I now think my reply in #3 did not answer so clearly so here is more about this.
Some one told me that because of the hexes in the TOAW are not regular, the map in TOAW can not perfectly match with actual map (e.g. Google Maps), and the length of Y axis in TOAW map vs Actual map is about 1.1:1.
Is that true?
There is a 1.1:1 ratio (see also below) but it is not really important for gameplay. And a hex map cannot perfectly match with a real map (because in real life things don't come in hexes), but it can match reasonably well and I would say good enough for good gameply if the scenario writer designed the map well (for example using a tool for map making such as in the article I linked to above).
The hexes in both TOAWIII & IV are not regular hexagons, meaning you can not reliably measure the distance between two arbitrary points on a game screen.
If you wish to compare a game map with a google map, always try dilational transformation along, for example, the x-axis first.
Well, yes and no.
How to measure and transform distances depends on the scale and projection of the underlying map (and how the grid relates to that) rather than on the hex grid itself. This is not a specific issue of TOAW but a property of maps and grids in general.
If you are not concerned with maps you can probably skip the rest because this is not very important for most players most of the time.
See also Bob Cross' article about map projections and TOAW: http://www.matrixgames.com/forums/tm.asp?m=2231107 (especially posts #3, #4).
In a simple case, if the map area is small or not significantly deformed, you can ignore the grid layout when you measure on the map. In such case you can think of the grid as an overlay only; it does not change the distance between two arbitrary points on the map. You can verify this in imaging software or with a paper map on which you overlay a grid.
What you cannot do in most cases of course is count hexes or use "normal" X/Y coordinates of hexes to calculate map distances. For example hex spacing is different horizontally and vertically, meaning you'll find a little bit different distance between the centers of the two hex pairs [0,0]-[0,3] and [0,0]-[3,0], even though in both cases you count three hexes. See also this article for geometry of regular hexagons and programming examples: http://www.redblobgames.com/grids/hexagons/
This vertical/horizontal ratio in TOAW IV is about 1.1:1 (the number mentioned in the first post above). But that is a property of this hex grid, not necessarily of the map under the hexes. If you want to measure distances directly on the map at all you must take into account how the map was projected in the first place and how the grid relates to that, and then apparently simple things can become hairy quickly.
However, this is probably not very important most of the time. First of all, in TOAW we don't plan and calculate movement in kilometers but in MP per hex, which is more practical for our purpose. And a well-designed map should take reasonably care of spatial distortions for gameplay. Also, you probably don't use TOAW maps instead of Google maps to measure real distances. So my car navigates much better with Google maps but my divisions much better with TOAW maps.
As for choices in realism and hex scale of the map projection, that is up to the scenario designer. Again, see Bob's article on maps for some considerations and there is his LatLong program (same link as in my post #3 above) for map/hex/scale calculations for a sinusoidal map projection. By the way, LatLong seems to be a good old 16 bit program; it does not run on 64 bit Windows 10 but you can still run it on 64 bit in an emulator today.