Determining New Forest Road Gradients

[nickvial]

Hello all,

I'm trying to build roads in ArcMap at a maximum of 15% gradient in ArcGIS Pro 10.7.1. I know where the roads should go, and areas I need to avoid, but I am having trouble determining whether or not the roads are at acceptable gradients without doing it the old school way of hand drawing them out using a ruler, topography lines, and constant field verification. I can do it that way, but was curious if the community knew of a better way to narrow it down. 

Background: these roads need to be at a certain gradient to facilitate log truck traffic in varied forest topography in order to avoid certain land features (Rocks, streams, etc) and still reach the Units. There are also existing roads I need to connect to but no old road prisms. The roads will be excavated (expensive) and it's important for the project to determine if the potential road location is even feasible with all variables considered. 

Thanks for the Help!

[jcarlson]

Do you have a good DEM of the area? And is it that no point of the road should exceed 15%, or are shorter >15% distances likely to be smoothed over during construction?

If you've got a good elevation model there are a number of ways to approach this problem (and all the more so if you have Spatial Analyst), but it gets trickier if you don't.

[nickvial]

John,

I have decent DEM of the area currently displayed as a hillshade. Unfortunately I wont have Lidar in this  particular area until next year. Short segments >15% are acceptable though would be best to keep the length and frequency of >15% segments to a minimum.  I also have spatial analyst. 

Thanks,

Nick

[jcarlson]

So, I'd check out Surface Parameters against your DEM. Obviously, Slope is the main thing you want, but it might also be useful to have Profile curvature.

Once you have your raster(s), I'd visualize the slope raster on a color ramp with discrete intervals, in order to highlight >15% cells. Here's my home area, admittedly, a bit flat. But I can still easily see which areas are a "no go" with discrete symbology.

If you output the curvature raster, too, you could symbolize that on a centered black-to-white-to-black color ramp and set the layer blend mode to multiply, and you'll see the slope classes shaded darker as the cell's curvature diverges from flat. Then you can see where, if you had to cross a high-slope area, whether the ground there is flat or not.

Once you've got that, I'd just eyeball it and start drawing potential routes. Draw more than you need, maybe a few alternatives for each case, if that's feasible.

My Spatial Analyst skills are a bit rusty, but what I think you want here is Zonal Statistics, which can give you the statistical breakdown of the intersected cells. However, I believe that will only work with an input polygon, so:

1. Run Buffer on the lines, not a large amount, but maybe the width of a raster cell, to ensure that the polygons will cross contiguous raster cells in the analysis.
   Make sure that you do not merge the buffered shapes, you want a separate polygon for each line.
2. Run Zonal Statistics, but have it run per-feature. You may have to do this in Model Builder and use an iterator.

There might be other, better options. Possibly using "statistics as table", to get all stats in a single output, then joining that back to the lines? Or "zonal histogram", so you can see how much and how often the values exceed 15%.

In any case, this can at least identify which lines have a median slope of less than 15%, and if they exceed 15%, by how much.

[jcarlson]

Alternate idea: interactive elevation profiles. These could give you a nice at-a-glance view of your line, and includes both average and maximum slope. And if the slope itself is more important than the elevation, you could also use your output slope raster (see other post) in place of a DEM, and your "elevation" profile could show you the slope over the entire line.