
• Satellite and airborne imagery help us detect
  unique biological patterns that help us
  understand underlying processes.
• Satellite and airborne imagery have the potential
  to revolutionize biogeography and make it a
  leading biological discipline in the 21st century
  (restoring some of the luster it held in the 19th
  century for Darwin, Wallace, Hooker, Bates, and
  others).
• We need to develop and institute tools that allow
  researchers to bridge the gaps in scale (and
  related knowledge gaps) between biome and
  organisms and the molecular components of
  organisms. Observations and associated models
  are our principal tools.
• Future missions to support biodiversity research
  should measure vegetation structure, plant and
  plankton chemistry, and physiology in a fully
  integrated observation approach.