A Canopy Radiative Transfer Model Considering Leaf Dorsoventrality

Abstract

Adaxial and abaxial leaf surfaces have different anatomical structures and varied chemical compositions. This asymmetry between leaf sides dictates that the spectral properties (reflectance, transmittance, and emissivity) of adaxial and abaxial surfaces are not identical. Laboratory measurements have demonstrated that this difference is significant over certain wavelengths. However, the influences of leaf dorsoventrality on such parameters (e.g., reflectance and brightness temperature) at the canopy scale have received little attention from the remote sensing community. One of the reasons is the lack of canopy radiative transfer models that can handle leaf dorsiventral properties. Although 3-D ray-tracing- or Monte Carlo-based models can achieve this, they are too complex to be implemented for massive tasks due to their low computational efficiency. Instead, they usually serve as benchmarks to evaluate other models. This study develops a unified optical–thermal canopy radiative transfer model considering leaf dorsiventral properties. It is based on the 1-D scattering by arbitrary inclined leaves (SAIL) model and, thus, has excellent efficiency and is easy to use. Evaluation of the proposed model by comparing it with the 3-D ray-tracing discrete anisotropic radiative transfer (DART) model shows that it is consistent with DART, with normalized root mean square errors (NRMSEs) of 0.013 and 0.005 within a reflective (for reflectance) and an emissive [for directional brightness temperature (DBT)] bands, respectively. Preliminary analyses ignoring leaf dorsoventrality within the optical and thermal spectral ranges by using the measured leaf spectra demonstrated that it induced significant errors in canopy reflectance (up to 40%) and DBT (up to 0.2 K) estimates under certain wavelengths. However, it should be noted that the influences of leaf dorsoventrality are determined by leaf adaxial/abaxial spectra, which still need to be explored due to limited measurements, especially under the thermal infrared bands.