Select to view content in your preferred language

Topo to Raster

4207
15
12-03-2009 03:54 PM
TedCronin
MVP Alum
Has this tool been updated from the old fortran code to something a bit speedier?  It currently is very difficult to use with our contour dataset.
0 Kudos
15 Replies
EricRice
Esri Regular Contributor
Hi Ted,

The Topo to Raster tool has not been updated for 9.4.  Its still the same version of ANUDEM as it is in prior releases.  Have you considered processing your data in several pieces utilizing the Margin in Cells option and then moaicking the adjacent rasters? You might also want to consider creating a Terrain from your large contour dataset.  Many of the surface analysis tools have been enhanced in 9.4 to directly analyze Terrains.  You can always convert the terrain to a raster later if necessary.

Regards,
0 Kudos
TedCronin
MVP Alum
Hi Ted,

The Topo to Raster tool has not been updated for 9.4. Its still the same version of ANUDEM as it is in prior releases. Have you considered processing your data in several pieces utilizing the Margin in Cells option and then mosaicking the adjacent rasters? You might also want to consider creating a Terrain from your large contour dataset. Many of the surface analysis tools have been enhanced in 9.4 to directly analyze Terrains. You can always convert the terrain to a raster later if necessary.

Regards,


BUMMER. Will have to consider the processing of chunks, just seems like a pain to have to do it this way.

Love it, perhaps you can talk to Clayton regarding contours to terrains as a viable workflow.
0 Kudos
EricRice
Esri Regular Contributor
Ted,

In 9.3 you can generate a Terrain/TIN from contour data.  For SF_Type choose mass points or softline.  The same applies to 9.4.  If you already have a Terrain and want to include the contours make sure the feature classes reside in the same feature dataset as the terrain otherwise you will not have access to them when executing Add Feature Class to Terrain.

Check out this page to learn more about surface feature types.

http://webhelp.esri.com/arcgisdesktop/9.3/index.cfm?id=3393&pid=3388&topicname=Representing_terrain_...

In 9.4 there is also some new renderers for Terrains; Terrain Contour and Terrain Point renderers.

On a side note, contours are generally a derivative of a surface so you should try to avoid a situation where contours are the only input to creating a surface.  Photogrammetrically derived mass points is probably one of the best sources of data to build a terrain with, but really you want a combination of SF Types to fully represent the surface.

Here are some common types of data used to build terrains.

http://webhelp.esri.com/arcgisdesktop/9.3/index.cfm?TopicName=Types_of_source_data_supported_in_terr...

Thanks for the feedback.
0 Kudos
TedCronin
MVP Alum
Ted,

In 9.3 you can generate a Terrain/TIN from contour data. For SF_Type choose mass points or softline. The same applies to 9.4. If you already have a Terrain and want to include the contours make sure the feature classes reside in the same feature dataset as the terrain otherwise you will not have access to them when executing Add Feature Class to Terrain.

Check out this page to learn more about surface feature types.

http://webhelp.esri.com/arcgisdesktop/9.3/index.cfm?id=3393&pid=3388&topicname=Representing_terrain_...

In 9.4 there is also some new renderers for Terrains; Terrain Contour and Terrain Point renderers.

On a side note, contours are generally a derivative of a surface so you should try to avoid a situation where contours are the only input to creating a surface. Photogrammetrically derived mass points is probably one of the best sources of data to build a terrain with, but really you want a combination of SF Types to fully represent the surface.

Here are some common types of data used to build terrains.

http://webhelp.esri.com/arcgisdesktop/9.3/index.cfm?TopicName=Types_of_source_data_supported_in_terr...

Thanks for the feedback.


Thank you Eric Rice.  First off, good to know, always good to be reminded of these things.

Second, will have to take a look at the renderers.
0 Kudos
KimOllivier
Honored Contributor
"The Topo to Raster tool has not been updated for 9.4. Its still the same version of ANUDEM as it is in prior releases." ie 4.5

What a pity, I was hoping for an updated version of ANUDEM, say 5.2, that now uses breaklines and has improved the model in other areas.

I find I have to tile up my topo data to build a good DEM but unfortunately there are a lot of artifacts when that is done.

1. You have to handle split lakes separately to find the level and mask them afterwards.

2. Islands in lakes are not handled, so these have to be put back in.

3. Large features on the edge do not get modelled properly unless you have an enormous cell buffer.

4. Floodplains are very hard to get looking plausible especially if you add stream lines that meander.

5. Every now and again there is a spike in the model caused by some unknown issue in the data.

6. Using a file based parameter list does not cut it because the parameters need to be dynamic to handle a series of tiles.

7. There is a limit to the number of tiles that can be merged, so it has to be done in a series of ever larger builds. Replacing a faulty tile is very difficult because of the complicated merge.

8. The limit command does not work to limit the data processed so I have to clip all participating layers for each tile.

9. Hillshade grids will be one pixel short all around the tile, so they cannot be merged later unless you build oversize tiles and then clip again.

10. Some tiles are too sparse to build, for example small islands off the coast, so the tiles have to be combined into a single non-standard tile, processed and then clipped out.

The process is extremely complex so I have reverted to a series of AML and Workstation that run to completion.
0 Kudos
briankelly
Deactivated User
Well something changed.  I have 9.4b installed on a clunker of a laptop with 7/32 and 2gb of ram.  Topo to Raster worked on it where TtR failed on the same contour dataset on a loaded workstation with 7/64, 4gb and 9.3.1 and 3d (failed to allocate enough memory error after fresh reboot).  I have yet to install the SP if that matters.

Maybe it ran because i let it save to the default gdb instead of as a grid on the file server...

I used default settings except the cell size (32.814 feet).  The contours covered an area of approximately 1 million acres.

Notes from the run:
it was relatively fast
at one point that status bar was over 200%

some interface feedback:
the geoprocessing status is kinda hard to read on a small screen (1200x800)
what happened to the ArcToolbox Index tab?  It was fast and you didn't have to hit a button to find a tool. 
can i vertical dock toolbars?

Sorry if these points have been addressed; haven't really had a chance to browse the forums yet.
0 Kudos
JasonMills
Occasional Contributor
I would like to see ANUDEM updated as well.
0 Kudos
PeterWilson
Frequent Contributor
It very frustrating that ESRI hasn't updated the version of Annudem that is used within "Topo To Raster". Annudem 5.2 has greatly improved the size of DEM that one can produce and generates DEM from contours very successfully. The "Topo To Raster" is becoming obsolete and our firm will be looking towards other software vendors if ESRI doesn't realise they need to keep up with the times and update their algorithms used in their packages. The look and feel of a software package is not nearly as important as the functionality. With the release of ArcGIS 10 I would have expected more.

Regards
0 Kudos
PeterWilson
Frequent Contributor
"The Topo to Raster tool has not been updated for 9.4. Its still the same version of ANUDEM as it is in prior releases." ie 4.5

What a pity, I was hoping for an updated version of ANUDEM, say 5.2, that now uses breaklines and has improved the model in other areas.

I find I have to tile up my topo data to build a good DEM but unfortunately there are a lot of artifacts when that is done.

1. You have to handle split lakes separately to find the level and mask them afterwards.

2. Islands in lakes are not handled, so these have to be put back in.

3. Large features on the edge do not get modelled properly unless you have an enormous cell buffer.

4. Floodplains are very hard to get looking plausible especially if you add stream lines that meander.

5. Every now and again there is a spike in the model caused by some unknown issue in the data.

6. Using a file based parameter list does not cut it because the parameters need to be dynamic to handle a series of tiles.

7. There is a limit to the number of tiles that can be merged, so it has to be done in a series of ever larger builds. Replacing a faulty tile is very difficult because of the complicated merge.

8. The limit command does not work to limit the data processed so I have to clip all participating layers for each tile.

9. Hillshade grids will be one pixel short all around the tile, so they cannot be merged later unless you build oversize tiles and then clip again.

10. Some tiles are too sparse to build, for example small islands off the coast, so the tiles have to be combined into a single non-standard tile, processed and then clipped out.

The process is extremely complex so I have reverted to a series of AML and Workstation that run to completion.


Hi Kim

I would like to find out if you would mind if I drop you an email. I'm currently completing my Masters in GIScience and I looking to use ArcHydro and Annudem (due to the limitations of "Topo To Raster") to produce quaternary catchments and river streams for South Africa. I'll be using 20m contours that cover the whole country and digitised and rectified river centre lines that will be 3D extracting the Z values from the contours and linearly interpolating Z values between.

Regards
0 Kudos