will invariably persist at subpixel scales in satellite imagery, given the coarse
nature of cloud detection algorithms and aerosol optical depth determinations.
High aerosol pixels will generally not be corrected to the same accuracy as low
aerosol pixels, and often there will be under- or over-estimations (and corrections)
for aerosols. Kaufman et al. (2005) noted difficulties in aerosol correction attributed to residual cirrus as well as defining cloud contamination versus aerosol
growth. Accuracy and precision values for the MODIS VIs are generally within
0.02–0.05 VI values (http://landval.gsfc.nasa.gov/ProductStatus.php?ProductID=
MOD13, accessed November 2013).
1.4.1 Ground Observation Networks
In-situ observation networks are evolving into highly calibrated and traceable
sensor systems that offer great potential in providing in situ temporally continuous
data to complement field and airborne validation data acquired as ‘spatial’ snapshots in time. These networks can provide higher quality (finer spatial and temporal resolution) measurements than the satellite, and can facilitate various
methods of quality, uncertainty, and cross-sensor continuity assessments involving
multiple satellites. They include the AERONET-based Surface Reflectance Validation Network (ASRVN), the baseline surface radiation network (BSRN), various
phenology networks, and FLUXNET. These provide in situ measurements of
surface and atmosphere conditions that enable independent assessments of vegetation canopy states and phenophases.
The ASRVN collects operational satellite data from over 100 AERONET sites,
equipped with automated sun photometers that provide atmospheric properties to
enable an independent rigorous atmospheric correction of satellite measurements
and allow computed VIs to be compared against their equivalent satellite products
(Wang et al. 2009; Holben et al. 1998). Compared to ground-based, aircraft, and
finer resolution satellite measurements, this approach has the advantage of identical spectral, spatial, and observation geometry sampling consistent with the
satellite measurement. The sun-photometer based atmosphere correction is
applicable to relatively large areas around an AERONET site for more rigorous
spatial analysis. Further, ASRVN also outputs top-of-canopy bidirectional reflectances that provide realistic assessments of accuracies and uncertainties in VIs due
to variability in atmosphere and sensor observation view angles.
In Fig. 1.7a, in situ VI measures from ASRVN results are compared with
simultaneous day MODIS VI values, while in Fig. 1.7b, the nadir view angle
ASRVN in situ results are compared with the MODIS 16-day composited VI
product that attempts to retrieve near-nadir values, hence such a comparison
allows one to ascertain the accuracy and uncertainty of such retrievals. These
comparisons are applicable to all compositing schemes, including the assessment
of NBAR nadir-view retrievals. The ASRVN data is not completely independent,
however, in that the same sensor is used to generate both the VI product and the
1 Indices of Vegetation Activity
13
nature of cloud detection algorithms and aerosol optical depth determinations.
High aerosol pixels will generally not be corrected to the same accuracy as low
aerosol pixels, and often there will be under- or over-estimations (and corrections)
for aerosols. Kaufman et al. (2005) noted difficulties in aerosol correction attributed to residual cirrus as well as defining cloud contamination versus aerosol
growth. Accuracy and precision values for the MODIS VIs are generally within
0.02–0.05 VI values (http://landval.gsfc.nasa.gov/ProductStatus.php?ProductID=
MOD13, accessed November 2013).
1.4.1 Ground Observation Networks
In-situ observation networks are evolving into highly calibrated and traceable
sensor systems that offer great potential in providing in situ temporally continuous
data to complement field and airborne validation data acquired as ‘spatial’ snapshots in time. These networks can provide higher quality (finer spatial and temporal resolution) measurements than the satellite, and can facilitate various
methods of quality, uncertainty, and cross-sensor continuity assessments involving
multiple satellites. They include the AERONET-based Surface Reflectance Validation Network (ASRVN), the baseline surface radiation network (BSRN), various
phenology networks, and FLUXNET. These provide in situ measurements of
surface and atmosphere conditions that enable independent assessments of vegetation canopy states and phenophases.
The ASRVN collects operational satellite data from over 100 AERONET sites,
equipped with automated sun photometers that provide atmospheric properties to
enable an independent rigorous atmospheric correction of satellite measurements
and allow computed VIs to be compared against their equivalent satellite products
(Wang et al. 2009; Holben et al. 1998). Compared to ground-based, aircraft, and
finer resolution satellite measurements, this approach has the advantage of identical spectral, spatial, and observation geometry sampling consistent with the
satellite measurement. The sun-photometer based atmosphere correction is
applicable to relatively large areas around an AERONET site for more rigorous
spatial analysis. Further, ASRVN also outputs top-of-canopy bidirectional reflectances that provide realistic assessments of accuracies and uncertainties in VIs due
to variability in atmosphere and sensor observation view angles.
In Fig. 1.7a, in situ VI measures from ASRVN results are compared with
simultaneous day MODIS VI values, while in Fig. 1.7b, the nadir view angle
ASRVN in situ results are compared with the MODIS 16-day composited VI
product that attempts to retrieve near-nadir values, hence such a comparison
allows one to ascertain the accuracy and uncertainty of such retrievals. These
comparisons are applicable to all compositing schemes, including the assessment
of NBAR nadir-view retrievals. The ASRVN data is not completely independent,
however, in that the same sensor is used to generate both the VI product and the
1 Indices of Vegetation Activity
13