基于无人机多光谱影像的苎麻光合参数反演

打开文本图片集

中图分类号：S563.1：S126 文献标识号：A 文章编号：1001-4942（2025）08-0160-08

AbstractPhotosynthetic parameters are important indicators reflecting the photosynthetic status of plants.In order to study the feasibility of inverting ramie（Boehmeria nivea L.）photosynthetic parameters from multispectral images captured by unmanned aerial vehicle （UAV），a multispectral UAV platform was utilized to collct the multispectral images of ramie at seedling，row closure，vigorous growth，and ripening stages under different nitrogen application levels and topdressing periods，and the net photosynthetic rate （Pn） ，stomatal conductance（Gs），intercellular CO2 concentration（Ci），and transpiration rate （ ）of the ramie leaves were measured at the same time. Through correlation analysis，seven vegetation indices （NDVI，GNDVI， RVI，SIPI，WDRVI，MSR，MCARI）were selected which had higher correlation with the four photosynthetic parameters. Inversion of the photosynthetic parameters was performed using three machine learning models of Random Forest（RF），Support Vector Machine（SVM），and Back-Propagation Neural Networks（BPNN）， and the model validation and accuracy comparison were also conducted.The results showed that the best models for inverting ramie photosynthetic parameters of Pn ，Gs，Ci and Tr from UAV multispectral images were RF models at ripening，ripening，seedling and ripening stages，respectively，with the coefficient of determination （ R2 ）values of 0.640，0.790，0.790 and 0.720，and the root mean square error （RMSE）values of 1.040， 0.070，9.190 and 1.380，respectively.Therefore，UAV multispectral images combined with RF model could effectively invert the photosynthetic parameters of ramie leaves.

KeywordsRamie ; Photosynthetic parameters； Multispectral images captured by unmanned aerial vehi-cle； Vegetation indices ； Machine learning

苎麻（Boehmeria nivea L.）是荨麻科（Urticace-ae）苎麻属（Boehmeria）多年生草本纤维植物，是中国重要的纺织原料[1]。（剩余10147字）