PARENT SESSION
Oral Session # 72: GIS and Remote Sensing I.
Presiding: J Drake
Thursday, August 7. 8:00 AM to 11:30 AM, SITCC Meeting Room 203.

Estimating stand structural attributes of ponderosa pine forests with discrete return lidar.

Hall, Sonia ^*,1, Burke, Ingrid¹, Box, David², Kaufmann, Merrill³, Stoker, Jason⁴, ¹ Colorado State University, Fort Collins, CO² 3Di Technology Inc., Boulder, CO³ Rocky Mountain Research Station, Fort Collins, CO⁴ EROS Data Center, Sioux Falls, SD

ABSTRACT- Stand and biomass structure are dynamic attributes of forests. Reliable spatially explicit estimates of these characteristics are needed periodically for many objectives, such as ecosystem modeling, carbon budget estimation, timber production, biodiversity studies and habitat mapping. Of growing importance is the estimation of stand-level (i.e. per hectare) live and dead fuels, to determine fire hazard and for fire behavior modeling. Optical remote sensing has provided estimates of stand attributes, but saturating relationships limit the usefulness of two-dimensional data. Continuous return lidar has been used to estimate three-dimensional stand structural variables. We used discrete, multiple return lidar data to estimate stand structural variables in ponderosa pine forests in Colorado. We used an information-theoretic approach to select the best simple models to estimate mean and maximum heights, tree biomass and its components, tree and biomass densities, basal area and bole volume. Most predicted vs. observed values were highly correlated, with r² ranging from 30% to 86%. The mean height estimates had the lowest r², but it increased substantially when measured tree heights were weighted by basal area (r²71%). We combined the model selection techniques with validation with new data and a comparison of regressions at two spatial scales to select a subset of the 45 initial metrics calculated from the lidar data. This subset is sufficient to predict forest and biomass characteristics with reasonable accuracy. Fusion of lidar and Landsat data will help provide periodic, spatially explicit estimates of these variables at low cost/ha, providing necessary inputs for fire behavior models. Modeling efforts will provide insight needed to determine the critical scale or scales at which fuels influence fire behavior.

Key words: ponderosa pine, information-theoretic approach, stand and biomass structure, lidar