基于真实场景的树木冠层BRF模拟与分析

发布时间：2018-05-13 15:56

本文选题：辐射度模型 + 二向反射因子　；参考：《吉林大学》2017年硕士论文

【摘要】：早期遥感应用中,在研究地表反射率时,人们通常假设地表为朗伯体,地表在各个方向为朗伯反射,与入射辐射方向、观测方向无关。虽然这种假设在建模计算过程中简单方便,但大量实践证明,自然界中的地表均存在不同程度的非朗伯特性,即二向性反射特性,尤其是在植被反射中二向性反射特性更加明显。随着遥感技术的发展,尤其是多角度遥感卫星的应用及二向反射特性理论研究的深入,人们在植被二向性反射特性研究取得的成果也日益成熟,比如辐射传输模型、几何光学模型、混合模型和计算机模拟模型等,并得到了国内外的广泛认可。在植被遥感领域中,植被二向性反射特性一直是植被多角度遥感研究的理论基础。由于植被本身结构的复杂性、实验观测仪器与卫星数据的局限性以及模型的自适性使得植被二向性反射特性研究难以达到理想精度。尤其对于较大形体的树木而言,根据树木真实生长结构尽可能减少抽象简化从而精确计算其冠层二向反射因子(BRF)相对更加困难。鉴于此本文采用辐射度模型基于树木真实结构场景的三维模拟基础上完成冠层二向反射特性的模拟与分析。辐射度模型主要包括两部分:一是基于L-系统模拟三维真实场景,二是在此基础上采用辐射度方法完成冠层BRF模拟。本文研究内容如下:选取植被冠层辐射度模型RGM为研究工具,模拟太阳主平面与垂直主平面方向上的二向反射率因子。输入参数包括三维结构文件、光谱文件、太阳和天空定义文件、土壤文件、温度场文件等,计算出可见光及近红外四个波段的树木冠层BRF值。为研究模型参数变化对冠层BRF影响,分别模拟了不同组分、叶子结构、太阳位置、场景参数(如视场角、天空光比例、下垫面类型)条件下各个波段的冠层BRF,并分析相应的变化规律。结果表明不同的模型参数变化时,冠层BRF会呈现不同的变化规律,如叶面积指数增大,相应的BRF值降低;太阳天顶角的变化会导致冠层BRF热点移动;视场角越大,热点效应越弱;天空漫射光比例越大,冠层BRF越低;下垫面类型的不同会导致冠层BRF发生相应的变化。本文基于以上结论,结合野外实测树木结构参数采用扩展L-系统完成树木三维场景模拟,继而计算整个场景冠层BRF,并与实测冠层BRF数据进行验证。结果表明,冠层BRF模拟值与实测值平均相关系数均在0.7以上,说明基于RGM模型冠层BRF模拟计算较为准确,同时验证了本次研究实验的可靠性。
[Abstract]:In the early application of remote sensing, when studying the surface reflectivity, people usually assume that the surface is Lambert body, and the surface is Lambert reflection in all directions, which has nothing to do with the direction of incident radiation and the direction of observation. Although this assumption is simple and convenient in modeling and computing, a large number of practices have proved that the surface of the earth in nature has varying degrees of non-Lambert characteristics, that is, bidirectional reflectivity. Especially in vegetation reflection, bidirectional reflection is more obvious. With the development of remote sensing technology, especially the application of multi-angle remote sensing satellite and the theoretical research of bidirectional reflectance, the achievements of bidirectional reflectance of vegetation, such as radiation transfer model, are becoming more and more mature. Geometric optics model, hybrid model and computer simulation model are widely accepted at home and abroad. In the field of vegetation remote sensing, the bidirectional reflectance of vegetation is always the theoretical basis of vegetation multi-angle remote sensing. Due to the complexity of vegetation structure, the limitation of experimental observation instruments and satellite data, and the self-adaptability of the model, it is difficult to achieve the ideal precision in the study of the bidirectional reflectance of vegetation. Especially for trees with larger bodies, it is more difficult to accurately calculate the canopy bidirectional reflectance factor (BRFs) by minimizing abstract simplification according to the true growth structure of trees. In this paper, the radiance model is used to simulate and analyze the two dimensional reflection characteristics of the canopy based on the 3D simulation of the tree real structure scene. The emissivity model consists of two parts: one is to simulate 3D real scene based on L- system, the other is to use radiance method to simulate canopy BRF. The main contents of this paper are as follows: the radiance model of vegetation canopy (RGM) is selected as a tool to simulate the bidirectional reflectivity factors in the main plane and vertical direction of the sun. The input parameters include three dimensional structure file, spectral file, solar and sky definition file, soil file, temperature field file, etc. The BRF values of tree canopy in four bands of visible light and near infrared are calculated. In order to study the effect of model parameters on canopy BRF, different components, leaf structure, solar position, scene parameters (such as field of view, sky light ratio) were simulated. The canopy BRFs of each band were analyzed under the condition of underlying surface type. The results show that when the model parameters vary, the BRF of the canopy will show different changes, such as the increase of the leaf area index, the decrease of the corresponding BRF value, the change of solar zenith angle will cause the hot spot of BRF in the canopy to move, the larger the angle of view, the greater the angle of view. The weaker the hot spot effect is, the lower the canopy BRF is, and the higher the ratio of diffuse light in the sky is, the lower the type of underlying surface is, the corresponding change of canopy BRF will occur. Based on the above conclusions, the extended L- system is used to simulate the 3D scene of trees in the field, and then the canopy BRFs of the whole scene are calculated and verified with the measured canopy BRF data. The results show that the average correlation coefficient between the simulated and measured values of the canopy BRF is more than 0.7, which indicates that the simulated calculation of the canopy BRF based on the RGM model is more accurate, and the reliability of the experiment is verified.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S758

【参考文献】