Simultaneous Extraction of Temporal, Spatial, and Spectral Information from Multi-Wavelength Lidar Data

Analysis of lidar data is generally done through parametric analysis where the lidar range dependent backscattering and extinction are deduced from the analytical lidar equation, and the range dependent concentration is deduced from the backscattering (or extinction). In previous work we have treated lidar measurements as a hyperspectral vector and applied traditional anomaly detection to determine (from one wavelength) the presence of a cloud of interest. Multi-wavelength lidar data, however, is not a matrix (range-by-time), but instead is a tensor: a multidimensional w-way array (range-by-time-by-wavelength). Matrix analysis techniques such as anomaly detection will not extract all the information available within this array. This work describes the use of n-way analysis techniques to simultaneously extract temporal, spatial, and spectral information from multi-wavelength lidar data.