Close Window

Validation of Near-Annual Resolution Forest Change Maps Derived from Landsat Data

Warren B. Cohen, Forest Service, Pacific Northwest Research Station, Corvallis, OR 97331, warren.cohen@oregonstate.edu (Presenting)
Maria Fiorella, Northrop Grumman, Oregon State Office, Bureau of Land Management, Portland, Oregon 97208, mfiorell@blm.gov
Robert E. Kennedy, Forest Service, Pacific Northwest Research Station, Corvallis, OR 97331, robert.kennedy@oregonstate.edu
Zhiqiang Yang, Department of Forest Science, Oregon State University, Corvallis, OR 97331, zhiqiang.yang@oregonstate.edu
Scott L. Powell, Forest Service, Pacific Northwest Research Station, Corvallis, OR 97331, scott.powell@oregonstate.edu
Melinda Moeur, U.S.D.A. Forest Service, Pacific Northwest Region, Portland, OR  97208, mmoeur@fs.fed.us

Trajectory-based Change Detection (TBCD) uses near-annual time series of Landsat imagery to track spatially fine-grained changes in vegetation. TBCD has evolved considerably since it was first published by Kennedy et al. (2007), to take fuller advantage of the information content in as much as 25 years of TM/ETM+ spectral response, and is now in the process of being automated for operational use. Automation is being prototyped over a contiguous 18-scene area encompassing all of western Washington and Oregon for a monitoring update of the Northwest Forest Plan (NWFP). Once operational, TBCD will be applied to a large number of scenes across North America within the context of the North American Forest Dynamics project funded by NACP. Needed is a way to independently validate output from the TBCD algorithm. In this poster we describe results from one approach for TBCD validation that is being tested within the context of the NWFP monitoring update for 2008. This particular validation approach is the most temporally resolved and comprehensive possible, in that is based on the full spectral time series, relying on the power of the human brain to integrate complex visual patterns with ancillary data and knowledge. The approach is unique in that it allows for a thorough assessment of each and every segment for any given trajectory. By linking our visual observations of changes in spectral space to digital orthophotos, we tie observed spectral changes directly to changes in forest cover and type.


NASA Carbon Cycle & Ecosystems Active Awards Represented by this Poster:

  • Award: NNH05AA36I
    Start Date: 2005-02-08
     
  • Award: NNH08AH54I
     
  • Award: 281945.02.61.01.05
    Start Date: 2003-10-01
     

Close Window