Slope Calculation Problem

KenBerglund · ‎02-11-2011

Hello,

I am having problems simply calculating % slope from a USGSDEM. Even though it had a GCS_North_American_1983 projection, I defined its projection to be that as well after reading some posts on the same issue. The resultant slopes have values into the millions. I then tried to project the DEM into UTM to ensure the units would be in meters. This caused an error. Under the spatial reference in the DEM, the "Linear Units" are blank. This is why I think the project to UTM failed as it does not have linear units defined. Anyone have any ideas? This should have been a simple task but has been troublesome. I guess I am asking, why would the slope % be in the millions? Or any ideas on why the project to UTM failed? Or how to define linear units?

Thanks for any help

Ken

PatrickTaurman · ‎03-02-2011

http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//009z000000v2000000.htm

Also I am curious about your answer "now my values are reasonable (0-100% or 1-90 degrees)". 45 degrees = 100% slope. So if you are getting an answer of greater than 45 degrees, your slope percent has to be greater than 100%.

As for the reprojecting, you used the correct workflow (Project Raster tool) to go from NAD 1983 to State Plane.

You may want to contact Support so someone can take a closer look at everything.

Patrick

RyanHollamby · ‎03-02-2011

Reading through that article, I have to wonder if there is something wrong with my Z units. Mine is always 1, not "zFactor = 0.3043" as described. I thought if my x,y,z are in meters that I leave my z value as 1 (?).

Good point about the degree problem. I will continue you looking and trying.

Thank You!
Ryan

PatrickTaurman · ‎03-02-2011

One thing you need to take care of is that everything is in the same units. If elevation values (Z) are in meters, but you define the Projection (XY) as feet, you will not get the correct slope. This is where the Z Factor comes into play. Are you positive your Z value are in meters?

RyanHollamby · ‎03-02-2011

One thing you need to take care of is that everything is in the same units. If elevation values (Z) are in meters, but you define the Projection (XY) as feet, you will not get the correct slope. This is where the Z Factor comes into play. Are you positive your Z value are in meters?

Both XY & Z are in meters.

Attached is what is looks like now. The values are correct, the red is steeper terrian and that is what the terrian looks like in the field, but has the grid within it which is ruining the overall projection. The values in the Table of Contents is between 0-90.

Seriously thankyou

RyanHollamby · ‎03-02-2011

These stripes look like artifacts produced by some forms of tiled compression followed by projection. This suggests the problem is inherent in the original DEM. Take a close look at it (hillshade it if you can) to verify, then go back to the source to track down where the corruption occurred. If it's in the source, the best way to eliminate the stripes is by constructing a custom spatial filter from the FFT of the original grid (which, unfortunately, is not something ArcGIS can help you do).

Hmm...I am going to try and select several other of the formats from the USGS's nationalviewer website and see if I can correct that way. I'll see if the other formats that I can download work

The hillshade also have the artifacts embedded.

Thank you!
Ryan

PatrickTaurman · ‎03-02-2011

Bill is correct, you will need to apply a Fast Fourier Transformation to your DEM. There are several articles describing the striping you are seeing (including from the USGS website).

http://ned.usgs.gov/Ned/about.asp

http://seamless.usgs.gov/faq/ned_faq.php#seven

http://www.ctmap.com/assets/pdfprojects/destripe.pdf

RyanHollamby · ‎03-02-2011

Bill is correct, you will need to apply a Fast Fourier Transformation to your DEM. There are several articles describing the striping you are seeing (including from the USGS website).

http://ned.usgs.gov/Ned/about.asp

http://seamless.usgs.gov/faq/ned_faq.php#seven

http://www.ctmap.com/assets/pdfprojects/destripe.pdf

Well all of those links have images exactly how mine are turning out. I am working with the Pikes Peak region, so I am surprised that it falls along with "In areas where the only available DEM is produced by older methods, then "striping" may still occur."

Is there someplace else that I should be getting my DEM from? Seems strange to have this many issues just from downloading a DEM from USGS and getting the slope values.

When I have the raster in my data set, before performing the slope stuff, there is no artifacts and very clear.

Thanks again,
Ryan

RyanHollamby · ‎03-02-2011

So I went and grabbed some DEM's from other parts of the country that are 'updated' and wouldn't be subjected to these artifacts and got the same dang thing. So that means I am still doing something wrong.

I appreciate everyones help. So I am going to run through a bunch of different ways and try to figure out what the heck is going on.

Thanks

RyanHollamby · ‎03-02-2011

WOW! I finally figured it out!

So after you set me straight on defining the projection and projecting the raster, getting the x,y, and z to meters, I figured out the missing piece.

The data from the USGS was fine as a ArcGrid. The link I was missing was when Projecting the Raster. The artifacts that I was generating was from the resampling techinque. I didn't know the difference between nearest, bilinear, cubic or majority. Since it was optional I brushed it off (MISTAKE). As I worked back through each step...researching each step I came across the meaning of the techniques. Went back in to the toolbox and then read the little help description. For the reasons in bold I went with Bilinear.

From the help:

"The NEAREST option, which performs a nearest neighbor assignment, is the fastest of the four interpolation methods. It is primarily used for categorical data, such as a land-use classification, because it will not change the cell values. Do not use NEAREST for continuous data such as elevation surfaces.

The BILINEAR option, bilinear interpolation, determines the new value of a cell based on a weighted distance average of surrounding cells. The CUBIC option, cubic convolution, determines the new cell value by fitting a smooth curve through the surrounding points. These are most appropriate for continuous data and may cause some smoothing; also, cubic convolution may result in the output raster containing values outside the range of the input raster. It is not recommended that BILINEAR or CUBIC be used with categorical data because the cell values may be altered. "

Anyways now I have a perfect slope file!!

I appreciate you helping me through this. Little did I know I had 3 big mistakes. I think that this sample technique is a common contributor to people's problems (seems how I read every single slope thread in this forum). The people that state the projection is set, values are the same (meters, feet, etc) yet they still get grids.

This thread is where I started: http://forums.arcgis.com/threads/24718-Slope-Values-Projection-Issue?p=82179#post82179

Thanks again. I hope someone that is having this problem tries this as a solution.

Ryan

LauraShaffer · ‎04-07-2011

Was this ever resolved? I'm having the same problem, but I'm still confused about what I'm supposed to do with the nice fresh NED tif I just downloaded from the USGS Seamless Server. I've never had trouble calculating slope etc. from DEM files downloaded elsewhere.

Without doing anything to the NED, ArcCatalog says that the Spatial Reference for it is GCS_North_American_1983. Is that the same as the projection? When I downloaded it, USGS specified the output format as "ArcGRID NAD 83 Geographic." What am I supposed to project it to? Everything else in my map is in NAD_1983_UTM_Zone_12N (it's northern Arizona).

I understand that the horizontal coordinates are probably in decimal degrees and the vertical are probably in meters. I'm not sure how to convert these to something usable (and I'm not sure where to look to confirm the units on these).

I tried projecting the NED to the 12N projection, then doing a raster calculation to convert the elevation (z) from meters to feet, then tried getting slopes (via both 3D and Spatial Analyst, is there a difference?) but the % slopes were still in the thousands. After all that I downloaded a fresh NED in case I messed up the original in the process somehow (superstitious, maybe but this is driving me crazy!).

So . . . how do I get from an NED from USGS to slope? Thanks in advance!!

-Laura