Simulating reflectance for phenological stage detection of Chinese cabbage using voxel-based virtual fields and Sentinel-2 imagery

Crop phenological information is crucial for agricultural activities such as fertilizer management, irrigation, disease prevention, and yield estimation. While optical Vegetation Indices and Synthetic Aperture Radar estimate crop phenology using satellite time series data, their universal application remains limited. Recently, simulating reflectance of virtual scenarios has proven effective for detecting crop biophysical parameters. This study simulated extensive voxel-based virtual Chinese cabbage fields under various seasons, growth stages, cultivation patterns, and survival rates across East Asia. A lookup table was created using reflectance simulations from the Simple Model of the Atmospheric Radiative Transfer of Sunshine (SMARTS) model, linked to Sentinel-2 Red band reflectance with shadow correction. Our key findings include: (1) Red band reflectance decreases from the rosette stage to heading, rebounding slightly during dormancy due to biophysical changes; (2) higher Sun Zenith Angles (SZA) in winter increase atmospheric path length, leading to greater radiation scattering and absorption, reducing Red band reflectance; (3) a 0.6-meter cultivation spacing is optimal, as 0.4 meters may hinder growth due to leaf overlap, while 0.8 meters could lead to inefficient land use; and (4) increased survival rates elevate vegetation fraction, reducing Red band reflectance. The lookup table was validated against Sentinel-2 imagery from 13 sites across Northeastern China, Japan, and the Koreas, achieving an R2 value of 0.75. This methodology, while focused on Chinese cabbage, is also expected to enhance phenological stage detection for other vegetable crops using high-resolution satellite imagery.