【作者】 王安琪；

【导师】 宫辉力； 施建成；

【作者基本信息】 首都师范大学 ， 地图学与地理信息系统， 2013， 博士

【摘要】 土壤水分是控制陆地和大气间水热能量交换的关键因子之一,作为水循环体系的重要组成部分,土壤水分在地球生态系统中起着重要的作用,是陆地植物以及其它土壤生物得以生存的物质基础。此外,土壤水分也是各类气候模型、水文模型、生态模型以及陆面过程模型的核心输入参数,其变化会影响水热过程,从而改变地表参数并影响气候变化。因此,定量化获取准确的土壤水分数值对农业生产、应对全球变化、生态环境保护等众多领域都有着重要的意义。不同的遥感平台可提供不同时间和空间尺度的土壤水分,它们分别具有各自的优势与不足。被动微波数据较热红外、主动雷达数据而言,对土壤水分信息敏感,重返周期在1到3天,可以捕捉到土壤水分的时间变化信息,但是其空间分辨率较低(25km-40km),一般适用于大尺度研究；星载可见光/热红外数据可以达到中等分辨率到高分辨率(100m-1km),但是其应用可能受到天气条件限制,而且对土壤水分的敏感性不够理想。为了克服上述问题,就需要将被动微波数据和光学数据结合,从而同时提高数据的时间和空间分辨率,以得到可靠的土壤水分。本文计划利用被动微波辐射计AMSR-E土壤水分数据和MODIS数据,发展大尺度土壤水分降尺度算法,并选取亚洲蒙古地区与青藏高原地区,作为方法验证区域。论文的研究内容具体包括以下四点：(1)被动微波辐射计AMSR-E反演大尺度土壤水分选取被动微波辐射计AMSR-E数据,通过引入Qp模型对已有算法进行改进计算得到的。该算法在Jackson的单通道算法的框架下,利用37GHz V极化亮温计算地表温度,采用植被含水量和归一化植被指数的经验关系,获取植被光学厚度,消除植被对微波信号的影响,最后采用基于Qp模型发展的双通道算法消除地表粗糙度的影响从而直接获得25km土壤水分。(2)利用空间频谱信息方法分解大尺度土壤水分引入频谱降尺度方法,即通过傅里叶变换将土壤水分数据的灰度变化函数分解为一系列周期函数的叠加,再通过对上述周期函数的统计转化为由振幅和相位表示的频率域图像。从土壤水分遥感图像频域出发,在不同空间分辨率的土壤水分图像其空间频率与功率谱密度的之间存在固定关系的前提下,结合AMSR-E低分辨率土壤水分图像,建立该图像功率谱密度和空间频率的幂指数拟合关系。其中作者对M. Montopoli (2012)等人提出的方法上进行了部分改进,优化了幂指数拟合关系,得到高分辨率振幅信息,同时利用克里金插值方法随机生成了高分辨率相位信息,经过傅里叶逆变换,选取蒙古地区,实现大尺度土壤水分降尺度方法,并分析了插值方法的数据局限性,从而提出了结合高分辨率数据的必要性和方法大致思想。(3)基于蒸散原理土壤水分相关参量响应机制分析利用被动微波辐射计AMSR-E土壤水分和MODIS地表温度LST、归一化植被指数NDVI与反照率Albedo产品,以基于Ts-NDVI三角形特征空间的蒸散理论为基础,采用多元线性回归的统计方法对研究区内长时间序列中,被动微波辐射计空间尺度下土壤水分对地表温度、植被指数与地表反照率等光学遥感参量的响应机制进行分析,从而得到土壤水分数据与遥感获取的以上光学因子之间的经验关系。(4)基于高分辨率相位信息频谱降尺度方法提出、实现与研究区验证利用前文得到的AMSR-E土壤水分和MODIS地表温度LST、归一化植被指数NDVI与地表反照率Albedo等光学因子之间的经验关系式,由MODIS数据推导得到高分辨率相位信息,并代入替换土壤水分数据中的低分辨率相位信息。之后通过傅里叶逆变换算法,即可得到空间分辨率优于被动微波数据分辨率的降尺度土壤水分数据。通过与蒙古实验区及青藏高原实验区内CEOP计划地面实测数据的比较表明,使用本文方法获取的高分辨率土壤水分数据的数值与变化取值与实测数据吻合较好,证实论文提出的降尺度方法的准确性及可信度较高。更多还原

【Abstract】 Soil moisture is a key parameter in regulating the hydrothermal energy exchange between land and atmosphere, as well as an important part of the hydrologic cycle of the ecological system on ground surface and an important material source for terrestrial plants and soil organisms. Soil moisture is also an important input parameter of many models, such as climate model, hydrological model, ecological model, land surface model etc. The change of soil moisture will affect the hydrothermal process, then change the surface parameters and impact the climate. Therefore the quantitative monitoring of accurate soil moisture value is of great significance to agricultural production, global change strategy, ecological and environmental protection and many other fields.Different types of remote sensing platforms are currently used to infer soil moisture at different spatial and temporal scales, each with its specific characteristics and limitations. Compared to thermal infrared and active radar data, passive microwave data gives more accurate inversion of ground surface soil moisture. The passive microwave data, being sensitive to soil moisture information and having a repeat cycle of1to3days, can capture the temporal variation information of soil moisture, although its relatively low spatial resolution (25km-40km) makes it generally suitable for large-scale research. Average soil moisture information of large-scale grid can be obtained from a variety of active and passive microwave remote sensor platform with much ease. While the satellite-borne visible light and thermal infrared data can achieve a medium to high level of resolution (100m-lkm), it’s likely to be affected by weather conditions and has a less-than-ideal sensitivity to soil moisture. To overcome these issues, some of the current researches incorporate passive microwave data and optical data, with the purpose of improving both temporal and spatial resolution while achieving reliable soil moisture information.Choosing Mongolia and the Qinghai-Tibet Plateau region as the study area, this paper improves the algorithm of decomposing large-scale soil moisture via passive microwave radiometer AMSR-E soil moisture data and MODIS surface temperature, NDVI and albedo products.The four main study points of this paper are shown as follows:I. Retrieving soil moisture from passive microwave radiometer AMSR-EThe author calculates the25km AMSR-E soil moisture data from passive microwave radiometer AMSR-E data through an improved version of the Qp model. Based on Jackson’s single channel algorithm, the author uses37GHz V polarized brightness temperature to calculate the surface temperature, and get vegetation optical thickness from the experience relationship of vegetation water content and NDVI, in order to eliminate the influence of vegetation on the microwave signal. Finally, based on the dual-channel algorithm developed from Qp model, this paper eliminates the impact of surface roughness and obtains the25km soil moisture directly.II. Decomposing large-scale soil moisture utilizing space spectrum methodThis paper uses the method of space spectrum downscaling, which is improved from M. Montopoli’s algorithm, to decompose AMSR-E soil moisture data. By breaking down the gray-scale changes function of the soil moisture data into a series of superimposed periodic function using the Fourier transform, the author converts the soil moisture data to frequency domain images of the amplitude and phase through statistical conversion on periodic functions. Due to the fixed relationship between the soil moisture and the spatial frequency of the power spectral density in different spatial resolution images, the exponent relationship of the power spectral density and spatial frequency can be created.III. Analyzation of the empirical relationship of soil moisture parameters based on the evapotranspiration theoryThe paper analyzes the long sequence response mechanism between soil moisture and optical remote sensing parameters, such as surface temperature, vegetation index and surface albedo. This method is based on the triangle Ts-NDVI feature space learning from the evapotranspiration theory and statistical method of multiple linear regression analysis, from which the author gets the empirical relationship between soil moisture and the remote sensing optical factors above. IV. The method of space spectrum downscaling via high-resolution phase information with its application and field verificationThe author retrieves high-resolution phase information from MODIS data via the relationship between AMSR-E soil moisture and MODIS surface temperature, NDVI and albedo, and then substitution to replace soil moisture data the low-resolution phase information. Then by the Fourier inverse transform, spatial resolution better than the resolution of passive microwave data can be obtained by downscaling soil moisture data.By verifying the results with the CEOP ground data in Mongolia and Qinghai-Tibet Plateau study area, it shows that the high-resolution soil moisture data obtained using the downscaling method introduced in this paper matches the measured data in good agreement. Thus the high accuracy and credibility of the method is confirmed.更多还原