针对热红外和微波数据的地表温度和土壤水分反演算法研究.pdf

发布时间：2022-05-30 发布人：admin 分类：说明书资料大小：4.73M 资料格式：pdf 举报版权申诉

zuo1188-2322690-16359647427522227253.pdf-第1页.png

第1页 / 共139页

zuo1188-2322690-16359647427522227253.pdf-第2页.png

第2页 / 共139页

zuo1188-2322690-16359647427522227253.pdf-第3页.png

第3页 / 共139页

zuo1188-2322690-16359647427522227253.pdf-第4页.png

第4页 / 共139页

zuo1188-2322690-16359647427522227253.pdf-第5页.png

第5页 / 共139页

zuo1188-2322690-16359647427522227253.pdf-第6页.png

第6页 / 共139页

zuo1188-2322690-16359647427522227253.pdf-第7页.png

第7页 / 共139页

zuo1188-2322690-16359647427522227253.pdf-第8页.png

第8页 / 共139页

abstract.pdf

ABSTRACT

第�

1.1研究目的和意义

1.2国内外研究现状

1.4本章小结

第二章热红外被动微波地表温度和土壤湿度反演的基本理论与方法

2.5 微波模型简要介绍

2.6被动微波地表温度反演算法简要介绍

2.7被动微波土壤水分反演算法简要介绍

第四章热红外地表温度反演

A.利用大气模型和神经网络参数反演参数

第五章热红外地表温度反演

5.3.2地球物理参数之间相互联系

5.3.3 RM-NN 解病态反演问题

5.3.3.1为什么使用神经网络

5.3.3.2多层神经网络

5.3.4 RM –NN地表温度和发射率反演

4\) 大气水汽含量参数敏感性分析

5.3.5 与MODIS LST产品比较分析及精度评价

分类号密级 UDC 编号中国科学院研究生院博士学位论文针对热红外和微波数据的地表温度和土壤水分反演算法研究毛克彪指导教师施建成研究员中国科学院遥感应用研究所申请学位级别理学博士学科专业名称地图学与地理信息系统论文提交日期 2007.1.1 论文答辩日期 2007.4.9 培养单位中国科学院遥感应用研究所学位授予单位中国科学院研究生院答辩委员会主席李增元

Ph.D. Dissertation Submitted to The Graduate School of the Chinese Academy of Sciences The Study of Algorithm for Retrieving Land Surface Temperature and Soil Moisture from Thermal and Microwave Data Candidate: Kebiao Mao Adviser: Jiancheng Shi, Professor Institute of Remote Sensing Applications, The Chinese Academy of Science January 2007 in Beijing

摘要地面温度和土壤水分是反映土壤—植被—大气系统能量流动与物质交换以及土地资源和环境管理的重要参数，也是地球表面能量平衡和温室效应的两个重要指标，是区域和全球尺度地表物理过程的一个关键因子。因此，地面温度和土壤水分在气候、水文、生态学和生地化学等许多领域中是非常重要的。但是依靠地面观测站的观测来大面积地获取地表温度和土壤水分参数是不现实的。然而借助于热红外遥感、微波遥感影像，可以方便快捷地获得大面积，甚至全球的地温资料，且数据更新快，成本低廉。在简要介绍有关热红外、微波遥感发展和概念的基础上，分析和探讨常用的地表温度和土壤水分的反演方法，并对反演精度的影响因素进行了分析。针对 ASTER、MODIS、AMSR-E 传感器的波段设置特点和传统反演算法中参数求取的局限性。我们对地表温度和土壤水分的反演算法做了一些研究，本文具体内容如下：（1）热红外波段的大气透过率是地表温度反演的基本参数。本项研究分析近红外波段估计大气水汽含量的方法，并进一步研究大气水汽含量与热红外波段的大气透过率关系，计算得到 MODIS/ASTER 热红外波段的透过率。（2）现有的地表温度反演方法主要有劈窗算法和单窗算法。这些算法都是针对具体的传感器和特定的参数条件开发的。本项研究在分析这些算法的推导过程的基础上，对 Planck 函数进行线性简化，提出了适合于 ASTER 和 MODIS 热红外波段的地表温度反演方法。其中最关键的是，我们对这些算法的参数进行了重新确定，使之适合于 ASTER 和 MODIS 数据，从而形成了从 ASTER 和 MODIS 数据中反演地表温度的劈窗算法。（3）在以往的单窗和劈窗算法中，通常假定发射率已知，这使得地表温度的反演精度在先验知识不够的地区受到限制。由于发射率在 8.475~11.65µm 范围内发射率变化很小，而且在局部范围内近似线性，因此本文针对这情况。对 ASTER 的第 11、12、13 和 14 波段，MODIS 29、31、32 建立辐射方程组，同时对相应的发射率建立线性方程组。联立方程从而形成针对 ASTER/MODIS 数据的地表 I

温度和发射率同时反演的多波段算法。为了提高反演精度，我们用神经网络对反演算法进行了优化计算。利用 MODTRAN4 模拟数据精度分析评价表明精度很高，与 NASA 的标准产品和地表实测数据比较分析表明算法精度也很高。（4）在分析 Aqua 卫星多传感器特征的基础上，利用 MODIS 的温度产品和 AMSR-E 不同通道之间的亮度温度建立反演地表温度的反演方程，从而克服了以往需要测量同步数据的困难。通过各通道的回归系数分析表明，不同的地表覆盖类型的辐射机制是不同的。要精确地反演地表温度，至少对地表分成三种覆盖类型，即雪覆盖的地表、非雪覆盖的地表和水覆盖的地表。以 MODIS 地表温度产品作为评价标准，对于验证的样本数据，本文建立的统计方法的平均精度在 2-3°C 左右。由于地面非常的复杂，由理论模型模拟的数据不可能非常好的描述反演方程。为了提高算法的实用性，还需要进一步对云覆盖和不同辐射机制的地表类型的混合像元进行研究。神经网络不需要推导具体的反演规则，这些条件决定了神经网络是被动微波地表温度反演的最佳选择之一。分析结果表明神经网络能够被很好地用来从被动微波数据 AMSR-E 中反演地表温度。（5）用 AIEM 模型针对被动微波数据 AMSR-E 进行了模拟分析结果表明：在给定粗糙度条件下，土壤水分和发射率存在很好的线性关系；在不同的土壤水分条件下，均方根高度和相关长度对发射率的影响基本相同。我们定义了极化指数，模拟数据表明， 18.7GHZ 与 10.7GHZ 的垂直极化指数与土壤水分有很好的关系，而且部分地消除了土壤粗糙度的影响，R-Square 大约 0.98。同时，我们推导了标准化微波指数近似等于标准化亮温指数。分析表明通过标准化发射率指数和标准化微波指数建立土壤水分反演算法是可行的。同时我们对算法进行了敏感性分析，分析表明当有降雨时，算法比较敏感。用 SMEX02 的实验数据验证分析表明，相对于实验数据, 算法精度大约是 25.9% 。算法低估了土壤水分，而且算法受植被的影响比较大。因此要提高算法的适用性，需要根据当地的实测数据对反演结果做进一步修正。对 SMEX02 实验场，修正后的精度为 6.5% 。最后，我们对中国地区的两景 AMSR-E 进行了实际反演分析，结果表明反演结果符合实际土壤水分分布情况，表明算法可行。关键词：亮度温度，地表温度，土壤水分，ASTER，MODIS，AMSR-E II

ABSTRACT The extensive requirement of temperature and soil moisture information on a large scale for environmental studies and management activities of the Earth’s sources has made the remote sensing of land-surface temperature (LST) and soil moisture two important issues in recent decades. Many studies have shown that land surface temperature and soil moisture are two key parameters in numerical weather-prediction model that lead to significant forecasting improvement in the physics of land surface processes on regional and global scales, combining the results of all surface-atmosphere interactions and energy fluxes between atmosphere and the ground. In this paper, we introduce the theoretical basis for the remote sensing of LST and soil moisture, some concepts of thermal and microwave remote sensing. We will propose new algorithm for retrieving land surface temperature and emissivity from ASTER, MODIS, AMSR-E, and for retrieving soil moisture from AMSR-E. The details are as following: (1) Atmospheric transmittance is a critical parameter that affects the accuracy of LST retrieval. The thermal radiance is attenuated in its way to remote sensor. Transmittance depicts the magnitude of the attenuation of the radiance transferring through atmosphere which varies with wavelength and viewing angle. water vapor transmittance strongly depends on the dynamics of water vapor content in the profile. In order to improve the accuracy of the retrieval, according to characters of MODS bands, we get experience expression between the transmittance and the water vapor content. At first we get the water content of atmosphere by using the NIR bands, and then get the transmittance of the ASTER/MODIS thermal bands through the relationship between transmittance and water content of atmosphere. (2) Every split window algorithm or mono-window algorithm is derived for special sensor. In this paper, we derive a split window algorithm for ASTER and MODIS data. The main character of split-window algorithm is that it only needs two parameters, which are transmittance and emissivity. Accord to the character of the MODIS bands, we obtain the transmittance of ASTER/MODIS thermal bands through water vapor content retrieved from MODIS NIR. (3) The emissivity is made as known parameter, which produce large retrieval error when we can not obtain enough prior knowledge of land surface type. So LST and emissivity retrieval is a typical ill-posed problem. There has some potential information between band emissivities III

because the emissivity in thermal band is almost constant. The extra two equations can be found by linear or non-linear relationship between neighbor band emissivities for every land surface type. The neural network (NN) can make full use of this potential information between band emissivities through training data because NN simultaneously owns function approximation, classification, optimization computation and self study ability. The training database can be built through simulation according to the emissivity characteristics of land surface type by MODTRAN4 or obtained from the reliable measured data. We use RM-NN retrieve LST and emissivity from ASTER/MODIS, and the analysis indicates that the accuracy is very high. (4) MODIS and AMSR-E are on board the Aqua satellite, which overcome the difficulty for AMSR-E to obtain the land surface temperature data. The regression analysis results between ASMR-E brightness temperature and MODIS LST product indicates that the radiation mechanism of surface covered snow is very different from others. In order to retrieve land surface temperature more accurately, the land surface should be at least classified into three types: water covered surface, snow covered surface, and non-water and non-snow covered land surface. In order to improve the practicality and accuracy of the algorithm, we built different equations for different range of temperature. We use neural network to retrieve LST from AMSR-E which can overcome some shortcomings of previous algorithm. The average land surface temperature error is under 2 ℃ relative to the MODIS LST product. (5) According to simulation analysis of AIEM, there is a good linear relationship between emissivity and soil moisture under given roughness. The normalized difference between emissivity of 18.7GHz and 10.7GHz can eliminate partly the influence of roughness and the R-Square is about 0.98. This paper intends to propose an algorithm for retrieving soil moisture by brightness temperature from AMSR-E data. The normalized difference of emissivity approximately equal to the normalized brightness temperature between 18.7GHz and 10.7GHz has been derived with details. This method avoids parameter land surface temperature which is the key parameter for the computation of emissivity. We make some sensitivity analysis for the atmosphere which is the main factor for our method. The analysis indicates that our method is very good in clear day but is not very well when it has rainfall. Key words: Bright temperature, LST, Soil Moisture, ASTER, MODIS, AMSR-E IV

目录摘要...........................................................................................................I Abstract..................................................................................................... IV 第一章绪论 ..........................................................................................1 1.1 研究目的及意义..........................................................................................................1 1.2 国内外研究现状..........................................................................................................3 1.3 主要研究内容和技术路线 .....................................................................................10 1.3.1 主要研究内容................................................................................................10 1.3.2 主要研究技术路线.......................................................................................10 1.4 本章小结 .....................................................................................................................14 1.5 参考文献 .....................................................................................................................14 第二章热红外和被动微波地表温度和土壤水分反演的基本理论与方法...........................................................................................................21 2.1 电磁波谱 .....................................................................................................................21 2.2 热辐射基本定律及基本概念.................................................................................22 2.3 大气窗口与热红外遥感..........................................................................................26 2.4 热红外地表温度和发射率反演常用方法 ..........................................................27 2.5 微波模型简要介绍 ...................................................................................................28 2.6 被动微波地表温度反演算法简要介绍...............................................................30 2.7 被动微波土壤水分反演算法简要介绍...............................................................31 2.8 本章小结 .....................................................................................................................31 2.9 参考文献 .....................................................................................................................31 第三章实验数据选择与分析 ..............................................................37 3.1 ASTER ............................................................................................................................37 3.2 MODIS ............................................................................................................................39 3.3 AMSR-E..........................................................................................................................41 第四章针对ASTER数据的地表温度和发射率反演算法研究............47 4.1 引言 ..............................................................................................................................47 4.2 劈窗算法 .....................................................................................................................47 4.2.1 劈窗算法设计思想........................................................................................48 4.2.2 Planck函数线性简化..................................................................................48 4.2.3 大气透过率估计 ...........................................................................................50 4.2.4 比辐射率估计................................................................................................53 4.2.5 劈窗算法推导................................................................................................54 4.2.5 算法验证.........................................................................................................56 4.2.6 参数敏感性分析 ...........................................................................................58 4.3 多波段算法.................................................................................................................59 4.3.1 多波段算法推导 ............................................................................................59 4.3.2 算法求解..........................................................................................................64 4.3.3 实际反演分析与评价..................................................................................68 4.4 本章小结 .....................................................................................................................73 4.5 参考文献...................................................................................................................74 第五章针对MODIS数据的地表温度和发射率反演算法研究..............67 V

5.1 引言 ..............................................................................................................................77 5.2 劈窗算法 .....................................................................................................................78 5.2.1 劈窗算法技术路线.......................................................................................79 5.2.2 Planck函数线性简化和参数估计 ...........................................................80 5.2.3 劈窗算法推导................................................................................................86 5.2.4 参数敏感性分析及算法精度评价 ...........................................................88 5.2.5 算法应用.........................................................................................................98 5.3 同时反演地表温度和发射率的RM-NN算法........................................................99 5.3.1 同时反演地表温度和发射率的病态问题分析...................................101 5.3.2 地球物理参数之间的关系.......................................................................102 5.3.3 RM-NN解病态反演问题..............................................................................112 5.3.4 RM-NN地表温度和发射率反演................................................................114 5.3.5 与MODIS地表温度产品比较分析及精度评价.....................................117 5.4 本章小结 ...................................................................................................................123 5.5 参考文献 ...................................................................................................................124 第六章针对被动微波数据AMSR-E地表温度反演研究....................129 6.1 引言 ............................................................................................................................129 6.2 被动微波地表温度反演的理论基础 .................................................................131 6.3 被动微波地表温度反演传统经验方法.............................................................133 6.4 针对被动微波AMSR-E数据反演地表温度的物理统计算法 ........................135 6.4.1 MODIS LST与AMSR-E亮度温度数据统计分析.....................................135 6.4.2 AIEM模拟分析及被动微波反演算法.....................................................138 6.4.3 算法验证及应用 .........................................................................................141 6.5 利用神经网络从被动微波数据AMSR-E中反演地表温度.............................143 6.5.1 微波辐射计 ..................................................................................................144 6.5.2 利用神经网络从AMSR-E中反演地表温度............................................145 6.6 本章小结 ...................................................................................................................157 6.7 参考文献 ...................................................................................................................158 第七章针对被动微波数据AMSR-E土壤水分反演研究....................162 7.1 引言 ............................................................................................................................162 7.2 被动微波土壤水分反演的理论基础 .................................................................163 7.3 针对AMSR-E数据的AIEM模拟分析......................................................................165 7.4 土壤水分反演算法及敏感性分析......................................................................168 7.5 算法验证及应用......................................................................................................176 7.6 本章小结 ...................................................................................................................180 7.7 参考文献 ...................................................................................................................180 第八章结语与展望 ............................................................................183 8.1 主要结论 ...................................................................................................................184 8.2 主要问题 ...................................................................................................................188 8.2.1 混合像元问题..............................................................................................189 8.2.2 地形问题.......................................................................................................189 8.2.3 植被结构与分布 .........................................................................................189 8.3 展望 ............................................................................................................................190 8.4 参考文献 ...................................................................................................................192 VI

分享到：

赞收藏

资料库

针对热红外和微波数据的地表温度和土壤水分反演算法研究.pdf

相关推荐

开发技术

热门标签

最新资料