Elsevier

Ecological Informatics

Volume 7, Issue 1, January 2012, Pages 7-18
Ecological Informatics

Retrieval of seasonal variation in photosynthetic capacity from multi-source vegetation indices

https://doi.org/10.1016/j.ecoinf.2011.10.004Get rights and content

Abstract

Seasonal information on photosynthetic-capacity parameters (maximum carboxylation velocity, Vcmax; and maximum rate of electron transport, Jmax) plays an important role in accurate simulation of carbon fixation in gas-exchange models. Exact inclusion of seasonal information on photosynthetic-capacity parameters into the models has been an irresolvable challenge. This paper investigated the relationships between vegetation indices (from multiple sources) and photosynthetic-capacity parameters of three beech forest stands (Fagus crenata) along an elevation gradient in the cold-temperate zone of Japan, over the entire growing season of 2006. Diverse vegetation indices were examined in terms of spectral, spatial and temporal scales; ranging from meteorological sensor-based broadband indices to hyperspectral data-based narrowband indices, to simulated MODIS (MODerate-resolution Imaging Spectroradiometer) indices based on hyperspectral data, and finally satellite-borne MODIS vegetation indices. Regression analysis revealed that all examined indices, with the exception of the downloaded MODIS products, had significant regression relationships with photosynthetic parameters (P < 0.001) when all data were pooled. Among the different indices, the simulated MODIS NDVI (Normalized Difference Vegetation Index) performed the best for both Vcmax and Jmax (R2 = 0.81 and 0.73, respectively). Site differences were apparent, as the simulated MODIS NDVI performed the best in exponential regressions for the 550 m site, while broadband NDVI performed best in exponential regression models for the 900 m site. The broadband SR (Simple Ratio) in relation to Vcmax performed best with respect to a linear model, whereas the broadband NDVI with Jmax performed the best in an exponential model for the 1500 m site. The results reveal that vegetation indices which are obtained across different scales nevertheless retain tight relationships with canopy-scale photosynthetic-capacity parameters. The established relationships were inversely applicable to derive seasonal trajectories of photosynthetic-capacity parameters. Thus, new insight and confidence is gained for using remotely estimated photosynthetic parameters, even though most previous research works were limited on linking of vegetation indices with biophysical parameters. The control effect of physiological capacity on reflectance and further on vegetation indices has not been adequately established and thus needs further orientation for rigorous research work.

Highlights

► Close relationships between multi-source VIs with seasonal Vcmax and Jmax in deciduous forests have been identified. ► The simulated MODIS NDVI performed the best for both Vcmax and Jmax. ► VIs across different scales maintained close relationships with canopy scale Vcmax and Jmax. ► The relationships potentially can be integrated into gas-exchange models to provide seasonal information of Vcmax and Jmax.

Introduction

Leaf “photosynthetic capacity” refers to the existing enzymatic complex which determines the momentary flux rates of carbon assimilation processes in vegetation at leaf level. Monitoring of changes in photosynthetic capacity is required to achieve a clear understanding of leaf carbon dioxide uptake, and via scaling, vegetation or ecosystem carbon gain. It is therefore a crucial factor in ecosystem carbon cycle studies (e.g. Field et al., 1995, Sellers et al., 1997). Maximum substrate saturated rate of rubisco carboxylation (Vcmax) and maximum rate of electron transport (Jmax) have been generally accepted as two pivotal parameters in photosynthetic-capacity simulations, since the pioneering work of Farquhar et al. (1980). In their biochemically-based model of leaf CO2 uptake and subsequent conversions, overall photosynthetic assimilation rate is either limited by the rate of carboxylation velocity or by electron transport rate. Farquhar's model has emerged as the most often utilized photosynthesis model at various scales in ecosystem carbon studies, ranging from leaf (Harley and Tenhunen, 1991, Leuning, 1995), to canopy (Chen et al., 1999, Tenhunen et al., 1999), to landscapes and regions (Kimball et al., 2000, Liu et al., 1999), and to continental (Bonan, 1998, Sellers et al., 1996) scales. The wide application of Farquhar's model, as pointed out by Xu and Baldocchi (2003), lies in the ease by which leaf or ecosystem photosynthetic processes may be accurately described in response to environmental conditions and in agreement with chamber or eddy covariance gas-exchange measurements.

Vcmax has been traditionally treated as a constant in gas-exchange models, while Jmax is maintained in a set proportion to Vcmax. In the work of Wullschleger (1993), as reported by Bonan, 1995, Chen et al., 1999, Vcmax was set to 33 μmol m 2 s 1, while Sellers et al. (1992) set Vcmax at 100 μmol m 2 s 1, possibly due to the differences in species and locations. However, increasing evidence indicates that large seasonal variations occur in both deciduous (Wang et al., 2008, Wilson et al., 2000, Xu and Baldocchi, 2003) and evergreen (Misson et al., 2006) forest ecosystems. Ignoring this basic principle will result in a large deviation from reality in simulation outputs (Wang et al., 2004a, Wilson et al., 2001). Furthermore, it appears that a general seasonal pattern for these critical photosynthetic-capacity parameters also does not occur. We base this conclusion on extremely time and manpower consuming direct gas-exchange measurements, where leaves or whole plants are periodically examined in specially designed cuvettes and where the results allow calculation of Vcmax and Jmax. Finding a simpler alternative method capable of providing seasonal trajectories of photosynthetic-capacity parameters remains very important to allow the inclusion of such information into gas-exchange and carbon cycle models.

Vegetation spectral reflectance, especially in the visible and near-infrared domains, contains a high volume of information on vegetation physiological processes. Vegetation spectra in these regions are characterized by very low reflectance in the red portion, followed abruptly by high reflectance in the near-infrared. The reflectance change is related to changes in photosynthetic activity in various narrow bands (Carter, 1998). Thus, non-destructive remotely sensed reflectance should provide a useful data source with respect to photosynthetic activity. It should be possible to derive important information from such data for gas-exchange models, since many sensors already exist with capability in providing information at different spatial and temporal scales. The availability of information on vegetation indices focused on the red/far-red transition at high spatial and temporal resolution is continually improving.

To date, many efforts have been devoted to linking reflectance information with canopy photosynthesis, and a variety of vegetation indices have been designed (e.g. Carter, 1998, Choudhury, 2001, Dobrowski et al., 2005, Gamon et al., 1995). Among the developed vegetation indices, the “photochemical reflectance index” (PRI) (Gamon et al., 1992) and “simple reflectance index” in the red-edge spectral region (Dobrowski et al., 2005) have been shown to vary diurnally with photosynthetic rate. On the other hand, commonly used vegetation indices such as simple ratio (SR) or “normalized difference vegetation index” (NDVI) may vary linearly or curvilinearly with observed photosynthesis flux rates over long periods (Choudhury, 2001, Gamon et al., 1995, Myneni et al., 1992, Running and Nemani, 1988, Sellers, 1985). However, while there is broad acceptance that vegetation indices may potentially track seasonal variations in photosynthetic capacity, only Wang et al. (2009) have to our knowledge directly demonstrated the correlations found between vegetation indices and photosynthetic-capacity parameters.

At present, most common data archives on vegetation indices originate from satellite data sources, such as AVHRR (Advanced Very High Resolution Radiometer) GAC data sources (James and Kalluri, 1994); or 1 km-data sources (Eidenshink and Faundeen, 1994); SPOT4-VEGETATION data sources (Duchemin et al., 2002); and MODIS (MODerate-resolution Imaging Spectroradiometer) data sources (van Leeuwen et al., 1999). Although satellite-borne data can provide information at large spatial scales, the data quality is strongly influenced by atmospheric conditions. Where coarse resolution information in both spatial and spectral contexts is utilized, we may expect that a weaker linkage to canopy photosynthetic activity as will be found. On the other hand, field spectrometers and aerial hyperspectral sensors, which sample electromagnetic spectra in very narrow contiguous bands, have special features that allow careful evaluation of spectral indices. Thus, information from such instrumentation may improve our understanding of the relationships between remotely sensed signals and photosynthetic processes. Nevertheless, the observation frequency of field hyperspectral measurements is still limited by high manpower requirements and costs at the present time.

Broadband vegetation indices obtained from tower-mounted photosynthetically active radiation (PAR) and global radiation (both incoming and reflected above canopies) sensors also have close links to the seasonal trajectory of canopy physiological activity (Wang et al., 2004b). Broadband indices were first proposed by Huemmrich et al. (1999) and applied at four BOREAS (Boreal Ecosystem-Atmosphere Study) sites. Wang et al. (2004b) used their methodology and found a close relationship between broadband NDVI (NDVIb) and Gross Primary Production (GPP) in a Scots pine forest (Pinus sylvestris L.). Compared with satellite remote-sensing, broadband vegetation indices are independent of prevailing weather at the ground surface and, therefore, can automatically monitor vegetation canopies over the seasonal course of development (Wang et al., 2004b).

In this paper, we have tried to improve our understanding of the correlations between vegetation indices and leaf physiological parameters further. We have investigated the correlations in photosynthetic-capacity changes and in vegetation indices obtained from different sources, including satellite-borne sensors, tower-mounted meteorological sensors, and a field portable spectroradiometer. The vegetation indices include MODIS satellite 250 m NDVI and Enhanced Vegetation Index (EVI); tower hyperspectral data-based narrowband NDVI and SR; tower hyperspectral data-based “simulated MODIS NDVI and EVI”; and meteorological data-based broadband NDVI and SR. The measurements were made at three beech forest sites (Fagus crenata) at different altitudes in the cold-temperate mountain zone of Japan. The main objective of the paper was to determine which vegetation index and data source can be most closely linked with changes in photosynthetic parameters and, therefore, offer potentials to aid in improving gas-exchange models. A comparison of the strengths and weakness of each of the vegetation indices is reported.

Section snippets

Research region

The study was carried out in natural beech forest (F. crenata) stands on the northern slope of Naeba Mountains in Japan (36º51′N, 138º40′E, see Fig. 1 for details). F. crenata is dominant in forest ecosystems from 550 m to 1500 m above sea level at this location in the cold temperate climate zone of Japan. In the Naeba Mountains, 15 plots plus four core sites with tower infrastructure have been observed since the 1970s with over 35 years of continuous data on stand biomass, leaf area index (LAI),

Seasonal patterns of photosynthetic parameters (Vcmax, Jmax)

Distinctive inverted parabolic patterns in seasonal trajectories were noted for photosynthetic parameters at all three sites over the growing season, with high values during mid-growing-season and low values at both the beginning and end of the growing season (Fig. 2A and B). Even though the distinctive patterns were site specific, three common stages can nevertheless be deduced from the patterns:

  • 1)

    Spring leaf-expansion period: This period occurred at the beginning of the growing season,

Spectral, spatial and temporal upscaling

Vegetation indices (from multiple sources) utilized in this study represent multiple scale information of spectral, spatial and temporal reflectance. If narrowband vegetation indices were treated as baseline data representing simultaneous and in situ reflectance of photosynthetic-capacity measurements, then simulated MODIS time-series data would represent an upscaled spectral outcome, and downloaded MODIS data would be further upscaled in the spatial context. Compared with narrowband vegetation

Conclusion

Canopy reflectance of deciduous forest is jointly governed by physiological (photosynthetic activity) and physical (LAI) factors. The effect of photosynthetic activity on reflectance has been overlooked for a long time, since previous research was biased by only linking canopy reflectance to physical factors. Our research results suggest that there is a close correlation between photosynthetic capacity and canopy reflectance, which may be applied inversely to derive photosynthetic parameters

Acknowledgments

We thank Prof. Y. Kakubari and C. Irie and other members of the Institute of Silviculture, Shizuoka University for their field supports. This research was supported by the JSPS Grant-in-Aid for Young Scientists (A) (Grant no. 18688007) to Quan Wang.

References (48)

  • A.A. Gitelson et al.

    Signature analysis of leaf reflectance spectra: algorithm development for remote sensing of chlorophyll

    Journal of Plant Physiology

    (1996)
  • A. Huete et al.

    Overview of the radiometric and biophysical performance of the MODIS vegetation indices

    Remote Sensing of Environment

    (2002)
  • R.B. Myneni et al.

    Remote Sensing of vegetation canopy photosynthetic and stomatal conductance efficiencies

    Remote Sensing of Environment

    (1992)
  • J. Ross et al.

    Sources of errors in measurements of PAR

    Agriculture and Forest Meteorology

    (2000)
  • S.W. Running et al.

    Relating seasonal patterns of the AVHRR vegetation index to simulated photosynthesis and transpiration of forests in different climates

    Remote Sensing of Environment

    (1988)
  • P.J. Sellers et al.

    Canopy reflectance, photosynthesis, and transpiration. III. A reanalysis using improved leaf models and a new canopy integration scheme

    Remote Sensing of Environment

    (1992)
  • C.J. Tucker

    Red and photographic infrared linear combinations for monitoring vegetation

    Remote Sensing of Environment

    (1979)
  • W.J.D. van Leeuwen et al.

    MODIS vegetation index compositing approach: a prototype with AVHRR data

    Remote Sensing of Environment

    (1999)
  • Q. Wang et al.

    Similarities in ground- and satellite-based NDVI time series and their relationship to physiological activity of a Scots pine forest in Finland

    Remote Sensing of Environment

    (2004)
  • Q. Wang et al.

    On the relationship of NDVI with leaf area index in a deciduous forest site

    Remote Sensing of Environment

    (2005)
  • G.B. Bonan

    Land–atmosphere CO2 exchange simulated by a land surface process model coupled to an atmospheric general circulation model

    Journal of Geophysical Research

    (1995)
  • G.B. Bonan

    The land surface climatology of the NCAR Land Surface Model coupled to the NCAR Community Climate Model

    Journal of Climate

    (1998)
  • A. Brooks et al.

    Effects of temperature on the CO2/O2 specificity of ribulose-1, 5-bisphosphate carboxylase/oxygenase and the rate of respiration in the light

    Planta

    (1985)
  • J.C. Eidenshink et al.

    The 1 km AVHRR global land data set: first stages in implementation

    International Journal of Remote Sensing

    (1994)
  • Cited by (19)

    • Diverse photosynthetic capacity of global ecosystems mapped by satellite chlorophyll fluorescence measurements

      2019, Remote Sensing of Environment
      Citation Excerpt :

      Currently, Earth System Models (ESMs) heavily rely on meta-analysis based Vcmax databases which are derived from the correlation between leaf N content and Vcmax (Kattge et al., 2009). Vcmax also has strong seasonal variations (Jin et al., 2012; Wang, 2008; Wang et al., 2004; Wang et al., 2009a, 2009b). The value of Vcmax is often set as a constant for each plant function type (PFT) in terrestrial biosphere models, and the uncertainty in Vcmax has hampered our understanding of global carbon and water cycles (Bonan et al., 2012; Harper et al., 2016; L. M. He et al., 2017; Madani et al., 2018).

    • Photochemical reflectance ratio for tracking light use efficiency for sunlit leaves in two forest types

      2017, ISPRS Journal of Photogrammetry and Remote Sensing
      Citation Excerpt :

      However, large variations of Vcmax within the same PFT have been widely observed (Kattge et al., 2009; Groenendijk et al., 2011a, 2011b). In order to acquire more accurate Vcmax for carbon cycle simulation over large areas, multiple attempts have been made to estimate Vcmax using remote sensing indices (Wang et al., 2008; Jin et al., 2012; Zhou et al., 2014) through the parametric regression methods (Verrelst et al., 2015). However, the relationship between Vcmax and VIs seems to vary from site to site, year to year, and even month to month.

    View all citing articles on Scopus
    View full text