The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B6b. Beijing 2008 
converted into at-sensor spectral radiance following the 
equation below (Chander & Markham, 2003): 
L X Grescale * Qcal R rescale 
where Lx=at-sensor spectral radiance in W/(m 2 -srpm) 
Grescaie, 5 reica/e =band-specific rescaling factors 
2cai = qiiantized calibrated pixel value in DNs. 
surface in W/m 2 
a=the surface broadband albedo 
e=land surface emissivity 
£ a =atmospheric emissivity 
(1) o=Stefan-Boltzmann constant 
r a =atmospheric temperature near ground 
surface in K 
r ç =land surface temperature in K 
3.2 Albedo 
The Second Simulation of the Satellite Signal in the Solar 
Spectrum (6S) model was further applied for atmospheric 
correction of bands 1~5 and 7. For the thermal band, at-sensor 
brightness temperature was calculated following Chander & 
Markham (2003): 
According to Liang (2000), the surface broadband albedo can 
be retrieved based on the spectral albedo for Landsat TM data: 
a = 0.356a, + 0.130a 3 
+0.373a 4 +0.085a 5 + 0.072a 7 -0.0018 
(4) 
T = 
1 D 
K. 
ln(AT, / L À +\) 
(2) 
where a,—surface reflectance in band i for Landsat 
TM data 
where 7’ B =at-sensor brightness temperature in K 
AT 1 =607.76W/m 2 -srpm 
K 2 =1260.56K 
Each image was classified into seven types of land cover: forest, 
crop, soil, grassland, high-rise building surface, low-rise 
building surface and water, through the combined method of 
maximum-likelihood and visual interpretation. Cirrus clouds 
and their shadows were masked out for the image of July 6, 
2004. Random samples were utilized to assess the classification 
accuracy, it was found that the resultant classification accuracy 
for the summer and winter images was 83.67% and 86.90%, 
respectively. 
The ground conventional meteorological data including 
atmospheric temperature and relative humidity used in this 
study were acquired from observations of the automated 
weather stations (AWSs) managed by Beijing Meteorological 
Bureau. In order to account altitude, atmospheric temperature at 
each station was interpolated following the routine proposed by 
Kato & Yamaguchi (2005), while the relative humidity was 
interpolated with the Inverse Distance Weighting (IDW) to the 
entire study area. The hourly integrated of the total incoming 
solar radiation (MJ/m 2 ) (sum of the direct solar radiation and 
the downward solar diffuse radiation at the ground surface) 
measured at Beijing Weather Observatory 116°28'E) 
was converted to hourly averaged data (W/m 2 ), which was 
assumed to be constant throughout the study area because of its 
limited extent. 
3. METHODOLOGY 
3.1 Calculation of net radiation 
The net radiation flux can be calculated by (Sheng et al., 2003): 
R n = Æ,(l - a) + ££ a crT a 4 - £crT s 4 (3) 
Where R„=net radiation at the ground surface in W/m 2 
R s =the total incoming radiation at the ground 
3.3 Land surface emissivity 
In this study, the land surface emissivity was estimated 
following the NDVI Thresholds Method (NDVI THM ) proposed 
by Sobrino et al. (2001). The readers are encouraged to refer to 
Sobrino et al. (2004) and Stathopoulou et al. (2007). 
3.4 Land surface temperature (LST) 
Land surface temperature is one of the most important 
parameters in land surface processes. Qin et al. (2001) proposed 
a mono-window algorithm for estimating land surface 
temperature: 
T s =[a(\-C - D) + (b(l-C - D) + C + D)T B -DTJ/C (5) 
C = £T ( 6 ) 
D = (l-r)[l + (l-Or] (7) 
where a=-67.355351 
6=0.458606 
r=the total atmospheric transmissivity of the 
thermal band 
4. RESULTS 
4.1 Validation of the net radiation flux 
The hourly integrated net radiation fluxes measured at Beijing 
Weather Observatory between 10:00am and 11:00am of local 
time at these two dates were converted to hourly averaged data 
(W/m 2 ) and were used to validate the accuracy of the estimated 
net radiation fluxes. The result is displayed in Table 1. It can be 
concluded that the estimation of the net radiation flux reached 
high accuracy. 
202