Plants cover more than 70% of our earth's land surface and play an important role in the
exchange of materials and energy between land and atmosphere. The biochemical
substances existing in leaves have been proven to be critical factors in influencing, or even
controlling, this exchange by means of various physiological processes, including
photosynthesis, evaportranspiration, litter decomposition, etc. Compared to traditional wet
chemistry methods, estimation of their contents through leaf and canopy reflectance spectra
has become a rapid, efficient, and promising method over the past few decades, which had
been physically based on the selective absorption features of a particular substance. This
paper discusses the responses of plant biochemical substances to the reflectance spectra at
both leaf and canopy scales. We use a leaf optical model PROSPECT and a leaf-canopy
coupled optical model PROSAIL to generate large amounts of leaf and canopy spectra.
Both of these models contain three biochemical parameters: cab-chlorophyll concentration,
Cw-equivalent leaf water thickness, and Cm-dry matter concentration. Local and global
sensitivity analysis (SA) methods are used on the simulated spectra to differentiate their
contributions to the outcome spectra at both scales so as to investigate their scale effects.
Results show that the maximum sensitivities of chlorophyll and water are much higher than
that of dry matter at leaf scale, and they tend to decrease at canopy scale because canopy
shape, soil, and incident/reflected geometric factors have a large influence on canopy
spectra. However, the sensitivity of dry matter exceeds the maximum sensitivity of
mesophyll structure parameters at canopy scale and becomes the top contributor at the
beginning of near infrared.