Estimation of Photosynthetic Rate of Plant
from Hyper-spectral Remote Sensing of Biochemical Content
Takahiro Endo*, Toshinori Okuda**, Masayuki Tamura**, and Yoshifumi Yasuoka*
*Institute of Industrial Science, University of Tokyo
4-6-1 Komaba, Meguro-ku, Tokyo, 153-8503 Japan
Tel: (81)-3-5452-6410 Fax: (81)-3-5452-6415
E-mail: tendo@iis.u-tokyo.ac.jp
**National Institute for Environmental Studies
16-2 Onogawa, Tsukuba, 305-0053 Japan
Keywords: hyperspectral remote sensing, carbon absorption, photosynthetic rate, chlorophyll, nitrogen
Abstract
Carbon absorption of plant is one of the essential parameters in assessing terrestrial ecosystem functions with respect to global warming. It is, however, not easy to measure carbon absorption directly on the ground. In this study remote sensing method was investigated to estimate photosynthetic rate of plant from the measurement of biochemical content. If net photosynthetic rate may be estimated, carbon absorption and NPP can be predicted. Firstly, we measure the relationship between biochemical concentrations and parameters of "Blackman" photosynthetic model. Secondly, we measure the relationship between biochemical concentration and hyperspectral characteristics. High-resolution reflectance over a range of 333 -2507nm with resolution of about 1.5-10 nm and net Amax (maximum assimilation rate) - photon flux density (PFD) were measured respectively by the GER 2600, LI-6400. Also, chlorophyll a, chlorophyll b, chlorophyll a+b and nitrogen concentrations were quantitatively analyzed from in situ measurement of cucumber's fresh leaves that were cultivated to have different biochemical concentration in a greenhouse chamber. Correlation between saturated Amax and chlorophyll a and nitrogen concentration was r
2=0.90, and 0.91, respectively. Both chlorophyll a and nitrogen concentrations were estimated by the first derivative spectral reflectance (RF') of fresh leaf. RF' at 678.011nm correlated best with chlorophyll a concentration (r
2=0.81). RF'at 732.122nm correlated best with nitrogen concentration (r
2=0.86). Finally, net Amax at given PFD was estimated by the photosynthetic rate model. A correlation between the actual net Amax and the estimated net Amax was r
2=0.74.
1. Introduction
In recent years, remote sensing has made a rapid progress in developing methods to estimate of a number of essential forest ecosystem variables, such as leaf area index (LAI), absorbed fraction of photosynthetically active radiation, canopy temperature, and plant community type. These variables are useful for predicting ecological fluxes. Further insights to ecosystem function, such as biochemical fluxes and processing require additional variables. The prospect of biochemical composition of plant canopy is relevant to the potential elucidation of biochemical fluxes, such as nitrogen cycle and carbon cycle components. A kind of terrestrial ecosystem function is descried as an enormous sink of carbon. The terrestrial vegetation can assimilate carbon dioxide, a kind of greenhouse gas, through photosynthesis, which is a basic process for biochemical fluxes. Photosynthetic rate, which is a kind of biosynthesis, depends on biochemical content in leaf and environmental condition, such as temperature, water and light condition. For example, in case of interspecies, a single leaf with high photosynthetic rate has high biochemical concentration, such as chlorophyll or nitrogen concentration.
Estimation of optical and biochemical properties of fresh leaves have been reported, especially using short-wave infrared wavelengths. Lee F. Johnson et al.
1. found correspondence between the first derivative spectral reflectance from the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) and concentrations of both chlorophyll and total nitrogen (R
2=0.71 and R
2=0.85, respectively) along a vegetation transect in Oregon. Paul J. Curran et al.
2. also used the first derivative reflectance of AVIRIS data to predict chlorophyll, nitrogen, lignin, and cellulose concentrations (R
2=0.96 R
2=0.94 R
2=0.93, and R
2=0.61, respectively) at a slash pine plantation, Florida.
The objective of this study is to estimate net photosynthetic rate based on physiological parameters. Our study examines whether chlorophyll and nitrogen concentrations could be estimated using spectral reflectance and net photosynthetic rate could be estimated by estimating chlorophyll and nitrogen content at the single leaf in the laboratory darkroom and the greenhouse chamber. We tried to investigate net photosynthetic rate models suitable for remote sensing and to statistically analyze relationship between parameters of photosynthetic rate model and biochemical concentrations of fresh leaves. Finally, the authors estimated the net photosynthetic rate using spectral reflectance of fresh leaves measured by a hyperspectral spectrometer.
2. Photosynthetic Rate Models
2.1. Light-Photosynthetic Rate Curve
At present, several net photosynthetic rate models have been developed based on the physiological properties in the plant physiological studies. There are mainly two groups of models based on either the photon flux density (PFD) or CO
2 concentration. One set of models defines PFD as incident light energy, while the other models define CO
2 concentration and rate of electron transport system in the leaf. The estimation of net photosynthetic rate of the latter group of models is more accurate than that of the former. But, these parameters can't be directly measured using remote sensing data. Hence, the authors selected the former set of models, such as " light-photosynthetic rate model ".
The light-photosynthetic rate models are mainly three types; (1) the Michaels-Menten type models; (2) the asymptotic exponential equation type models; (3) the Blackman type models3.. Based on a review of the available literature, the Blackman type was the best fit to an actual net photosynthetic rate. The Michaels-Menten type doesn't fit to the actual photosynthetic rate, when incident light energy is low. In case of using the asymptotic exponential equation type, the slope of the curve raises slowly in the region of curve inflexion. The Blackman type model estimated the actual fresh leave's photosynthetic rate more effectively from low incident light energy levels to high incident light energy levels. So, the authors selected the Blackman type.
2.2. Blackman Type
The "Blackman type" estimation model of net photosynthetic rate of a fresh leaf has five parameters. Amax, the maximum assimilation rate same as maximum photosynthesis rate is estimated as in Eq. (1):
where, Amax
[mmol CO
2 m
-2 s
-1] is the maximum assimilation rate, PDF [
mmol photon m
-2 s
-1] is Photon Flux Density, Amax' [
mmol CO
2 m
-2 s
-1] is the gross maximum photosynthetic rate,
q is the convex value of this model's curve, which stands for the Rubisco activity level. Rubisco is the enzyme that traps CO
2 gas from atmosphere into Calvin cycle.
F is the initial slope of the curve under low incident level. Amax
sat is the saturation value of the Amax-PFD curve, R [
mmol CO
2 m
-2 s
-1] is the respiration rate of flesh leaf.
As a result of literature reviews, Amax
sat and R have high correlation with chlorophyll and nitrogen content. These two parameters can be estimated by hyperspectral characteristics.
q and
F is said to be within a range of 0.7-0.95 and 0.04-0.06, respectively. But, we used the measured value in this study, as we had the results of experiment.
