Van R. Kane

1
Calibrating Landsat/ASTER with LiDAR for Forest
Studies

• Van R. Kane
• PhD Candidate
• College Forest Resources
• University of Washington
• Jerry Franklin, advisor
• Key collaborators
• Bob McGaughey (PNW Research Station)
• Alan Gillespie (UW)
• Supported by NASA Fellowship, City of Seattle,
  The Nature Conservancy, King County

2
Motivation

• Measuring forest structure essential to analyzing
  habitat potential and biomass
• To date, spectral analyses have been moderately
  successful at classifying forests into broad
  structural classes
• Can Landsat/ASTER classification accuracy be
  improved through calibration with LiDAR?

Study areas located in Pacific Northwest
3
Relating Field and Canopy Structure Measurements
LiDAR ordinations 0.81 0.87 correlated with
field ordinations
4
Forest Classification
8 2209361128
7 2109142128
5 2207142128
6 2110073028
4 2208313128
3 2208024128
2 2208142128
1 2111073128
Classes based on 95th percentile height, rumple,
canopy closure
5
Classification vs. Age Forest Zone
Kane et al. (in review)
6
Rumple as Stand Structure Predictor

Rumple canopy self shadowing
Kane et al. (2008)
7
Adaptive Shade Correction (ASC)

• Topographic correction using both illumination
  geometry and measured canopy self shadowing
• Coefficients determined from modeling self
  shadowing of LiDAR canopy surface models

Correction method Adjusted R2
ASC 0.83-0.94
SCSC (Soenen et al. 2005) 0.36-0.73
SCS (Gu and Gillespie 1988) 0.0-0.14
Solar zenith angles 29-49 degrees
Kane et al. (2008)
8
Spatial Variability
9
Conclusions and Next Steps

• LiDAR data can be used to classify stands with
  similar results as ground-level plot data would
  produce
• Methods can be used to directly study forests or
  to produce canopy truth for satellite
  measurements
• Next steps
• Compare Spectral Mixture Analysis and Tasseled
  Cap
• Demonstration projects Biomass estimates and
  forest structure variation along suburban -gt wild
  land transect