517
Soil and Vegetation Moisture Changes for Forest Fire Events
worse drought event than the one in 1954, which was identified as the “Drought of
the Century” (http://southwestfarmpress.com/peanuts/080707-peanut-crop/).
21.4.1.2 Methods
In addition to NMDI, MODIS NIR band 2 (0.86 µm) was combined with SWIR
bands 5 (1.24 µm), 6 (1.64 µm), and 7 (2.13 µm), respectively, to derive vegetation
water indices according to the following equations:
NDWI 2
m
m
m
m
,
.
.
.
.
5
0 86
1 24
0 86
1 24
=
−
+
R
R
R
R
µ
µ
µ
µ
(21.3)
NDWI 2
m
m
m
m
,
.
.
.
.
6
0 86
1 64
0 86
1 64
=
−
+
R
R
R
R
µ
µ
µ
µ
(21.4)
NDWI 2
m
m
m
m
,
.
.
.
.
7
0 86
2 13
0 86
2 13
=
−
+
R
R
R
R
µ
µ
µ
µ
,
(21.5)
where R represents the reflectance at the wavelengths denoted by the subscripts. By
combining information from NIR and SWIR channels, these indices are expected
to be positively correlated with vegetation moisture conditions: the higher the index
values, the wetter the vegetation (Wang and Qu 2007; Wang et al. 2008).
The relatively wetter and drier years were first identified using the 7-year time
series of PDSI and the percentage area under droughts. These results were then
compared with drought conditions represented by MODIS-derived vegetation water
indices to evaluate the capability of each index for drought monitoring. Since the
PDSI and the percentage area under droughts were area-average values, the MODISderived vegetation water indices were averaged over the entire study area in the comparison. To further assess the drought monitoring performance of each water index,
a scaled index has been employed to examine the index sensitivity corresponding to
the wet and dry conditions. Finally, monthly fire activity data, including fire number and burned area, were used to investigate the relation between drought and fire
occurrences.
21.4.2 ReSultS and diScuSSion
21.4.2.1 Vegetation Water Indices and Drought Condition Analysis
Time series of monthly PDSI and the percentage of area under severe drought categories of D2–D4 from 2001 to 2007 are shown in Figure 21.6a and b, respectively.
Both figure panels illustrate that Georgia experienced the most severe drought in
2007: the averaged monthly PDSI values are less than –3.0 from April to December,
and starting in May, more than 50% of areas were under the severe drought category
of D2–D4. Following closely behind 2007 as the driest year, 2002 and 2006 are
also relatively dry, represented by lower PDSI values and many areas under severe
Soil and Vegetation Moisture Changes for Forest Fire Events
worse drought event than the one in 1954, which was identified as the “Drought of
the Century” (http://southwestfarmpress.com/peanuts/080707-peanut-crop/).
21.4.1.2 Methods
In addition to NMDI, MODIS NIR band 2 (0.86 µm) was combined with SWIR
bands 5 (1.24 µm), 6 (1.64 µm), and 7 (2.13 µm), respectively, to derive vegetation
water indices according to the following equations:
NDWI 2
m
m
m
m
,
.
.
.
.
5
0 86
1 24
0 86
1 24
=
−
+
R
R
R
R
µ
µ
µ
µ
(21.3)
NDWI 2
m
m
m
m
,
.
.
.
.
6
0 86
1 64
0 86
1 64
=
−
+
R
R
R
R
µ
µ
µ
µ
(21.4)
NDWI 2
m
m
m
m
,
.
.
.
.
7
0 86
2 13
0 86
2 13
=
−
+
R
R
R
R
µ
µ
µ
µ
,
(21.5)
where R represents the reflectance at the wavelengths denoted by the subscripts. By
combining information from NIR and SWIR channels, these indices are expected
to be positively correlated with vegetation moisture conditions: the higher the index
values, the wetter the vegetation (Wang and Qu 2007; Wang et al. 2008).
The relatively wetter and drier years were first identified using the 7-year time
series of PDSI and the percentage area under droughts. These results were then
compared with drought conditions represented by MODIS-derived vegetation water
indices to evaluate the capability of each index for drought monitoring. Since the
PDSI and the percentage area under droughts were area-average values, the MODISderived vegetation water indices were averaged over the entire study area in the comparison. To further assess the drought monitoring performance of each water index,
a scaled index has been employed to examine the index sensitivity corresponding to
the wet and dry conditions. Finally, monthly fire activity data, including fire number and burned area, were used to investigate the relation between drought and fire
occurrences.
21.4.2 ReSultS and diScuSSion
21.4.2.1 Vegetation Water Indices and Drought Condition Analysis
Time series of monthly PDSI and the percentage of area under severe drought categories of D2–D4 from 2001 to 2007 are shown in Figure 21.6a and b, respectively.
Both figure panels illustrate that Georgia experienced the most severe drought in
2007: the averaged monthly PDSI values are less than –3.0 from April to December,
and starting in May, more than 50% of areas were under the severe drought category
of D2–D4. Following closely behind 2007 as the driest year, 2002 and 2006 are
also relatively dry, represented by lower PDSI values and many areas under severe