Proceedings of the 7 th International Conference on Functional-Structural Plant Models, Saariselkä, Finland, 9 - 14 June 2013. Eds. Risto Sievänen, Eero Nikinmaa, Christophe Godin, Anna Lintunen & Pekka Nygren. http://www.metal.fi/fspm2013/proceedings. ISBN 978-951-651-408-9. 52 tLiDAR methodologies can overcome limitations in estimating forest canopy LAI from conventional hemispherical photograph analyses Eric Casella 1 , Mat Disney 2 , James Morison 1 & Helen McKay 1 1 Centre for Sustainable Forestry and Climate Change, Forest Research, Farnham, Surrey, GU10 4LH, UK 2 Department of Geography, University College London, London, WC1E 6BT, UK Correspondence: Eric.Casella@forestry.gsi.gov.uk Highlights: The hemispherical photography technique has been widely used to assess the three-dimensional reconstruction quality of virtual plant canopy architectures (Casella & Sinoquet 2003). High-resolution terrestrial Light Detection And Ranging (tLiDAR) has recently been applied for measuring the 3-D characteristics of forest vegetation (Omasa et al. 2006.) and specifically the extraction of canopy directional gap fraction (Danson et al. 2007). In contrast with the digital hemispherical photography method, sky conditions appear to have little influence on the quality of the data collected by the tLiDAR technique. This study considers the resolution used during both point cloud data acquisition and the computation of equiangular hemispherical images, which may influence the resolving power of this technique in estimating gaps in a forest environment. Keywords: TLS, laser, point cloud, gap fraction, equiangular projection, composite hemispherical picture, resolution INTRODUCTION Leaf area index (LAI) is defined as the one-sided leaf blade area per unit ground area. It is a key parameter in ecophysiology for scaling-up from leaf to canopy level gas exchange and energy fluxes between the vegetation and the atmosphere. LAI is one of the most difficult parameter to quantify in situ although many non- destructive methods have been proposed (Bréda 2003). The upward-looking hemispherical photography technique has been used extensively to map and quantify canopy gaps for LAI computation. However, photographs must be taken under uniform overcast sky conditions and image analyses involve complex and critical steps for pixel classification between sky and canopy components despite recent increases in the resolution of digital cameras. Terrestrial laser scanners (TLS) have the potential to provide detailed information about forest canopy architecture by collecting 3-D point clouds of several million data points (Tab. 1) that can be transformed into hemispherical images by an equiangular projection procedure (Steyn 1980). However, the resolving power of this technique in estimating gaps in a forest environment may be affected by the resolution used during both point cloud data acquisition and the computation of hemispherical images as shown in Figure 1. Fig. 1. Modelled mean number of laser return hits per pixel in each 5 o zenith band from hemispherical images generated by an equiangular transform projection procedure (Fig. 2.) of a x, y, z coordinate data set computed from an opaque hemisphere for the Low (), Medium (), High () and Ultra-high () TLS resolution levels (Tab.1). METHODS The TLS used was the pulsed time-of-flight HDS 6100 (Leica Geosystems Ltd.) which has a rotating mirror system that covers a 360º (horizontal) x 310º (vertical) field of view with a range of about 79 m at 90% albedo (Tab. 1). The laser beam wavelength is 670 nm with a 3 mm spot size at its source and a 0.22 mrad beam divergence. Hemispherical photographs were taken using a Nikon Coolpix 995 camera (2.25 M pixel) with the Nikon FC-E8 hemispherical lens. Spatially and temporally coincident point-cloud data and digital hemispherical photographs were collected in a six-year-old stand of Eucalyptus spp. in southeast England. The stem density was 700 trees ha - 1 , with an average tree height of 11±3 m. For each position (n=8) and for scan zenith angles of 0-90º, one point-cloud was recorded at each of the four predefined resolution levels of the TLS (Tab. 1).