Estimating the leaf area index of bell peppers according to growth stage using ray-tracing simulation and a long short-term memory algorithm

The leaf area index (LAI), which represents crop growth characteristics, is used to calculate canopy photosynthetic rates, set irrigation standards, and predict crop growth. The LAI can be non-destructively and continuously estimated using the light-intensity ratio of the upper and lower crop canopy, but it is affected by solar altitude and external weather conditions. The objective of this study was to develop a method to estimate the LAI of bell peppers (Capsicum annuum L.) using the light-intensity ratio of the upper and lower crop canopy via solar altitude and weather conditions. Growth stages and weather conditions with solar altitude were set using 3D-scanned plant models and ray-tracing simulation, respectively. The light intensities at each location of the canopy for given conditions were calculated using ray-tracing simulation. The relationship between the light-intensity ratio and the LAI was analyzed using a long short-term memory (LSTM) algorithm, which is a type of artificial neural network. According to our results, the ratio varied depending on solar altitude and external weather conditions and exponentially decreased with increasing LAI. This LSTM algorithmic approach was able to quantitatively analyze this complex relationship; compared with a greenhouse experiment for validation, the algorithm was highly accurate (R2 = 0.808). Accuracy further increased when solar altitude and weather conditions were added to the model. Therefore, we conclude that, using this method, the LAI can be accurately measured in a non-destructive and continuous manner.