| 引用本文: | 徐洪刚,陈震,程千,等.无人机多源光谱数据反演夏玉米LAI研究[J].灌溉排水学报,0,():-. |
| Xu Honggang,CHEN Zhen,CHENG Qian,et al.无人机多源光谱数据反演夏玉米LAI研究[J].灌溉排水学报,0,():-. |
|
| |
|
|
| 本文已被：浏览 1255次 下载 0次 |
|
|
|
| 无人机多源光谱数据反演夏玉米LAI研究 |
|
徐洪刚, 陈震, 程千, 李宗鹏, 李鹏, 范永申
|
|
中国农业科学院农田灌溉研究所/河南省节水农业重点实验室
|
|
| 摘要: |
| 【目的】为研究光谱反演大田夏玉米叶面积指数(LAI)的效果,【方法】以大田夏玉米为研究对象,利用无人机获取试验区不同生育期热红外以及多光谱图像,提取热红外冠层温度(TC)以及多光谱植被指数,结合地面实测LAI数据,分析光谱数据与实测LAI之间的相关关系,并将TC与筛选出的11种植被指数作为输入变量,LAI作为输出变量利用多元线性回归(Multiple linear regression, MLR)、支持向量机(Support vector machine, SVM)和随机森林(Random forest, RF)3个算法模型训练学习,建立夏玉米LAI的反演模型。【结果】结果表明,多光谱植被指数以及TC均与夏玉米LAI在P < 0.0001水平上极显著相关,相关系数均在0.5以上；RF算法于拔节期、喇叭口期、大喇叭口期以及吐丝期4个生育期的 LAI 预测值与实测值 R2分别为0.707、0.834、0.794、0.849,均高于(除大喇叭口期外)MLR 算法(0.446、0.434、0.803、0.763)和SVM算法(0.511、0.569、0.838、0.812),对应的RMSE为0.092、0.182、0.224、0.158,均低于MLR算法(0.186、0.183、0.294、0.171)和SVM算法(0.127、0.32、0.227、0.177),对应的NRMSE为12.04%、13.65%、11.61%、12.14%,均低于MLR算法(24.24%、13.71%、15.21%、13.16%)和SVM算法(16.52%、24%、11.73%、13.64%)；融合热红外TC后的反演模型精度均有不同程度的提升,各生育期LAI 预测值与实测值 R2分别为0.788、0.874、0.81、0.862,均大于同时期未融合TC的LAI反演模型且具有更低的RMSE和NRMSE。【结论】多光谱植被指数以及TC均与夏玉米具有较强的相关性,且RF算法构建的夏玉米LAI反演模型精度优于MLI和SVM算法,同时TC的加入可以有效提升夏玉米LAI反演精度。该研究为快速准确的大田夏玉米LAI遥感监测提供了技术和方法。 |
| 关键词: 夏玉米;多光谱;热红外;植被指数;叶面积指数;随机森林回归 |
| DOI： |
| 分类号:S252；S274 |
| 基金项目:中国农业科学院科技创新工程重大产出培育项目“天空地农田精准灌溉信息智能感知技术与装备研发”、中央级科研院所基本科研业务费专项资助项目（FIRI2019-01-01）、中央级公益性科研院所基本科研业务费专项“基于无人机多种影像融合的作物信息提取”（FIRI202002-03） |
|
| UAV multi-source spectral data inversion for summer corn LAI |
|
Xu Honggang, CHEN Zhen, CHENG Qian, Li Zongpeng, LI Peng, FAN Yongshen
|
|
Farmland Irrigation Research Institute of Chinese Academy of Agricultural Sciences/ Henan Key laboratory of water-saving agriculture
|
| Abstract: |
| 【Background】Leaf Area Index (LAI) can reflect the health status of crops, and appropriate LAI is of great significance to ensure the growth and yield of maize. The traditional LAI monitoring method is time-consuming and laborious, and it is difficult to achieve large-scale rapid monitoring. UAV remote sensing has become a research hotspot of LAI monitoring in recent years because of its advantages of flexibility, portability and high-precision image acquisition.【Objective】The objective is to invert the effect of summer maize LAI by spectrum.【Method】In sunny and windless weather, the DJI M210V2 UAV equipped with Micasense Red Edge MX and ZenmuseXT2 dual photothermal imaging sensors was used to obtain the multispectral and thermal infrared images of Summer Maize in different growth stages in the experimental area. Through Pix4D to complete the image mosaic operation, import ArcGIS to extract the thermal infrared canopy temperature (Tc) and multispectral vegetation index. Combined with the measured LAI data on the ground, the correlation between the spectral data and the measured LAI was analyzed. Then, the inversion model of summer maize LAI was established by using three algorithm models: Multiple Linear Regression (MLR), Support Vector Machine (SVM) and Random Forest (RF).【Result】(1) The results showed that the multispectral vegetation index and Tc were significantly correlated with summer maize LAI at P < 0.0001 level, and the correlation coefficients were all above 0.5. (2) The predicted LAI values and measured R2(0.707、0.834、0.849) of RF algorithm were higher than those of MLR algorithm and SVM algorithm at the jointing stage, trumpet stage, trumpet stage and silking stage, and the RMSE and NRMSE of RF algorithm were lower than those of MLR and SVM algorithm. (3) After fusion of Tc, the inversion accuracy of jointing stage model was significantly better than that of the other three growth stages. With the development of growth period, the promotion effect gradually decreased. However, the inversion accuracy of the RF model with the integration of thermal infrared Tc in each growth period was improved in varying degrees, indicating that the addition of Tc can improve the LAI inversion accuracy of summer maize.【Conclusion】Multispectral vegetation index and Tc have strong correlation with summer maize, and the accuracy of summer maize LAI inversion model constructed by RF algorithm is better than MLI and SVM algorithm. At the same time, the addition of Tc can effectively improve the accuracy of summer maize LAI inversion. This study provides the technology and method for the rapid and accurate remote sensing monitoring of summer maize LAI, which is of great significance to improve the accuracy of LAI low altitude remote sensing monitoring and realize the intelligent production management of agriculture. |
| Key words: Summer maize; Multispectral; Thermal infrared; Vegetation index; LAI; RF. |
|
|
|
|
