Geoscience Reference
In-Depth Information
(Balzter et al., 2007b) and describes the vertical location from which most of the
back scattered radiation originates. Canopy gaps smaller than the radar resolution
cell will infl uence the interferometric height of the scattering phase centre, which
will be a measurement of the area-weighted average of the canopy top and the
proportion of open ground seen by the radar (Hagberg et al., 1995). Some studies
have exploited the wavelength-dependent penetration depth of microwaves to esti-
mate canopy top and underlying terrain height, e.g., with the airborne dual-wave-
length SAR system TOPOSAR containing a single-pass X-band and a repeat-pass
P-band sensor system (Andersen et al., 2004), or from the single-pass X-band and
repeat-pass L-band E-SAR system (Balzter et al., 2007b). The method of polarimet-
ric interferometry (Cloude et al., 2001) makes use of different polarisations to esti-
mate contributions of different scattering mechanisms and of interferometry to
locate the scattering phase centre heights of these mechanisms. A limitation of this
method is that it relies on suffi cient scattering phase centre separation to derive
canopy height estimates.
An example of a large-scale mapping project using two wavelengths (C-band
interferometry from ERS-1 and 2, and L-band backscatter from JERS-1) is the
Siberian forest cover map (Balzer et al., 2002b), which classifi ed open water, smooth
surfaces (agriculture) and four forest growing stock volume classes (
<
20, 20-50,
50-80 and
80 m
3
/ha) over an area of about a million km
2
at 50-m spatial resolu-
tion. The map was used in operational forest management by over 40 Russian forest
enterprises.
>
Vegetation phenology
The seasonal cycle of vegetation greening during the photosynthetically active
growing seasons followed by dormancy during winter or the dry season is com-
monly described as vegetation phenology. Obtaining spatially explicit quantitative
information on vegetation phenology is important because the year-to-year differ-
ences in the carbon fl ux from terrestrial metabolism have almost been as large as
variations in the growth rate of atmospheric CO
2
concentrations (Houghton, 2000).
A statistical framework for the analysis of AVHRR time-series data was presented
by de Beurs and Henebry (2005). The same authors relate changes in NDVI to
annual growing degree days (GDD), a statistic which is defi ned as the average of the
daily maximum and minimum temperatures compared to a base temperature. GDD
are linked to the metabolic activity of plants and can indicate the onset of fl owering
and other phenological events. A number of studies have found that phenological
indicators from remote sensing are correlated with climatic indicators. For the
Amazon basin, the seasonal amplitude (amount of temporal change within one
season) of NDVI was observed to increase during El Niño periods with concurrent
low rainfall anomalies, and to decrease during wet La Niña episodes (Asner et al.,
2000). For Europe, spring phenology correlates with winter temperature anomalies
and the winter North Atlantic Oscillation (NAO) index (Stockli and Vidale, 2004).
For Siberia, the timing of leaf appearance in spring shows a strong correlation with
sea surface temperatures over the equatorial Pacifi c of the previous summer, which
are related to El Niño-Southern Oscillation patterns (Vicente-Serrano et al., 2006).
Delbart et al. (2005) studied the timing of the onset of greening-up and leaf
senescence over Central Siberia and compared three spectral indices from the SPOT-
VEGETATION sensor. They conclude that in the boreal biome, NDVI-based
