Advice for using the slope tool in Spatial Analyst

ZacharyLehmann · ‎08-27-2014

Hey all,

I'm by no means an expert with GIS, but I have been using the software for over 8 years and know my way around pretty well. I've been having some trouble with the slope tool that I'm hoping this community might be able to help with.

I'm using 1-meter LIDAR data to create this slope data set. The result is the image attached here. Now at first I thought there was some error with the tool since it is making these lines all over the data set that are basically where the pixel values jump from one value to another. That's how the tool is supposed to work and apparently there are a lot of areas where the adjoining cells are the same value leading to this weird topo line look.

My question is does anyone know of any tips for smoothing this surface out. The model I'm trying to create needs smoother surface than this to really work and I'm not sure how to go about it considering that it's technically doing everything it's supposed to.

Does that make any sense/can anyone help?

Appreciate it.

Ross_PaulMartin · ‎08-27-2014

Zachary,

Assuming your are deriving your slope from a raster elevation model, which was derived from a point cloud - the output should be a continuous surface. If you have classified your slopes into small range values, the inevitable output will be a notchy or stair step like vectorized slope model. It is best to consider the purpose of the slope tool or question and start from the continuous surface model and use Grid Reclass to make derivative products that answer just these specific questions. Always start from the Raster as your base input layer and then work the question accordingly. If you wish to create slope classifications as a standard map, you can generate a smooth output more easily if you choose fewer classes over a large range of slope. The rasterized cell steps will be less visible then. It can be further abstracted for aesthetic purposes if you run smoothing operations over the vector line work, but there is hazard to making classified discrete data into good looking pictures. This is especially true if someone then uses the data beyond the NMAS standard or appropriate scale (since it looks good).

I looked at your image sample and the "topo like" lines looks a lot like the base surface you have is not showing continuous values (perhaps already classified). The reason for this is that you have linear features in the image that have a consistent slope change of 40%. Unless your LIDAR collects are of rice pattys, the slope values will never jump on linear features unless they are cliffs or man made objects. Your image was full of them and that makes me think the base raster of the slopes was already classified in some fashion.

View solution in original post

Ross_PaulMartin · ‎08-27-2014

Zachary,

Assuming your are deriving your slope from a raster elevation model, which was derived from a point cloud - the output should be a continuous surface. If you have classified your slopes into small range values, the inevitable output will be a notchy or stair step like vectorized slope model. It is best to consider the purpose of the slope tool or question and start from the continuous surface model and use Grid Reclass to make derivative products that answer just these specific questions. Always start from the Raster as your base input layer and then work the question accordingly. If you wish to create slope classifications as a standard map, you can generate a smooth output more easily if you choose fewer classes over a large range of slope. The rasterized cell steps will be less visible then. It can be further abstracted for aesthetic purposes if you run smoothing operations over the vector line work, but there is hazard to making classified discrete data into good looking pictures. This is especially true if someone then uses the data beyond the NMAS standard or appropriate scale (since it looks good).

I looked at your image sample and the "topo like" lines looks a lot like the base surface you have is not showing continuous values (perhaps already classified). The reason for this is that you have linear features in the image that have a consistent slope change of 40%. Unless your LIDAR collects are of rice pattys, the slope values will never jump on linear features unless they are cliffs or man made objects. Your image was full of them and that makes me think the base raster of the slopes was already classified in some fashion.

ZacharyLehmann · ‎09-09-2014

I appreciate the answer! I couldn't figure out how the data was becoming classified. Turns out when I was extracting a portion of the DEM I was using I had the bit number set too low. I changed it to 64-bit so the raster remained continuous after extraction and now it works perfectly!

Ross_PaulMartin · ‎09-09-2014

Great news. Was my answer helpful?

Sent from my Verizon Wireless 4G LTE semi-smartphone

ZacharyLehmann · ‎09-09-2014

Indeed!