Upscaling seasonal phenological course of leaf dorsiventral reflectance in radiative transfer model

Abstract

Forest top-of-canopy (TOC) reflectance observed by remote sensing instruments is driven by a wide range of parameters, including optical properties of leaves being the major driver, understory and bark together with structural properties of the forest - spatial distribution of leaves and branches in the canopy contributing also to variable extent. In addition, the instantaneous observation geometry (e.g. the sun and sensor zenith and azimuth angles) largely influences the observed TOC reflectance due to the changes in shadow fraction in the canopies. As a result, forest TOC reflectance is highly dynamic, following the seasonal patterns of optical and structural properties of vegetation.From the above-mentioned factors, the leaf optical properties are the most important. They hold an information on leaf biochemical (e.g. chlorophyll and carotenoid contents), water content and structural (e.g., leaf thickness, distribution of pigments within a leaf) properties at the leaf-level. Whereas biochemical effects on leaf reflectance are well understood, the role of dorsiventral asymmetry leaf internal structure on two-sided leaf optical properties has not been paid much attention, assuming equal optical properties on both leaf sides. However, the internal structure of a dorsiventral leaf is highly asymmetric often having different surface structure, tissue density and pigment distribution. Moreover, leaf optical properties are rarely measured during phenological development with high time resolution. Typically, studies focus on peak growing season neglecting the dynamic seasonal change of optical properties for broadleaved vegetation of temperate forests.In this study, we measured the seasonal course of leaf-level two-sided optical properties, biochemical and structural traits of five common broadleaved tree species of Central Europe (Betula pendula, Fagus sylvatica, Acer pseudoplatanus, Acer platanoides, and Sorbus aucuparia) with a typical dorsiventral, leaf anatomy without heliotropism. We link their leaf optical properties (e.g. reflectance difference of adaxial and abaxial sides, reflectance to transmittance ratio, fraction of specular component of leaf reflectance) with the laboratory analyses of inner leaf structure for contrasting leaf phenological stages from spring to fall. This allowed us to quantify both, species and seasonal differences of leaf dorsiventral optical properties and the role of inner leaf structure on leaf dorsiventral optical properties. Next, we up-scaled measured leaf-level dorsiventral optical properties to canopy level using Discrete Anisotropic Radiative Transfer model (DART) for three contrasting seasonal phenological stages - early spring (day of the year; DOY 115), summer (DOY 212) and fall (DOY 285). The aim was to quantify the impact of leaf optical properties parametrization (i.e. assuming equal optical properties on both sides, or dorsiventral leaf optical properties asymmetry) for forest of different species composition, vegetation phenological stage and observation geometries. The biggest differences between simulations were observed in VIS and NIR regions, where the neglecting differences in abaxial side leaf reflectance may introduce relative difference up to 20%, causing the underestimation of “one-sided” scenario compared to “two-sided” one, especially for off-nadir backscattering viewing directions. This may lead to increase in uncertainty when interpreting remote sensing observations, namely the chlorophyll content retrievals from back-scattering observation geometries and canopy structural retrievals (e.g. leaf area index) for all viewing geometries. We, thus, suggest to account for leaf reflectance dorsiventral differences in the next generation of radiative transfer models.