| 徐亚薇,王珍,栗岩峰,李久生.基于无人机多光谱遥感的滴灌单元尺度棉花生长指标反演[J].干旱地区农业研究,2025,(5):224~233 |
| 基于无人机多光谱遥感的滴灌单元尺度棉花生长指标反演 |
| Inversion of cotton growth indexes at drip irrigation subunit scale based on UAV multi\|spectral remote sensing |
| |
| DOI：10.7606/j.issn.1000-7601.2025.05.22 |
| 中文关键词: 棉花 生长指标 反演模型 无人机多光谱遥感 滴灌单元 |
| 英文关键词:cotton growth index inversion model UAV multispectral remote sensing drip irrigation subunit |
| 基金项目:国家重点研发计划项目（2022YFD1900404）；国家自然科学基金（52279053） |
|
| 摘要点击次数: 268 |
| 全文下载次数: 96 |
| 中文摘要: |
| 基于搭载RTK模块和4个波段多光谱相机的无人机遥感平台，采集棉花蕾期、花期及铃期冠层表面高程和多光谱影像数据，同时在田间实测获取滴灌单元尺度棉花株高、叶面积指数、干物质量等生长指标。构建数字表面模型用以棉花株高反演；采用偏最小二乘回归、支持向量机、随机森林3种模型，建立棉花叶面积指数及干物质量反演模型，以最优模型对滴灌单元尺度作物生长指标进行预测并评估模型的准确性。结果表明：基于数字表面模型方法的棉花遥测株高与实测株高决定系数R2达0.59，随机森林模型对棉花叶面积指数和干物质量的反演效果最优，R2分别达到0.56和0.74。滴灌单元尺度株高实测值与预测值的决定系数R2达0.83；蕾期、花期及铃期叶面积指数实测值与预测值的平均差异分别为0.17、0.46、0.97，干物质量实测值与预测值的平均差异分别为0.12、1.37、3.17 t·hm-2，数据点基本落在95%一致性界限，滴灌单元作物生长指标实测值与预测值一致性较高。 |
| 英文摘要: |
| Based on the UAV remote sensing platform equipped with RTK module and 4 bands multi\|spectral camera, canopy surface elevation and multi\|spectral image data at cotton bud stage, flowering stage, and bolling stage were collected. The cotton plant height, leaf area index, and biomass at drip irrigation subunit scale were determined. The Digital Surface Model was constructed for retrieving cotton plant height. Partial Least Squares Regression, Support Vector Machine, and Random Forest models were used to establish inverse models of cotton leaf area index and biomass. The optimal model was used to predict the crop growth index at drip irrigation subunit scale. The results showed that the determination coefficient (R2) of the telemetry cotton height and measured plant height based on the digital surface model method was 0.59. The random forest model had the best inversion effect for cotton leaf area index (LAI) and biomass, while the model R2 were 0.56 and 0.74, respectively. The coefficient R2 between measured and predicted plant height for the drip irrigation subunits was 0.83. The average difference between the measured and predicted leaf area index at bud stage, flowering stage, and bolling stage were 0.17, 0.46, and 0.97, respectively. The average difference between the measured and predicted biomass was 0.12 t·hm-2, 1.37 t·hm-2, and 3.17 t·hm-2, respectively, and the data points basically fell within the 95% agreement limit. The measured value of crop growth index in drip irrigation subunits were consistent with the predicted value. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| | |