基于PROSAIL模型的青海湖流域草地叶面积指数反演

发布时间：2018-07-07 07:04

本文选题：PROSAIL模型 + 草地　；参考：《青海师范大学》2014年硕士论文

【摘要】：草地是陆地生态系统的重要组成部分，是地球表面的天然屏障，提供了畜牧业的原料，，对人类的生存和发展有着不可估量的作用。叶面积指数作为指示作物生长状况的生物物理参数,与植被各种生物物理过程有着密切的联系。因此准确地获取草地的叶面积指数对青海湖流域牧草产量估算具有重要意义。本文选取青海湖流域为研究区,选用MODIS影像和Landsat-8影像，结合实地采集叶面积指数数据、实测光谱数据,利用基于辐射传输模型的PROSAIL模型对草地叶面积指数进行遥感反演研究。论文主要研究包括以下几个方面： 1、数据的获取与预处理：包括样地实测光谱数据、叶面积指数数据、叶绿素浓度数据的采集和整理；遥感影像的大气校正。 2、 PROSAIL模型模拟分析：结合实测光谱数据将叶片反射率转化为冠层反射率，并对PROSAIL模型在青海湖流域草地的适用性进行了分析； 3、PROSAIL模型参数敏感性分析：根据实测数据分析了PROSAIL模型输入参数的敏感性。并依据模型敏感度计算公式定确定模型参数的敏感度。 4、查找表的建立：将敏感参数按照一定的步长进行取值，得到叶片不同情况下的冠层反射率，建立LAI与冠层反射率的查找表； 5、叶面积指数LAI反演：将进行过大气校正过的遥感影像像元反射率按照代价函数与查找表进行匹配查找，得到相应的冠层叶面积指数LAI，然后用青海湖流域样地实测数据对反演结果进行验证。通过研究,得到以下结论： 1、PROSPECT模型在反演草地叶面积指数方面有着较好的适用性:PROSPECT模型反演的的草地叶片反射率与实测叶片反射率绝对偏差小于0.015。 2、 PROSAIL模型输入参数敏感度由高到低为LAICabCmSLNCw，确定LAI和Cab两个最敏感的参数用于建立草地的LAI-冠层反射率查找表，对应选择Landsat-8影像4、5、6波段参与LAI反演；MODIS影像选择1、2波段进行反演。 3、反演LAI结果与实测的LAI具有很好的一致性，Landsat-8影像两者的相关系数R2=0.855，均方根误差RMSE=0.63；MODIS两者的相关系数R2=0.809，均方根误差RMSE=0.86。
[Abstract]:Grassland is an important part of terrestrial ecosystem and a natural barrier on the surface of the earth. It provides raw materials for animal husbandry and plays an inestimable role in the survival and development of human beings. As a biophysical parameter indicating crop growth, leaf area index is closely related to various biophysical processes of vegetation. Therefore, it is of great significance to obtain the leaf area index of grassland accurately for the estimation of forage yield in Qinghai Lake basin. In this paper, the Qinghai Lake Basin is selected as the study area, MODIS image and Landsat-8 image are selected, combined with the field data of leaf area index and the measured spectral data, the remote sensing inversion of grassland leaf area index is carried out by using PROSAIL model based on radiative transfer model. This paper mainly includes the following aspects: 1. The acquisition and pretreatment of data: including the collection and collation of the measured spectral data, leaf area index data and chlorophyll concentration data; Atmospheric correction of remote sensing image. 2. Simulation and analysis of Prosail model: the leaf reflectivity was transformed into canopy reflectance based on measured spectral data, and the applicability of PROSAIL model in Qinghai Lake basin grassland was analyzed. 3 sensitivity analysis of PROSAIL model parameters: the sensitivity of input parameters of PROSAIL model is analyzed based on the measured data. According to the model sensitivity formula, the sensitivity of the model parameters is determined. 4. The establishment of the lookup table: the sensitive parameters are calculated according to a certain step size, and the canopy reflectivity of the leaves under different conditions is obtained. Building the look-up table of Lai and canopy reflectivity. 5. Lai inversion of leaf area index: matching pixel reflectivity of atmospheric corrected remote sensing image according to cost function and lookup table. The corresponding canopy leaf area index (Lai) was obtained, and the inversion results were verified by the measured data of Qinghai Lake basin. Through research, The conclusions are as follows: 1 the project model has good applicability in retrieving the grassland leaf area index. The absolute deviation between the measured leaf reflectance and the grassland leaf reflectance obtained by the 0. 10% project model is less than 0. 015.2, and that of the project is less than 0. 015.2, and the absolute deviation of the measured leaf reflectance is less than 0. 015.2. The model input parameter sensitivity is from high to low to LAICabCmSLNCw. the two most sensitive parameters Lai and Cab are determined to establish the LAI- canopy reflectance lookup table of grassland. Landsat-8 images are selected for inversion in 4 and 5 ~ 6 bands. 3. The inversion results are in good agreement with the measured Lai. The correlation coefficients of Landsat-8 images and Landsat-8 images are 0.855, RMSE 0.63 and RMSE = 0.63 respectively. The correlation coefficient between MODIS and MODIS is 0.809, and the root mean square error (RMSE) is 0.86.
【学位授予单位】：青海师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：S812;TP79

【参考文献】