| 刘雨,慈宝霞,高雪松,罗洪洋,张吕夥,刘扬,马富裕.无人机不同飞行高度对滴灌棉花花铃期冠层温度提取精度的影响[J].干旱地区农业研究,2026,(2):79~93 |
| 无人机不同飞行高度对滴灌棉花花铃期冠层温度提取精度的影响 |
| Influence of different UAV flight altitudes on the accuracy of canopy temperature extraction during drip\|irrigated cotton flowering and boll\|setting stage |
| |
| DOI：10.7606/j.issn.1000-7601.2026.02.08 |
| 中文关键词: 无人机 飞行高度 棉花 花铃期冠层温度 提取精度 滴灌 |
| 英文关键词:unmanned aerial vehicle flight altitude cotton canopy temperature during the flowering and boll\|setting stage extraction accuracy drip irrigation |
| 基金项目:新疆生产建设兵团重点领域科技攻关计划项目（2025AB081）；石河子大学自主立项科研项目（ZZZC2023011 2024-2025） |
|
| 摘要点击次数: 154 |
| 全文下载次数: 15 |
| 中文摘要: |
| 以花铃期滴灌棉花为研究对象，重点探讨飞行高度对冠层温度提取精度的影响。设置12、20、30、50、70 m的飞行高度分别获取3、5、7、12 d·次-1水分处理下花铃期棉花冠层热红外图像。采用首尾1%剔除法、低频0.5%剔除法以及低频1%剔除法提取冠层温度信息，分析不同剔除异常值的方法对冠层温度提取精度的影响。结果表明，12~30 m低频0.5%剔除法作为低空提取精度最佳方法，两年冠层温度与实测温度拟合模型R2分别为0.874和0.934，RMSE分别为2.435和2.171；50~70 m首尾1%剔除法作为高空提取精度最佳方法，2023年和2024年冠层温度与实测温度拟合模型R2分别为0.833和0.914，RMSE分别为3.904和3.859；30 m飞行高度为棉花冠层温度提取最佳高度，2023年和2024年冠层图像温度与实测温度拟合模型R2分别为0.863和0.720，RMSE分别为2.424和3.664。综上，针对不同飞行高度采用差异化的温度异常值剔除方法开展棉花花铃期冠层温度提取研究，结果表明30 m飞行高度结合低频0.5%剔除法为该时期棉花冠层温度提取的最优方法组合，可显著提升无人机热红外影像中棉花冠层温度的提取精度，2023年和2024年的预测精度分别提升70.71%和4.03%。 |
| 英文摘要: |
| Taking drip\|irrigated cotton at the flowering and boll\|setting stage as the research subject, this study focuses on the impact of flight altitude on the accuracy of canopy temperature extraction. The study acquired thermal infrared images of the cotton canopy at the flowering and boll\|setting stage under water treatment intervals of 3 days, 5 days, 7 days, and 12 days per cycle at flight altitudes of 12 m, 20 m, 30 m, 50 m, and 70 m. Canopy temperature information was extracted using the top and bottom 1% elimination method, the low\|frequency 0.5% elimination method, and the low\|frequency 1% elimination method to analyze the effects of different outlier removal approaches on extraction accuracy. The results showed that for low\|altitude extraction (12~30 m), the low\|frequency 0.5% elimination method yielded the highest accuracy, with the fitted models between extracted canopy temperature and measured temperature achieving R2 values of 0.874 and 0.934, and RMSE values of 2.435 and 2.171 over two years, respectively. For high\|altitude extraction (50~70 m), the top and bottom 1% elimination method performed best, with R2 values of 0.833 and 0.914, and RMSE values of 3.904 and 3.859 over two years, respectively. A flight altitude of 30 m was identified as the optimal height for cotton canopy temperature extraction, with R2 values of 0.863 and 0.720, and RMSE values of 2.424 and 3.664 for the fitted models between image\|derived canopy temperature and measured temperature in 2023 and 2024, respectively. In summary, a study on extracting cotton canopy temperature during the flowering and boll\|setting stage was conducted using differential temperature outlier removal methods for different flight altitudes. The results indicated that combining a flight altitude of 30 m with the low\|frequency 0.5% removal method was the optimal approach for cotton canopy temperature extraction during this period. This method significantly improved the extraction accuracy of cotton canopy temperature from UAV thermal infrared imagery, with prediction accuracy increasing by 70.71% in 2023 and 4.03% in 2024. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| | |