Next-generation 3D models of the globe depend on high-quality surface texturing derived from satellite imagery. Such imagery is acquired at different times, under varying sun angles and atmospheric conditions, and with different sensors, leading to radiometric inconsistencies across the data. Without a common color reference, these differences result in visible seams between building blocks, inconsistent textures, and a non-physical appearance in the final model. This work investigates methods for constructing a color reference from Sentinel satellite imagery and evaluates their suitability for radiometric normalization across large-scale textures. The aim is to enable consistent, visually appealing, and physically plausible texturing of global 3D models.