Discussion: Getting an old map on georeferenced new basemap

How do you do it?
Old map and new map are both OCAD.

Currently I don't have a good idea. Projection does not match. Old one is no-projection map. Plan is to start with lidar contours and then put items one by one. But that does not feel efficient.

Discard the old map.

Seriously, this issue is exactly why I'm fighting the windmills here about georeferencing. It's such a downer to realize that this old map that you think is a somewhat valuable accumulation of data would take longer to fit to real world than it would take to make an altogether new survey. It's exactly the situation with many maps in the Bay Area.

T/D is right, it always looks feasible, but unless the whole mapping process for the old map, from basemap to OCAD input, was done exceptionally well, it takes longer than starting again.

To answer the question: F9 (adjust background map). T/D will object vociferously (see this thread.

If it is a map for a serious competition T/D is probably right. Most of the maps I update are of schools, and sometimes I discard the old map, but mostly it saves time to keep it and use F9.

OCAD learning video 'georeference a map'

http://www.ocad.com/en/support/learn-video

scroll down to 'miscellaneous'

if you have a georeferenced base map (rather than a scan as in the example) you shouldn't need to use F9

There is an OCAD 9 and 10 version of the video. I just watched the 10 version. For some reason it doesn't show the 'rubber sheeting' option in Transform Map - but that is the one to use. I believe 'affine' is similar to what F9 does (not 100% sure about that).

When you are finished you can keep the map and background open at the same time and check how accurate (or not) the process has been. At the very least you might get some idea of which parts of the map that are OK or which are so distorted you need to start over.

I just noticed - there is another video for 'rubbersheeting' lower down. So maybe the way to go is use 'affine' first off then 'rubbersheeting' for any distortion that remains after that.

Start with LIDAR contours and then put items one by one. The chances that the old map is wrong in a way F9-fixable by affine transformation are near-zero.
(i.e. what T/D said)

No doubt someone is about to tell me the reason why what I'm about to suggest is a bad idea (well, maybe not since it isn't quite transforming the old map to match the new one) but....

In leaf off season, take a good handheld GPS, preferably one with the ability to record a waypoint as the average of all the coordinates observed while you stand at the feature for a minute or so, take waypoints on all the point features on the existing map, and add them to your new basemap. In areas of high point feature density, take waypoints on only one or a few of the most prominent features that you're confident were mapped well the first time and use waypoints taken on those as a guide to position features from the old map. If you think the existing mapping of vegetation is very good, maybe transfer it too. If the new basemap doesn't show trails well, at least not everywhere, record tracks of the trails where necessary and add those to the new basemap. Then fieldcheck the new map, including checking that what you transferred and recorded by GPS looks right, using whatever combination of aids to fieldchecking (GPS, laser rangefinders, etc) and compass and pace, are part of your fieldchecking system.

So basically, yes, junk the old map and survey everything anew, except for what I suggest in the areas of high point feature density. The advantage over starting over from scratch is you know where the features are when you start and people who can record a waypoint with a GPS are more common than experienced field checkers.

Note: eddie may have something to say about a more sophisticated way to transform an existing map to match georeferenced information if and when he drops in.

Bring the old map with you to look at when you do your fieldwork.

A few years ago, I remapped a fairly large area (almost 10Km noth to south, not nearly as wide). I was astounded by the distortion of the old map (circa 1980) made from aerials. There was simply no way to line it up to the orthorectified photos or any other georeferenced data source. However, I was able to salvage a good bit of the data by using much smaller sections of the map. Basically, I would determine the area that I was going to field check on a given day (usually around 1Km^2) and carefully align the map to the georeferenced data sources using the affine transformation. Within that small area, it was not difficult to line things up. I'd then do the field checking with the georeferenced data as the primary base map, but I'd flip to the old map to help with identifying features (this if fairly easy to do if you're using OCAD 10 or 11 in the field on a tablet). I agree with those who have indicated that using an old map as the primary base source is dicey, but I think you can still get a lot of useful information off it by using this technique.

Has anyone used Ocad Transformer?
http://oapp.se/Applikationer/OCAD_Transformer.html
I think it can do rubber sheet transformation.

I had a situation similar to Eric's last year, except it was a new photogrammetric plot (not LIDAR) which I had to georeference first. It was useless trying to match the old (circa1980) map, but I fit it as best as I could. I then used it only as a secondary reference for classifying features (mostly roads and trails) found on the new basemap, and then occasionally in the field only for identifying and locating, but not plotting some features.

So, yes I'd agree, scrap the old map (unless it is very, very good) and start from scratch with the new base map or data. There's no substitute for fieldwork....

Rubbersheeting works really well. I was actually impressed at how closely some of our old maps will line up with LIDAR basemaps using this technique. There are always a few complex areas that no amount of rubbersheeting will save. These I usually clean up using the basemap after transforming.

Looks like the standard and pro versions of OCAD11 have it.

OCAD`s really been upping their suite of tools for this kind of digital wizardry. Annoyingly we just upgraded to 10 and less then a year later it`s obsolete. Seems like they decided to rip everyone off by moving to 11 instead of updating 10.

Correction: As it turns out OCAD 10 has been updated to include Rubbersheeting. Also my licence was upgraded to 11 for free because it was purchased after Feb 1st during the OCAD 11 beta testing. My apoligies to OCAD.

Rubber-sheeting is also in OCAD10 standard (I think it was introduced in an update, it may not have been in the first release of 10).

I had no joy trying to transform a whole map with it, but I might retry on a piece by piece basis.

Eric I sympathise, I just upgraded my T to a Model A and less than a year later they're offering 8 airbags, traction control and a builtin navman.

...and a builtin navman

Not sure I'd consider that an upgrade -- for orienteers... :-D

bigE: OCAD gave me a free upgrade from 10 to 11. I'm not sure exactly how long before the OCAD11 release that I bought OCAD10 (I can check it if you really want to know), but I was surprised they actually DIDN'T rip me off.

Correction above...

I recently built a Rogaine map by combining three existing orienteering maps using OCAD. In the beginning I wasn't exactly sure if I was going to be able to pull it off but after reading a bunch of documentation and looking through some tutorials I was able to stitch everything together by rubbersheeting over one giant LIDAR basemap. I am very proud of the finished product. OCAD is a great software capable of far more then people realize.

is a great software capable

... if you pony up €€€€ for versions 10+, which we'd be glad to do if only the attendees weren't so cheap.

Nice recovery Eric

Thanks Carl. Damn internet's getting me in trouble again... time to lay low for a bit.