Main content area

Bidirectional texture function of high resolution optical images of tropical forest: An approach using LiDAR hillshade simulations

Barbier, Nicolas, Proisy, Christophe, Véga, Cédric, Sabatier, Daniel, Couteron, Pierre
Remote sensing of environment 2011 v.115 no.1 pp. 167-179
canopy, carbon dioxide, deforestation, emissions, image analysis, lidar, light, monitoring, ordination techniques, reflectance, remote sensing, shade, simulation models, texture, topography, tropical forests, Amazonia, French Guiana
Quantifying and monitoring the structure and degradation of tropical forests over regional to global scales is gaining increasing scientific and societal importance. Reliable automated methods are only beginning to appear; for instance, through the recent development of textural approaches applied to high resolution optical imagery. In particular, the Fourier Transform Textural Ordination (FOTO) method shows some potential to provide non-saturating estimates of tropical forest structure, including for large scale applications. However, we need to understand more precisely how canopy structure interacts with physical signals (light) to produce a given texture, notably to assess the method's sensitivity to varying sun-view acquisition conditions. In this study, we take advantage of the detailed description of canopy topography provided by airborne small footprint LiDAR data acquired over the Paracou forest experimental station in French Guiana. Using hillshade models and a range of sun-view angles identical to the actual parameter distributions found for Quickbird™ images over the Amazon, we study noise and bias in texture estimation induced by the changing configurations. We introduce the bidirectional texture function, which summarizes these effects, and in particular the existence of a textural ‘hot spot’, similar to a well-known feature of bidirectional reflectance studies. For texture, this effect implies that coarseness decreases in configurations for which shadows are concealed to the observer. We also propose a method, termed partitioned standardization, that allows mitigating acquisition effects and discuss the potential for an operational use of VHR optical imagery and the FOTO method in the current context of international decisions to reduce CO₂ emissions due to deforestation and forest degradation.