Monday, February 15, 2010

2005 NAIP Finally Done

This past week:

Tracy was only able to meet with me for 30 minutes today during her office hours but she was able to help me work out a few conversion and geodatabase creation issues. She said she doesn’t do much work with Rasters in her research which is unfortunately mostly what I do in mine. When I email Dawn she usually refers me to Tracy as the campus GIS assistant so if Tracy can’t help me I’m on my own to find a way to figure things out. Right now I’m having issues converting a MrSID compressed NAIP of Rainier to a Raster Grid that I can import into a Rainier Geodatabase I’m creating. My Hood geodatabase is all set up and working great. Having all my work within the geodatabase means that I can relate files easier and do more in depth analysis. It also means that my entire project is self-contained which is great for backup as I only need to back up one folder rather than a million linked folders as Arc usually does with files not in a geodatabase. I’ve got all of the Landsat band ratio’s ready to import into Arc and I have established a great reclassification scheme based off the brightness value I pick for the threshold. I’m going to start with the low snow years of 87, 92, and 2005. The 2005 year corresponds perfectly with the 2005 NAIP so I have a real clear view of the debris-free ice extent for picking the threshold value. It’s a little tough comparing a 30m resolution band ratio to the .5m resolution NAIP so I’m looking at several different threshold values and comparing. Right now a value of .3 is looking the best it seems.

The next issue I’m having (at least with Hood) is that the only outlines I have are from USGS maps digitized for GLIMMS by Andrew Fountain. I’m not sure off the top of my head what year he did this and from what year USGS map he did it from but if it wasn’t from orthophotos or a hi-res sat image then I can’t take it as highly reliable. My step after I classify based on the threshold is to “clip” out the reclassified area that is clearly not glacier outside of the mountain (extra purple area around Hood in image on right). For Rainier I did this by using the very large 1913 outline (which was definitely the furthest extent of the glaciers) but for Hood I now only have the GLIMMS outlines to use to clip the reclassified image and if those outlines were from glacier extents later than 87 I’d be clipping out parts of the 87 band ratio extent when I do the clip function. I suppose I could use a buffer of a couple hundred meters (there is no way the glaciers could have changed that dramatically) to create another clip layer based on the GLIMMS outline but I’d rather wait until I hear back from Andrew Fountain about what he may provide for me. The nice thing about the the GLIMMS outlines is that they come with a great attribute table and are composed of 115 different selectable polygons so I can select certain glaciers like the Elliot (outlined in torquoise in the purple on right). Being able to select the individual areas I can then perform analysis on each. Based on what I’ve seen so far it looks like the south facing glaciers have undergone more ablation than the north on Hood at least.

My next step is trying to find a way to measure the area change and elevation change between decades. The elevation change may be easier but the area is harder. Before when I measured the area change it was in 2D but I want to do it in 3D now incorporating the slope because when the 2D sat image is draped over the LiDAR hillshade the area of coverage is different. I talked to Tracy about this and she suggested a few things but she wasn’t sure how to do it either. I’m going to look into this on Monday but if I can’t figure it out I’m going to stop in during Dawn’s office hours next week.

Also, there is a 2009 NAIP for Clackamas county.... The link is here:

You don’t have any connections to be able to download that NAIP for free would you? Is it possible to use any funds to get it? For some reason the 2005 NAIP on the south side of the mountain above Timberline is “hazy” compared to the other sides... I’ve double checked and it is definitely the NAIP from Oregon explorer, not anything I’ve done... Not sure why they allowed that as Hood seems like a pretty important oregon feature to leave crappy looking on the NAIP... I could use the 2005 NAIP to help locate the threshold value but for any “eye candy” produced with the research, having it draped over the most recent (and I assume clearer) 2009 NAIP would would look much nicer...

So, right now I’m waiting for Andrew’s better outlines and I’m researching and looking through help files how to do the spatial analysis within Arc. Once I’ve figured that out I can look specifically at the different glaciers of both mountains. The 2nd image up from the bottom is a close-up of the GLIMMS glacier outlines... as you can see I question their accuracy. The last image is of the elliot glacier over the hillshade. The green in the last two, and first is 2005 while the blue is 1987. In the middle image the light blue is 2005 and the dark is 1987. You’ll also see that I found a great rivers layer that I’ve added to the figures as well.

From your email you sounded busy. Good luck with everything. After weeks of rain this weekend is supposed to be sunny and warm so my buddy Chris and I are going to attempt a ski descent from the summit of Hood tomorrow morning. I’ll tell you how it went and provide a link to the helmet cam footage next week if you’re interested...


No comments: