Hyperspectral imagery to monitor crop nutrient status within and across growing seasons

Imaging spectroscopy provides the opportunity to monitor nutrient status of vegetation. In crops, prior studies have generally been limited in scope, either a small wavelength range (e.g., 400–1300 nm), number crop cultivars, single growth stage or growing season. Methods that are not time- site-specific needed use imaging for routine monitoring status. Using data from four cultivars potatoes (Solanum tuberosum L.), three stages and two seasons, we demonstrate capacity full-range (400–2350 nm) quantify (petiole nitrate, whole leaf vine total nitrogen) predict tuber yield across seasons. We specifically tested capabilities of: (1) ordinary least-squares regression (OLSR) using traditional hyperspectral vegetation indices (VIs); (2) partial (PLSR) full spectrum VNIR- (visible-to-near infrared: SWIR-only (shortwave 1400–2350 wavelengths; (3) predictive models developed one potato type planting season on withheld different Our results show OLSR produced poor predictions with all dates pooled together (validation R2 < 0.01). Single-date performed better (R2 = 0.20–0.60, relative RMSE 15–30%). PLSR well were comparable spectral regions (full-spectrum, VNIR-only SWIR-only), validation 0.68–0.82 RRMSE 12–25%. Testing types, reliable 0.45–0.75, 13–30%), but some bias. Cross-season had 0.46–0.75 17–100%, more significant bias than cross-potato models. To achieve generalizable robust, recommend: obtaining ground measurements capture plant conditions developmental stages, ensuring image processing approaches minimize discrepancies among dates.