In my office we have a copy of the LiDAR data by county, and I was wondering if the DEMs were available as HUC8s (mosaics(?)). If not, is there a recommended method for creating DEMs by HUC8s. I've tried a couple different ways I know, but encountered some computer problems. One issue I am having is that I am examining a watershed the spans two counties, and at the county borders, County A's DEM overlaps with County B's DEM, and at the same location, the DEM's have different values (+/- 1 meter (?)), Any advice on how to handle this?
Thank you very much,
your approach of mosaicking DEMs together was the right one but you ran into the typical issues. If the overlapping areas of the neighboring DEM's (at the county border) would have the correct elevation for both DEM's you could use a MEAN as mosaic operator, which calculates the average elevation for each overlapping cell based on the input elevations. If you trust one DEM more regarding its elevation values then add that one as last DEM to the toolbox and use LAST as mosaic operator (takes only the DEM values from the last DEM and ignores those of others).
But now you have to find out which DEM has the correct elevation values (by comparing with other elevation data sources). An elevation difference of +/- 1 meter is way to high, the vertical accuracy for Lidar-based DEMs should be less than 0.18 meter! Look up the RMSE in the metadata! Are the DEM's from the same Lidar acquisition project?
Also when mosaicking DEM's together it might happen that you have a gap between neighboring DEM's instead of an overlap area and that gap would result in NODATA values in your mosaicked DEM and might cause all kind of trouble in later analysis (for example hydrological applications). Here it is recommended to apply a conditional statement to the mosaicked DEM to remove all NODATA values (IsNULL) with a focal statistic (for example a 3 by 3 neighborhood and the MEAN again).
Thanks Grit! I just noticed that one of my counties was a different 'collect'/acquisition than the other ones, so I think that is likely the problem. I was really stumped as to how there could be different values!
I concur with Grit on the approach - use a blend/mean approach when mosaicing overlapping data from different counties that come from different collects.
For those counties that are in the same collect the overlapping data should be the same.
Powered by mwForum 2.27.3 © 1999-2012 Markus Wichitill
Net::CIDR::Lite module not available.