TransDEM Forum

To start I downloaded all 1/9 dem's for the state of delaware. I then imported the dem into arcmap 10.1 and processed the dem's for the USGS quadrant tiles 7.5. My plan was to work one quadrant at a time and merge them together as I finish them. As you know states in many cases do not have a perfect 90 degree plane, so some of the dem is being clipped by transdem. I noticed when I was exporting the dem I was loosing some area (tried expanding the area by changing the area in the in the north, south, east, and west boxes), but it would not let me expand the area to capture the complete dem. To counter this my plan is to obtain some of Maryland's 1/9 dems merge them together with files of Delaware, then extract information is smaller proportions if needed. Then reprocess the dem in transdem to counter the lost area.

(Vector File)
In this process I kept getting an error when trying to export a rail line shape file that I made in ArcMap 10.1. The issue was the vertex crossed over the area that transdem did not draw, and to get it to finally draw I had to reposition the vertex inside the outer bounding limits of the exported dem to get it to export the vector file(rail lines).

(Dems)
The usgs ground representation information:
1 Sec usually has raster pixel representing of an area of 30 x 30 meters. The files are not big but you loose lots terrain features.
1/3 Sec usually has a raster pixel representation of an area 15 x 15 meters. The files are not all that big, and I would consider this for anyone just looking to make and over all general layout where many terrain features will be captured. (To produce a trainz terrain at 5m with this file involves alot of inturpolation of data that was not precisely measured or missing data that needs to be filled in by the computer, however a 10m terrain setting would involve much less inturpolation). (Some road and rail beds can be seen.)
1/9 sec usually is derived from Lidar. I seen rasters that represented from 1 x 1 meter square to 3 x 3 meter square. If you want detail in your route these are the files of choice, but remember this data is highly intensive. Along with this kind of detail comes along lots of file storage needed. (If looking for a detailed 5m terrain for trainz I recommend this). (Depending on detail the raster was derived from you could see a culvert or even small narrow water drainage, road, and rail beds are usually very distinguished terrain features).


I have to do things manually as I am trying to model railroad layouts in the late 1890's to 1950's. I have to obtain historic aerial photo imagery, and georeference it if need be in ArcMap10.1, then I can finally start to draw in road and rail line vector files as needed.

Maybe someone can answer this question I have been looking for a image or atleast the location where the roundhouse was located in the Delmar area on the border of Deleware & Maryland. I read a document that Delmar had a roundhouse, but could not find any aerial historic photo of it. The document just touched on it that Delmar had a roundhouse, but no detail was given to lead me anywhere. Main thing I saw was that Delmar was pretty much destroyed in early 1900's, and again by 1910's by a major city fire. I was hoping to atleast find remains of the structure in the 1937 photos if there was one. I checked all of delaware's historic aerial photos from 1937, 1954, and to current looking for even building remains after being burned , but having no success. I however did not look over historic aerial photos of Maryland yet to see if it may have been placed on the Maryland side of the city.

Hope some of this info helps others.

Thanks
Greg L.

G'day greg4933,

Surely, Greg, given that the 1 arcsec data has a resolution of 30 meters (and it does), would not the 1/3 arcsec data, therefore, have a resolution of 10 meters (30*1/3=10, when I went to school, although I will admit that was a looong time ago)? Every 1/3 arcsec set of data that I have ever downloaded from the USGS has shown THAT resolution (10 meters) when opened in TransDEM, without failure! This resolution fits into the standard 10 meter Trainz baseboard resolution PERFECTLY, as documented in the "paperwork" accompanying TransDEM and should, therefore, require NO intErpolation (don't you youngsters know how to use Spell Check?)...

...although I have never used an external vector file to provide the 'data' for a route polyline, I have noticed that it is 'normal' for TransDEM to ignore any portion of a polyline that extends beyond the area of the loaded DEM (if any part of it is outside the 'bounding box')...

...incidentally, if the amount of data for a 1 arcsec DEM is "x", then the amount data for the same area in 1/3 arcsec resolution will be 9 tines that (9x) and the data for the 1/9 arcsec resolution will be another 9 times that (81x)...

Jerker {:)}

Just finished doing a GSD on a couple 1/3 files. For the new GIS people GSD = (Ground Sampling Distance). I found a range between 8.2 -11.7 meters with their current saved format. Many professors just said that for the most part you can assume that 1/3 will have at least a resolution of 15 x 15 meters. From the test samples that I did they are right! To say that 1/3 is a true 10 x 10 meter is false as a couple of samples that I did would have failed having a value over 10.5.

G'day greg4933,

...ooops! Indeed, Greg, you found me out, there. I had meant to use the word "times", the error being nothing more than a typographical one but it should be noted that there is an actual word spelt "tine" (for which the plural form exists, as in; "the tines of the fork were bent"), so no Spell Checker would have picked up that error (that one "escaped" the proof reader). There was never any rudeness intended in my question - it was asked civilly and I was expecting an equally civil response...

...now, if your measurements and calculations are correct, then all I can say is that the USGS had 'got it' horribly wrong! As I said, every download of 1/3 arcsec data I have ever made from their sources has returned a value of 10 in the resolution text box readout in TransDEM. Even if the true value is slightly less or slightly greater than this, FOR OUR PURPOSES, it is close enough to 10 to be called "10" (and any inaccuracies will go unnoticed)...


...I'm afraid I'm not at all sure of that which you mean by ..."...Many professors just said..."..., could you kindly explain this?...

Jerker {:)}

USGS (and many other agencies) publish DEMs in Plate Carrée pseudo projection. This is a Cartesian coordinate system with the units being fractions of arc degrees, usually identical for both principal axes. Unfortunately, spacing between DEM points depends on latitude, when converted into metres. It will only be identical for lat and long at the equator. At moderate latitudes the aspect ratio is more like 5:4 or 4:3.

When TransDEM opens a Plate Carrée DEM, it computes DEM point spacing for the given longitude and southern and northern edge latitude. It then determines a default value for the new uniform grid width in metres, based on the data gathered. The exact conversion results are shown in the import dialog. For example, for Chicago, IL, at N 50°, 1/3 arc sec convert into 8 m east/west (and 10m north/south). TransDEM will suggest 10m for the metric grid width. In the next step TransDEM will then re-sample the Plate Carrée DEM to a metric one, applying UTM coordinates.