Characterizing non-pigment canopy biochemistry from imaging spectrometer

data for studying ecosystem processes

Gregory P. Asner,Gregory P. Asner,

Mary E. Martin,Mary E. Martin,

Scott V. Ollinger, &Scott V. Ollinger, &

Carol A. Carol A. WessmanWessman

Raymond F. KokalyRaymond F. Kokaly

U.S. Geological SurveyU.S. Geological Survey

Ohio State Univ.Ohio State Univ.

General Research Goals

• To quantify non-pigment biochemistry at the leaf and canopy levels using reflectance spectroscopy

• Apply knowledge of leaf spectroscopy for detailed characterization of vegetation

• Quantify canopy biochemistry to increase understanding of fine-scale variations in ecosystem processes over large areas

Types of Remote Sensing

Electromagnetic

Wavelength (m)

Passive

Systems

Active

Systems

0.1 - 0.4 Ultraviolet (UV)

0.4 – 0.7 Visible

0.7 – 1.4

1.4 – 3.0

Near-Infrared

Shortwave Infrared

3.0 – 1,000 Infrared

1,000-

300,000

Microwave

Airborne CamerasAirborne Cameras

RadiometersRadiometers

SpectrometersSpectrometers

LIDARLIDAR

Radar (SAR)Radar (SAR)

•Satellite-basedSatellite-basedImaging Spectroscopy Data: Imaging Spectroscopy Data:

EO-1 HyperionEO-1 Hyperion

• AirborneAirborne Imaging Spectroscopy Data:Imaging Spectroscopy Data:

HyMapHyMapProbeProbeCASICASI AVIRISAVIRIS

Imaging SpectrometersImaging Spectrometers

EO-1EO-1 SpacecraftSpacecraft

NASA ER-2NASA ER-2

Trends in quantifying non-pigment leaf biochemistry

• Greater understanding of lab, field and remote sensing level spectra in relation to biochemical composition

• Increased use of full spectrum or specific biochemical absorption features to characterize vegetation

• Greater application of imaging spectroscopy to understand ecosystem processes

Understanding vegetation spectra in relation to biochemical composition

• Water

• Nitrogen (in chlorophylls and proteins)

• Lignin and Cellulose

• Non-Photosynthetic Vegetation

Kokaly, REMOTE SENS. ENVIRON. 84:437-456 (2003)

Water• Water content/potential is a limiting factor

for plant transpiration (photosynthesis and carbon gain) and thus growth and development

Kokaly and Clark, REMOTE SENS. ENVIRON. 67:267–287 (1999)REMOTE SENS. ENVIRON. 67:267–287 (1999)

Spectroscopic Estimates of Leaf/Canopy Water

• Leaf/Canopy Liquid Water Thickness• Gao & Goetz, 1994

• Canopy Equivalent Water Thickness• Green et al., 1991 & Roberts et al. 1997

• Canopy Relative Water Content• Serrano et al. 2000

• Foliar Water Potential • Stimson et al. 2005

Serrano et al., REMOTE SENS. ENVIRON. 74:570–581 (2000)

NDWI = NDWI = (R860 - R1240)(R860 - R1240)

(R860 + R1240)(R860 + R1240)

NDII = NDII = (R819 - R1649)(R819 - R1649)

(R819 + R1649)(R819 + R1649)

Stimson et al., Remote Sensing of Environment 96 (2005) 108– 118

Foliar Water Potential (MPa) Foliar Water Potential (MPa)

980

nm

1200

nm

Leaf

(Lab)

Canopy

(AVIRIS)

Clark et al., Remote Sensing of Environment 96 (2005) 375 – 398Remote Sensing of Environment 96 (2005) 375 – 398850 1350 1850

Nitrogen

• By dry weight, very low abundance in leaves, only 0.5 to 4%

• Present in important biochemical constituents of plants– Chlorophyll– Protein

• Linked by field studies to rates of ecosystem functioning (carbon fixation) and widely used in ecosystem models

Nitrogen in Proteins

• Proteins composed of amino acids ranging from a 100 to 100,000 in number

RuBisCO

• Ribulose-1,5-bisphosphate carboxylase/oxygenase, catalyzes the first major step of carbon fixation

• RuBisCO is the most abundant protein in leaves (maybe the most abundant on Earth).

Kokaly, REMOTE SENS. ENVIRON. 75:153–161 (2001)

Kokaly, REMOTE SENS. ENVIRON. 75:153–161 (2001)

Kokaly and Clark, REMOTE SENS. ENVIRON. 67:267–287 (1999)REMOTE SENS. ENVIRON. 67:267–287 (1999)

Nitrogen inChlorophyll

Mutanga et al., Remote Sensing of Environment 96 (2005) 108– 118

Biochemical Components of Plants

• Cellulose– Cellulose monomers (β-glucose) are linked

together through 1→4 glycosidic bonds– Cellulose is a common material in plant cell

walls– Cellulose is the most abundant form of living

terrestrial biomass (R.L. Crawford 1981).

Biochemical Components of Plants

• Lignin– 10% to 40% by dry weight– Lignin is a large macromolecule with molecular

mass in excess of 10,000 amu.– Structural component of plant cell walls.– It is hydrophobic and aromatic in nature.– Lignin concentration in litter has strong

influence on ecosystem processes

CelluloseCellulose

LigninLignin

PineCellulose:Lignin = 1.5

= 3.8

GrassCellulose:Lignin

Kokaly et al., Remote Sensing of Environment, in press

Apply knowledge of leaf spectroscopy for detailed

characterization of vegetation

Cerro Grande FireCerro Grande FireLos Alamos,Los Alamos,New MexicoNew Mexico

PineCellulose:lignin =1.5

=3.8

GrassCellulose:lignin

Kokaly et al., Remote Sensing of Environment, in press

AVIRIS Dry Grass

AVIRISDry Pine

Kokaly et al., Remote Sensing of Environment, in press

Results: AVIRIS Maps

GreenGreenVegetationVegetation

Mineral/AshMineral/Ash

Mineral-1Mineral-1mm

Mineral-2Mineral-2mm

Dry ConiferDry Conifer

Dry & GreenDry & GreenConiferConifer

Straw mattingStraw matting

Ash/CharcoalAsh/Charcoal

Straw mattingStraw matting & Green grass& Green grass

Kokaly et al., Remote Sensing of Environment, in press

GreenGreenVegetationVegetation

Mineral/AshMineral/Ash

Mineral-1Mineral-1mm

Mineral-2Mineral-2mm

Dry ConiferDry Conifer

Dry & GreenDry & GreenConiferConifer

Straw mattingStraw matting

Ash/CharcoalAsh/Charcoal

Straw mattingStraw matting & Green grass& Green grass

Kokaly et al., Remote Sensing of Environment, in press

High Burn SeverityHigh Burn Severity

Kokaly et al., Remote Sensing of Environment, in press

30m

Analysis of spectral shape (full spectrum and/or absorption features) for vegetation

characterization

• Multiple Endmember Spectral Mixture Models:– Roberts, et al. (1998) Remote Sens. Environ. 65:267–279

• Monte Carlo spectral unmixing model:– Asner and Heidebrecht (2002) Int. J. Remote sensing 23:

3939–3958

• Tetracorder, absorption feature analysis:– Clark, et al. (2003) . J. Geophys. Research 108 (E12): 5-1

to 5-44

Understanding ecosystem processes by quantifying canopy biochemistry

• Wessman, et al. (1988) Nature 333: 154–156• Asner and Heidebrecht (2003) IEEE Trans. on

Geoscience & Remote Sensing 41: 1283-1296.• Smith, et al. (2002) Ecological Applications 12:

1286–1302• Ollinger and Smith, (2005) Ecosystems 8: 760–778• Asner et al. (2005), Remote Sensing of Environment

96: 497–508

Asner and Heidebrecht, Int. J. remote sensing, 2002, vol. 23, no. 19, 3939–3958

Asner and Heidebrecht, IEEE Trans. on Geoscience & Remote Sensing (2003), 41:1283-1296.

Non-Photosynthetic Vegetation and Shrubland Ecosystems

Kokaly, REMOTE SENS. ENVIRON. 75:153–161 (2001)

Mutanga et al., Remote Sensing of Environment 96 (2005) 108– 118

• V

Asner and Vitousek,

Proc. Nat. Acad. Sci.,

Vol 102 , 2005.

Smith et al., Ecological

Applications, 12(5), 2002,

1286–1302

Smith et al., Ecological

Applications, 12(5), 2002,

1286–1302

Ollinger and Smith,

Ecosystems (2005) 8: 760–778

Ollinger and Smith, Ecosystems

(2005) 8: 760–778

Ollinger and Smith,

Ecosystems (2005) 8: 760–778

Remaining Challenges/Future Directions

• Overlapping of biochemical absorption features– Leaf water masking effect– Understanding protein/lignin/cellulose changes which

all affect the 2.1 and 2.3 m features

• Quantification of other biochemicals (starch, tannin, hemi-cellulose, phosphorous)

• Independent-site/Multi-site validation• Atmospheric correction needs improvement to

lessen impact of residual features