k-Nearest
Neighbor (kNN)
Forest inventories provide forest estimates from county
to national levels, but do not provide maps of specific
forest resources. We are working with a new estimation
method, known as k-Nearest Neighbor (kNN), which has
the potential to generate precise local estimates and
maps of forest attributes.
We believe this is the first
application in the U.S. of the kNN approach to extrapolate
field data to landscape
levels using satellite remote sensing. After a three-year
collaboration with our eForest project, the U.S. Forest
Service is beginning to implement and test the approach
in the
North Central
region.
Below is a kNN estimate
of forest cover type and timber volume in Northeastern
Minnesota in cubic meters with calculations based on
1,835 FIA subplots
and three
Landsat dates, including two thermal bands for each
date. Click on the map to enlarge.
Please visit
the k-Nearest
Neighbor project website to
learn more about our research in this area. |
