Characterization of water stress and prediction of yield of wheat using spectral indices under varied water and nitrogen management practices

There is a need to characterize the water stress in wheat using suitable indices, which will help us to find out the water stress sensitive period for efficient use of irrigation water. Recently indices based on canopy spectral reflectance, which are non destructive, fast and reliable, are being used effectively to characterize the water stress. A field experiment was carried out during the year 2010-2012 in split plot design with four levels of irrigation (irrigation at 0.4 IW/CPE, 0.6 IW/CPE, 0.8 IW/CPE and 1.0 IW/CPE, IW = 6 cm) as main plot factors and three sources of nitrogen (100% N from urea, 50% N from urea and 50% N from farmyard manure (FYM) and 100% N from FYM) as subplot factors. The objective of the study was to find out the water stress indices best correlated with wheat grain and biomass yield, to determine the optimum growth stage for measurement of water stress indices and to predict the grain and biomass yield of wheat based on water stress indices. The canopy reflectance was measured in the spectral range of 350-2500 nm with 1 nm bandwidth with the help of hand held ASD FieldSpec Spectroradiometer at seven phenostages, viz., crown root initiation (CRI), tillering, booting, flowering, milk, soft dough and harvesting stage. Then different water stress indices were computed as: water index (WI) = R970/R900, normalized water index-1 (NWI-1) = (R970 − R900)/(R970 + R900), normalized water index-2 (NWI-2) = (R970 − R850)/(R970 + R850), normalized water index-3 (NWI-3) = (R970 − R920)/(R970 + R920), normalized water index-4 (NWI-4) = (R970 − R880)/(R970 + R880), where R and the subscript numbers indicate the light reflectance at the specific wavelength (in nm). It was observed that spectral reflectance based water indices recorded at the milk stage, WI and NWI-1 were significantly negatively correlated with the grain yield and NWI-1 and NWI-3 were significantly negatively correlated with the biomass yield of wheat, having maximum correlation coefficients. Validation of regression model based on NWI-1 could account for the maximum 87.5% variation in the observed grain yield and the regression model based on WI could account for maximum 89.2% variation in the observed biomass yield of wheat with minimum root mean square errors. So the regression models based on NWI-1 and WI recorded at milk stage can be successfully used to predict the grain and biomass yield of wheat in advance.