Correlation of Spectral Reflectance to Growth in Rice Vegetation
From narrow-band reflectance spectra, broad-band RED and NIR reflectance were computed. The reflectance simulated for RED and NIR for HRV sensors of French SPOT program were derived from the 2-nm-wide narrow-band LI-1800 spectra in bandpasses of 610-680 nm and 790-890 nm. For MSS sensor system of Landsat series, bandpasses in the range of 600-700 nm and 700-800 nm were computed to simulated Band 2 and Band 3, respectively. The averaged values were presented for RED and NIR bands of the broad-band sensors. Linear regression was performed to compare the difference in NDVI between values calculated from single reflectance and band reflectance.
Results and Discussion
For a typical reflectance spectrum, wavebands in the visible region were mostly absorbed by chlorphylls while wavelengths in NIR were reflected significantly because of
the absence of absorption (Gausman st al., 1969). It was shown similar in seasoanl pattern
of reflectance spectrum but was shifted by weather and soil impact (Fig 1). Variations
were found much greater in the early growth and maturing stages than during the long
middle period of the growing season. The greater standard errors graphed in the 1st crop
of 1997 inferring a stronger climate oscillation occuring in the growing period. These
'environmental' and 'growth' effects influencing the time-seqential performance of
canopy reflectance spectra were also found in many field crops (Masoni st al., 1996;
Sinclair et al., 1971; Yang Ko, 1998).
Figure 1: Seasonal variations in reflectance spectra of rice vegetation cover over the cropping seasons of
1996 and 1997. The upper and lower levels are the intervals of standard errors.
Fig. 2 shows that NDVI curve reached the peak near heading, and then fell off as
the growing season continued. It can be accounted for by the seasonal trends of reflectance
at 674 and 764 nm, the increasing or decreasing difference between RED and NIR
reflectance giving a corresponding upward or downstream NDVI. RED reflectance
decreased with time beacause of increased chlorophyll absorption by increased green
vegatation (Tucker et al., 1979). As growth diminished and senescence began, RED
reflectance started to increase. It was reversed for NIR reflectance. The NIR bands are
essentially unabsorbed by foliage and are mostly transmitted and reflected with vegetation
structure (Sinclair st al., 1971). Ashley and Rea (1975) observed the similar results
following the ground foliage production and used it to depict phennological change.
Figure 2: Seasonal changes in leaf area index (LAI) and leaf dry weight and
the normalized difference vegetation index (NDVI) of rice vegetation over the cropping seasons of
1996 and 1997.
The increase and decrease of NDVI of NDVI followed the changes of growth traits (Fig. 2).
It suggests that NDVI is sensitive to the amount of photosynthetically active vegetation
and may be a superior variable for monitoring vegetation. Fig. 3 further shows that
relations of these growth traits and NDVI were height. Changes of NDVI agree well with the cnages of canopy cover (up to LAI of ca. 5.5) and foliage biomass (up to LDW of ca. 240 gm-2), indication a strong association of percent vegetation cover and plant chlorosis with NDVI. Tucker (1979) had a similar result, showing that NDVI was sensitive primarily to the green leaf area of green leaf biomass. Curran (1983) estimated LAI from RED and NIR reflectance measurements by correlating LAI with ratio vegetation index.
Figure 3: The curvilinear relationships between leaf area index (LAI) and leaf
dry weight and the normalized difference vegetation index (NDVI) of rice vegetation over the
cropping seasons of 1996 and 1997.
