Monitoring of corn leaf area index based on multispectral remote sensing of UAV
-
摘要:目的
探究更高效估测玉米 LAI 的无人机多光谱遥感监测模型，实现对玉米叶面积指数(Leaf area index，LAI)的快速预测估算。
方法以全生长周期的玉米植株为研究对象，通过多光谱遥感无人机获取玉米植株影像并实地采集玉米LAI，利用多光谱信息研究植被指数与玉米LAI之间的定量关系，并选择相关的植被指数；分别使用多元线性逐步回归、支持向量机回归算法(Support vector machine regression，SVM)、随机森林回归算法(Random forest regression，RF)和基于鲸鱼算法(Whale optimization algorithm，WOA)优化的随机森林算法(WOA-RF)构建玉米LAI预测模型，通过分析对比，选择最优预测模型。
结果筛选出的植被指数NDVI、NDRE、EVI、CIG与LAI呈极显著相关(P<0.01)，构建了多元线性回归模型、SVM模型、RF模型和WOA-RF模型的预测模型，R2分别为0.873 2、0.878 0、0.917 7和0.940 8，RMSE分别为0.277 5、0.236 5、0.209 0和0.128 7。
结论基于WOA-RF的玉米LAI预测模型的预测精度能够满足玉米生产的需要，对玉米生长期间的种植管理具有指导意义。
Abstract:ObjectiveIn order to achieve a rapid estimation of the leaf area index (LAI) of maize, this study explores a more efficient monitoring model for maize LAI estimation based multispectral remote sensing of unmanned aerial vehicle (UAV).
MethodThis study focused on maize plants throughout their entire growth cycle. Multispectral imagery of maize plants was acquired using UAV, and maize LAI were collected in field. The quantitative relationship between vegetation index and maize LAI was investigated using multispectral information to select relevant vegetation indices. Multiple linear stepwise regression, support vector machine regression (SVM), random forest regression (RF), and a random forest algorithm optimized using whale optimization algorithm (WOA-RF) were used to construct maize LAI prediction models, respectively. The best prediction model was selected on the basis of comparison.
ResultThe vegetation indices of NDVI, NDRE, EVI and CIG were highly correlated with LAI (P < 0.01). The models of multiple linear regression, SVM, RF, and WOA-RF were constructed, with R-squared values of 0.873 2, 0.878 0, 0.917 7, and 0.940 8 respectively, and the root mean square error (RMSE) values of 0.277 5, 0.236 5, 0.209 0, and 0.128 7 respectively.
ConclusionThe prediction model of maize LAI based on WOA-RF provides a high level of accuracy, which can meet the requirement for maize production. It can be used to guide planting management of maize during the growth period.
-
Keywords:
- Unmanned aerial vehicle (UAV) /
- Remote sensing /
- Multispectral /
- Corn /
- Leaf area index (LAI) /
- Monitoring
-
![]()
图 5 6个试验区玉米LAI变化图
Figure 5. The variation chart of maize LAI in six experimental areas
![]()
图 6 支持向量机模型训练集和测试集预测结果
Figure 6. The predict results of training and test sets of the SVM model
![]()
图 7 最佳决策树和最小叶子节点数的拟合系数(R 2)
Figure 7. The fitting coefficients (R2) of the ntree and Min_samples_leaf
![]()
图 8 随机森林回归模型训练集和测试集的预测结果
Figure 8. The predict results of training and testing sets of the random forest regression model
![]()
图 9 WOA-RF回归模型训练集和测试集的预测结果
Figure 9. The predict results of training and testing sets of the WOA-RF model
表 1 植被指数与叶面积指数相关性分析
Table 1 Correlation analysis between vegetation index and leaf area index
植被指数
Vegetation index皮尔逊相关系数
Pearson correlation coefficient显著性
Sig.EVI 0.884 0.001 CIG 0.850 0.001 NDRE 0.836 0.003 NDVI 0.824 0.002 SAVI 0.793 0.003 OSAVI 0.774 0.004 PVI 0.765 0.007 GNDVI 0.763 0.007 RDVI 0.643 0.010 TSAVI 0.571 0.012 
下载: 导出CSV
表 2 多元线性回归模型最优植被指数1)
Table 2 Optimal vegetation index for multiple linear regression model
模型Model B Sig. VIF R2 RMSE 常量 Intercept −1.021 0.092 0.873 0.277 EVI 3.314 0.001 3.916 CIG 0.263 0.001 3.082 NDVI 3.308 0.007 2.828 NDRE 2.871 0.020 3.262 1) B：回归系数，Sig.：显著性水平，VIF：多重共线性检验中的方差膨胀因子
1) B：Regression coefficient, Sig.: Significance level, VIF: Variance inflation factor in multicollinearity test
下载: 导出CSV
表 3 不同模型性能指标对比
Table 3 The comparison of performance metrics of various models
模型 Model 样本 Sample R2 RMSE MAE 多元线性回归 MLR 0.873 2 0.277 5 0.193 2 SVM 训练集 Training set 0.895 5 0.222 6 0.151 1 测试集 Test set 0.878 0 0.236 5 0.182 1 RF 训练集 Training set 0.963 0 0.132 0 0.151 2 测试集 Test set 0.917 7 0.209 0 0.134 9 WOA-RF 训练集 Training set 0.966 8 0.125 0 0.124 3 测试集 Test set 0.940 8 0.128 7 0.134 5
下载: 导出CSV
-
[1] 李志坚, 谷云松, 贺云林, 等. 湘南地区甜玉米优质高产栽培技术[J]. 农业科技通讯, 2023(2): 177-179. [2] 徐春光. 诸城市植保无人机飞防作业快速发展[J]. 农机质量与监督, 2022(1): 28. [3] 许鹤. 农作物灾害损失评定遥感方法研究[J]. 农业与技术, 2023, 43(8): 12-15. [4] CASANOVA D, EPEMA G F, GOUDRIAAN J. Monitoring rice reflectance at field level for estimating biomass and LAI[J]. Field Crops Research, 1998, 55(1/2): 83-92.
[5] MORIONDO M, MASELLI F, BINDI M. A simple model of regional wheat yield based on NDVI data[J]. European Journal of Agronomy, 2007, 26(3): 266-274. doi: 10.1016/j.eja.2006.10.007
[6] 杨贵军, 李长春, 于海洋, 等. 农用无人机多传感器遥感辅助小麦育种信息获取[J]. 农业工程学报, 2015, 31(21): 184-190. [7] 邵国敏, 王亚杰, 韩文霆. 基于无人机多光谱遥感的夏玉米叶面积指数估算方法[J]. 智慧农业(中英文), 2020, 2(3): 118-128. [8] 张瑾. 基于无人机多光谱影像的夏玉米叶面积指数估测模型研究[D]. 太原: 山西农业大学, 2022. [9] 孙诗睿, 赵艳玲, 王亚娟, 等. 基于无人机多光谱遥感的冬小麦叶面积指数反演[J]. 中国农业大学学报, 2019, 24(11): 51-58. doi: 10.11841/j.issn.1007-4333.2019.11.06 [10] HE B, JIA B, ZHAO Y, et al. Estimate soil moisture of maize by combining support vector machine and chaotic whale optimization algorithm[J]. Agricultural Water Management, 2022, 267: 107618 doi: 10.1016/j.agwat.2022.107618
[11] 刘帅兵, 金秀良, 冯海宽, 等. 病害胁迫下玉米LAI遥感反演研究[J]. 农业机械学报, 2023, 54(3): 246-258. [12] 王圆, 毕玉革. 基于无人机高光谱的荒漠草原地物精简学习分类模型[J]. 农业机械学报, 2022, 53(11): 236-243. doi: 10.6041/j.issn.1000-1298.2022.11.023 [13] 齐钊. 基于高分辨率遥感影像的湖滨带土地覆被变化检测方法研究[D]. 连云港: 江苏海洋大学, 2022. [14] 张静, 倪金, 马诗敏, 等. 基于GIS的大连市金普新区洪水淹没分析[J]. 地质与资源, 2021, 30(5): 590-594. [15] ROUSE J W, HAAS R H , SCHELL J A , et al. Monitoring vegetation systems in the Great Plains with ERTS [C]//Third ERTS Symposium. Washington DC: NASA Spec Publ, 1974: 309-317.
[16] GITELSON A A , KAUFMAN Y J, MERZLYAK M N. Use of a green channel in remote sensing of global vegetation from EOS-MODIS[J]. Remote Sensing of Environment, 1996, 58: 289-298
[17] 彭燕, 何国金, 张兆明, 等. 中国区域Landsat遥感指数产品[J]. 中国科学数据(中英文网络版), 2020, 5(4): 83-90. [18] BURGES C J C. A tutorial on support vector machines for pattern recognition[J]. Data Mining and Knowledge Discovery, 1998, 2(2): 121-167. doi: 10.1023/A:1009715923555
[19] BIAU G, SCORNET E. A random forest guided tour[J]. Test, 2016, 25(2): 197-227. doi: 10.1007/s11749-016-0481-7
[20] NASIRI J, KHIYABANI F M. A whale optimization algorithm (WOA) approach for clustering[J]. Cogent Mathematics & Statistics, 2018, 5(1): 1483565.
[21] 张伟萍, 付民, 张海燕, 等. 改进的WOA-VMD算法在水声信号去噪中的应用[J]. 中国海洋大学学报(自然科学版), 2023, 53(1): 138-146. [22] 赵丙秀, 董宁. 基于WOA-BP神经网络下马铃薯产量预测分析模型[J]. 农机化研究, 2024, 46(3): 47-51. doi: 10.3969/j.issn.1003-188X.2024.03.008 -
期刊类型引用(4)
1. 周文杰, 李柯嫱, 孙太安, 周其秀, 曾小林, 舒会生, 普建才. 6种杀菌剂对露地辣椒炭疽病的田间防治效果. 湖北植保. 2025(03) 
百度学术
2. 李明瑾，霍建飞，韩长志，姚玉荣，贲海燕，郝永娟，王万立. 辣椒炭疽病病原菌中分泌蛋白的预测及特性分析. 江苏农业学报. 2025(01): 41-50 . 百度学术
3. 张河庆，宋占峰，韩帅，吴婕，苗明军，席亚东，乔昕. 四川辣椒炭疽病菌鉴定及防治药剂筛选研究. 中国蔬菜. 2024(04): 93-99 . 百度学术
4. 周锋，安家琳，苏海川，张胜，王国权，逯盼盼，刘润强. 月季黑斑病的病原特征及其对杀菌剂的敏感性分析. 河北农业科学. 2024(03): 57-61 . 百度学术
其他类型引用(0)
