HYPERSPECTRAL MONITORING BASED ON THE NET PHOTOSYNTHETIC RATE OF BROOMCORN MILLET UNDER VARYING PHOTOPERIODS

Photoperiod regulates photosynthetic electron transport and influence carbon assimilation in broomcorn millet.This study employed hyperspectral technology to decode coupling mechanisms of photoperiod-net photosynthetic rate (Pn) using light-sensitive (LM5) and light-insensitive (YR01) cultivars under three treatments: natural light (CK), short-day treatment before heading (G1/G2), and short-day treatment throughout the growth period (G3).Pn prediction models were constructed using partial least squares (PLS), support vector machine (SVM), and back-propagation neural network (BPNN).Results showed that G3 increasing Pn by 15.76% (LM5) and 5.58% (YR01) (P<0.05).The 1ST treatment exhibited the best correlation between the spectral data and Pn (r = -0.535),with the 1ST-RSI (716, 715 nm) showing the highest correlation (r = 0.658).The full-spectrum PLS model showed superior performance (validation R=0.669,RMSE=3.226,RPD=1.642).These findings confirm that the synergistic optimization of spectral preprocessing and modeling algorithms can precisely quantify the photoperiodic response of broomcorn millet Pn, providing technical support for the application of hyperspectral technology in assessing photosynthetic productivity and guiding crop management.