Making Products from High Resolution DEM
August 17 2011 |
I am having trouble creating good cartographic products now that data is becoming more and more detailed. I am using high resolution DEM data (created from 1 m resolution LiDAR data), with this I have two problems;
1) Is it possible or does it make sense to create a 1 m contour layer from this data. I've tried to do some generalizing with focal stats etc. but the contours just keep coming out too messy. This even happens with 5 m intervals. Any thoughts?
2) I'm trying to create a layer showing slopes and eventually mapping hazardous slopes (above a certain percentage). The problem is that this data is mapping some sharp slopes in corn fields which in turn will show up as a hazard (i.e. the till rows from tractors). I tried smoothing the data with focal stats, then converting the raster to polygon to delete slope polys under a certain area BUT all the tiny hazard slopes are so close that they become joined and therefore have a large area. Any thoughts?
With both of these questions, is there any way to generalize or clean the data without compromising too much data integrity but allowing for a visually nice cartographic product?
Thank you very much,
Mapping Center Answer:
For your first question, with Lidar data the vertical accuracy is not particularly good relative to the horizontal density of points so creating contours is not straightforward. This is the reason why you see poor quality contours when generated using lidar data. You can, however, improve the contours by first creating a Terrain dataset with your las points using a z-tolerance pyramid technique. Z- tolerance pyramiding allows you to set up multiple pyramids that honor vertical accuracy requirements (see Terrain Pyramids). Then with a coarser pyramid level (don’t use the full resolution pyramid), create a raster using the Terrain To Raster tool (use natural neighbor interpolation). Then use this raster to extract the contours. This approach will give you a smoother set of contours.
For the second question, performing some resampling on the slope raster may help. A low pass filter or perhaps a median filter type of reclassification with some other conditions could be set up to ignore noisy slope values. This will be a matter of trial and error given your specific dataset. The same might apply to your overall question about generalizing the las data.
If you would like to post a comment, please login.