FactChecker said:
I'm having trouble seeing how the projection would have a North/South component.
The angular velocity is ##\vec \omega = \omega \vec e_z = \omega [\cos(\theta) \vec e_r - \sin(\theta) \vec e_\theta]##. A westward velocity vector is given by ##\vec v = -v \vec e_\varphi##. It follows that
$$
\vec a_c = 2\vec v \times \vec \omega = - 2v\omega \vec e_\varphi \times [\cos(\theta) \vec e_r - \sin(\theta) \vec e_\theta]
= - 2v\omega [\cos(\theta) \vec e_\theta + \sin(\theta) \vec e_r].
$$
The projection onto the tangent plane of the sphere (spanned by ##\vec e_\theta## and ##\vec e_\varphi## at every point) is therefore ##-2v\omega \cos(\theta) \vec e_\theta## and ##\vec e_\theta## is a unit vector in the south direction. For ##\theta > \pi/2##, ##\cos(\theta) > 0## and this vector is therefore pointing towards the north and for ##\theta > \pi/2##, ##\cos(\theta) < 0## and this vector is therefore pointing to the south.
At the equator ##\theta = \pi/2## and therefore ##\cos(\theta) = 0##, leading to no Coriolis force in the tangent plane at all (as you would expect).
Edit: Removed some typos (1 -> 0).
Edit 2: Some clarifications.
Edit 3: The perhaps easiest way to see that the force
must have a component in the tangent plane (everywhere but at the equator) is to note that the angular velocity is
not in the tangent plane. Hence, its component orthogonal to the tangent plane will give a vector in the tangent plane when crossed with any velocity vector in the tangent plane.