A Leaf Optical Properties Model Accounting for Differences between the Two Faces

Modeling radiative transfer in canopies allows efficient use of remote sensing observations to quantify vegetation state variables and functioning. Canopy radiative transfer models use requires accurate description of leaf optical properties. Although widely used, the PROSPECT leaf optical properties model is based on simplifying assumptions that limit its performances. One of these assumptions is that scattering and absorbing materials are evenly distributed in the leaf thickness, leading to the same optical properties for both faces. However, plants have developed particular adaptations resulting sometimes in strong differences in the properties of each faces, in relation to surface characteristics as well as leaf internal structure and distribution of chlorophyll and water. The objective of this paper is to develop a new model accounting for the difference between leaf upper and lower faces. The QSPECT model is an improved version of PROSPECT where the leaf considered made of four layers corresponding to the upper and lower epidermis, the palisade and spongy mesophylls. This advanced model requires four additional parameters describing the distribution of main scattering and absorbing materials. Results acquired over few reflectance and transmittance measurements show that QSPECT is able to describe accurately the typical differences observed between upper and lower leaf faces. Of particular importance are the surface reflectivities as well as chlorophyll content distribution * Corresponding author. This is useful to know for communication with the appropriate person in cases with more than one author. Introduction Leaf optical properties are key input variables to understand and model radiative transfer in canopies. Radiative transfer models are very useful for exploiting remote sensing observations and transform the signal collected onboard satellite into surface characteristics such as leaf area index or chlorophyll content. Several leaf optical properties models heave been proposed in the past as listed in a recent review (Jacquemoud et al., 2001). They allow to simulate leaf reflectance and transmittance from a limited set of state variables describing the content of absorbing materials such as chlorophyll, water or dry matter, and the scattering occurring at the interfaces between materials with difference in refraction index value. Most leaf optical properties models assume the leaf as an homogeneous layer with evenly distributed absorbing and scattering materials. As a consequence, reflectance and transmittance of both faces are the same. However, leaves have to ensure a number of functions under very wide environment conditions. They have therefore adapted their structure to better answer these challenges. A typical dicotyledon leaf section shows the palisade and spongy mesophyll tissue layers, bounded by two layers of epidermal cells as illustrated in Figure 1. The epidermis is made up of a single layer of colorless cells, with few if any chloroplasts. Its main role is to protect the inner layers form the outside environment, contributes to maintain leaf port, while controlling exchanges of gas and radiation with the cuticle (waxy, crystallized), its roughness, and the possible presence of hairs and their characteristics (length, density, shapes). Palisade cells are elongated, generally densely packed, and arranged in one to several layers. They contain the largest proportion of chloroplasts where most of photosynthesis is taking place. The spongy mesophyll is made up of highly lobed irregularly shaped cells of variable size, separated by large intercellular air-filled spaces that facilitate the circulation of gases inside the leaf. As a consequence, the small amount of absorbing material in the spongy mesophyll and the high number of discontinuities in the refraction index values, a large fraction of light incoming from the palisade mesophyll is scattered back, increasing light absorption efficiency by chloroplasts (Raven et al., 1996). Because of the particular organization of leaves, important differences may be observed between the optical properties of upper and lower faces particularly in spectral domains where stronger absorption occurs: in the visible domain characterized by chlorophyll pigment absorption, reflectance of upper faces is generally lower than that of lower faces where more scattering occurs. Transmittance and reflectance in the near infrared domain are generally less affected. Plants will exploit these differences in leaf optical properties between faces to better suit environmental conditions. This is reported for olive trees that have leaves with more reflective lower faces. In case of severe water stress, olive trees orientate their leaves in a more erectophile way, exposing the lower faces towards the incoming light which reduces the amount of radiation absorbed by the canopy, hence transpiration and photosynthesis. Apart from the effect of differences between faces on canopy functioning, impact is also expected on canopy reflectance. This may induce additional uncertainties in remote sensing retrieval of some key biophysical variables such as leaf area index (LAI) or chlorophyll content. This study will attempt to document differences between leaf optical properties faces: as a matter of fact, although differences between optical properties of the two faces are known qualitatively, very little studies documented their magnitude. More precisely, the objective of this study is to develop a leaf optical properties model that explicitly accounts for inhomogeneity in the leaf and describes the corresponding differences in leaf reflectance of both faces. The modeling will be based on the PROSPECT model proposed by Jacquemoud and Baret (1990). This model offers the advantage to be relatively simple with a limited number of variables allowing to use it in a more operational way. PROSPECT was already validated several times with relatively good performances although not describing the effects due to inhomogeneous leaf structure (Jacquemoud and Baret, 1990). In this study, measurements of leaf optical properties are acquired to quantify the possible differences between upper and lower faces. Then, the PROSPECT model is adapted to account for the layer structure description according to Figure1. It will be named hereafter “QSPECT”. A sensitivity analysis is conducted, with emphasis on the variables that drive the differences between both faces. Finally, some validation elements are presented and discussed.