【作者】 辛晓洲；

【导师】 田国良； 柳钦火；

【作者基本信息】 中国科学院研究生院（遥感应用研究所） ， 地图学与地理信息系统， 2003， 博士

【摘要】 系统分析总结了遥感计算地表蒸散的意义和目的、研究历史和现状、以及面临的主要困难。遥感替代点上的蒸发测量，实现大范围地表能量和水分动态监测，可以在气候、生态、农业、水文等领域中发挥重要作用，过去几十年的研究虽然取得了许多重要成果，但依然面临很多有待突破的重要问题。本文的主要特色是方法研究，对遥感计算地表蒸散的方法和模型做了较多的理论分析，数据验证，以及改进和创新，并对其空间和时间尺度问题做了一定的探讨。 针对遥感中最常用的单层模型做了深入的理论探讨，在学术关键问题——遥感表面温度与空气动力学温度的关系上，提出了新思路和新方法，为理解这个复杂的问题增添了新的视角。该方法利用表面热辐射方向性和显热通量都是源于土壤和植被温度的贡献这一共同点，发现在一定倾斜角度的观测下，视场中植被与土壤的比例可以较好地反映植被和土壤与大气进行湍流热交换的贡献率，所以称该方法为最佳组分面积比法，用最佳组分面积比可以将任意角度下观测到的辐射温度订正为空气动力学温度，经过地面和遥感数据验证表明，用该方法计算的通量精度高于普通的单层模型。 双层模型虽然提出很多年了，但在应用中一直存在信息不足，求解困难的问题，本文用最新多角度热红外遥感数据———AMTIS系统反演的组分温度，首次实现了双层模型在遥感中的应用，结果表明，在输入参数精度相近的情况下，双层模型模拟的通量误差远小于单层模型，尤其在土壤干旱，表面温度较高的地表，双层模型的理论优势在计算中表现得更加突出。在组分温度的帮助下，实现了土壤蒸发和植被蒸腾的准确分离，并且用植被蒸腾速率反算出冠层气孔阻抗、二氧化碳通量、以及作物群体水分利用效率等农田关键参数，是遥感获得此类参数的新思路。本文还提出了一种新的双层模型的简化求解方法，在没有组分温度信息的情况下，可以利用经验参数和表面温度实现双层模型计算。 针对遥感基于像元计算的特点，提出一种全新的通量计算方法，模型的结构和思路专为遥感像元面上计算的特点而设计，主要考虑由地表非均匀和不连续性引起的像元内部热量交换——局地微平流对显热和潜热通量的影响，验证表明，这种考虑平流的非均匀模型可以较好地模拟地表非均匀状态下的热量通量。 论文还分析了蒸散模型的空间尺度效应，通过一系列的模拟实验检验了双层模型中国科学院遥感应用研究所博士学位论文在不同地表非均匀状态下的尺度误差，并且根据通量尺度扩展的基本法则推导了双层模型参数的尺度扩展公式。结果表明亚像元状态和结构是像元尺度误差的主导因素，风速变化对尺度误差有明显的影响，不论何种形式的非均匀性，都须遵循相同的尺度扩展步骤，以达到消除误差的目的。将AMTIS和ASTER图像降低分辨率后计算通量，检验其尺度误差，结果表明，在不同覆盖类型的交界处通量误差比较明显，由于算法本身的不连续性，其误差规律较为复杂，另外，道路、建筑等对通量计算无用的信息也是造成大像元尺度误差的原因之一。 论文的最后部分讨论了从瞬时蒸散速率推算日蒸散量的方法，介绍了简化法和蒸发比率法的基本思路，分析了存在的主要问题，并用地面数据和遥感数据验证了蒸发比率法中以长波上行辐射作为分母的方法，结果表明方法可行，精度基本可以满足要求。风速变化和云的干扰依然是时间尺度扩展中遇到的主要问题，由于难度较大，这方面的研究还需要更多，更细致的工作。更多还原

【Abstract】 Remote sensing provides an approach to monitor land surface energy and water balance over a large area simultaneously, which is very important and useful in researches and applications in global climate change, hydrology, ecology and agriculture etc. In this paper, methodologies and models on remote sensing of land surface evapotranspiration were investigated in details and validated using field and remotely sensed data, several novel concepts and models were proposed as the center part of this paper, and spatial as well as temporal scale problems were discussed in the second half of the paper.Single-layer model is convenient to apply but the unclear relationship between radiative and aerodynamic temperatures is still a bottleneck in this field. A new method was developed to derive reliable surface heat fluxes from radiative temperature viewed from arbitrary zenith angle. Aerodynamic and radiative temperatures are connected through a so-called Optimum Component Fraction (OCF) parameter-The fraction of vegetation in the field of view when the two temperatures are equivalent in oblique viewing. Heat fluxes estimated from radiative temperature by this model is more accurate than other regular corrective methods.Two-layer model has been proposed for many years but was difficult to apply in remote sensing because component temperature were unavailable in traditional thermal sensors. A new airborne multi-angular thermal sensor system and retrieved soil and canopy temperatures were used to solve two-layer model, and the simulated heat fluxes show much better accuracy than the results from one-layer model especially above dry surfaces. And this is the first full application of two-layer model in remote sensing. From separated evaporation and transpiration through the model some important field parameters can bederived, such as canopy resistance, CO2 flux and crop water use efficiency. A simplified two-layer model was also presented in case of that only radiative temperature is available.A very much different fluxes model was suggested for remote sensing estimation, which takes account of the effects of advections in vertically or horizontally anisothermal vegetations. The inter-exchange of heat can result in lower total sensible heat flux and higher total latent heat flux of the considered areas. This method is designed technically for the simulations of pixel, is a try of new generation effluxes model.Scaling-up of patch model is necessary in the calculation of surface energy fluxes and evapotranspiration from remote sensing data. The simulation error of two-layer model caused by sub-pixel heterogeneity and discontinuity of surface geometry and physics were investigated using a number of data experiments. It is shown that the error could be rather remarkable in some extreme situations and could be neglected in the others. The variance of parameters inside pixel, contexture of the pixel and the surface wind speed are the controlling factors of the scaling error. AMTIS and ASTER images were scaled up using pixel aggregation algorithm to find scaling error of surface flux estimation. The results show that the largest error appears at the interface of different coverage types, and the error show much complexity because of the discontinuity of algorithms at these boundaries. Additionally, the contamination of building, highway and other ground information also add some error in the estimation of land surface evapotranspiration.To derive the accumulated daily evapotranspiration from remotely sensed instantaneous evaporation rate is a key step to use this kind of information in other domains. So-called Simplified Methods and Self-Preservation Methods were introduced and compared, and a Self-Preservation Method was validated using field and remote sensing data. The estimated daily evapotranspiration is mostly affected by diurnal variations of surface wind speed and overpass of clouds, and more detailed and intensive research works are to be carried out to obtain more reliable daily water evaporation loses.更多还原