Assessing the utility of GreenSeeker® based Normalized Difference Vegetation Index (NDVI) for weed variability differentiation in Bt Cotton

An experiment was conducted to assess the applicability of GreenSeeker® based Normalized Difference Vegetation Index (NDVI) for weed variability identification in Bt cotton. Various weed management practices consisting of combinations of hand weeding, inter-cultivation, pendimethalin, quizalofop-ethyl, pyrithiobac-Na including a weedy check and weed free check were used as treatments. Temporal and spatial variability of weed densities was noticed due to treatment effects in respective plots. NDVI value was calculated using a handheld GreenSeeker® for crop plants as well as weeds. All statistical comparisons were tested at the 0.05 level of significance (p < 0.05) using appropriate statistical methods. Results indicated that apart from weed free check the treatment with pre-emergence application of pendimethalin 678 g a.i. ha-1 + one hand weeding at 20 DAS and inter-cultivation two times at 45 and 60 DAS recorded significantly less weed population than other treatments at 20 (2.90 m-2), 40 (3.95 m-2), 60 DAS (2.49 m-2) and at harvest (4.02 m-2). It also recorded significantly higher NDVI for cotton at 60 (0.72) and 75 DAS (0.73). The NDVI values calculated for Bt cotton could not differentiate the effect of weed management practices in early stage and showed significant difference only at 60 DAS. Whereas NDVI calculated over weed plants showed higher value with increase in density of weeds. The study identifies the limitations of solely relying on GreenSeeker® based NDVI for weed variability identification. The future research can investigate if other indices from multispectral or hyperspectral data improve weed differentiation.