To describe the dynamics of the global carbon cycle requires an accurate determination of the spatia

1
USING SPECTRAL REFLECTANCE TO DETERMINE LIGHT USE
EFFICIENCY
Karl F. Huemmricha, Lawrence Corpb, Andrew Russc,
Elizabeth Middletond, William Kustasc, John
Pruegerc, and Yen-Ben Chengd aUniversity of
Maryland Baltimore County, Code 614.4, Goddard
Space Flight Center, Greenbelt, MD 20771, USA,
karl.f.huemmrich _at_ nasa.gov bSSAI, cUnited States
Department of Agriculture Agricultural Research
Service, dNational Aeronautics and Space
Administration

• Introduction
• To describe the dynamics of the global carbon
  cycle requires an accurate determination of the
  spatial and temporal distribution of
  photosynthetic CO2 uptake. Vegetation stress
  results in photosynthetic down-regulation,
  decreasing ecosystem carbon uptake. Present
  approaches to determine ecosystem carbon exchange
  rely on meteorological data as inputs to models
  that predict the relative photosynthetic function
  in response to environmental conditions inducing
  stress (e.g., drought, high/low temperatures).
  Also associated with photosynthetic
  down-regulation are changes in the apparent
  spectral reflectance of leaves. This study
  examines the determination of ecosystem
  photosynthetic light use efficiency (LUE) based
  on vegetation spectral reflectance changes
  associated with physiologic stress responses.
• Light use efficiency (LUE) models, driven by
  remotely sensed inputs, have been used to
  estimate productivity for a number of ecosystems.
  At it simplest, a LUE has the form
• G ? fapar Qin ??APAR
• where
• G is gross ecosystem production (GEP), the uptake
  of carbon through photosynthesis measured using
  eddy covariance techniques
• Qin is the incoming photosynthetically active
  radiation (PAR), directly measured at the flux
  towers
• fapar is the fraction of PAR absorbed by green
  vegetation
• APAR is the PAR absorbed by vegetation or fapar
  Qin
• is the light use efficiency.
• Leaves have multiple responses to stress that
  have specific effects on leaf spectral
  reflectance. These include
• Xanthophyll cycle pigments (531 nm)
• Solar Induced Fluorescence (685, and 740 nm)
• Altering the amount of photosynthetic pigments
  chlorophyll a and b, anthocyanins and carotenoids
  (multiple wavelengths)
• Leaf water content (970, 1240, 1630 nm)

Methods Data were collected at the Optimizing
Production Inputs for Economic and Environmental
Enhancement (OPE3) fields (39.03N, 76.85W) at
USDA Beltsville Agricultural Research Center.
Agricultural Research Service researchers grew
corn (Zea mays L., 'Pioneer 33A14') and collected
CO2 flux throughout the 2007 growing season.
Throughout six days 21 June (day of year 172), 2
July (183), 9 July (190), 31 July (212), 9 August
(221), 14 August (226) hyperspectral reflectance
measurements were made using an ASD FieldSpec
along a transect in the corn field. The length
and position of the transect were chosen to be
representative of the tower footprint. The net
CO2 flux was partitioned into gross ecosystem
production (GEP) and ecosystem respiration. Two
half-hourly flux values were averaged for
comparison with average reflectance from
transect. LUE was calculated as the ratio of
GEP and the PAR absorbed by the canopy (APAR).
APAR is the product of incident PAR and fapar
estimated from the Normalized Difference
Vegetation Index (NDVI).
Results We examined the use of spectral
reflectance indexes to detect LUE at hourly
intervals and between days. We found a number of
different approaches can be used to estimate LUE
from reflectance. In the plots each colored line
connects observations from the same day. The
Photosynthetic Reflectance Index (PRI) detects
changes in Xanthophyll cycle pigments from
reflectance at 531 nm compared to a reference
band at 570 nm (Figure 1). We used the ratio of
the first derivatives of the reflectance spectra
at 685 and 720 nm to detect effects of solar
induced fluorescence on reflectance curve (Figure
2). To detect leaf water content the Normalized
Difference Infrared Index was used, NDII is the
normalized difference of reflectances at 858 and
1629 nm. We used Gitelsons model for detecting
concentrations of a leaf pigment (ap) in the
presence of multiple other pigments using three
different wavelength bands R(?1)-1-
R(?2)-1 x R(?3) ? ap For carotenoid
concentrations the following wavelengths were
used ?1 515 nm ?2 565 nm ?3 790 nm,
(Figure 4).
Figure 1
Aerial View of Study Area
Figure 2
Flux Tower CO2 fluxes measured using eddy
covariance techniques. The tower also measured
metrological conditions and incident PAR.
Transect Hyperspectral reflectance measurements
using an ASD field spectrometer were collected at
101 points along a 300 foot transect at hourly
intervals through each measurement day.
Calibration Panel A second ASD measured a
calibration panel when transect measurements were
collected. These data were matched with transect
measurements to calculate reflectance factors.
Figure 3
a
b
c
Conclusions All of the indices were able to
detect most of the variance in LUE within and
between days. The fluorescence index did the
best (r20.82), followed by PRI (r20.70) and the
Gitelson carotenoid index (r20.66), with the
NDII doing the poorest (r20.43), although
relationship between leaf water from the NDII and
LUE would be much higher if data from June 21
were removed. These indices required the use of
narrow band and hyperspectral data, highlighting
the importance of these types of data to detect
physiological changes in vegetation. Further,
this study shows that frequently measured
reflectances can be used to determine fluxes at
hourly time scales over a number of days.
Reflectance data collected multiple times in a
day provide a method of early detection of plant
stress indicating the need to develop methods for
measuring sites frequently.
Conceptual diagrams showing how leaf optical
properties can vary due to changes in a)
xanthophyll pigments, b) fluorescence, and c)
chlorophyll concentration.
Figure 4