Department of Biosystems Engineering, College of Agriculture and Life Sciences, Kangwon National University, Chuncheon 24341, Kangwon, Korea.
Department of Agricultural and Biological Engineering, Mississippi State University, Starkville, MS 39759, USA.
Sensors (Basel). 2020 Dec 11;20(24):7098. doi: 10.3390/s20247098.
Accurate and reliable calibration methods are required when applying unmanned aerial vehicle (UAV)-based thermal remote sensing in precision agriculture for crop stress monitoring, irrigation planning, and harvesting. The primary objective of this study was to improve the calibration accuracies of UAV-based thermal images using temperature-controlled ground references. Two temperature-controlled ground references were installed in the field to serve as high- and low-temperature references, approximately spanning the expected range of crop surface temperatures during the growing season. Our results showed that the proposed method using temperature-controlled references was able to reduce errors due to ambient conditions from 9.29 to 1.68 °C, when tested with validation panels. There was a significant improvement in crop temperature estimation from the thermal image mosaic, as the error reduced from 14.0 °C in the un-calibrated image to 1.01 °C in the calibrated image. Furthermore, a multiple linear regression model ( = 0.78; -value < 0.001; relative RMSE = 2.42%) was established to quantify soil moisture content based on canopy surface temperature and soil type, using UAV-based thermal image data and soil electrical conductivity (ECa) data as the predictor variables.
在精准农业中应用基于无人机 (UAV) 的热遥感进行作物胁迫监测、灌溉规划和收获时，需要准确可靠的校准方法。本研究的主要目的是通过使用温度控制的地面参考来提高基于 UAV 的热图像的校准精度。在田间安装了两个温度控制的地面参考，作为高温和低温参考，大致涵盖了生长季节作物表面温度的预期范围。我们的结果表明，使用温度控制参考的提出的方法能够将由于环境条件引起的误差从 9.29°C 降低到 1.68°C，当用验证面板进行测试时。从热图像镶嵌图中可以看出，作物温度估计有了显著的改善，因为未校准图像中的误差从 14.0°C 降低到校准图像中的 1.01°C。此外，建立了一个多元线性回归模型（ = 0.78；P 值 < 0.001；相对 RMSE = 2.42%），使用基于 UAV 的热图像数据和土壤电导率 (ECa) 数据作为预测变量，基于冠层表面温度和土壤类型来量化土壤水分含量。
