3
THEORY
SRS Sensors
Figure 7: A subset of the data displayed in Figure 6, showing a
single day of NDVI data. Notice the concave pattern that is typical
in diurnal NDVI measurements. The concave pattern is due to
changing sun-sensor-surface illumination geometry throughout the
day.
4. Calculating daily averages, using values acquired only during
the noon hour, significantly reduces the amount of data vari-
ability (Figure 6b). A smoothing algorithm applied to the daily
averages reduces variability even further. In this example, data
were filtered and averaged by time, but you can also use solar
zenith and azimuth angles to filter you data. For example, Ryu
et al. (2012) sampled across a consistent solar elevation angle
(60) each day, ignoring all other values. Using solar zenith an-
gle as a filter ensures that data from each day are collected
under similar sun-sensor-surface illumination conditions.
5. If you are comparing measurements acquired under different
sun-sensor-surface configurations (e.g., comparing PRI mea-
surements made during the morning and afternoon), it may
be necessary to first calculate a bidirectional reflectance distri-
bution function (BRDF). An empirical BRDF model, derived
from NDVI or PRI measurements and canopy-specific param-
14
