Fusion of optical, radar and waveform LiDAR observations for land cover classification

Land cover is an integral component for characterizing anthropogenic activity and promoting sustainable land use. Mapping distribution coverage of at broad spatiotemporal scales largely relies on classification remotely sensed data. Although recently multi-source data fusion has been playing increasingly active role in classification, our intensive review current studies shows that the integration optical, synthetic aperture radar (SAR) light detection ranging (LiDAR) observations not thoroughly evaluated. In this research, we bridged gap by i) summarizing related assessing their reported accuracy improvements, ii) conducting own case study where first time waveform LiDAR associated improvements are assessed using collected spaceborne or appropriately simulated platforms case. Multitemporal Landsat-5/Thematic Mapper (TM) Advanced Observing Satellite-1/ Phased Array type L-band SAR (ALOS-1/PALSAR) imagery acquired Central New York (CNY) region close to collection airborne LVIS (Land, Vegetation, Ice Sensor) were examined. Classification was conducted a random forest algorithm different feature sets terms sensor seasonality as input variables. Results indicate combined spectral, scattering vertical structural information provided maximum discriminative capability among types, giving rise highest overall 83% (2–19% 9–35% superior two-sensor single-sensor scenarios with accuracies 64–81% 48–74%, respectively). Greater improvement achieved when combining multitemporal Landsat images LVIS-derived canopy height metrics opposed PALSAR features, suggesting contributed more useful thematic complementary spectral beneficial task, especially vegetation classes. With Global Ecosystem Dynamics Investigation (GEDI), launched instrument similar properties now operating onboard International Space Station (ISS), it hope research will act literature summary offer guidelines further applications multi-date multi-type improved classification.