General Category => Model Formulation & Parameters => Topic started by: jmkassak on February 02, 2011, 04:35:40 PM

Title: Problem with elevation analysis?
Post by: jmkassak on February 02, 2011, 04:35:40 PM

We are having an issue with the elevation analysis where several of the "5th percentile" results are identical to one another. Specifically, inland fresh marsh, tidal fresh marsh, mangrove, inland open water, ad irreg. flooded marsh all have a value of 8.2744. Additionally, the "min" for inland fresh marsh and tidal fresh marsh are also 8.2744. I looked more closely at elevation analyses from previous runs at other subsites, and this issue actually came up once before. It's unclear why it only occurs some of the time.

I am wondering if there's possibly an issue in the elevation analysis code. I did consider the possibility that this issue was a result of error in our input files. But thinking conceptually, that doesn't seem possible. The inputs are processed separately from one another; the values within the habitat dataset and the elevation dataset wouldn't influence one another during the input creation. Even if there were errors in our elevation data or the habitat data, it still wouldn't lead to identical values for the 5th percentiles. To confirm, I did take the final input text files and convert them back to raster, and compared them to the original rasters from which we make the input text files, and they're identical.

Any thoughts on what might be going wrong here?  I'd like to use the elevation analysis to see if we need to make any revision to the conceptual model ranges, but I'm not feeling confident at this point that I can rely on the current analysis for that purpose.

Title: Re: Problem with elevation analysis?
Post by: Jonathan S. Clough on February 07, 2011, 07:50:08 AM
We have seen this before.

This may stem from the fact that over five percent of your data for these cover classes cover an area which has "no-data" which has then been assigned an arbitrary value.  Often LiDAR over water is assigned a value of zero or close to zero and this then gets changed by the NAVD correction factor to something else.

I'm not ruling out the possibility of a bug in the analysis but I don't think it's likely.  See if you can find a large flat area with that elevation value. 

Good luck!