Simulation and machine learning models for bias assessment and reduction in leaf area density estimators in tropical forest

Leaf Area Index (LAI) is a key parameter regulating water and carbon fluxes. Current LAI estimation methods have inherent biases, particularly from woody components. In this presentation, I will present how our work employs trial emulation to quantify these biases. I will also introduce SOUL, a deep learning model that uses only point coordinates to perform leaf/wood segmentation.