generate a raster for a multipart polygon - state of Hawaii - using contour lines

[ChristiNelson1]

Hi there,

Im working with a multipart polygon.  Lets say its the state of Hawaii.  So one feature class called Hawaii (state) that contains a few polygons (whicha are the island boundaries) inside of it...all called Hawaii.

Now lets say I have 2 ft contour data for the islands only.  No bathymetric data.  So I want the output raster to reflect only where the contour data is (land only, nothing beyond the exact polygon boundary for each island).

How do I do this?  Most of the time the raster is generated using an extent boundary, but I only need data for the land (only within each polygon boundary).  The state polygon boundary does not have elevation assigned to it, as it is only a polygon.

Further, how do I get it so that the raster truely reflects the contour data, so that when I click on a contour line, the raster elevation is the same as that of the line?

Thanks,
Christi
I also want the data

[curtvprice]

Before you interpolate values from the contours using a tool like Topo To Raster, set the geoprocessing Mask environment to your Hawaii polygons.

Areas outside the polygon should not have values. This should work with either single or multipart polygons.

[SteveLynch]

Christi

Data are available from various sources, eg. http://hawaii.wr.usgs.gov/hawaii/data.html

TopoToRaster has 6 input data types, PointElevation, Contours, Streams, Sinks, Lake and Boundary.

Boundary will clip the result.

Also, you have to make sure that a representative output cell size is used.

Steve

[ChristiNelson1]

Thanks all,


@Steve -might you know how I would determine what a reasonable cell size is for 2-foot contour data?  Is this what will determine whether the elevation of a point on my raster will match the contour line elevation?

Thx,
Christi

[SteveLynch]

Christi

TopoToRaster is based on a 3rd party piece of software, anudem (http://fennerschool.anu.edu.au/research/publications/software-datasets/anudem). It is based on thin plate splines and is in most cases not an exact interpolator.

If more than one spot height or contour line is present within a cell then averaging is done. You should therefore make sure that you pick the correct or representative cell size.

TopoToRaster is also computationally intensive and the largest output raster is roughly 36 million cells. If a larger raster is required the MARGIN parameter can be used.

With all this said, I've had a quick look at the Hawaii contours and some of them are 50 ft apart.

Hope this helps.
Steve

[ChristiNelson1]

Steve,

You have been incredibly helpful!  FYI, I only used the state of Hawaii as an example, to make my situation clear.  Im really using planning pressure zone polygons (I work at a water district for water resources planning) that are each part of the same pressure zone. 

The contours I am using are 2 ft contours.  These contours were generated from county digital orthophotos that had a 2 ft resolution

Cell size recommendations now?  Im thinking 2 ft.

[EricRice]

Christi,

Once you get a DEM what will you be using it for?  What are the spatial resolutions/accuracies required for your project?  When you say you have 2ft contours, remember that it's only in the Z that you have 2ft resolution.  Steve's point in the Hawaii data (which I haven't looked at) is that in the X,Y some of those contours are 50ft apart. To further his point - If you have don't have 2ft resolution in x,y within your real data, then a raster with 2ft cellsizes is probably overkill. 

You can get 10m DEM's for free for pretty much all of America from USGS if that resolution would fit your requirements.  If you have access to the professional photogrammetrist who derived the data, you should ask them what they think a good resolution is for the data. Also, many digital photogrammetry applications output surfaces as rasters or TIN's directly, then contours are derived from those. They may already have the raster your looking for.

Another option is to create a TIN/Terrain from your contours, then convert that to raster at varying cellsizes to find an appropriate one for your needs.

Regards,
Eric

[EricRice]

Another way to evaluate the spacing would be to convert your contours to points via the Feature Vertices to Points tool.  Each point would have the z value from the contour.  Then take the points, and convert them to raster using the Point to Raster tool, with Cell Assignment Type set to Count.  The resulting raster values will show you how many points were in each cell.  Inspect the result to look for variance in density and data voids.  If the density is to high, reduce the cellsize.  If data voids are too big, increase the cellsize.  It's all about finding balance in this iterative process, but should help determine a close to optimal cellsize for interpolating.

Eric