bands for crop discrimination using hyperspectral data in VNIR 
region are centred at 547, 675, 718 and 904 nm. 
For studying the crop biophysical parameters and crop variability 
five vegetation indices were generated using the hyperspectral data 
with 15 nm width. Those are presented in table 2. 
Table 2. Formulation for computation vegetation indices using 
hyperspectral data ‘ 
  
  
  
  
  
  
Vegetation Index Formula 
NDVI(552-687) (R552-R687)/(R552+R687) 
NDVI(927-687) (R927-R687)/(R927+R687) 
RVI(552-687) R552/R687 
RVI(927-687) R927/R687 
MSI R1662/R927 
  
  
  
  
NDVI-Normalised Difference Vegetation Index, RVI- Ratio 
Vegetation Index, MSI-Moisture Stress Index 
For studying soil variability through spectral data various 
radiometric indices related to soil colour were computed spectral 
property. These indices, as described in Table3, were adopted 
from Mathieu et al (1998). Since soil olour is influenced by the 
presence of many soil constituents, such as iron oxides, limestone, 
organic matter, and water content (Taylor, 1982), it was assumed 
the colour related indices would properly highlight the soil 
variability in spectral data. For computing colour related indices 
the narrow band spectral values were averaged over the broadband 
regions of blue, green and red, as defined in IRS LISS I sensors. 
(Pandya et al, 2002). 
Table 3. Radiometric indices calculated using hyperspectral data 
for soil property study (adopted from Mathieu et al, 
IAPRS & SIS, Vol.34. Part 7, “Resource and Environmental Monitoring”, Hyderabad, India,2002 
  
  
  
  
  
  
  
  
  
1998) d 
Index Formula Index Property 
Brightness Index, BI | (R>+G*+B?)/3.0)*° | Average reflectance 
magnitude 
Saturation Index, SI (R-B)/(R+B) Spectra Slope 
Hue Index, HI (2*R-G-B)/(G-B) Primary colours 
Coloration Index, CI | (R-G)/(R+G) Soil Colour 
Redness Index, RI R2/(B*G?) Hematite Content 
  
  
  
  
   
   
    
    
3.4 Analysis of Satellite Data 
IRS 1D LISS III (multi-spectral, 23-m resolution) and 
Pan (panchromatic, 5.8-m resolution) data were acquired on 
February 21, 2002. To get combination of high spectral resolution 
of IRS LISS III and high spatial resolution of IRS Pan, both the 
data were fused using the Brovey technique. Four indices were 
computed for cropped field using LISS III data (table 4). For 
studying the soil variability the Brightness index (BNI) is defined 
by Leone et al. (1995). For this study the BNI was computed as 
the sum of the DN values of all the three layers of merged data. 
Table 4. Indices derived from IRS LISS III data. 
Index Formula 
NDVI -R +R 
NDVI GR-R)/(GR+R 
RVI 
MSI 
NIR, R, GR and MIR are DN values in Near Infrared, Red, Green 
and Middle Infrared bands, respectively. 
  
3.5 Statistical analysis and surface mapping 
The statistical analysis of the measured parameters, both 
conventional and spectral, were carried using standard Windows 
based statistical software, such as SPSS for Windows, Releases 
6.0. The statistical analysis included computation of standard 
deviation and coefficient of variation of each parameter to know 
the variability; correlation analysis of spectral and conventional 
parameters; and empirical model fitting. 
Surface maps were generated, for various parameters, using Surfer 
V 7.0, the surface mapping system of Golden Software Inc. The 
surface fitting was carried out using Krigging interpolation 
technique. 
4. RESULTS AND DISCUSSION 
The spectroradiometer provided reflectance curves for 35 
locations for the fallow plot and 30 locations for the wheat 
cropped field. The average reflectance pattern of all these 
locations, separately for soil and crop, are plotted in the figure 2. 
  
  
  
0.60 
——Crop  — Soil 
050 | 
0.40 
9 
© 
= 
3 
oe 030 
© 
e 
9 
~ 
020 
0.10 + 
0.00 T T T T T T T 
300 500 700 900 1100 1300 1500 1700 
  
Wavelength(nm) 
Figure 2. Average spectral reflectance pattern of crop and soil 
classes. 
The reflectance curve showed the typical soil and crop spectra, 
except some erroneous values between 1350-1390 nm due toe 
error in detector response. The average crop spectra showed a 
reflectance peak of 0.053 at 565 nm (green reflectance), dip of 
0.021 at 675 nm (chlorophyll absorption), strong peak of 0.52 at 
865 nm (leaf structure caused reflectance) and a dip of 0.052 at 
1454 nm (water absorption). The soil reflectance consistently 
increased from 0.078 at 355 nm to 0.40 at 1810 nm. 
    
| I» lo: 
[ 
[ 
  
fo 
na 
5C 
so