卫星像元尺度地表能量平衡与水热通量遥感模型建模及算法研究

Estimation of surface energy budget and evaporatranspiration using satellite data has been one of the most important and tough tasks in the science of quantitative remote sensing. For decades, various models and methods have been developed that are based on different theories and techniques. Some of them are quite successful in the sence of linking remotely sensed data with surfaces models, and the usefulness and reliability of these methods have been proved in the applications and validations at different regions. However, it was also found that the scale effect is one of the the largest uncertainties in the methods that use remotely sensed data to drive models developed at the scale of field observations. Enven though techniques that apply multi-scale data to correct spatial scale error were developted, but it still can be said that the spatial scale issue is in the way of land surface energy balance and heat fluxes estimation from remote sensing data. In this study, we try to think in a different way and establish models of energy balance and heat fluxes at the scale of satellite pixels instead of seeking a method of error correction. The model at pixel scale is different from the model at field scale that it considers the possible spatial distribution, variation and complexity of land surface inside the pixel area. Once the model at pixel scale is built, the spatial scale issue will be readily solved since the information of the spatial heterogenety has been taken into account as model parameters. By doing so, the ability and accuracy of remote sensing of land surface energy balance and heat fluxes will be enhanced. First of all, a series of scenes with different contexture, heterogenety and degree of complexity will be built as typical pixels using RGM model. Then, the processes of surface energy balance and heat fluxes will be modeled using technique of computer simulation model based on real scene and SVAT model such as CUPID. The simulation will be run for many times till the number of results is big enough for next step analysis. And then parameter sensitivity analysis will be carried out based on the simulation results. Parameters are categorized into two types, one type includes the parameters of energy balance and heat fluxes that are same with traditional models, and the another type is to describe the contexture, variation and complexity of the pixel. The most sensitive and important parameters will be added into the new model.After the model is built, we will develop method that apply the model with satellite data. The most important thing is the way to obtain model parameters from satellite data. At last, a series of methods for several satellite sensors to estimate surface energy balance and heat fluxes will be proposed and validated for the applications of the new model. By the end of project, the ability and accuracy of surface energy balance and heat fluxes remote sensing will be enhanced for sure.

从遥感像元的视角出发，重新梳理和分析现有遥感能量平衡与蒸散模型原理及方法，提出新的研究思路和方法，给出解决遥感蒸散模型与尺度问题的综合方案。重点在于通过面向对象、随机动力、地统计等方法，研究遥感像元层次，以及图像层次的模型和方法，建立与遥感观测像元和图像空间尺度相符合的"面"模型。将传统的点上建立的模型和方法，从原理上上推到面层次，既提高了遥感蒸散的模拟能力，同时也能解决空间尺度问题。在此基础上，研究新模型在不同尺度上的参数化问题，同时针对当前的一些主流传感器建立模型的实用化算法方案，并且发展利用多源卫星数据反演下行辐射和计算地表蒸散的方法。通过本项目的研究，有望突破当前中低分辨率卫星数据估算地表能量平衡和蒸散的瓶颈问题，提高估算精度和能力，为相关研究和应用提供新思路，新方法，新技术。

遥感蒸散模型的尺度问题是困扰遥感蒸散精度的一个世界性难题。本项目开展了遥感观测像元和图像空间尺度相符合的“面”模型研究，探索了遥感像元层次以及图像层次的模型和方法，试图将模型机理和建模过程上推到像元代表的“面”层次。本研究取得成果丰硕，资助发表了学术论文20篇，其中，SCI期刊8篇，EI期刊7篇，中国科学引文数据库（CSCD）期刊5篇；申请发明专利6项（授权1项）；获得软件著作权3项。项目取得的主要成果有：1）基于HiWATER试验获取的地面观测网络以及航空、卫星遥感数据，全面分析地表景观和参数异质性对蒸散遥感估算带来的影响；2）利用数学工具研究地表参数非均匀性的表达方法，并利用实测数据分析了地表参数分布和变化的时空特征，由此分析发现与显热和潜热通量时空特征最为相似的地表变量分别为地表辐射温度和土壤表层含水量。此发现完全由观测数据分析得出，对今后地表水热通量和蒸散遥感的空间尺度分析和建模都具有十分重要的意义；3）对于地表景观和参数异质性分别提出相应的建模策略，发展混合像元蒸发比计算方法，以及混合像元日蒸散估算模型，该模型是一个考虑地表异质性的算法框架，具有很好的可扩展性，可将不同的景观和参数异质性带入模型； 4）发展综合考虑地形和云效应的地表下行辐射估算方法，可大幅减小山地下行辐射卫星反演的误差，为青藏高原下行辐射遥感估算精度不高问题提出解决思路；5）发展了针对不同卫星数据的像元蒸散估算方法，如HJ-1, MODIS, VIRSS等，在模型中考虑了像元内的非均匀情况；6）基于地表参数异质性分析结果，发展考虑多变量的地表温度降尺度算法，利用中低分辨率卫星数据得到较高分辨率的亚像元地表蒸散估算结果。与此同时，如何有效地将本项目研究成果应用于全球地表蒸散时空尺度扩展和产品生产，尚有许多工作要做。