Unable to produce usable contours & hillshade from GeoBase 1:50K dem

[JakubSisak]

Hello,
The only source of decent country-wide coverage of 1 arc-second (0.75 - 3arc) or roughly 30m resolution dem in Canada seems to be GeoBase (http://www.geobase.ca/geobase/en/index.html)

Unfortunately, this is the only dem I am NOT ABLE to produce smooth contours and hillshade from. The contours come out jagged (much like stairs almost as if the contour followed around 30m pixels) and the resulting hillshade appears as if it was forming platforms. (see attached)

I never have any issues with other dem; SRMT, LiDAR, LIO (Ontario). 

Can someone provide a suggestion?

[DonovanCameron]

I work with Geobase dems often.

The 'staircase' you are seeing is normal, and is almost unavoidable. It can be corrected using a simple smooth or generalize tool on the selected data. Tedious, as you have to visually inspect your data after creating the contours. Smaller scale = more work

[JakubSisak]

Thanks. At least I know am not the only one who has this problem. I keep coming across posts pointing people to GeoBase for higher resolution DEM but frankly it just isnt usable for hydrology, hillshade, etc.

I believe the data is erroneous; i was in contact with some support people from GeoBase and i explained the problem in detail but they do not have the budget to fully investigate such requests.
As for contours they suggested smoothing, but we all know that does not work for contours.

It's almost better to use the source 3 arc-second srmt data then using the data at the existing state.  This data should be comparible to the US 1 arc-second srmt but it is not...

For some areas Canvac/NTDB contours can be used with the Topo to Raster tool but for the most part the contour interval is too sparse to produce anything half realistic...

Could you post an example of a smoothed/generalized GeoBasae 50K dem?

[JakubSisak]

Looks like i am out of luck; see attached map search window from Geobase. It shows the same anomaly.

[EricRice]

Jakub,

The pattern you see in the hillshade is likely inherent to the DEM data.  The striping is described as a product of data collection technique known as manual profiling, and a solution for eliminating the striping is to use the Fast Fourier Transform (FFT).

http://seamless.usgs.gov/faq/ned_faq.php#seven
"GPM2 DEMs can have an appearance of a grid pattern when a shaded relief graphic is produced from the DEM."

http://www.ctmap.com/assets/pdfprojects/destripe.pdf 
Check out page 3; Los Indios Canyon, NM Unfiltered DEM (Shaded Relief) vs the Destriped DEM result.

These links discuss USGS DEM's, but the same probably applies to the GeoBase data.

As for getting better (non blocky) contours with the data you have, I would recommend setting a very small offset to the Base contour parameter.  For example, if you're base was 0, then you should input 0.0001 for this parameter.  The contours will no longer pass exactly through the cell centers which was the probable cause for the stair step look.  Check out the section, "Controlling contour quality" in the link below.  This method should work better than generalizing the contours after creation.

http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#/How_Contouring_works/009z000000vq00000...

[JakubSisak]

Thank You for your response.  I tried both approached with unsatisfactory results. I also received a response from GeoBase explaning that the anomaly has to do with rounding of elevation values.  They also provide some sugestions such as settings for Focal Statistics but it simply does not solve the problem....

Here is the response from Geobase:
Thank you for contacting GeoBase for a follow up on improvements to avoid 'stair stepping' effects found in certain CDED data sets.  Presently there is no change attributed to the status of CDED specification requirements.  CDED data originates from multiple partners and data sources which make it difficult to manage a homogeneous product throughout.
As you may now be aware the Z resolution in CDED files is causing the stair stepping effect.  According to CDED specification standards for this scale, all elevations readings are rounded to the nearest meter of elevation for the entire Canadian coverage with approximately an elevation point every 23 metres by 16-11metres for the 1:50K scale.   In flat areas, the rounding effect of the elevations causes homogeneous elevation plateaux instead of having a very gentle slope because the grid height values varies just a few centimetres.  It's these adjacent plateaux that cause the stair stepping effect. 
In certain cases some smoothing effect may be achieved by using the Focal Statistics tool in ArcMap Spatial Analyst toolbox, neighbourhood tools before processing CDED files.  Depending on this software application you may be able to adjust viewing features with the tool menu for '' Neighbourhood=Circle, Radius=5, Units=Cell and Statistics Type=MEAN''.  But I don't think there is a perfect setting for all CDED files. There is a part of trial and error in the settings for each tile.  Because of this situation and the software you are using you may need to request assistance for your software provider to smooth out the look of you files.
Because no perfect solution was arised from our tests, GeoBase did not provide a universal user guide for smoothing out CDED elevation data, in the Frequently Asked Question segment.