【作者】 赵利民；

【导师】 田庆久；

【作者基本信息】 南京大学 ， 地理学， 2011， 博士

【摘要】 地表热辐射场景生成机理与遥感成像过程的模拟分析一直是热红外遥感领域高度关注的研究热点。新型航天红外遥感成像系统研发与设计需要大量的前期论证,传感器参数的设计离不开多种场景、成像条件下红外辐射场景特性分析及所成图像的评估,这就需要对热红外遥感过程进行模拟。同时,正确认识和建立热红外遥感前向模型也是进行地表温度反演的基础,通过研究地表能量的产生机理和变化规律,以及成像过程对红外影像的影响,可以为提高地表温度反演精度提供参考依据和解决途径。本文以地表热辐射的产生、传递为主线,以热辐射动态场景的建模、分析为重点,围绕热红外遥感成像过程链路进行了系统的研究。首先结合一维地表导热方程与地表能量平衡模型,模拟植被、裸土、柏油路面和混凝土地面四种典型地表组份热辐射日变化规律；在此基础上,利用随机分形的思想建立三维地形场景,基于能量线性混合的假设,以高空间分辨率光学遥感图像(IKONOS、AVIRIS)丰度分解为基础,构建了5m/10m分辨率的热辐射背景场,并结合辐射换热理论探索表面间能量的交换,解析地描述了表面间多次散射引起的目标有效辐射增量,并建立多次散射模型MSM(Multi-scattering Model)以分析目标红外辐射特性；基于热辐射大气传输模型和湍流MTF模型分析大气对地表热辐射场景的影响,最后根据能量传递的物理过程模拟红外传感器对地表热辐射场景的响应。通过本文的研究,得到以下主要结论和认识：(1)地表温度日变化规律受物侯特性影响,且与近地表气象条件相关二次谐波形式可以较好地表达裸露地表温度日变化特征,模拟值与实测值最大标准偏差在3℃以内,误差主要来源于地表热力学参数测量、估算方面。裸露地表的导热系数越低,昼夜温度起伏越大；对流换热系数越小,温度出现峰值的时刻越接近太阳辐照度峰值时刻；对于植被冠层,白天温度还受叶面积指数的影响。近地面大气相对湿度、温度与地表温度之间具有较高的线性关系,尤其是大气温度对地表温度很高的指示作用。植被和水体温度在昼夜周期内的变化相对平缓,裸露地表日温度变化相对剧烈,且受入射辐射影响较大；日出前和日落后一段时间内各类地表温度差异最小,正午前后一段时间内温度差异最大。(2)地表热辐射场景随地表温度、类型分布以及观测方位产生动态变化基于分形地形和地表丰度/覆盖度数据,结合地物温度与发射率特性,可简单有效地模拟出地表热辐射场景分布与方向亮温变化,且不丧失地表纹理细节信息MSM模型考虑了像元组分间多次散射辐射的影响,可提高混合像元有效辐射的模拟精度,亦可用来定义非同温混合元的有效发射率。地表热辐射场景细节丰富程度以及亮温方向特性均随时间发生变化：温差越大时地物纹理越清晰,场景亮温方向特性越明显。地表单元的有效辐射与单元内组分温度、材质、几何结构以以及组分细分程度等有关,并随观测角度而发生改变；组分间的多次散射增强了地表有效辐射,但同时又对辐射方向特性产生平滑作用；大气下行辐射对地表辐射方向特性无明显影响,但增加了像元的有效辐射,使像元亮温更接近表面温度。(3)大气对地表热辐射场景产生模糊作用大气辐射与消光作用削弱了传感器入瞳处地表热辐射场景的动态范围,大气湍流使热辐射传递路径发生抖动,进一步模糊了传感器观测到的热辐射影像,且空间分辨率越高,大气湍流对影像的模糊作用越强。夜间条件下大气辐射一般为正效应,增强了地表辐射表观辐亮度,而白天时大气一般起负作用。大气水汽、观测角度、通道设置以及地表比辐射率是地表热辐射场景大气作用模拟需要重视的关键参量,而对大气湍流作用的模拟需要注意垂直廓线方向上湍流折射结构系数变化。(4)场景热辐射信号经探测器系统成像被进一步退化增大红外相机的孔径可以增加探测器焦平面上聚集到的能量,有效降低光学系统衍射造成的模糊效应；探测器对辐射能量的响应受积分时间、平台振动的干扰,使得获得的图像质量进一步退化；噪声产生于成像系统整个过程中,对于集成度较高的当代红外遥感成像系统,可以用等效噪声温差来表征系统的噪声水平系统产生噪声对低温条件下成像的图像质量的影响显著。更多还原

【Abstract】 The modeling of land surface thermal radiance scene and the simulation of the remote sensing imaging procedure play an important role in the studies of thermal remote sensing fields. Recently, with active renewals for remote sensors, the application fields meet more actuate observation precision for the Earth land. In this situation, the sensor designers need to develop various thermal imaging systems to meet the different and complex land radiance conditions, and the plentiful early demonstrations of the new remote sensing system are desired. The definitions of the sensors’parameters depend on various evaluations for the land surface radiance and imaging characteristics in different natural conditions, which require lots of field experiments. An economic and effective way is computer modeling and simulation. Meanwhile, with the correct understanding of the origin, reactive, and atmospheric transfer of surface energy flux senses, and the systematic analysis of the factors affecting the final thermal imagery, it is useful to improve the accuracy of land surface temperatures retrieval from thermal remotely sensed imageries.This research is made around the whole link of the thermal remote sense imaging system, and the main purpose of this paper is to investigate the thermal radiant characteristics of the typical land surface and to analyze the main factors for thermal remote sense imaging procedures. The main line of this paper is the origin and transfer of the land surface energy flux, where the focal points are the modeling and analyzing of the dynamic thermal radiance scenes.Assuming that the land surface background is composed by 5 typical components, including the vegetation canopy, the bare soil, the water body, the asphalt pavement and the concrete surface. Land surface temperatures for different land covers are the key parameters for land thermal radiance scenes modeling. Firstly, the temperature daily variations for four typical solid surfaces are modeled and analyzed, which are the vegetation canopy, the bare soil, the asphalt pavement and the concrete surface. The harmonic model is built to simulate the daily change of the temperature for bare soil, asphalt pavement and concrete surface, and a model of SVAT (soil-vegetation-atmosphere transfer), CUPID, is employed to simulate the vegetation canopy temperature. The biological and thermal proprieties of the foresaid four kinds of land surface are measured or evaluated, which are then used as inputs of the temperature models respectively, and 24-hour observations of the surface temperatures are carried to calibrate the modeled results. The sensitive analysis for the harmonic and CUPID model is made to exhibit the main impact factors for surface temperature.Based on the radiation heat exchange equation over surfaces and the linear-energy-mixed theory, the thermal radiance of land scenes with 5m/10 m spatial resolutions are modeled. The main inputs of the model are surface temperature, emissivity and spatial distribution of the components. The fractal-based method is applied to simulate the 3-Dimentional natural terrain; the proportion of each component is evaluated or extracted from multi/hyper-spectral optical remote sensing imageries. Daily variation of the directional brightness temperatures (DBT) for a 5 m resolution scene is simulated, and the influence of downwelling atmospheric radiant to the scene DBT is analyzed.The thermal reflectance for bare surface is not negligible during the scene radiance modeling, since the thermal radiance that emitted and reflected synchronously at each component are at the similar levels.. The concept of configuration factor is applied, and the multiple scattering effects between heterogeneous non-isothermal surfaces are described rigorously, based on which a directional thermal radiance model named Multi-Scattering Model (MSM) is built, and the numerical calculation of the MSM is discussed. The MSM is applied to modeling the DBT of row crops, and the results are compared with measured DBTs. The MSM is also used to describe the effective emissivity over non-isothermal targets, and a simulation of the remotely sensed pixels with "V" structure is performed, respectively.The atmospheric attenuation of land surface thermal radiation is simulated with 3 ways according to the abundance of the meteorological data, which is the radiative transfer method, the empirical method and the lookup table (LUT) method. The atmospheric effects simulations are carried out with a Gaussian-Triangular filter as the sensor channel response function, which takes the radiation wavelength range of 10.5μm-12.5μm. The degeneration of the land surface radiance by the atmospheric turbulence is simulated using very short exposure atmospheric modulate transfer function (MTF), and then the land scene thermal radiance at the top of the atmosphere is obtained. The physical procedure of the at-sensor radiance transfer in the thermal remote sensing camera system is analyzed briefly, and the halo and diffractive effects caused by the optical imaging system, and the image movement that is brought by the dither of the remote sensor platform are simulated. The stochastic noise of imaging system is indicated by the noise equivalent temperature difference (NETD) and the SiTF. The NETD is modeled and calculated according to the parameters of a remote sensor, and then the system noise is injected to the digital image. Finally, the response and degenerate of the radiance scene by the thermal remote sensor is simulated.With the modeling, simulation and analysis of the thermal radiance scene and the imaging procedure, some of results and conclusions can be drawn as follows: (1) The daily variation of the land surface temperatures are fluctuated by the solar radiance and the near surface metrological conditions. Meanwhile, they are restricted by the surface thermal/biological properties.The analytical temperature model with second harmonic terms presented in this paper can be used to predict the daily bare ground surface temperature variation with sufficient accuracy, and the result is restricted by the measurement or estimation veracity of the ground thermal properties. The solar radiance is the basic reason that makes the ground temperatures changing temporally. For bare surfaces, the daily diversity of the temperatures increases with the rise of the thermal conductivity, while the time that the peak value of the daily temperatures occursclose to midday as the convective heat transfer coefficient decreased.In addition, for vegetation canopies, the temperature in daytime is also affected by the leaf area indexes (LAI). There is a high correlation between surface and the air temperature, and a good correlation between temperatures and the relative humidity. In daily scale, the fluctuation of surface temperature is smoothly for vegetation canopy and water body, and rough for bare and dry surfaces; the minimized differs of all kinds of ground surface occurs at a short period of time before sunrise or after sundown. (2) The land radiance scene dynamically changed with the distribution of surface temperature, land cover and the observation directions.Base on the fractal terrain simulation and proportion information extraction of surface components, the land radiance scene and the DBT variation of it can be modeled easily and effectively, which the texture details of the land radiance distribution is remained. The multi-scattering effects over components of pixels can be completely described by the MSM model, with which the accuracy for effective radiance modeling of mixed pixels is enhanced. The effective emissivity for non-isothermal targets can also be defined by the usage of MSM. The results of thermal scene modeling show that both the detailed grade of the scene texture and the DBT characteristic of the whole scene are changing with solar-time, especially in a time period after sunrise. The more acuteness of the thermal diversity over pixels, the more details of the scene radiant textures and the variations of the scene DBTs. The effective radiance of land surface takes directional property, which is correlated with the temperatures, emissivity, spatial structure and subdivision extent of the components in each pixel. The effective radiance is aggrandized because of the multi-scattering effects, whereas the change range of DBT is smoothed. At the same time; the effective radiance is also enhanced by the downward atmospheric irradiation, which makes the brightness temperature close to the surface temperature, while the influence of the atmosphere for the scene directional brightness character can be neglected. The effective emissivity is magnified by the multi-scattering effects; the distinctness of the directional effective emissivity for non-isothermal pixel increases with the rise of the diversion for the components temperatures, the isomerous and subdivide state of the substructure, and decreases with the emissivity of components. (3) The land radiance scene is fainted by the atmospheric effects.The dynamic range of the thermal radiance scene is reduced by atmospheric extinction and radiation. The surface radiance transfer path is shuddered by the turbulent diffusion of atmosphere, and then the image that is projected to the infrared focal plane assembling (FPA) of sensor is shortly moved, which blurs the detected scene. The blurring effect drought by the atmospheric turbulence is increased with the rise of the spatial resolution for land scene and the atmospheric refractive index turbulent structure constant. The contribution of the atmospheric effects for ground radiance at TOA is positive at night while negative in daytime. Some parameters should be regarded during the simulation of atmospheric effects, which are the total water vapor contents, the observation angle, the setting of the response channels and the ground surface emissivity. (4) The land radiance scene image is deteriorated further after the response, transmission, processing of the signals by the sensor system.With the increase of the aperture size of the infrared camera, the energy focused to the FPA increases accordingly, and the image blurring brought by diffraction of the optical system is reduced consequently. The radiance response of the sensor is disturbed by detector integral time and vibration of the platform, and the remotely sensed imagery is deteriorated. The noise originated from the entire process of the land radiance imaging can be characterized by the NETD for highly integrated thermal infrared remote sensing system. With the simulation of the main steps of the sensor imaging procedures, the results show that image drift by the movement of platform is the key reason for which the image is blurred by sensor; the system noise is influenced by the ground radiant levels and the atmospheric transmittance, therefore the noise has significant effect on the image quality for night thermal scene.更多还原