3) Mr. Dong Zhe

Soil moisture is a key state variable for hydrology and climate. The Optical Trapezoid Model (OPTRAM) estimates soil moisture from reflective bands at high spatial resolution, but most implementations (i) lack systematic, cross-biome validation and (ii) rely on manual parameter tuning. We present an automated OPTRAM (AOPTRAM) workflow that integrates an extended Adaptive Sliding Window (ASW) edge-detection algorithm to derive STR-VI envelope functions without calibration. Using Sentinel-2 imagery, we evaluate AOPTRAM against in-situ measurements from ten networks (130 stations) spanning the bare soil, shrubland, grassland and tree cover. We compare vegetation indices (NDVI, SAVI), edge-fitting forms (linear, exponential and polynomial), and spatial aggregation (network- vs. station-scale), and assess accuracy using Kling-Gupta efficiency (KGE) and mean absolute error (MAE). Results show that performance is jointly controlled by VI choice, edge form, land cover, and spatial scale: SAVI with a linear edge yields the strongest skill in shrublands, whereas NDVI with an exponential edge performs best in grassland and tree cover. Bare soil remains challenging due to a narrow STR dynamic range that compresses moisture separability. Constructing STR-VI scatterplots at the station scale improves accuracy relative to the network scale and better captures seasonal dynamics. Overall, AOPTRAM delivers accuracy comparable to optimally tuned OPTRAM while offering a fast, calibration-free pathway for regional soil moisture monitoring across diverse ecosystems.