Dynamic Range-based Intensity Normalization for Airborne, Discrete Return Lidar Data of Forest Canopies.

A novel approach for improving the consistency of intensity measurements using range-based normalization is introduced. The normalization is data-driven, can be fully automated, and involves scaling differences in observed intensity between returns collocated in space but registered to different laser scanning swaths. The scaling is proportional to the overall rate of attenuation f of laser energy. The utility of this approach for applications of lidar over forests was evaluated by examining classification results of broad cover types obtained using observed and normalized intensity measurements in an Oregon study area. The normalization was more effective for single returns, leading to a 53 percent reduction in intensity coefficient of variation between observed and range-normalized measurements. The poor cover type classification accuracy (44.4 percent; kappa 0.167) obtained by using observed intensities of first returns improved substantially to 75.6 percent (kappa 0.624) when using above-ground, single returns and f = 2.04. [ABSTRACT FROM AUTHOR]