This work presents a bathymetric data generalization algorithm based on depth labels rendered at scale. It aims to facilitate the final cartographic sounding selection for chart portrayal through the process referred to as hydrographic sounding selection. Currently, automated algorithms for hydrographic soundings selection rely on radius- and grid-based approaches; however, their outputs contain a dense set of soundings with a significant number of cartographic constraint violations, thus increasing the burden and cost of the subsequent, mostly manual, cartographic sounding selection. As technology improves and bathymetric data are collected at higher resolutions, the need for automated generalization algorithms that respect the constraints of nautical cartography increases, where errors in the hydrographic sounding selection phase are carried over to the final product. Thus, we propose a novel label-based and shoal-biased, generalization algorithm that utilizes the physical dimensions of the symbolized depth values at scale to avoid the over-plot of depth labels. Moreover, we define validation tests for assessing adherence to cartographic constraints for nautical charts, namely functionality, legibility, spatial, and shape. We describe the limitations of current radius- and grid-based approaches with respect to these constraints, and detail our algorithm. Each approach is implemented in Python, and we use our validation tests to compare the results of our approach with the results of current approaches. Utilizing four datasets, it is shown that our label-based generalization performs the best regarding the cartographic constraints of functionality and legibility, and is equal to the other approaches in adhering to the spatial constraint.