日光诱导叶绿素荧光(SIF)是指示植被光合作用过程的无损探针，在不同时空尺度上对植被进行SIF的观测可以反映植被的实际光合作用及生理状态。然而在观测、分析和利用SIF的过程中，仍存在很多不确定因素。SIF的发生具有较为复杂的机理，从机理出发理解SIF与植被结构的相互作用，并分析影响SIF激发的主要因素将有助于更好地理解SIF与光合作用以及生物量的内在联系。因此，植被SIF辐射传输模型在解释和利用SIF遥感信号方面具有重要的作用。植被SIF信号相对较弱，且受环境、植被和生理等多种因子的影响，需要定量化描述，这为SIF辐射传输模型的构建带来挑战。近年来，大量学者已经发展一系列SIF辐射传输模型，为SIF遥感的发展提供了坚实的理论基础。本文回顾了叶片、冠层和生态系统尺度的SIF模型，从建模机理出发，对比模型优劣势，并对未来SIF模型的发展前景进行了展望。

Sun-Induced chlorophyll Fluorescence (SIF) has been recently used as a novel indicator of photosynthesis of vegetationdue toits direct relation to vegetation photosynthesis. The mechanism behind the SIF signal is rather complicated. Thus, the physical process and the interaction between SIF and vegetation structureshould be understoodfor better interpretation of SIF data. In this respect, the development of relevant SIF models is the key to improve the understanding and use of SIF signals. In this review, we introduce various SIF models in different scales by clarifying the mechanism behind each model andpropose the prospects in the future work.(1) At the leaf level, fluorescence models are generally based on leaf optical properties models and focus on the simulation of leaf reflectance and transmittance. Several theories canexplain the propagation of light in a turbid medium or a simplified blade. The simplest one assumes that light decays exponentially within the blade according to Beer’s law.The Kubelka–Munk differential equations arealsoused to solve radiation propagating in a turbid medium, which isfollowed by the Plate model fordiving the blade into several homogenous layers. The most popular leaf fluorescence model called Fluspect is based on the PROSPECT model, which follows Plate’s theory. The key objective is to simulate the re-absorption effect accurately due to the band overlap between SIF emission and chlorophyll absorption.(2) At the canopy level, fluorescence models incorporate the canopy radiative transfer and leaf fluorescence models, which can be characterized as 1D and 3D models. 1D models, such as FLSAIL, FluorSAIL,andSCOPE, incorporate SAIL model with a leaf fluorescence model and assume the canopy as several horizontally homogenous layers. 3D models, such as DART, FluorWPS, andFluorFLIGHT, simulate the canopy fluorescence in a realistic scene using ray tracing method. They are suitable to heterogeneous vegetation canopies.(3) At the ecosystem level, these fluorescence models help reduce the uncertainties in simulating carbon cycle and predicting ecological system response to global change by incorporating them with land surface models, such as NCAR CLM4, BEPS, and BETHY.Despite the advances in the SIF models across multiple scales, further studies are still needed with respect to model development, validation, and inversion. For example, accurate tree positions and canopy structure parameters can be derived from light detection and ranging data, which enables thereconstruction of 3D scenes based on the real landscape. With the development of insitu SIF measurement techniques, SIF models can be validated with the measurements except cross-validation through various models. SIF model inversion is a prospective research area to derive vegetation structure and biochemical parameters through spectral data. The novel machine learning approaches may also provide new opportunities to be incorporated with SIF models to solve inversion problems.