Estimation of Fv/Fm in Spring Wheat Using UAV-Based Multispectral and RGB Imagery with Multiple Machine Learning Methods

The maximum quantum efficiency of photosystem II (Fv/Fm) is a widely used indicator photosynthetic health in plants. Remote sensing Fv/Fm using MS (multispectral) and RGB imagery has the potential to enable high-throughput screening plant agricultural ecological applications. This study aimed estimate spring wheat at an experimental base Hanghou County, Inner Mongolia, from 2020 2021. images were obtained flowering stage Da-Jiang Phantom 4 multispectral drone. A total 51 vegetation indices constructed, measured on ground was simultaneously Handy PEA analyzer. performance 26 machine learning algorithms for estimating compared. findings revealed that majority approximately half demonstrated strong correlation with Fv/Fm, as evidenced by absolute coefficient greater than 0.75. Gradient Boosting Regressor (GBR) optimal estimation model RGB, important features being RGBVI ExR. Huber MS, feature MSAVI2. Automatic Relevance Determination (ARD) combination (RGB + MS), SIPI, ExR, VEG. highest accuracy achieved ARD test sets (Test set MAE = 0.019, MSE 0.001, RMSE 0.024, R2 0.925, RMSLE 0.014, MAPE 0.026). combined analysis suggests extracting (SIPI, VEG) remote UAV input variables can significantly improve stage. approach provides new technical support rapid accurate monitoring Hetao Irrigation District.