Problem

@cgentemann and I were discussing about implementing a new feature for matplotlib for a 3D plot formed by three separate 2D surface contour plots, like a box. She thinks box3d is a nice name for this feature to conform with matplotlib naming.

Here is a sample image of what I am asking for

Proposed Solution

The example of the implementation is the code in
https://github.com/iuryt/Panthalassan/blob/main/Tutorials/box3D.ipynb
and
https://github.com/cgentemann/cloud_science/blob/master/Butterfly/ecco_3d_eddy_image.ipynb

The idea is to have a way to do the same plot just doing

fig = plt.figure(figsize=(7,4))
ax = fig.add_subplot(111,projection='3d')
ax.box3d.pcolor(X,Y,Z,data,surfaces=[surfacex,surfacey,surfacez],offsets=[offsetx,offsety,offsetz],**kw)
#or
ax.box3d.contour(X,Y,Z,data,surfaces=[surfacex,surfacey,surfacez],offsets=[offsetx,offsety,offsetz],**kw)
#or
ax.box3d.contourf(X,Y,Z,data,surfaces=[surfacex,surfacey,surfacez],offsets=[offsetx,offsety,offsetz],**kw)

The default values for the offsets are the same as surfaces, but the user may want to plot a surface for a different offset.
The kw arguments are the same as for the used function.

Additional context and prior art