Relationships among airborne scanning LiDAR, high resolution multispectral imagery, and ground-based inventory data in a ponderosa pine forest

Estimating forest structure and stand density using remotely sensed data is important for a wide range of scientific and management goals, including assessing biogeochemical budgets (e.g. aboveground carbon storage) and determining the susceptibility of an area to catastrophic fires. The objective of this study is to determine relationships among ground-collected forest structure data, high resolution IKONOS imagery, and airborne scanning LiDAR collected at a ponderosa. pine (Pinus ponderosa) dominated site in the Black Hills of South Dakota. Ground data were collected in the summer of 2001 along four 10×140 meter belt transects. IKONOS imagery was obtained over the site on July 28, 2000, and airborne scanning discrete-return LiDAR was acquired at a nominal 2 meter post spacing (56 cm beam footprint diameter) on October 26, 2001. No thinning or fire activity occurred at the site between data collection dates. Transect data were subdivided into 10×10 meter plots and co-registered with the IKONOS and LiDAR data for analyses. A combination of IKONOS multispectral and panchromatic data was used to select image endmembers (i.e. spectrally "pure" components) of bare soil, open grass, and tree/shade. In 80% of the plots, LiDAR-derived first return canopy height agreed with field-measured maximum tree height to within 20%. On average, LiDAR-derived first return canopy height underestimated field measured maximum tree height by 3.7%. Effective tree canopy leaf area index (LAI_e, a measure of canopy cover fraction) ranged from 0.3 to 2.5 among the plots. The fraction of LiDAR tree canopy returns were significantly correlated with LAI_e at the plot level (r=0.55; p<0.001). In addition, a significant positive correlation existed between the IKONOS-derived tree/shade endmember fraction for pixels falling within the transect plots and LAI_e (r = 0.76; p<0.001), as well as between the tree/shade endmember fraction and LiDAR tree canopy fraction (r=0.76; p<0.001). The Enhanced Vegetation Index (EVI) was significantly negatively correlated with all measures of tree density, including the LiDAR tree canopy fraction, tree basal area, and IKONOS-derived tree/shade endmember fraction. NDVI correlations with all measures o- f tree density were weaker than those found using the EVI. These results indicate that in this ecosystem, IKONOS data can serve to make the important distinction between tree canopy coverage and exposed understory grasses near peak summertime greenness.