Hyperspectral Indices, Correlation and Regression Models for Estimating Growth Parameters of Wheat Genotypes

Abstract

A field experiment was conducted for three culti- vars of wheat i.e. PBW 343, PBW 550 and DBW 17 with five nitrogen levels i.e. 0, 75, 125, 175 and 225 kg N/ha in split plot design. Growth parameters like LAI, chlorophyll content, nitrogen, chlorophyll concentration index (CCI) and plant height were recorded. Spectral indices such as Normalized difference Vegetation Index (NDVI), Ratio Vegetation Index (RVI), Moisture Stress Index (MSI), Green Index (GI), Leaf Chlorophyll Index (LCI), Plant Senescence Reflectance Index (PSRI), were computed using the multiband spectral data. Among the different phenological phases of crop growth, best correlations were observed between different growth parame- ters and vegetation indices at booting stage of wheat. The spectral indices selected for the study recorded at booting stage were correlated better with all the crop parameters having 'R 2 ' values varied between 0.68 and 0.83. All the spectral indices were having good correlation (R 2 )r anging between 0.58 and 0.82 for grain yield and 0.63-0.80 for 1000 grain weight, But best correlation were noticed between NDVI and grain yield having R2 value 0.78, NDVI and 1000 grain wt having R 2 value 0.80, LCI and grain yield having R 2 value 0.82, LCI and 1000 grain weight having R2 value 0.79. The LCI and NDVI were having best correlation with grain yield and 1000 grain wt. It was concluded that amongst the various indices selected, Leaf color index (LCI) was the only index which was correlated better, with all the plant growth parame- ters and yield at booting stage of the wheat crop.