Extracting maize canopy temperature based on unmanned aerial vehicle thermal and RGB imagery and its application to water stress monitoring

Authors: ZHANG, LIYUAN - Northwest A&f University; NIU, YAXIAO - Northwest A&f University; Zhang, Huihui; HAN, WENTING - Northwest A&f University; LI, GUANG - Northwest A&f University; TANG, JIANDONG - Northwest A&f University; PENG, XINGSHUO - Northwest A&f University

Submitted to: Frontiers in Plant Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/11/2019
Publication Date: 10/9/2019
Citation: Zhang, L., Niu, Y., Zhang, H., Han, W., Li, G., Tang, J., Peng, X. 2019. Extracting maize canopy temperature based on unmanned aerial vehicle thermal and RGB imagery and its application to water stress monitoring. Frontiers in Plant Science. 10:1270. https://doi.org/10.3389/fpls.2019.01270.
DOI: https://doi.org/10.3389/fpls.2019.01270

Interpretive Summary: The accurate extraction of canopy temperature is critical for effectively monitoring crop water stress status with unmanned aerial vehicle (UAV) thermal remote sensing imagery. In this study, we proposed a method of using high-resolution UAV-based RGB imagery as the supplement to UAV thermal imagery for the extraction of canopy temperature (Tc) of maize crop at the late vegetative growing stage. To reduce the number of parameters and the application difficulty required for crop water status monitoring, four water status indicators were compared. The ground-truth Tc obtained by a handheld infrared thermometer was used to calibrate the UAV thermal imagery and evaluate the Tc extraction results. This study demonstrates using high-resolution UAV RGB imagery could improve the accuracy of extraction of canopy temperature from thermal imagery.

Technical Abstract: The accurate extraction of canopy temperature is the prerequisite for effectively monitoring crop water stress status based on unmanned aerial vehicle (UAV) thermal remote sensing imagery. High-resolution (1.25 cm) UAV optical remote sensing imagery was used as the supplement to UAV thermal imagery for the accurate extraction of canopy temperature (Tc) of maize at the late vegetation stage. To reduce the number of parameters and the applied difficulty required for crop water stress monitoring, Tc, degrees above non-stress, standard deviation of canopy temperature and canopy temperature coefficient of variation which only need canopy temperature were chosen to monitor maize water stress. The ground-truth Tc obtained by a handheld infrared thermometer was used for temperature calibration of the UAV thermal imagery and evaluation of the Tc extraction results. Leaf stomatal conductance was used to evaluate the monitoring effect of the four Tc-based crop water stress indicators. The results showed there was a high correlation between Tc extracted by the co-registration approach (RGRI-Otsu method) proposed in this study and groundtruth Tc with R2 value of 0.94 (n=30) and RMSE value of 0.8 °C. When ground-truth Tc was less than 32.9 °C, a relative lower Tc was obtained by RGRI-Otsu method. However, when ground-truth Tc was greater than 32.9 °C, a relative higher Tc extraction result was obtained. The change of the proportion of shadow soil and shadow leaves, and lighted soil in the UAV thermal orthophoto which was caused by the change of maize LAI maybe the reason for this phenomenon. Tc had the best correlations with leaf stomatal conductance with R2 value of 0.76, indicating that Tc obtained by RGRI-Otsu method could effectively monitor maize water stress. This study demonstrates the potentiality of using high-resolution UAV optical imagery as the supplement to UAV thermal imagery for the accurate extraction of canopy temperature.