Shouyang Liu

　　INRA EMMAH, UMR 1114 Domaine Saint-Paul, Site Agroparc, 84914 Avignon Cedex 9, France

　　Green area index (GAI) has been difficult to estimate accurately at large scales due to the cost prohibitive nature of classical in-situ methods. Using the passive remote sensing technique of light detection and ranging (LiDAR), the difficulty of mapping GAI on ecosystems could be overcome. However, it has not yet been addressed on agricultural systems, especially for conditions with large GAI values. Through this work, we proposed a self-learning method to estimate GAI using LiDAR-derived metrics over a wheat field. Specifically, we developed a LiDAR simulator to carry out scanning on digital 3D objects, mimicking the measuring principle and setups of actual LiDAR sensors. Further, coupling the LiDAR simulator with the 3D ADEL-Wheat model made it possible to produce in silico scan experiments over varied canopies. We then propose to use a machine learning algorithm to correlate LiDAR-derived metrics to GAI over synthetic datasets. Eventually, the model performance was evaluated with both independent synthetic datasets and in-situ destructive measurements. We achieved a mean RMSE of 0.41 and mean rRMSE of 0.11. The model is further compared with the classically used Beer-Lambert approach. Our method is preferable for its better accuracy and capability of generalization. This study illustrated how LiDAR data could provide interesting interpretation for agricultural system.