Advances in canopy radiation and scattering characteristic three-dimensional joint simulation models

Abstract

Canopy radiation and scattering signal contain high amounts of vegetation information. Biophysical parameters can be quantitatively retrieved by establishing a canopy radiation and scattering model and inverting this model. Thus far, models in visible/near infrared (VIS/NIR), thermal infrared (TIR), and microwave (MW) regions have been developed. Each model suitable to a specific spectral (frequency) domain is characterized by advantages and disadvantages in terms of parameter inversion. Joint simulation models can produce complementary advantages and improve inversion precision. Although joint simulation and inversion research area has been improved, studies are mostly based on semi-empirical models or analytical models. A three-dimensional model is more suitable to characterize multiple scattering caused by vegetation structures and specified distribution of different components. This article presents a review on advances in three-dimensional canopy radiation and scattering characteristic simulation models specifically used for VIS/NIR, TIR, and MW regions, as well as a joint simulation model. The three-dimensional VIS/NIR Bidirectional Reflectance Distribution Function (BRDF) model can be divided into two categories, namely, Monte Carlo ray tracing and radiosity models. The model is improved in terms of the ability to simulate remote sensing images with kilometer pixels and to simulate non-lambert characteristic of components. The three-dimensional TIR directional radiation model can be extended from the three-dimensional BRDF model by considering self-emission. The three-dimensional MW backscatter model includes incoherent and coherent models. A coherent model can export phase information and is more accurate, especially at low frequencies, than an incoherent model. MW emission can be simulated by calculating bistatic scattering with a three-dimensional MW backscatter model. In addition to this VIS/NIR and TIR joint simulation model and the active and passive MW joint simulation model, VIS/NIR and active MW joint simulation models were discussed. These models were listed and described in this paper. Based on our analysis, three scientific issues and corresponding solutions were proposed using the three-dimensional joint simulation models. One issue involved scene unification. Models suitable for different spectral (frequency) domains describe a specific scene in different details. A standard should be established to perform simulations based on the same scene. Another issue involves the unity of physical and chemical parameters of components. Driving parameters vary. Therefore, transformation is required, and some models (e. g., PROSPECT and dual-dispersion models) should be coupled to convert various parameters into few but unified parameters. The last issue is the choice of efficient inversion methods. three-dimensional models require more calculation than other models. Look-up table and neural network methods are two of the commonly used approaches. However, these methods should be improved to satisfy three-dimensional joint simulation models. Other prospects are developed for further advancements of the joint three-dimensional simulation model. Three-dimensional joint simulation systems should be developed and used by researchers in future studies, considering that remote sensing platforms of the mechanism models (http://210.72.27.51:83/ModelCategory/All/) developed by the State Key Laboratory of Remote Sensing Science are constantly improved.