952 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
CD Bimiillr.if nop 
CZ3 Oiral nop 
m Forest 
I I Qm 
HI > ! fhw 
I Water 
• 
product Fig. 3 The classification map of BJ-1 imaj 
Fig.2 The reprojection image of M0D12Q1 
2) the generation of Beijing-1 LAI maps: LAI can be 
estimated at a variety of spatial scales and with different 
spacebome sensors(Jing M. Chen 1996) using techniques 
ranging from regression models to canopy reflectance model 
inversions with varying success, which includes: (1) statistical 
models that relate LAI to band radiance(G. D. Badhwar 1986; 
Mehul R. Pandya 2006) or development of LAI-vegetation 
index relationships(Jing M. Chen 1996; Mehul R. Pandya 
2006); (2) biophysical models like the Price model; and (3) 
inversion of canopy reflectance using a numerical model or 
LUT-based modeles(Mehul R. Pandya 2006). The empirical 
approach is common in remote sensing of LAI estimation. It is 
based on the idea that foliage reflectance is low in the red 
portion of the spectrum because most is absorbed by 
photosynthetic pigments, whereas much of the NIR is reflected 
by foliage. Then an empirical vegetation index such as 
normalized differences vegetation index(NDVI) based on the 
red and NIR is used to estimated LAI. 
In this paper, we use the experiment data of Winter wheat in 
Shunyi, Beijing and maize in Luan cheng, in Heibei province. 
And the computer simulation model—Radiosity model is used 
to simulate the canopy regime of different LAI for grass and 
forest. Basic on these experiment data and the measured data, 
we can get the statistical relationship of NDVI and LAI, shown 
in Table.3. And using above relationship of NDVI and LAI, we 
can obtain the LAI distribution map of three periods: 16, Mar; 
14, May; 17,Jun, shown in fig.4(Xiao Yueting 2008). 
Type 
Formula 
R2 
Winter Wheat 
NDVI=0.83*(1 -exp(LAI/-1.155)) 
0.623 
maize 
NDVI=0.9*(l-exp(LAI/-2.111)) 
0.638 
Grass 
NDVI=0.87*( 1 -exp(-LAI/0.873)) 
0.961 
Forest 
NDVI=0.9*( 1 -exp(-LAI/l .702)) 
0.564 
Table.3 The relationship of NDVI and LAI 
(a) Winter wheat 
0« 05 \6 15 78 75 38 » 
UM 
(c) Grass 
Tree 
^—H 
t 1 
1 
8 9-(1-*xpt-1*x/1.702)) 
R fc *0 564 
70 t$ ** 35 40 4.5 SO 
(d) Forest 
v.-T A'' 
' * 
10.147 048 
I 0.48 ■ 0.70 
I 0.70 • 096 
1000- 1.t5 
I 1 15-153 
' 1 tS-7 03 
Œw.» 
Fig.5 The generation of LAI Map of Beijing-1 image(Xiao 
Yueting 2008) 
Fig 4. The relationship of LAI and NDVI 
2.2.2 MODIS Products Processing: 
LAI can descript the vegetation growth, and in the growth time 
series, LAI variations of different land cover types have the 
significant difference. MODIS LAI products can offer stably 
time series LAI products, and can obtain the LAI variations of