Tell me more ×
Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. It's 100% free, no registration required.
up vote 5 down vote favorite
1

Given optical satellite or aerial imagery of a resolution similar to the image below, is it possible to accurately detect clearcut areas? Is their research or documentation that supports a specific methodology (e.g. object-oriented classification which considers contrast and geometry)?

How can this process be automated over areas ~1000km^2? Clearcuts

share|improve this question
How would you distinguish between clearcut areas and naturally clear areas like meadows? Or are you only looking for areas with no trees with no regard to how they got that way? Do the areas need to have low lying vegetation, or can they be bare ground as well? – Baltok Nov 29 '12 at 17:43
Just looking for clearcut areas (logging). I'm not sure how one would distinguish them from meadows. Low lying vegetation or bare ground, as long as they are clearcuts. I suspect there is no perfect answer, but I'd like to get as close as possible. – Radar Nov 29 '12 at 19:08
2  
Do you have imagery from before and after the clearcutting took place or just the one time? – blah238 Nov 29 '12 at 19:44
Good question, unfortunately the imagery only covers the period after the clearcutting took place. – Radar Nov 29 '12 at 19:53
Assuming the clearcutting happened in the past couple decades, you might be able to view the uncleared forestry using landsat imagery and perform image differencing on NDVI or something similar. – Fezter Nov 30 '12 at 1:42
show 3 more comments

3 Answers

up vote 2 down vote

It is hard to answer this without knowing what data you will be using and what software you have access to. I have done this using Landsat TM/ETM+ satellite imagery with Feature Analyst extension in ArcMap. You can build a signature file, which will allow you automatically classify other images that have a simular spectral signature.

share|improve this answer
up vote 2 down vote

For a fantastic way to detect, visualize, and report your findings to the public, check out the Landtrendr (Landsat-based Detection of Trends in Disturbance and Recovery) program from OSU. The Landtrendr program is one of the most exciting recent developments in change detection research. There is very good documentation on the methods, and Landtrendr code will likely be available in the near future or by special request. Here is a link to a NASA video describing the process: Landsat Senses a Disturbance in the Forest.

Landsat 8 will likely be the best available (free) data for detecting clearcuts at very large spatial extents. Data should be available sometime after June 2013. In the meantime, there is plenty of data available from previous Landsat missions at Glovis and EarthExplorer.

More traditional approaches include digital processing of multispectral imagery through a variety of methods:

  • Contrast thresholding (aka Density Slicing)
  • Pixel based classification: ISODATA, Maximum Likelihood, Random Forests
  • Object-oriented image anaysis (OBIA): Image Segmentation, Feature extraction

Landtrendr resources:

share|improve this answer
1  
this is a very good recommendation. Bob and Warren have identified distributions associated with specific forest change. Based on a multi-temporal image stack it is possible to automatically detect changes such as clearcuts. It is some work to pull the data together and is specific to Landsat. In continuous canopy forest, using high resolution imagery, you could also employ a texture based analysis. – Jeffrey Evans Apr 11 at 16:35
up vote 2 down vote

You could use MODIS LAND imagery. The best resolution however is at 250m which may be a little bit coarse for you.

There is several tools provided that can be used such multispectral analysis and everything is free.

http://modis.gsfc.nasa.gov/

You can see Modis near real-time imagery here: http://rapidfire.sci.gsfc.nasa.gov/realtime/ but use these as demo only, you cant do analysis on those images because they are not gridded

share|improve this answer
+1 for useful links. 250m is a little coarse for this purpose though as some clear cuts would only be represented by a few pixels. – Radar Apr 12 at 15:59

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.