Research ArticleENVIRONMENTAL SCIENCE

Canopy near-infrared reflectance and terrestrial photosynthesis

See allHide authors and affiliations

Science Advances  22 Mar 2017:
Vol. 3, no. 3, e1602244
DOI: 10.1126/sciadv.1602244
  • Fig. 1 SIF relates to NIRV through surface vegetated fraction.

    (A) The correlation between NIRT and SIF increases with vegetated fraction. The upper bounds of the NDVI quartiles are as follows: 0.17, 0.27, 0.37, and 0.72. (B) NIRV closely proxies multiyear monthly averaged SIF. All data calculated using 2008–2010 GOME-2 data averaged monthly and regridded to 0.5°. Shading indicates the logged number of pixels within each bin.

  • Fig. 2 Comparison of multiyear monthly mean (A) SIF and (B) NIRV against global data-driven GPP estimates.

    SIF estimates come from GOME-2 data averaged monthly and regridded to 0.5°. MODIS NIRV estimates were aggregated to 0.5° from 500-m scenes of BRDF-corrected reflectances. GPP estimates come from the Max Planck Institute upscaling approach (16). Shading indicates the logged number of pixels within each bin.

  • Fig. 3 Convergence of scaling between NIRV and GPP.

    (A) For each site (two represented here), we fit a linear regression against average monthly GPP and the monthly value of MODIS NDVI, fPAR, GPP, and NIRV (shown). (B) NIRV explains more of the variation in monthly observed GPP than MODIS NDVI and fPAR. There is no significant difference in the performance of NIRV and MODIS GPP across 105 FLUXNET sites. Whiskers denote the 5th and 95th percentile of site-level R2, with markers indicating outlying sites. (C) Slope parameter of site-level regressions (normalized between 0 and 1) of annual-average monthly NIRV and annual-average monthly GPP from FLUXNET sites. Black lines represent the slope parameters of individual sites.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/3/3/e1602244/DC1

    text S1. Sellers’ two-stream approximation

    text S2. 2D reflectance model

    text S3. NIRV and SIF simulations using SCOPE

    table S1. Parameter values for two-stream approximation model.

    table S2. Model parameters used in 2D reflectance model.

    table S3. Parameter ranges used in 2D reflectance model sensitivity test.

    table S4. SCOPE parameter values.

    table S5. Multiyear monthly average NIRV, fPAR, and GPP, by land cover classification for 105 FLUXNET sites.

    fig. S1. NIRV more strongly predicts canopy fPAR than does NDVI.

    fig. S2. NIRV as a function of NDVI and NIRT.

    fig. S3. Linear relationship between modeled NIRV and SIF.

    fig. S4. Comparison of the monthly MODIS-derived (A) NIRT and (B) NIRV against GOME-2 measurements of SIF.

    References (37, 38)

  • Supplementary Materials

    This PDF file includes:

    • text S1. Sellers’ two-stream approximation
    • text S2. 2D reflectance model
    • text S3. NIRV and SIF simulations using SCOPE
    • table S1. Parameter values for two-stream approximation model.
    • table S2. Model parameters used in 2D reflectance model.
    • table S3. Parameter ranges used in 2D reflectance model sensitivity test.
    • table S4. SCOPE parameter values.
    • table S5. Multiyear monthly average NIRV, fPAR, and GPP, by land cover classification for 105 FLUXNET sites.
    • fig. S1. NIRV more strongly predicts canopy fPAR than does NDVI.
    • fig. S2. NIRV as a function of NDVI and NIRT.
    • fig. S3. Linear relationship between modeled NIRV and SIF.
    • fig. S4. Comparison of the monthly MODIS-derived (A) NIRT and (B) NIRV against GOME-2 measurements of SIF.
    • References (37, 38)

    Download PDF

    Files in this Data Supplement:

Navigate This Article