The Potential of Discrete Return, Small Footprint Airborne Laser Scanning Data for Vegetation Density Estimation

We evaluate the potential of deriving a vegetation leaf area index (LAI) from small footprint airborne laser scanning data. Based on findings from large area histograms of discrete laser returns for two contrasting plots, LAI is estimated from the fraction of first to last and single returns inside the canopy. The canopy returns are classified using thresholding of LIDAR raw data heights subtracted by interpolated digital terrain model heights. This should yield LAI estimates being independent of fractional vegetation cover, an ambiguity many passive optical approaches suffer from. Validation is carried out using 78 georeferenced hemispherical photographs, with LAI and gap fractions for a range of zenith angles being computed using the Gap Light Analyzer (GLA, Frazer et al. [1997]). Since the range sensitivity of the hemispherical photographs is not a priori known, we use variable LIDAR data trap sizes to find a suitable diameter. This is achieved searching the maximum R value of the regression for the trap size range from 5 to 50 m diameter. Larger diameters ( 30 m ) provide best results for our canopy types. Regressions of LIDAR estimates shows a moderate agreement with field data based on hemispherical photographs, with R 0.6 for LAI . Due to either heterogenity of the canopy or geolocation errors, a quite large amount of noise seems to be attributed to the regression, explaining the somewhat low values of R.