The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
summer maize were extracted from the Spectral Database
System of Typical Objects in China, while grass and forest
datasets were obtained using Computer Simulation Model.
Since the ground based minimum NDVI value is higher the
minimum NDVI value in remote sensing scale, we advanced a
simpler semi-empirical relationship between NDVI and LAI as
shown in Table 3. The advanced arithmetic for transformation is
seen in Equation (3)
VI = VI œ -(F/qo-0)exp(rKyjLAI) (3)
where VI^ and K V i are consistent with Equation (2).
The original and advanced semi-empirical regression functions
are shown in Table 3.
After all these work, with the foregoing classification map (Fig.
3), we applied the relationship between NDVI and LAI for
estimating the LAI of different species, and to see how these
LAI change in different crop growth stages. As shown in Figure
4, we could see that the LAI value of crop is the highest in May
14, since the winter wheat grows best in May 14 and are
harvested in June 17. And LAIs value of grass and forest are
the highest in June 17.
The LAI temporal distribution characteristic is shown in Fig.5;
there are several high LAI values in March and June, which
locate in the forest area. In 16-March-06 LAI image, 99% of
LAI values are below 1.6, in 14-May-06 LAI image, 99% of
LAI values are below 1.8, and in 17-June-06 LAI image, 99%
of LAI values are below 4.2. After checking the whole data, we
found that there were a few high NDVI value points in March
and June, that’s why the LAI got high value in those areas.
Linear Regression Function
R 2
F
Sig.
Non-Linear Regression Function
R 2
F
Sig.
LA1=0.397*RVI
0.939**
3077.24
0.000
LAI=0.487*exp(0.236 *RV1)
0.655**
376.101
0.000
LAI=3.599*NDVI
0.902**
1833.453
0.000
LAI =0.218 *exp(3.345 ♦NDVI)
0.661**
385.627
0.000
LAI=T7.201*PVI
0.934**
2816.534
0.000
LAI =0.629*exp(8.554 *PVI)
0.641**
353.635
0.000
LAI=5.439*SAVI
0.927**
2525.258
0.000
LAI =0.290*exp(4.369*SAVI)
0.678**
416.311
0.000
LA1=5.493 *MS AVI
0.938**
3005.292
0.000
LAI =0.361 *exp(3.904*MSA VI)
0.676**
414.030
0.000
Table 1: Regression fimctions between VI (x) and LAI (y).
RVI is Ratio Vegetation Index, NDVI is Normalized Differenced Vegetation Index,
PVI is Perpendicular Vegetation Index, SAVI is Soil Adjusted Vegetation Index, MSAVI is Modified SAVI.
RVI- M . NDVI-SAVI- *«*-* Q + + +
R NIR + R NIR + R + L 2
Regression Function
R 2
SSE
SSY
F
10.67-1.45
-
336.559
280.210
-
LAI =-1.221 *Ln(°™- NDV j
0.83-0.18
0.622
105.983
280.210
325.495
0 29 - PVI
LAI- 2.706 *Ln{ )
0.29 + 0.01
0.402
167.588
280.210
133.059
LA,=-2.m-L„(° M - SAr j
0.66-0.12
0.629
103.991
280.210
335.523
LAI = -2.010 */./?( °'^9 - MSAVI
0.69-0.10
0.496
141.299
280.210
194.654
Table 2: Semi-empirical regression functions between VI (x) and LAI (y)