|
landsat-fia
|
Trained directly on the most recent set of FIA plots across the state
|
None
|
Yes
|
|
1
|
Non-forested pixels excluded from sample. 20 equal interval bins constructed across acceptable pixels from all LiDAR-AGB surfaces. 0.05% of each LiDAR-AGB surface is sampled broken out evenly across bins. ~25k pixels.
|
Non-forested AGB pixels are set to 0 and included in map accuracy assessment.
|
No
|
|
2
|
LiDAR-AGB pixels in Developed, Water, or Barren LCMAP classes are excluded. 20 equal interval bins constructed across acceptable pixels from all LiDAR-AGB surfaces. 0.05% of acceptable pixels sampled from each surface broken out evenly across bins. ~40k pixels.
|
AGB pixels in Developed, Water, or Barren classes are set to 0 and included in map accuracy assessment.
|
No
|
|
3
|
All landcover classes are included in sample. 20 equal interval bins constructed across all pixels from all LiDAR-AGB surfaces. ~45k pixels.
|
None
|
Yes
|
|
4
|
0.01% of each LCMAP class x LiDAR-AGB surface sampled randomly. No stratification by AGB. ~10k pixels.
|
None
|
Yes
|
|
5
|
LiDAR-AGB pixels in Developed, Water, and Barren LCMAP classes forced to 0 and included in the sample. 20 equal interval bins constructed across pixels from all LiDAR-AGB surfaces. 0.05% of pixels sampled from each surface broken out evenly across bins. ~45k pixels.
|
AGB pixels in Developed, Water, or Barren classes are set to 0 and included in map accuracy assessment.
|
Yes
|
