| Wheat above ground biomass(AGB) serves as a crucial indicator for assessing its growth status. Currently, unmanned aerial vehicles(UAVs) equipped with various optical sensors have been extensively utilized for monitoring wheat AGB. However, in the late stages of wheat growth, high vegetation coverage hinders optical sensors from capturing information within the canopy, thereby diminishing the accuracy of AGB monitoring. To mitigate this issue, this study incorporated point cloud characteristic parameters(PCs), which are related to canopy height, as representative indicators of canopy structure for estimating AGB using spectral indices, including color indices(CIs) and vegetation indices(VIs). The accuracy of AGB estimation models was then compared before and after the inclusion of PCs. The results revealed variations in the effectiveness of AGB estimation using different characteristic parameters during the jointing, heading, and filling stages of wheat growth. In the training set, the effect of CIs was slightly lower than that of VIs, while in the validation set, it was marginally higher. Although PCs did not exhibit the most remarkable performance across the three key periods, their stability was notably strong, and certain canopy structural characteristics demonstrated potential for AGB prediction. When individual features were used for AGB prediction, the r2 values ranged from 0.47 to 0.75. In contrast, integrating the three features of CIs, VIs, and PCs substantially enhanced the accuracy of the AGB estimation model. Specifically, during the three growth stages, the models r2 values were 0.79, 0.81, and 0.77, and the root mean square error(RMSE) was 0.42, 0.74, and 0.80 t·hm-2, respectively. Among different types of remote sensing features, support vector machine regression(SVM) provided superior estimation results compared to Gaussian process regression(GPR). Therefore, by integrating the three features(CIs, VIs, and PCs) and employing the SVM algorithm, an effective estimation of wheat AGB can be achi. |
基本信息
