An evaluation of LiDAR and optical satellite data for the measurement of structural attributes in British upland conifer plantation forestry

Abstract

This study evaluates the ability of LiDAR, IKONOS and Landsat ETM+ data to provide estimates of forest structure in British upland conifer plantations. Little use has so far been made of these technologies in the UK, whereas in some other countries remote sensing has become integral to forest management systems. The aim of this thesis is to demonstrate the application of the selected remote sensing systems to provide up-to- date and accurate information on key forest variables such as tree height, volume and density. Two upland conifer areas, located in south-west Scotland and north-east England, were used to develop and validate the regression models used to estimate these forest variables. The ability of LiDAR to provide an accurate measurement of the ground and canopy surfaces was investigated in densely stocked plantations, typical for commercial forestry in the U.K. The results show that, despite the dense nature of the forest canopy, sufficient laser pulses penetrate through to the ground to generate an accurate Digital Terrain Model (DTM). Provided that the ground surface is accurately defined, a point density of 2 returns/m(^2) will enable measurement of tree height to be made. LiDAR-derived top heights were found to be as accurate as field-based measurements (RMSE of 0.57 m). LiDAR-derived top height is easily integrated with established Forestry Commission models to provide volume estimations. Tree density is not accurately estimated using LiDAR data (RMSE of 434 trees/ha). Results strongly suggest that predictive equations developed for top height can be transferred to other conifer forests. Furthermore, the relationship between field-measured top height and laser-derived top height appears to be stable across different conifer species. LiDAR data can be used to identify tree species in pure and mixed stands. Two methods were developed: the first used summary measures based on the laser height distribution and the second the near infrared intensity. These measures when mapped spatially can be used to classify areas by species and to identify areas of anomalous growth and wind damage. At a larger spatial scale. Landsat ETM+ and IKONOS data can provide height estimates up to the point of canopy closure (approximately 10 m). LiDAR-derived height can be used in place of field-based measurements to drive reflectance-based models to estimate height from optical satellite data. The methods developed are transferable to other conifer forests that are managed in a similar way. The results from this thesis show that LİDAR, IKONOS and Landsat ETM+ data provide valuable and complementary information at a_ range of scales and can assist managers to make more informed resource management decisions.