Some slides associated to the book Handbook of Marine Craft Hydrodynamics and Motion Control (2021 edition) are provided by the author of the book here.
At page 12 of the slides for Chapter 4, the moment due to the restoration forces (gravity, buoyancy) of a floating vessel is computed, expressed in the body frame of the vessel. According to page 14, such moment is computed in the Center of Flotation; however, the lever arm computed at page 12 is the following:
$$r_{\text{GM}}^b = \begin{bmatrix} -\text{GM}_L \sin\theta \\ \text{GM}_T \sin\phi \\ 0 \end{bmatrix}$$
where $GM_L$ is the longitudinal metacentric height, $GM_T$ is the transverse metacentric height, $\theta$ is the pitch angle of the vessel, $\phi$ is its roll angle, and the superscript $b$ indicates that the arm is represented in the vessel body frame.
The geometrical meaning of the second component of the arm is shown in a figure on page 10, as follows:
where $CB$ stands for Center of Buoyancy, and $CG$ is the Center of Gravity.
Then the moment is computed as follows: $$\mathbf{m}_r^b=\mathbf{r}_{GM}^b\times \mathbf{f}_b^b$$
where $\it{f_b^b}$ is the buoyancy force expressed in body frame.
By looking at this, it seems to me that the moment was computed about the Center of Gravity, not about the Center of Flotation, as claimed later.
Is this right, or is there any implicit assumption here that makes the assertion on the slides correct (for example the vessel is boxed-shaped)?