The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
455
changing information of the five crops canopy reflectance is
similar with others’ work.
2002 Early №ce 2002 Late Rice
2004 Rap«
2003 Maze
2003 cotton
2004 cotton
2003 Rice
2004 Rape
Wavelengtti /pm
2003 Cotton
2001 Winter Wheat
2004 Cotton
Wavelength /pm
2004 Winter 'Wheat
2004 Winter Wheat
05 ).o t s
Wavelength /pm
Figure 1: Statistical analysis of crop leaf reflectance
2003 Maàze
Figure 2: Statistical analysis of crop canopy reflectance
I Forest
Water
Unused
Winter wheat
Summer Maize
City
Figure 3: Study area classification map
Baret and Guyot put forward a semi-empirical regression
function between VI and LAI, as shown in Equation (2)
VI = VI ж -(VI X - VI g )exp(~K VI LAI) (2)
where VI w is the asymptotically limiting value of a specific VI
when LAI approaches a very large value; VI g is the index value
corresponding to bare soil conditions (LAI = 0). The dynamic
range of the VI (i.e., VI^ -VI g ) is the difference between its
maximum ( VI) and minimum value (VI g ). K V i is the
absorption-scattering coefficient that determines the sensitivity
of the VI to a unit increase of LAI (Baret, F. 1991).
4.2 “Beijing-1” LAI image
Table 1 lists the linear and non-linear regression functions
between LAI and several Vis created by spectral reflectance in
broad red and near IR bands of winter wheat. The best linear
regression function is the relationship between RVI and LAI
(R 2 =0.939), while the best non-linear regression function was
relationship between SAVI and LAI (R 2 =0.0.678). The
parameters in different VI are: PVI (a=0.96916, b=0.084726),
SAVI (L=0.5).
Table 2 lists the semi-empirical regression function between
LAI and VI created with broad red and near IR bands of winter
wheat. The best semi-empirical regression function is between
SAVI and LAI. However, the L parameter of SAVI is simplified
to 0.5, we couldn’t prove its applicability for the “Beijing-1”
images, and then we just used the NDVI instead since the
NDVI-LAI relationship is the best in relationships between
RVI-LAI, PVI-LAI and MSAVI-LAI except SAVI-LAI
relationship.
With the same method, we got the semi-empirical regression
functions for summer maize, grass and forest. The datasets of