Jump to content

 
Photo

Techniques in DEMs for Trails Illustrated (National Geographic) Maps

- - - - -

  • Please log in to reply
4 replies to this topic

#1
Greg Corradini

Greg Corradini

    Contributor

  • Validated Member
  • PipPip
  • 12 posts
  • United States

Hello,
I've been using National Geographic's Trails Illustrated maps to do hiking in the Seattle area and I'm a amazed by the DEM backdrop (You can get good zoom detail of a sample map here http://www.natgeomaps.com/ti_826 just by clicking on the 'zoom' icon).

Did the cartographers use image classification to generate the colorful areas of forest and rock basin to be part of the DEM (especially look at the Mt. Baker region of the map in the link above)? See those nice shades of green, brown and tan -- oh yeah!

I'm looking for someone familiar with these maps and raster processing to offer some high-level advice on the steps to approximate something like this on my own. Most of my work deals with vector data but I'm familiar with different analytical raster procedures. I have access to relatively good LIDAR data and hill shades. I'm assuming I could do an image classification using the hill shades and multispectral Landsat data. I'll be using GDAL to do most of the work (if possible).

Thanks
Greg

#2
CGIS

CGIS

    Contributor

  • Validated Member
  • PipPip
  • 37 posts
  • Gender:Male
  • Location:Canada
  • Canada

Hi Greg,

You're right about the image classification - I've done this before using LandSAT multispectral in IDRISI - using unsupervised classification. Feed in bands, and group the analysis results into logical groups. Take the groups (classification values on the pixels) and grab just the ones you want (coniferous, deciduous, rock, ice/snow, water, etc..) and separate these into TIF files, named according to class (rock.tif, forest.tif, etc..).

Now, in whatever software you like to use, and with whatever method you like to use, create your hillshade image, save as TIFF.

Ensure that the hillshade TIF and classified areas TIF have the same dimensions (pixelsX, pixelsY, pixel aspect) so that they align as they should.

Now, load the TIFs into photoediting software (pshop for example), and stack them. Use select-by-colour, or overlay methods to combine them, and use something like colour overlay to adjust the colours (green hue, tan, hue, etc..)

Now, save out a flattened TIF and bring that back into your GIS.



Andrew Schroeder

#3
Greg Corradini

Greg Corradini

    Contributor

  • Validated Member
  • PipPip
  • 12 posts
  • United States

Hi Greg,

You're right about the image classification - I've done this before using LandSAT multispectral in IDRISI - using unsupervised classification. Feed in bands, and group the analysis results into logical groups. Take the groups (classification values on the pixels) and grab just the ones you want (coniferous, deciduous, rock, ice/snow, water, etc..) and separate these into TIF files, named according to class (rock.tif, forest.tif, etc..).


OK, that's what I thought and I'm already doing the right next steps.

Now, in whatever software you like to use, and with whatever method you like to use, create your hillshade image, save as TIFF.

Ensure that the hillshade TIF and classified areas TIF have the same dimensions (pixelsX, pixelsY, pixel aspect) so that they align as they should.

Now, load the TIFs into photoediting software (pshop for example), and stack them. Use select-by-colour, or overlay methods to combine them, and use something like colour overlay to adjust the colours (green hue, tan, hue, etc..)

Now, save out a flattened TIF and bring that back into your GIS.

Andrew Schroeder


Great tips, thanks for the steps and advice Andrew! I'll post my finished product up here for others to see when it's done

#4
Greg Corradini

Greg Corradini

    Contributor

  • Validated Member
  • PipPip
  • 12 posts
  • United States

also, i came across this site which has some nice details about the different use of colors with shaded reliefs.

http://www.reliefshading.com

#5
CentralAmericaExpe

CentralAmericaExpe

    Contributor

  • Validated Member
  • PipPip
  • 16 posts
  • Gender:Male
  • Interests:"I do recognize risk, and having recognized risk, I spend a lot of time minimizing it." Tim Severin
  • Canada

....you may want to approach this in another way:

Ground cover maps (vegetation, surface geology, etc) for most of the contiguous USA are already available in (vector) shapefiles from either each State's GIS repository, or from the USGS itself.

You could also acquire hole-filled SRTM raster data at 10m, 30m and 90m grids of any region of the USA in HGT format. These can be ingested or converted (like with arc toolbox in ArcGIS, or other tools in open-source: from HGT to ArcINFO ASCII to GRID) to a usable raster format, which in turn could be processed into hillshade rasters, then tiled (or, permanently seamlessly stitched together) into the appropriate sized raster coverage. Place the gray-scale hillshade under your original DEM, and apply to your DEM a stretched colorshade of your choice to represent bare terrain (exposed rock: say 20% to 60% black stretched to 256 colorshades). Now, open your ground cover vector layer classified into vegetation types over your DEM and hillshade, and choose the appropriate color of each vegetation type in the legend, and make the entire vegetation type vector layer transparent and remove the borders of every vegetation type.

I've worked with LIDAR (typically only available in limited US coverage: coasts, and thin line along Mexican border with USA, and a hand-full of interior coverages), and those files are absolutely behemoth (at ~1-meter cell size) ! Just the Kittyhawk area of the North Carolina Outer Banks was brutal to manipulate using a 3D visualization software, i'll tell ya. If you can find interior LIDAR data, and have the horsepower to man-handle in your GIS system an area as large as those National Geographic coverages, more power to you :) If the LIDAR is too much for the ultimate scale you'll be outputting, try the SRTM 10-meter data sets...

Good luck,
Derek-
"I do recognize risk, and having recognized risk, I spend a lot of time minimizing it." Tim Severin




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->