Extremely Long Find Equal-Value Areas

digitalaviatordigitalaviator Global Mapper UserPosts: 76Trusted User
edited January 2014 in Technical Support
Hi Mike,

I've been doing some work on vegetation mapping lately using NLCD 2006 which is an 8 bit indexed palette raster (I have cut out Utah and exported to an 8 bit indexed raster), and turning the 30m x 30m pixel image into a vector by using Find Equal-Value Areas.

I've noticed it's taking an extraordinary amount of time to go through the final stage of Connecting Edges, it did the initial couple of stages within about 12 hours but it's now been connecting edges for 48 hours. It has been stuck saying estimated Finish: Jan 22 - 05:05 for 48 hours now and windows says GM is (Not Responding). However in the task manager it is showing about 25% CPU usage (i.e. one core of my quad core CPU) and about 50% of my system RAM of 8GB. The values on CPU and RAM change which indicates GM is still working. This happens both on GM 12.02 and GM 15 (in demo mode) both 64 BIT.

On smaller areas the render times are exponentially quicker (e.g. if I were to break the image into 1/4 of the size it's about 10 times faster all together).

I had a power outage after two days of rendering the last time which has meant I have to re-start the process again.

Projection is Albers (NAD83) from the default NLCD 2006 dataset.

Should it be taking this long for it to connect edges, when it's able to find the equal area values within hours? Why would connecting edges take many many many times longer?

Cheers,
Paul.

Comments

  • global_mapperglobal_mapper Administrator Posts: 17,238
    edited January 2014
    Paul,

    The 'Connecting Edges' step can be very slow for extremely complex areas not because it takes a long time to follow the edges but because it is also trying to determine the island layout at the same time and possibly fix any invalid polygons. Do you have the option checked at the bottom of the equal value area dialog to chop up areas if invalid results (usually self-interesting) are found? If not then the slowdown is only due to the numerous polygon-in-polygon tests used to determine which areas are holes/islands in others and how they are stacked (i.e. you could have a large area with a hole in it, then an area inside that hole, with a hole inside it, etc.

    Thanks,

    Mike
    Global Mapper Guru
    geohelp@bluemarblegeo.com
    Blue Marble Geographics for Coordinate Conversion, Image Reprojection and Vector Translation
Sign In or Register to comment.