| 基于星-机光谱融合的棉花叶片SPAD值反演 |
| |
| 引用本文： | 王淑婷,孔雨光,张赞,陈红艳,刘鹏.基于星-机光谱融合的棉花叶片SPAD值反演[J].中国农业科学,2022,55(24):4823-4839. |
| |
| 作者姓名： | 王淑婷 孔雨光 张赞 陈红艳 刘鹏 |
| |
| 作者单位： | 1土肥高效利用国家工程研究中心/山东农业大学资源与环境学院，山东泰安 2710182山东省国土空间规划院，济南 2500143鲁南高速铁路有限公司，济南 2500984山东农业大学农学院，山东泰安 271018 |
| |
| 基金项目： | 山东省自然科学基金(ZR2019MD039);山东省重点研发计划(LJNY202103) |
| |
| 摘 要： | 【目的】为提高棉花叶片叶绿素含量的反演精度，并掌握其在山东省夏津县的空间分布特征。【方法】本研究以山东省德州市夏津县为研究区，以夏津县大李庄棉田为试验区，通过SPAD（soil and plant analyzer development，SPAD）仪实地测定试验区棉花叶片叶绿素含量的相对值（SPAD值），并获取同期试验区无人机（unmanned aerial vehicle，UAV）近地多光谱图像和研究区Sentinel-2A MSI（MSI）卫星影像；然后分别基于UAV和MSI的光谱反射率，构建并筛选最优光谱参量，采用多元线性回归（multiple linear regression，MLR）建立SPAD值定量反演模型；最后采用二次多项式拟合法融合UAV和Sentinel-2A MSI对应的最优光谱参量，对比分析融合前后模型效果，优选最佳反演模型，实现研究区SPAD值反演。【结果】研究表明，(REG-R)/(REG+R)、R/G、CL(red edge)、NDVI可作为SPAD值的最优光谱参量；基于UAV图像的定量反演模型精度优于基于MSI影像的模型；基于二次多项式拟合后建模R 2提高了0.015—0.057，RMSE降低了0.457—0.638，验证R 2提高了0.040—0.085，RMSE降低了0.387—0.397，RPD提高了0.020—0.139；将融合后的MSI光谱参量代入基于UAV图像的反演模型（Fused MSI-ModUAV），也可获得较高的反演精度，建模R 2达0.672，RMSE为3.982，验证R 2达0.713，RMSE为3.859，RPD为1.685；基于上述模型进行研究区棉花叶片SPAD值反演分析，试验区整体呈南高北低的分布趋势，研究区呈中间低、四周高的分布趋势，均与实地情况一致，具有较好的预测效果。【结论】采用二次多项式拟合法融合无人机和卫星影像数据，可较好地实现区域高精度作物生长指标的定量反演，研究结果可丰富多源遥感融合理论与技术，为后续棉花长势监测与精准生产提供技术参考和数据支持。
|
| 关 键 词： | SPAD值 无人机 Sentinel-2AMSI 反演模型 二次多项式拟合法 |
| 收稿时间： | 2022-01-17 |
SPAD Value Inversion of Cotton Leaves Based on Satellite-UAV Spectral Fusion |
| |
| WANG ShuTing,KONG YuGuang,ZHANG Zan,CHEN HongYan,LIU Peng.SPAD Value Inversion of Cotton Leaves Based on Satellite-UAV Spectral Fusion[J].Scientia Agricultura Sinica,2022,55(24):4823-4839. |
| |
| Authors: | WANG ShuTing KONG YuGuang ZHANG Zan CHEN HongYan LIU Peng |
| |
| Institution: | 1National Engineering Research Center for Efficient Utilization of Soil and Fertilizer Resources/College of Resources and Environment, Shandong Agricultural University, Taian 271018, Shandong2Shandong Institute of Territorial and Spatial Planning, Ji’nan 2500143Lunan High Speed Railway Co., Ltd, Ji’nan 2500984College of Agronomy, Shandong Agricultural University, Taian 271018, Shandong |
| |
| Abstract: | 【Objective】The aim of this study was to improve the inversion accuracy of chlorophyll content in cotton leaves, and to grasp its spatial distribution characteristics in Xiajin county, Shandong province. 【Method】Taking Xiajin county, Dezhou city, Shandong province as the study area and Dalizhuang cotton field in Xiajin county as the test area, the relative value of chlorophyll content (SPAD value) in cotton leaves in the experimental area was measured by SPAD (soil and plant analyzer development), and obtained the near earth multispectral image of unmanned aerial vehicle (UAV) and Sentinel-2A MSI (MSI) satellite image in the study area in the same period; Then, based on the spectral reflectance of UAV and MSI satellite images, the optimal spectral parameters were constructed and selected, and the inversion model of SPAD value was established by multiple linear regression (MLR); Finally, the quadratic polynomial fitting method was used to fuse the optimal spectral parameters corresponding to UAV and Sentinel-2A MSI. By comparing and analyzing the model effects before and after fusion, the inversion model was optimized, and the SPAD value inversion of the study area was realized. 【Result】(REG-R)/(REG+R), R/G, Cl(red edge) and NDVI could be the optimal spectral parameters of SPAD value. The precision of cotton leaf SPAD inversion model based on UAV near ground image was better than that based on satellite image; After quadratic polynomial fitting, the calibration R2 was increased by 0.015-0.057, and RMSE was decreased by 0.457-0.638, while the validation R2 was increased by 0.040-0.085, RMSE was decreased by 0.387-0.397, and RPD was increased by 0.020-0.139. The fused spectral parameters based on Sentinel-2A MSI image were input to the inversion model based on UAV data (Fused MSI-ModUAV), the high inversion accuracy of SPAD value in cotton leaves could be obtained, with the model calibration R2 up to 0.672, RMSE of 3.982, validation R2 up to 0.713, RMSE of 3.859, and RPD of 1.685. Based on the above model, two inversion prediction maps of different scales were obtained. The SPAD value of cotton leaves in the test area showed the distribution trend of high in the south and low in the north, and the study area showed the distribution trend of low in the middle and high around, which were consistent with the field situation and showed the model had a good prediction effect. 【Conclusion】Therefore, the fusion of UAV and satellite image data by using quadratic polynomial fitting method could better realize the quantitative inversion of regional high-precision crop growth indicators. The research results could enrich the theory and technology of multi-source remote sensing fusion, and provide the technical reference and data support for cotton growth monitoring and precision production. |
| |
| Keywords: | SPAD value UAV Sentinel-2A MSI inversion model quadratic polynomial fitting method |
|
| 点击此处可从《中国农业科学》浏览原始摘要信息 |
| 点击此处可从《中国农业科学》下载免费的PDF全文 |
|
