Improved sugarcane LAI estimation using radiative transfer models with spatial constraint

Sugarcane crop, cultivated in subtropical and tropical regions, provides major sugar supply, and makes great contributions to human life and economic development. The sugarcane leaf area index (LAI) is highly related to the production. Our research aims at estimating sugarcane LAI through remote sensing observations. The physically-based radiative transfer model (RTM) inversion methods are widely applied in vegetation variable estimation. However, ill-posedness problem widely exists in the model inversion processes. Therefore, the study develops a spatial constraint method to regularize the RTM inversion, and LAI variable is estimated on object-level. The estimated object-level LAI variable is compared with the pixel-level, and validated using the SNAP biophysical processor. The results shows that the object-level LAI estimates show great performance.