434
Multiscale Hydrologic Remote Sensing: Perspectives and Applications
hemisphere very well. It also shows the seasonal vegetation growth (November
to February) in the southern hemisphere in Figure 18.7c and the growing season
between April and October in the northern hemisphere in Figure 18.7d.
In an earlier section, several other microwave indices were discussed. Figure 18.8
shows the 2003 polarization difference MPDT (Choudhury and Tucker 1987) in the left
column, the polarization index PI (Becker and Choudhury 1988) in the middle column,
and the MVI B parameters in the right column. The sites used here are the Coordinated
Enhanced Observing Period observation sites of the World Climate Research Program.
The site in Mongolia (http://data.eol.ucar.edu/codiac/dss/id=76.124), 46.55°N and
106.65°E, represents grassland (top row), and the other in Tibet, China (http://
monsoon.t.u-tokyo.ac.jp/camp/tibets/), 34.22°N and 92.44°E, represents shrubland
(bottom row). The green curves from the left to right columns are the field measurements of soil temperature (Mongolia site), soil moistures at 3–4 cm depth, and NDVI
from MODIS, respectively. There were no in situ soil temperature measurements available for the Tibet site; therefore, the 36.5-GHz-brightness-temperature measurements
were used to indicate temperature trend (Owe et al. 2001).
All three microwave vegetation indices MPDT, PI, and B parameters obtained
over the Mongolia site agree with the relations presented in the literature (Choudhury
and Tucker 1987; Becker and Choudhury 1988; Paloscia and Pampaloni 1988; Owe
et al. 2001). Values are generally inversely related to vegetation optical thickness (Figure 18.8, top row). The MVIs B parameters derived in this study showed
consistent characteristics for both sites (top right side and bottom rows of Figure
18.8). However, the MPDT and PI data for the Tibet site (Figure 18.8, bottom row)
showed opposite behavior; they increased as NDVI increased from May to June.
These results indicate that there are large uncertainties when using MPDT and PI
to present vegetation signals. These uncertainties might be explained by recalling
that—MPDT is affected by three factors: the surface emission signal in terms of
polarization difference, the vegetation transmissivity V t , and the surface temperature as shown in Equations 18.13 and 18.4. The other index PI is affected mainly by
the surface emission and vegetation. For a fixed surface emission and temperature
values, MPDT and PI will decrease as V t decreases, since they will have an inverse
relation with vegetation optical thickness (Paloscia and Pampaloni 1988; Owe et al.
2001). However, MPDT and PI will increase as soil moisture increases for a given
vegetation transmissivity. MPDT is also proportional to surface temperature. As a
result, whether MPDT and PI measurements could show an inverse relation with
vegetation optical thickness is dependent on which signal change is the dominant
signal in the time series plots. When the impact of the vegetation signal (optical
thickness) change is greater than that of soil moisture and/or temperature change,
MPDT and PI will show a negative relation with vegetation. This is the case for the
Mongolia site (Figure 18.8, top row). However, the opposite relation can be observed
when the impact of soil moisture and/or temperature change is larger than that of
vegetation change, as shown in bottom row of Figure 18.8 for the Tibet site.
All these results clearly demonstrate that significant new vegetation information
can be provided by the MVIs, because they reflect not only the leafy part of vegetation information but also the woody part of the vegetation information resulting
from the intrinsic differences between what microwave and optical sensors observe.
Multiscale Hydrologic Remote Sensing: Perspectives and Applications
hemisphere very well. It also shows the seasonal vegetation growth (November
to February) in the southern hemisphere in Figure 18.7c and the growing season
between April and October in the northern hemisphere in Figure 18.7d.
In an earlier section, several other microwave indices were discussed. Figure 18.8
shows the 2003 polarization difference MPDT (Choudhury and Tucker 1987) in the left
column, the polarization index PI (Becker and Choudhury 1988) in the middle column,
and the MVI B parameters in the right column. The sites used here are the Coordinated
Enhanced Observing Period observation sites of the World Climate Research Program.
The site in Mongolia (http://data.eol.ucar.edu/codiac/dss/id=76.124), 46.55°N and
106.65°E, represents grassland (top row), and the other in Tibet, China (http://
monsoon.t.u-tokyo.ac.jp/camp/tibets/), 34.22°N and 92.44°E, represents shrubland
(bottom row). The green curves from the left to right columns are the field measurements of soil temperature (Mongolia site), soil moistures at 3–4 cm depth, and NDVI
from MODIS, respectively. There were no in situ soil temperature measurements available for the Tibet site; therefore, the 36.5-GHz-brightness-temperature measurements
were used to indicate temperature trend (Owe et al. 2001).
All three microwave vegetation indices MPDT, PI, and B parameters obtained
over the Mongolia site agree with the relations presented in the literature (Choudhury
and Tucker 1987; Becker and Choudhury 1988; Paloscia and Pampaloni 1988; Owe
et al. 2001). Values are generally inversely related to vegetation optical thickness (Figure 18.8, top row). The MVIs B parameters derived in this study showed
consistent characteristics for both sites (top right side and bottom rows of Figure
18.8). However, the MPDT and PI data for the Tibet site (Figure 18.8, bottom row)
showed opposite behavior; they increased as NDVI increased from May to June.
These results indicate that there are large uncertainties when using MPDT and PI
to present vegetation signals. These uncertainties might be explained by recalling
that—MPDT is affected by three factors: the surface emission signal in terms of
polarization difference, the vegetation transmissivity V t , and the surface temperature as shown in Equations 18.13 and 18.4. The other index PI is affected mainly by
the surface emission and vegetation. For a fixed surface emission and temperature
values, MPDT and PI will decrease as V t decreases, since they will have an inverse
relation with vegetation optical thickness (Paloscia and Pampaloni 1988; Owe et al.
2001). However, MPDT and PI will increase as soil moisture increases for a given
vegetation transmissivity. MPDT is also proportional to surface temperature. As a
result, whether MPDT and PI measurements could show an inverse relation with
vegetation optical thickness is dependent on which signal change is the dominant
signal in the time series plots. When the impact of the vegetation signal (optical
thickness) change is greater than that of soil moisture and/or temperature change,
MPDT and PI will show a negative relation with vegetation. This is the case for the
Mongolia site (Figure 18.8, top row). However, the opposite relation can be observed
when the impact of soil moisture and/or temperature change is larger than that of
vegetation change, as shown in bottom row of Figure 18.8 for the Tibet site.
All these results clearly demonstrate that significant new vegetation information
can be provided by the MVIs, because they reflect not only the leafy part of vegetation information but also the woody part of the vegetation information resulting
from the intrinsic differences between what microwave and optical sensors observe.