地上生物量和叶绿素是紫花苜蓿生长过程中的重要指标, 可以为其生长的动态监测与管理提供有效的帮助。 紫花苜蓿作为最为重要的饲草作物, 如何利用现代光谱智能技术有效且准确地预测其状态是紫花苜蓿种植过程中的重要问题。 基于无人机多光谱对不同品种紫花苜蓿的地上生物量和叶绿素含量的估算结果进行研究并为此构建预估模型。 共研究了21个紫花苜蓿品种, 采用无人机搭载多光谱相机在天气晴朗无风时起飞并拍摄图像, 将无人机拍摄得到的多光谱图像采用ENVI 5.3软件进行分析, 挑选出NDVI、 EVI、 SAVI、 Green NDVI、 NDGI、 DVI、 NGBDI、 OSAVI、 NDRE 和MSR共10个植被参数和无人机多光谱相机自带的5个光谱波段（蓝、 绿、 红、 红边、 近红外）进行特征分析, 再使用Matlab 2020b软件, 采用支持向量机（SVM）构建不同品种紫花苜蓿的地上生物量和叶绿素含量的预测模型。 然而在实际操作的运行中, 发现使用SVM构建的预估模型其准确率不理想, 因此使用智能算法鲸鱼(WOA)和灰狼(GWO)对SVM预估模型进行优化, 发现使用SVM预估模型能预估不同品种的紫花苜蓿的地上生物量和叶绿素含量, 其中经WOA智能算法优化后的SVM预估模型在估算不同品种的紫花苜蓿的地上生物量和叶绿素含量时其准确率最高。 研究中构建的预估模型为筛选品质较好的紫花苜蓿品种有一定的指导意义, 同时也为今后无人机多光谱预估紫花苜蓿的生物量及其相关的生理生态指标提供了有效的帮助和合理的参考依据。

Above-ground biomass and chlorophyll are important indexes in alfalfas growth process, which can effectively help the dynamic monitoring and management of alfalfa growth. As the most important forage crop, how to effectively and accurately predict the status of alfalfa by using modern spectral intelligence technology is an important issue in the planting process of alfalfa. However, in the development process of spectroscopy, its progress in agriculture is relatively slow. Therefore, establishing a rigorous and accurate prediction model based on spectroscopy knowledge requires certain algorithms, training, testing and verification. Therefore, this experiment studied the estimation results of above-ground biomass and chlorophyll content of different alfalfa varieties based on UAV multi-spectrum and established the prediction model. In this experiment, a total of 21 alfalfa varieties were studied. The UAV equipped with a multi-spectral camera was used to take images in sunny weather without wind, and the images captured by the UAV were analyzed by ENVI 5.3 software. NDVI,EVI, SAVI, Green NDVI,NDGI, DVI, NGBDI, OSAVI, NDRE and MSR. These 10 vegetation indexes and 5 based bands (blue, green, red, red edge and near-infrared) which UAV cameras were analyzed, and then Matlab 2020b software was used to analyze these indexes. A support vector machine (SVM) was used to build the prediction model of above-ground biomass and chlorophyll content in the different alfalfa varieties. In the actual operation, it was found that the accuracy of the prediction model built by SVM was not ideal. Therefore, this experiment used intelligent algorithms whale (WOA) and Gray Wolf (GWO) to optimize the SVM prediction model. The results showed that all prediction models could roughly predict the above-ground biomass and chlorophyll content of different varieties of alfalfa. Among the three models, the SVM prediction model optimized by WOA intelligent algorithms had the highest accuracy in estimating above-ground biomass and chlorophyll content of different alfalfa varieties. Therefore, this experiment can provide certain guidelines for the selection of alfalfa varieties with better quality in the future agriculture. It also provides effective help and reasonable reference for the UVA multi-spectral estimation of alfalfa biomass and its related physiological and ecological indicators in the future.