Question for GIS people

[nvprospector]

I have a map that I am importing into Oziexplorer and I am having a few issues with the Datum Shift. The map was published in 1905 and it does not have a Map Projection. I do have a Lat/Long grid and a distance scale on the map. Because of the year of the map, I don't have any clear reference on which datum was used. So I am trying to calculate a Datum Shift to WGS84.

So far I have tried the Bursa-Wolfe using all 7 parmeters and just tried this evening Molondensky with 7 parmeters. I have been thinking about trying the Inverse Molodensky, if I can find better documentation, but I don't like being limited to a max of 3 reference points.

With what I have tried thus far I have been able to get my points on the map within a 1/2 inch north of the map point, so I am close. East-West is right on. I was wondering if any GIS people have any ideas on what to try next?

-Tim-

[craig]

When you do the registration, ignore the lat/lon lines completely. Find another map with a known projection of the same area. Find features on both maps and use the coordinates/projection off of the second map.

Craig

[nvprospector]

When you do the registration, ignore the lat/lon lines completely. Find another map with a known projection of the same area. Find features on both maps and use the coordinates/projection off of the second map.

Craig

Thank you for the reply. The issue with using another map is this map nearly takes up the whole state of Nevada. This brings to light the issue when this map was created geodesy still was only able to determine the datum for a relatively small area for the local origin on the surface. Maps that have projections on the collars for the same year have different aspects depending on which area of the state they where created, and who the surveyor was. Another issue I have encountered with this map is you can tell there is distortion of area between the globe and the map. So the a and b orientations are a little off. I have been able to correct this by adjusting the angular deformation. This is why I was able to pretty much get the map into Ozi. Now if the darn track points will lay correctly across the whole map instead of some areas being way off the mark.

Maybe it is time to sit down and read the book I have for North American Datum’s and formulas that where used before NAD27 was introduced.

-Tim

[nvprospector]

Should of read my emails first. A member of OGC is going to take a look at the map. I loaded it into GRASS GIS and it was sent to him with a pretty pink bow. He said from my description it sounds like the map is using multiple projections and I might have to split the map into three parts in order to create a datum that will work.

-Tim

[craig]

Thank you for the reply. The issue with using another map is this map nearly takes up the whole state of Nevada. This brings to light the issue when this map was created geodesy still was only able to determine the datum for a relatively small area for the local origin on the surface. Maps that have projections on the collars for the same year have different aspects depending on which area of the state they where created, and who the surveyor was. Another issue I have encountered with this map is you can tell there is distortion of area between the globe and the map. So the a and b orientations are a little off. I have been able to correct this by adjusting the angular deformation. This is why I was able to pretty much get the map into Ozi. Now if the darn track points will lay correctly across the whole map instead of some areas being way off the mark.

Maybe it is time to sit down and read the book I have for North American Datum’s and formulas that where used before NAD27 was introduced.

-Tim

Glad to hear you are getting it worked out. Your problem is much more complex than what was described in your initial question.

[nvprospector]

Your right, I did not explan it to well. Re-reading the question it almost sounds like I could not import the map as opposed to creating a custom datum to use with the map.

Playing with the map I can see what the GIS pro was saying, I can see the distortion in the map at a few levels. I really want this map in Ozi. It shows all the stage, wagon and freight routes as of 1905 with in the state of Nevada and Eastern California. This map will really help with my research.

Again, thank you for your reply.

-Tim

[DaveM]

It sounds like you are trying to do a datum shift on import rather than registering the image itself which, I believe, requires the original to have used a set datum (listed or not). Are you sure your map was created with a datum at all or is it possibly a "map graphic" using a fictional flat model of the earth? A lot of older maps (and even some modern road maps) don't use a datum or projection but simply scale out from one point and manipulate the coverage to match.

If that's the case a datum shift may never work to properly fit your map, you will have to rubber sheet it. Prominent marked points on the map and the lat long line intersections should give you a good number of registration points to use.

What format is your map image in? Can you post an image of the map here?

[DaveM]

He said from my description it sounds like the map is using multiple projections and I might have to split the map into three parts in order to create a datum that will work.

-Tim

This sounds like what I am talking about in terms of a "map graphic" or fictional projection model. I'd think that the easiest way forward here is the rubber sheeting method as opposed to trying to actually determine the datums that may have been in the original source material used to create your map.

[nvprospector]

Click here to download the map in tiff format. (http://www.epathfinder.net/pub/maps/dennys_1905.tif)

sheet 54 x 59 cm; Scale: [ca: 1:950, 400]
(W120°00-W114°00/N40°00--N36°00)

Rubber sheeting would work. I already used 9 calibration points for the map. I know that you can add many more if you edit the map file directly. I will have to give that a try.

Thanks

[DaveM]

Click here to download the map in tiff format. (http://www.epathfinder.net/pub/maps/dennys_1905.tif)

sheet 54 x 59 cm; Scale: [ca: 1:950, 400]
(W120°00-W114°00/N40°00--N36°00)

Rubber sheeting would work. I already used 9 calibration points for the map. I know that you can add many more if you edit the map file directly. I will have to give that a try.

Thanks

Cool map. Any chance you have a .tfw (tiff world file) to go with this image? I'd like to load this into a GIS base of Nevada and see what it looks like, how it sits under modern data.

[nvprospector]

This map is from my personal collection and I had it scanned at a place that has a deck scanner large enough for maps. So there is no tfw for the file. The only other place that has this map is the Univeristy of Reno and no one has georef the map yet.

If this map peaks your interests, I also have one that I just got back from the scanning company today that shows all of the military roads listing all minor camps, water points, large camps and forts for Nevada and California that is dated 1873. It also lists places that a unit can fortify if the need arises. What is nice about this new map is it was done by a military surveyor so the datum shift was easy to calculate. Plus, this is not the first map I have worked with from this surveyor so I already have the datum shift programed into Oziexplorer.

[winkosmosis]

I just downloaded the file. I really don't think datums have any relevance here-- it's just a map, not a high precision plat. For example you can tell by looking at it that the roads were hand drawn. You just need to georeference it with a some good points and find a polynomial warp that makes it match the current features. You'll never get it more precise than the scale and the original mapping precision allows.

I'd georeference it for you but I don't have any Nevada data. I'll see if I can find some tomorrow. Can Ozi read projection files or does the whole thing have to be resampled to a geotiff?

[craig]

This sounds like what I am talking about in terms of a "map graphic" or fictional projection model. I'd think that the easiest way forward here is the rubber sheeting method as opposed to trying to actually determine the datums that may have been in the original source material used to create your map.

That's what I was suggesting too, with one caveat.

The datum will become a problem if the lat/lon lines on the original map are used because those lines are tied to a datum. If map features other than the lat/lon grid are located on the original map are identified on a second map where the datum is known, he can rubber sheet to the datum/projection of the known map.

Craig

[DaveM]

That's what I was suggesting too, with one caveat.

The datum will become a problem if the lat/lon lines on the original map are used because those lines are tied to a datum. If map features other than the lat/lon grid are located on the original map are identified on a second map where the datum is known, he can rubber sheet to the datum/projection of the known map.

Craig

Yeah, I like your method. I was just separating out the registration operation into it's simplest form, no need for a referenced 2nd map. Simply find co-ordinates on the map with known points and use a GIS package to rubber sheet them into correct alignment with each other. In looking at his map (the scale, extent and precision) I don't see that much more would be gained by going beyond a simple rubber sheeting.

[winkosmosis]

How's this? I georeferenced with county and state boundaries. As you can see, the points don't really match up. I've georeferenced a lot of old maps and that's just the way it goes. I didn't do 2nd or 3rd degree polynomial transformations (what craig and DaveM call rubber sheeting I think) because there was too much distortion.

I can send you the JPG with the georeferencing file, assuming Ozi can use that data

http://img.photobucket.com/albums/v468/winkosmosis/nevadageoref.jpg

[nvprospector]

How's this? I georeferenced with county and state boundaries. As you can see, the points don't really match up. I've georeferenced a lot of old maps and that's just the way it goes. I didn't do 2nd or 3rd degree polynomial transformations (what craig and DaveM call rubber sheeting I think) because there was too much distortion.

I can send you the JPG with the georeferencing file, assuming Ozi can use that data



Ozi can read georef files. Looking at what you have laid out is the same results I ended up getting when bringing in the map to Ozi. Importing the map is not the issue. The issue is I need to create some sort of datum in order to use the map with moving maps and a GPS system and I am aware that it won’t be accurate.

Ozi only has the currently used datum’s for the globe and the oldest datum that is listed is NAD27. One is able to add different datum shift into the program and this is what I am working on. This isn’t the first time I have wrestled with an old map, it is for some reason this map is giving me trouble and I can’t figure out a datum shift that will work for the map and it is driving me crazy. Maybe I need to step back and take a different look at the issue.

Thanks for the advice thus far.

-Tim

[nvprospector]

That's what I was suggesting too, with one caveat.

The datum will become a problem if the lat/lon lines on the original map are used because those lines are tied to a datum. If map features other than the lat/lon grid are located on the original map are identified on a second map where the datum is known, he can rubber sheet to the datum/projection of the known map.

Craig

I will give this a try. I have a map that is from the 1910 census for the state that will work for this attempt. It is roughly the same size and I do have the projections and the datum shift programed into Ozi already. It will be very easy to move things around to make it work

Tim

[winkosmosis]

What do you mean by datum shift?

Why can't you use whatever datum is default?

[nvprospector]

I can send you the JPG with the georeferencing file, assuming Ozi can use that data

I would like to see that georef file if you don't mind. I would like to see how far I am off with your projections. I am still new with GIS and I still like to compare results.

-Tim

[nvprospector]

What do you mean by datum shift?

Why can't you use whatever datum is default?

Geographic coordinates based on a different datum (and spheroid/ellipsoid) then the one used to obtain the original coordinates. Or, when one converts NAD 27 to WGS 84.

[winkosmosis]

PM me your email address and I'll send the JPG and all the files that go with it (the original TIFF won't work with the georeferencing files).
I'm having weird issues though... It now shows up in Arcmap as an all black image. I'll try viewing it again this weekend.

A datum is really a model of the surface of the earth (sphere, ovoid, lumpy ovoid, etc), and with such an old imprecise map it doesn't have any relevance, especially after georeferencing. If you can, set OziExplorer to the state plane for Nevada, and georeference to that.

[nvprospector]

A datum is really a model of the surface of the earth (sphere, ovoid, lumpy ovoid, etc), and with such an old imprecise map it doesn't have any relevance, especially after georeferencing.
This makes sense on the issue being an imprecise map.


If you can, set OziExplorer to the state plane for Nevada, and georeference to that.
This I can do and will give it a try.

Thanks

[winkosmosis]

BTW, this is where I got the base data

http://www.blm.gov/nv/st/en/prog/more_programs/geographic_sciences/gis/geospatial_data.html

[nvprospector]

BTW, this is where I got the base data

http://www.blm.gov/nv/st/en/prog/more_programs/geographic_sciences/gis/geospatial_data.html

Nice site. Thanks.

[DaveM]

What do you mean by datum shift?

Why can't you use whatever datum is default?

Ditto this question. Once you register or rubber sheet the image it will be whatever datum and projection you reference it to. You are assigning a datum and projection. The image is warped by means of control points to match an existing, known datum and proj (what ever you pick). A datum shift is different, where you are updating the existing projection of a map to a newer datum. This involves displacement of the original co-ordinates on the map to (hopefully) more precise locations but is not a transformation by altering the map image shape itself.

Once your map is registered to a projection you don't need to worry about any datum shifts or what it's original datum was.

[craig]

I agree 100% with David's last response.