A tough map

Alpamayo

Hi. I found online this map and, for the first time, I'm having a real hard time in trying to georeference it. Is it me or there is something wrong with such map?
In general I use the utm grid and with OziExplorer I manage to solve the problem. This time though the georeferencing is completely off. What do you think?
Thanks
Mike

Alpamayo

Here is the correct link: http://journeytovalbona.com/archive/journeytovalbona/www.journeytovalbona.com/wp-content/uploads/2017/05/Peaks-of-Valbona-Loop-1.jpg

Mykle

This is an interesting map!  The contours were generated from an elevation grid, so the drainages don't drain.  The grid appears to be 1000m at an appropriate scale.  Compared to the low-resolution map at https://journeytovalbona.com/valbona-hiking-trails/ the grid looks shifted (perhaps a different datum than expected).  So you may need to georeference the map (roughly) using the grid on the map, then shift the map to some reference points from Google Earth or another source.  Even better, you may need to do a lot of personally checking the map on the ground!

Alpamayo

Hi Mykle, thanks for your prompt reply. I'm beginning to have some doubt about the "quality" of this map. If you notice the grid is not made of squares (like I would expect), but rectangles. They declare their datum is WGS84. I'll try to re-georeference it starting from the coordinates of the summits. But I'm a little skeptical ....

Ice Age Mark

Hello there Mike,

Without more information, like coordinates for the grid lines, how you proceeded, etc., it's difficult for anyone to do anything except believe you that it's not working.

Having said that, it's not you; something is amiss with the maps from this site.  I'm not sure Mykle's choice of the word "interesting" is the one I would use.

As you noticed, using the 1000m UTM gridlines shown on the jpg doesn't work; however, creatively using some geographic features and ignoring the grid does.  I did this using 4 mountain peaks in a diamond shaped pattern for control points.  I loaded the World Topo (changed to UTM zone 34N) online source and used the peak spot elevation dots for a reference map, and eyeballed the peak symbols on the jpg image.  No numbers, just clicking points.  I often use Helmert Similarity rectification for stuff like this because it minimizes text and symbol distortion from projecting, and it seems to do well outside the control net.

The result was pretty good, even well outside the control points.  I made the jpg semi-transparent with Google Earth imagery underneath to check things.  You can see in the first image by inspecting the roadway and the trail to "Guri Stakes" it's darn close (this was really quick and roughshod).  The second image is from well outside the control near the bottom, left of center.  Still pretty close.  You can see that a ravine visible on GE has been used as a trail to where poor Stan was abandoned, which is marked on the jpg map.  (I would use that tiny obvious kink in the trail, which is also visible on GE, as a control point too.)

I think with strategic/creative point picking of more points you could make it work.  (I once used uniquely identifiable large glacial erratics as control points to georeference some 1948 aerial imagery)  The grid is worthless.  Once rectified, the jpg's pixel size is over 6 meters, so you can't expect too much for accuracy.

So what happened?  I didn't look into it too much; but you can see from the third image that the jpg's border is aligned with the correct UTM grid (red lines), but the jpg's gridlines (cyan lines) are rotated about 1 degree clockwise, and shifted.  My first guess is that the image was rotated (maybe to make text in UTM be horizontal) then later exported with the original grid projection.  

Another thing I noticed is that with the lo-res map Mykle found, the gridlines are angled in opposite direction from your map.  This would mean that the two maps have a different central meridian, or are on opposite sides of the UTM zone's central meridian.  Yet both are obviously of the same area and in UTM 34.  This can't be correct.  The compass is aligned with the border, not the grid.  This is also incorrect.  Maybe someone "adjusted" these maps to suit printing and/or website display, without regard to any effect on georeferencing them.

Lastly, as Mykle noted, the contours were generated from an elevation grid and are not hydrologically correct.  I beleive this is "karst" (limestone with caverns) terrain, so many closed depressions are to be expected.  Programs that make contours (like GM) don't distinguish "depression" contours with a different linetype.  It will likely be impossible to sort this out.

It sounds like you're pretty seasoned at this, but if you need any further help or explanation, just let me know.

Best of luck to you,

Mark

Alpamayo

Thanks a lot, Mark, for your detailed explanation. I'm not sure - at this point - if the map is worth the time or effort to get something accurate out of it. After all it's clearly homemade and it seem it was mainly done as eye candy and/or to give hikers the rough gist of what to expect about the hiking trails of this park. Odd enough they sell it for 15 euros and even claim it's the best map of the area. For a minute I had thought I could make it work under MyTrails app for hiking in the area. But if the map is unreliable I'd better find myself a serious official map of this Albanian park or just use the OpenTopoMaps and forget about this "joke".
Thank you again for your time

Mike

Ice Age Mark

Hello again,

So it is.  I don't think I would use it on the ground.  Here is my control point file to save you time and give a look.

Best regards,

Mark

Alpamayo

Thanks a lot Mark.

Best regards
Mike

Alpamayo

PS. My compliments! You've done a pretty darn good job!!!