Answers are grounded in this paper and cite their sources.
Accurate remote yield prediction across planting densities is vital for sustainable management, yet intensified competition amplifies spatiotemporal phenotypic variability, compromising accuracy. This study integrated time-series plant height (PH) and soil and plant analyzer development (SPAD) from unmanned aerial vehicle (UAV) LiDAR/multispectral data via different strategies to improve prediction. PH was retrieved from LiDAR point clouds using cubic polynomial regression ( R² = 0.961), while SPAD was estimated from multispectral vegetation indices using the extreme gradient boosting (XGBoost) algorithm ( R² = 0.788). Both growth features were then used as inputs to construct yield prediction models via multiple linear regression (MLR), partial least squares regression (PLSR), and XGBoost, respectively, for single-feature models, the multi-stage SPAD-based XGBoost model showed better performance than the single-stage (grain‑filling stage) counterpart ( R ² = 0.776 vs. 0.722; RMSE = 480.75 vs. 488.35 kg ha⁻¹). Furthermore, integrating multi-stage PH and SPAD as fused features further improved prediction accuracy, with the XGBoost model attaining an R ² of 0.811 and an RMSE of 411.73 kg ha⁻¹, compared to 0.776 and 480.75 kg ha⁻¹ for the optimal single‑feature (SPAD‑only) model. This confirms that the synergy of structural (PH) and physiological (SPAD) dynamics across key stages provides a more robust yield estimation. While maintaining a high prediction accuracy ( R ²=0.805, RMSE=438.617 kg ha⁻¹), substituting full-season high-frequency monitoring with plant height and SPAD data from key growth stages reduces data acquisition cost and complexity, enables pre‑harvest yield estimation and improves the timeliness of agronomic decision-making. Overall, this study highlights the value of UAV-based multi-dimensional and multi-stage data fusion of PH and SPAD in yield estimation models, enabling a robust, effective strategy for in‑season precision management and sustainable intensification.
No takes yet. Share an insight, caveat, or question.
Li et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: