Multisensor data fusion from uav platforms: a robust approach for estimating photosynthetic capacity across vertical leaf layers in winter wheat
Аннотация
Context Accurate assessment of photosynthetic capacity is essential for monitoring wheat growth and physiological processes, yet there is limited research on estimating the photosynthetic capacity at each leaf layer of winter wheat. Aims This study aims to estimate the photosynthetic capacity of different leaf layers of winter wheat using UAV-based multi-source data fusion. Methods We combined multi-spectral, thermal infrared, and LiDAR data to estimate the photosynthetic capacity of each leaf layer (PC) and whole-canopy photosynthetic capacity at 0, 20, and 30 days after anthesis (DAA). We also explored the relationship between PC, grain yield, and key vegetation indices (VIs), canopy temperature (CT), and canopy vertical light distribution (CVLD). Key Results Strong correlations between whole-canopy photosynthesis and grain yield ( r = 0.80, 0.73, and 0.39 at 0, 20, and 30 DAA, respectively) were observed. VIs, CT, and CVLD were found to be key determinants of photosynthetic capacity ( r = 0.49, 0.66, and 0.86, respectively). Furthermore, multi-source data fusion significantly improved model performance, with validation-set accuracies of R^2 = 0.39–0.77, RMSE = 0.29–10.62 µmol·m⁻^2·s⁻^1, and MAE = 0.19–8.54 µmol·m⁻^2·s⁻^1. Compared with models based on a single feature, the best model accuracy increased from R^2 = 0.67 to 0.78 after integrating multi-source features. Conclusion The UAV-based multi-source data fusion approach provides robust and high-precision estimation of vertical leaf-layer photosynthetic capacity, with strong potential for improving precision agriculture practices. Implications and Impacts This study supports site-specific monitoring and resource optimization in wheat production, contributing to sustainable agricultural management through more accurate photosynthetic capacity estimation. This study advances Precision Agriculture by enhancing the estimation of vertical leaf-layer photosynthetic capacity in winter wheat through UAV-based multi-source data fusion. This approach improves site-specific monitoring, optimizing resource use, and supporting sustainable crop management. The first integration of multisource remote sensing data (VIs, CT, LiDAR) to estimate photosynthetic capacity across vertical leaf layers in winter wheat. Novel PC index (Pn×LAI) outperforms traditional Pn in correlating with grain yield (R^2 = 0.82 vs. 0.73). CVLD show strongest correlation with photosynthesis (R^2 = 0.86). Multi-sensor fusion model achieves field-scale accuracy across growth stages.