 This study develops a framework for estimating bathymetry using the multispectral instrument, MSI, on Sentinel-2 satellite and regression-based random forest models, with Ebon Ladar Bathymetry, ALB, as validation. The proposed model is trained with limited ALB data to expand its practicality and yields an RMSI of 8% or lower for the 0-13.5 m depth range. The study also evaluates the influence of training sample locations on model performance and proposes a preliminary methodology for selecting optimal satellite imagery based on spectral data embedded in Sentinel-2 imagery. This article was authored by S.S.J.D. A. Ebtelriman, B. Wilkinson, and others.