Influence of phenology on waveform features in deciduous and coniferous trees in airborne LiDAR

Information on forest structure is vital for sustainable forest management. Currently, airborne LiDAR remote sensing has been well established as an effective tool to characterize the structure of canopies and forest inventory variables. Radiometry and geometry are highly intertwined in LiDAR remote sensing of forest vegetation and phenology influences the geometric-optical properties of deciduous and evergreen trees causing seasonal variation in LiDAR observations. This variation may be considered as a nuisance or exploited in for example tree species identification. Airborne LiDAR data are also influenced by sensor functioning, acquisition settings, scan geometry and the atmosphere. Reliable estimation of subtle phenological effects calls for data in which the impact of the external factors is minimal. We experimented with such data and explored LIDAR waveforms (WFs) in boreal trees in winter, early summer and late summer. Our objectives were to i) assess the match of the multitemporal LiDAR data for observing true changes in vegetation; ii) quantify the influence of phenology in deciduous and evergreen trees; iii) study the effect of varying scan zenith angle (SZA) and canopy age on WF features in different phenostates; iv) assess the temporal feature correlation in individual living and dead standing trees. A WF-recording pulsed LiDAR sensor unit operating at the wavelength of 1550 nm was used in repeated acquisitions. WF attributes such as energy, peak amplitude and echo width were derived for each pulse and were localized vertically to crown, understory and ground components. Silver and downy birch, black alder, European aspen, Siberian larch, Scots pine, Norway spruce and dead standing spruce formed our strata. Results showed that phenology caused more variation in WF features of deciduous trees compared to evergreen conifers. Deciduous trees displayed substantial between-species variation that was linked with differences in branching pattern, leaf orientation and bark reflectance. Pine displayed a possible winter-early summer anomaly in canopy backscattering that may be linked with changes in foliage clumping or with the role of stamens in early summer trees. Trees displayed positive temporal correlation in WF features and correlations were the strongest in evergreen and deciduous conifers and decreased with time. SZA had minor influence on WF features whereas age exercised a strong effect on many features with parallel variation between species and phenostates. Structural changes following death, i.e. ‘aging’ changed the geometric WF features of dead standing trees. Our results provide new insights for enhancing tree species identification by using WF LiDAR and for LiDAR time-series analysis of vegetation.