Real-Time Intelligence, Sensor-based Nitrogen Estimation for Precision N- Management in Wheat

This paper deals with the significance of precision farming in addressing challenges faced by Indian agriculture. This study aims to advance our understanding of plant sensors and their effectiveness in predicting plant nitrogen (N) status and crop yields. It seeks to establish a reliable relationship between sensor measurements and actual leaf N content using machine learning algorithms, GreenSeeker, chlorophyll meter, and spectro-radiometer to provide a more accurate and efficient means of N management. The study focuses on variable N-fertilized wheat crops. Increments of 30 kg/ha in N application rate increased SPAD values and NDVI both indices showing peak values at 150 kg N/ha. There existed a nearly linear relationship between sensor indices with leaf N content within the boundaries of selected N rates (0 -240 kg N/ha). Further, a correlation analysis of different sensor (Chlorophyll meter, GreenSeeker) indices with leaf N content was positive and significant (R2=0.86-0.95). Spectro-radiometer-based spectral-derived indices gave a good prediction of plant N status as indicated by an R2 value of 0.71. Apart from these sensors, machine learning tools also revealed a very strong positive correlation (0.974) among studied parameters; the heat map generated depicted multicollinearity, which could be considered in further analysis for the predictive model. Overall, the N stress in the wheat crop can be monitored on a real-time basis for determining the N fertilization in this crop