In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B
201
Ghulam et al, in 2007, suggested a robust drought index which
takes into account both, the vegetation status and the soil
moisture condition. Therefore, they represented a new index
called MPDI which can be expressed as follows:
MPDI =
flfled+ K fyy/fl ~ Fy(fiv,Red + a fyff/fl)
(1 - f v )№TT
PDI - F V PDI V
(1 ~Fv)
(3)
Where, R v ,R C d and Rv,nir are vegetation reflectances in the red
and near infrared bands which in this study was considered 0.05
and 0.5 respectively (Ghulam, 2007). F v is the percentage of
vegetation cover which can be estimated using different
methods such as neural networks (Carpenter 1999), linear
spectral unmixing (Elmore 2000; Conghe 2005) and vegetation
indices (Baret 1995). In the present research, a semi-empirical
method presented by Baret in 1995, has been used which
expressed as (Baret 1995):
w-q-
' W„-hJ
(4)
Where, VI max and Vlmin are vegetation index for a surface with
100% vegetation (F v =l) and bare soils (F v =0), respectively. The
factor K is a constant value minimizing the estimation of RMSE
which was considered 0.6175 in this study (Ghulam, 2007). The
amount of vegetation index was calculated in this study based
on the most commonly used indices; they are presented in Table
2.
Vegetation Index
Reference
R sVr.123
RVI = —^
Pearson &
Miller. (1972)
R nir + R R eo
Rouse et al.
(1974)
„ f F NIR - a ^RED - a
Vl + a*
Richardson &
Wiegand, (1977)
SAVI = *»«-*“» (l+L)
*\Vi£ + Fred + L
Huete et al.
(1988)
^ a i R mR - g Preb ~ ß)
Fred + - ß)
Baret et al.
(1989)
Qi et al. (1994)
wr + 1 “ <J + fT _ _ R&ed)
2
SAVI 2 = P ' N!R
Fred + ( a /P)
Major et al.
(1990)
Rred, Rnir denotes reflectance in NIR and RED wavelengths
a and (3 are the soil line coefficients
Table 2. Vegetation indices Used in the study
4. RESULTS AND CONCLUSIONS
The results of run-test method indicated that a severe drought
with the magnitude of -134.36, the intensity of -67.17, the
severity of -121.98 and for the period of 2 years had occurred in
2000 and 2001 (Table. 3). On the other hand, meteorological
data showed that in April and May when the crops in the study
area are at the peak of the growing season and need enough
water, monthly rain fall had decreased considerably. Water
shortage causes a decrease in the vegetation cover as well as in
the seeding of crops. Consequently, the seed bank of the
rangeland would be decreased which will result in diminishing
of the F v in the following year (Jangjoo, 2001). Although in
2002, compared to previous years, the rainfall increased
considerably (Table 1) it did not affect the percentage of
vegetation cover due to the seed shortage in the rangeland.
Year
Annual rainfall
X-Xo
Drought
status
1999
405.5
145.1956
Wet
2000
134
-126.304
Dry
2001
242
-18.3044
Dry
2002
458.97
198.6656
Wet
Table 3. Drought status in the studying years
Changes in the vegetation cover (F v ) of the study area during
the time period of 1999 to 2002 has been evaluated using the
commonly used vegetation indices (Table. 2) and the equation
4. Among the indices used, only SAVI2 showed the F v had
decreased except in high-land areas where vegetation had no
visible changes (Figure 2). In the high-land areas, there is a
certain supply of snow until the end of the vegetation growing
season and it provides enough water for the vegetation.
Consequently, F v did not change in these areas.
1:200,000
Figure 2. Fraction of vegetation changes (Using SAVI2 and
Equation 4)
The study area, since is located in the arid areas of Iran, has
sparse vegetation cover and the soil background has a
noticeable effect on the recorded reflectance by the sensor.
Therefore, RVI and NDVI which are only based on the Red and
NIR reflectance are not appropriate indices for assessing
vegetation in the arid areas. The results of using these two
indices in the current study confirmed this fact and presented
vegetation changes as mixed pixels (Figure 3).
