According to the docs:
blocksize_y columns in the
raster_columns view represent the width and height (in pixels) of the raster blocks (often called tiles) in the raster column.
r_table_name = 'mytable'
AND r_raster_column = 'rast';
So if the raster is stored as one large block (meaning it's not split into smaller tiles), then the
blocksize_y values would match the overall width and height of the raster. On the other hand, if the raster is stored as multiple tiles (split into smaller blocks), the
blocksize_y values would be smaller than the overall width and height of the raster.