Estimating leaf area density of Osmanthus trees using ultrasonic sensing

Leaf area density is a key tree canopy parameter required for the intelligent profiling of spray applications. A method to calculate leaf area density of target trees based on the echo signals of ultrasonic sensors was proposed. The relationship between leaf area and leaf mass of Osmanthus trees (Osmanthus fragrans) was analysed. A strong linear relationship between leaf mass and leaf area was found, with the ratio between the mean leaf mass to the mean leaf area equal to 288. To validate the performance of the tree canopy leaf area density model, four different leaf area densities of the canopy were selected in the laboratory and tested using a test bench at three different distances. The results showed that relative error between observed and predicted mean ultrasonic echo signal voltages ranged from 1.02 to 16.8%, and MRE (mean relative error) was 6.55%. Therefore, the tree canopy leaf area density model satisfactorily estimated leaf area density in the laboratory. Six Osmanthus trees were used during field tests with nine test points on each tree being selected and measured. The results showed that relative error between observed and predicted leaf area density ranged from 0.23 to 35.5% (MRE = 15.7%), which suggested good performance for the model when estimating leaf area density in the outdoor environment.