AUTHOR=Liu Yuan , Nie Chenwei , Zhang Zhen , Wang ZiXu , Ming Bo , Xue Jun , Yang Hongye , Xu Honggen , Meng Lin , Cui Ningbo , Wu Wenbin , Jin Xiuliang 

TITLE=Evaluating how lodging affects maize yield estimation based on UAV observations

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 13 - 2022

YEAR=2023

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2022.979103

DOI=10.3389/fpls.2022.979103

ISSN=1664-462X

ABSTRACT=Timely and accurate estimation of crop yield before harvest is important. There have been many approaches were developed to estimate crop yields with remote sensing information. However, few existing approaches have been tested under lodging condition. The feasibility of existing approaches under lodging condition and the influences of lodging on crop yield estimation both remained unclear. In this study, an index (namely lodging index) was developed to represent the degree of lodging and a new crop yield estimation approach was developed to estimate maize yield after lodging occurred. The RGB and multispectral images obtained with low-altitude Unmanned Aerial Vehicle (UAV) after lodging occurred was used to accurately estimate maize yield with random forest regression (RFR) algorithm. The results showed that: 1) the lodging index can be used to represent the lodging degree of each plot well; 2) the model including lodging index yielded slightly better performance than the model without lodging index at all growth stages; 3) the model developed based on the UAV data collected at denting (R5) stage yielded the best performance, with R2, RMSE and rRMSE of 0.859, 1086.412 kg/ha, and 13.1%, respectively. This study provides valuable insight into the precise estimation of crop yield, and incorporating lodging stress-related variable into the model can provide relatively accurate and robust estimation of crops grain yield.