Using field spectroscopy to assess the potential of statistical approaches for the retrieval of sun-induced chlorophyll fluorescence from ground and space

https://doi.org/10.1016/j.rse.2013年01月01日7 Get rights and content

Abstract

Sun-induced chlorophyll fluorescence (Fs) is an electromagnetic signal emitted in the 650–800 nm spectral window by the chlorophyll-a of green leaves. Previous studies demonstrated the retrieval of Fs on a global scale using high spectral resolution measurements by the Fourier Transform Spectrometer (FTS) on board the greenhouse gases observing satellite (GOSAT). The retrieval of Fs from GOSAT-FTS data is based on the modeling of the in-filling of solar Fraunhofer lines by Fs. The first Fs retrieval methods for GOSAT-FTS measurements were based on physical formulations of the radiative transfer between the atmosphere, the surface and the instrument including the Fs emission. As an alternative, a statistical method was also successfully applied to GOSAT data. This method is based on a singular vector decomposition (SVD) technique producing a basis of spectral functions able to model the contribution of the reflected solar radiation to the top-of-atmosphere measurement in a linear way. The Fs signal is included in the forward model as an extra parameter adding to the reflected solar radiation. Here, we use field spectroscopy measurements to provide further experimental evidence on the retrieval of Fs with statistical approaches in both Fraunhofer lines and atmospheric oxygen and water vapor bands. The statistical retrieval method used with GOSAT-FTS data has been adapted to a set of ground-based spectro-radiometer measurements in the 717–780 nm range. Retrieval results in the 745–759 nm window, which contains only Fraunhofer lines, support the overall approach of estimating Fs from space measurements in that spectral window. Furthermore, the application of the method to broader fitting windows including both Fraunhofer lines and and (oxygen and water vapor) atmospheric bands atmospheric bands has been proven to be very effective to reduce the retrieval noise and has also shown a good comparison with reference O2A-based retrievals. This allows consideration of statistical methods as a powerful option for Fs retrieval from broad-band space-based measurements in the near-infrared.

Highlights

► We investigate the retrieval of fluorescence (Fs) from space with field data. ► The retrieval proposed is based on a novel statistical approach. ► We demonstrate the feasibility of Fs retrieval solely with Fraunhofer lines. ► Statistical approaches have been shown to be useful for retrievals in O2A.

Introduction

A small fraction (~ 1%) of the solar radiation absorbed by green leaves is emitted as chlorophyll-a fluorescence, which is widely known as sun-induced chlorophyll fluorescence (Fs). The Fs emission spectrum expands from around 650 nm to over 800 nm and includes two broadband peaks centered in the red (685 nm) and far-red (740 nm) spectral regions (e.g. Baker, 2008). Because Fs and photochemistry compete for excitation energy in photosystem II (the photosystem where most of the fluorescence emanates from), the Fs signal is widely regarded as a potential tool to track changes in photosynthetic activity of vegetation via remote sensing (Daumard et al., 2010, Flexas et al., 2002, Porcar-Castell, 2011, Soukupová et al., 2008).
Even though there has been an intense activity in the field of the remote sensing of sun-induced chlorophyll fluorescence in the last years (Meroni et al., 2009, and references therein), it has not been until very recently that measurements of s from space have been proven feasible (Guanter et al., 2007) and the first global maps of Fs have been produced (Frankenberg, Fisher, et al., 2011, Guanter et al., 2012, Joiner et al., 2011, Joiner et al., 2012). The initial analysis of those maps has shown a high correlation between Fs and gross primary production at the global and regional scales (Frankenberg, Fisher, et al., 2011, Guanter et al., 2012). This can represent a decisive step towards the global mapping of photosynthesis from space.
The main challenge for Fs retrieval is to disentangle it from the solar signal reflected by the atmosphere and the surface. At the ground level, this has been achieved with high spectral resolution measurements in the 600–800 nm window and the modeling of the in-filling of the O2A-band by Fs. This modeling has traditionally relied on the Fraunhofer Line Discriminator (FLD) principle (Alonso et al., 2008, Amorós-López et al., 2008, Corp et al., 2006, Damm et al., 2011, Daumard et al., 2010, Moya et al., 2004, Plascyk and Gabriel, 1975, Rascher et al., 2009, Zarco-Tejada et al., 2009, Zarco-Tejada et al., 2012), generally consisting in a measurement channel inside the O2A-band and two reference channels located outside the absorption.
In the last years, spectro-radiometers with a sub-nanometer spectral resolution providing several measurements inside the O2A- and O2B-bands have enabled the application of the so-called spectral fitting methods (SFMs) (Meroni and Colombo, 2006, Meroni et al., 2010). In both cases, the in-filling of the O2A- and O2B-bands (centered around 760 nm and 687 nm, respectively) by Fs is determined by comparison with a reference measurement of the solar irradiance arriving at the canopy. To avoid confusion, it must be noted that the FLD method was originally developed for measurements in solar Fraunhofer lines (Plascyk & Gabriel, 1975), but it is being used with telluric O2 bands for field-based Fs retrieval.
Both FLD and SFM techniques have been successfully used to retrieve Fs in O2 absorptions from ground measurements (Amorós-López et al., 2008, Meroni et al., 2008, Meroni et al., 2011, Middleton et al., 2008). However, the application of O2A-based methods on airborne and spaceborne instruments is not straightforward due to the absorption of the Fs signal by O2 between the surface and the top-of-atmosphere (TOA) and to the relatively strong impact of atmospheric scattering on the O2 absorption features (Frankenberg, Butz and Toon, 2011, Guanter et al., 2010). Satellite data have been successfully employed for O2A-based retrieval of Fs from space, although only over some particular areas containing non-fluorescent surfaces to constrain the retrieval (Guanter et al., 2007). Methods for the retrieval of Fs from space-borne spectrally-resolved measurements in O2A and O2B are being investigated (Guanter et al., 2010) in the framework of the FLEX mission concept (Drusch & FLEX Team, 2008) currently under development within the 8th ESA Earth Explorer Program.
The high spectral resolution measurements provided by the Fourier Transform Spectrometer (FTS) onboard the Greenhouse gases Observing SATellite "IBUKI" (GOSAT) (Kuze et al., 2009) represent the first space data from which the production of global Fs maps has been feasible to date. The band 1 of the FTS covers the 756–775 nm spectral window with a spectral resolution of around 0.025 nm at 760 nm. This high spectral resolution has enabled the application of the Fs in-filling approach to individual solar Fraunhofer lines located around the O2A-band. Different spectral components of the TOA signal are shown in Fig. 1 for a spectral resolution of around 0.13 nm. The 756–775 nm window covered by the GOSAT-FTS band 1 is in the center of the top-of-canopy radiance spectrum displayed. Both the O2A-band and the Fraunhofer lines in the near-infrared region are shown. All the spectral features between 745 and 759 nm in the reflected radiance spectrum are due to Fraunhofer lines. The fractional depth of Fraunhofer lines is practically not affected by atmospheric absorption or scattering, which makes Fs retrievals in this spectral window to be almost insensitive to atmospheric effects (Frankenberg et al., 2012). The same in-filling principle was later used by Joiner et al. (2012) for global Fs retrievals from SCIAMACHY (SCanning Imaging Absorption spectroMeter for Atmospheric CHartographY) measurements in the strong Fraunhofer line in 866 nm.
Here, we use top-of-canopy (TOC) radiance measurements to evaluate Fs retrieval methods applicable to space measurements. The main goals of this study are (i) to assess the potential of statistical Fs retrieval methods making use of broad fitting windows including both Fraunhofer lines and atmospheric bands, and (ii) to provide an indirect test of the consistency of Fraunhofer-line based Fs retrievals from space.

Section snippets

Generalities on Fs retrieval

The spectral radiance measured by a sensor at the TOA over a fluorescent target with a reflectance assumed to be Lambertian can be formulated asLTOA=L0+Egρsπ+FsT↑1−Sρs,where ρs is the surface reflectance, L0 is the atmospheric path radiance, Eg is the global (direct plus diffuse) at-surface irradiance, S is the atmospheric spherical albedo, and T↑ is the total atmospheric transmittance (for diffuse plus direct radiation) in the observation direction.
Assuming that Sρs ≪ 1 in the near-infrared

Trade-off between spectral resolution and fitting window for Fs retrieval in the Fraunhofer lines

Firstly, we investigated the precision of Fs retrieval with the Fraunhofer line approach by means of a sensitivity analysis evaluating the impact of spectral resolution and width of the fit window on the retrieval. This analysis is useful to interpret our results and to further determine the best trade-off between resolution and spectral coverage for an instrument intended for ground-based retrievals in Fraunhofer lines.
Simulations were performed for a typical vegetation target in the 700–759 nm

Summary and conclusions

We have used field spectroscopy to assess the potential of a statistical approach for the retrieval of sun-induced chlorophyll fluorescence from ground and space measurements. The proposed method relies on the modeling of the radiance spectra measured over the canopy with a set of spectral functions derived from non-fluorescent targets. Such spectral functions are produced with a singular vector decomposition technique applied to measurements over non-fluorescent targets. The resulting singular

Acknowledgment

This research has been funded by the European Commission through the 7th Framework Marie Curie Actions Programme and the Emmy Noether Programme of the German Research Foundation. The authors thank Albert Porcar-Castell from the University of Helsinki for his comments on an earlier version of the manuscript. We acknowledge the staff of the Remote Sensing of Environmental Dynamic Laboratory, DISAT, UNIMIB (S. Cogliati, M. Migliavacca) for their support during the field campaigns.

References (35)

  • P.J. Zarco-Tejada et al.

    Fluorescence, temperature and narrow-band indices acquired from a uav platform for water stress detection using a micro-hyperspectral imager and a thermal camera

    Remote Sensing of Environment

    (2012)
  • L. Alonso et al.

    Improved Fraunhofer line discrimination method for vegetation fluorescence quantification

    IEEE Geoscience and Remote Sensing Letters

    (2008)
  • J. Amorós-López et al.

    Evaluation of remote sensing of vegetation fluorescence by the analysis of diurnal cycles

    International Journal of Remote Sensing

    (2008)
  • N.R. Baker

    Chlorophyll fluorescence: A probe of photosynthesis in vivo

    Annual Review of Plant Biology

    (2008)
  • C. Buschmann

    Variability and application of the chlorophyll fluorescence emission ratio red/far-red of leaves

    Photosynthesis Research

    (2007)
  • L.A. Corp et al.

    Fluorescence sensing techniques for vegetation assessment

    Applied Optics

    (2006)
  • F. Daumard et al.

    A field platform for continuous measurement of canopy fluorescence

    IEEE Transactions on Geoscience and Remote Sensing

    (2010)

    • Cited by (133)

    View all citing articles on Scopus
    View full text
    Copyright © 2013 Published by Elsevier Inc.