The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
2.2 Image pre-processing
Landsat 7 Enhanced Thematic Mapper Plus (ETM+) image
(Oct 13, 2002) and Landsat 5 TM images (Oct 26, 1988; Oct 23,
1991; Oct 4, 1996) were used in this research. The data
acquisition date has a highly clear atmospheric condition. All
images bands 1-5 and 7 have a spatial resolution of 30m, and
the thermal infrared band (band6) has a spatial resolution of
120m for Landsat 5 TM images and 60m for Landsat 7 ETM+
images. The Landsat image were further rectified to a common
Universal Transverse Mercator coordinate system based on
1:24,000 scale topographic maps, and were resampled using the
nearest neighbor algorithm with a pixel size of 30 by 30 m for
all bands including the thermal band. The resultant RMSE was
found to be less than 0.5 pixels.
2.3 Derivation of LST from Landsat TM/ETM+ imagery
The LST were derived from the corrected TM/ETM+ TIR band.
The local time of satellite overpasses was in the morning
(approximately 11:00 AM) (this was the best image available),
so that the chance for detecting a weaken UHI is maximized.
The following equation was used to convert the digital number
(DN) of Landsat TM/ETM+ TIR band into spectral
radiance[12].For Landst-5 TM, Chen at al proposed a method
of deriving brightness temperature. First, the digital numbers
(DNs) of band6 are converted to radiation luminance
( R tm6 , m W * cm * sr ) by fo e following formula:
R TM6 =0.0068337 xDN + 0.1534 (1)
The next step is to convert the radiance luminance to at-satellite
brightness temperature (i.e., blackbody temperature, TB) under
the assumption of uniform emissivity. The conversion formula
is:
T =
\n(K 2 /L x +l)
(3)
K, =1282.71/:
Where, 1
K 2 = 666.09(mW * cm" 2 * sr~' * jum~')
constants; and ^ is the spectral
mW * cm ' * sr ' * /um '
and
are calibration
radiance in
2.4 Estimation of vegetation fraction using Dimidiate
Pixel Model
Dimidiate Pixel Model [13-15] assumes that a pixel consists of
two components: pure vegetation and non-vegetation, so the
reflectance of any pixel can be presented as follows:
R = R+R„
(4)
Where v is the reflectance of pure vegetation while 5 is
the reflectance of non-vegetation.
We assume that the vegetation coverage proportion of a pixel is
f c , that is the vegetation fraction, then the non-vegetation
1— fc
coverage proportion of the pixel is J . If the whole pixel
is covered by vegetation, the reflectance we gain is veg ; if it
R R
has no vegetation coverage, the reflectance is sml , so v
and s of a mixed pixel can be presented as a product of
Rveg and (Eq(5)), Rsoil and l ~f C (Eq(6)),
respectively:
T =
1 B
K,
ln( Kl
R TM J b
+ 1)
(2)
where TB is effective at-satellite temperature in K;
K K
1 and 2 are pre-launch calibration constants.
K x =1260.56K
and ^,‘60J66(mW * cm'’ * sr 1 * Jim' 1 ) _ b npKstnts
effective spectral range, when the sensor’s response is much
, no/ b = 1239(jum)
more than 50%, .
For Landsat-7 ETM+, it is also simplified to two separate steps.
First, the DNs of band6 were converted to radiance by the
following formula:
Radiance = gain * DN + offset . Then the effective
at-satellite temperature of the viewed Earth-atmosphere system
under the assumption of a uniform emissivity could be obtained
from the above spectral radiance by the following equation:
R v =fc*R veg
R s =(\-fc)*R soil
(5)
(6)
Through computing Eq (4), Eq (5) and Eq (6) together, we
acquire the equation of calculating vegetation fraction as
follows:
fc = (R-R soil )/(R veg - R soil ) (?)
R R
Where soil and vegi are two key parameters of dimidiate
pixel model. Obviously, if we get these two parameters, we can
compute the vegetation fraction by using remote sensing
information through eq (7).
ND VI j s fo e indication factor of vegetation growth state and
the vegetation. According to dimidiate pixel model, we can
express the NDV1 0 f ea( fo pixel as equation (8):
1380