非均匀地表水热通量遥感中的基于对象方法研究

Study on remote sensing of land surface heat and vapor fluxes has been confronted with a significant difficulty, i.e., the spatial scale problem over heterogeneous surfaces. In the last decade, it even become the most important issue in the area of remote sensing heat fluxes. In the study of the models and methods in this effort, three issues have been focused and emphasized: 1) how to define and calculate land surface heterogeneities accurately; 2) how to introduce factor of heterogeneities into the traditional radiation transfer model and its reversal methods; 3) how to obtain information of surface heterogeneities in the remote sensing data in applications. Compared with the complex and continuous land surface, remote sensing pixels are highly simplified and uncontinuous, i.e., each pixel is an independent unit in the chain of data processing to retrieving and application. This situation could be changed by the method of object-based. Object is a group of pixels that are next to each other and similar in some features. The object is more likely has a natural explanation than one pixel in the geometric or physical meaning. Moreover the behavior and characters of an object are easy to understand and describ than a single pixel.The object-based method can serve as a useful tool to deal with heterogeneous surface issues. In this project, the way of defining the object and its features will be investigated, and the method of obtainning useful spatial information in this way is the main interest. This method will be used in the study of land surface heat fluxes by remote sensing data of different resolution. By doing this, new prospects and techniques are hopefully inspired and developed in the remote sensing of heat fluxes, and the accuracy of remote sensing products are enhanced in the future.

非均匀地表水热通量遥感已经成为定量遥感研究的重要方向。如何准确定量表达空间异质性，如何在辐射传输和反演模型中加入此类异质性因子，以及如何在实际应用中获取（计算）异质性因子，是需要迫切解决的三大问题。本项目针对自然地表复杂连续和遥感像元简化割裂这一矛盾，提出基于对象（Object Based, 简称OB方法）的定量遥感参数空间信息分析方法，旨在借鉴面向对象方法的思路，在定量遥感空间异质性研究领域开辟新的方向，为非均匀地表水热通量遥感提供新方法。重点在于从自然地表出发，研究像元层次和对象层次结合的分析方法，提出地表非均匀信息提取有效方法。在此基础上，研究新方法在不同尺度上的适用性，提出多尺度数据结合的分析方法，发展利用多源卫星数据反演下行辐射和计算地表水热通量的方法。通过本项目的研究，将在遥感水热通量估算方法，以及遥感数据的应用等方面有所突破和发展，为相关研究和应用提供新思路，新方法，新技术。

项目提出的“基于对象”方法在水热通量定量遥感计算中的研究是首次将面向对象方法引入定量遥感建模和反演。项目研究以“地表能量平衡和水热通量遥感方法”为主题，以“基于对象”的方法探索为方向，以多源卫星数据的计算方法为出口，研究和发展了地表水热通量及能量平衡卫星遥感理论及方法。项目共取得各项成果20项，其中发表科技论文10篇（SCI检索论文7篇），出版专著1部，申请发明专利3项（授权1项），发布国家标准1项，取得软件著作权3项，获得科技奖励2项。培养研究生5名。项目取得的成果主要有：1）考虑到植被类型作为一种自然对象对冠层气孔导度有决定性的影响，通过引入不同植被类型叶片的气孔长度和密度等参数计算最大气孔导度，构建了考虑植被类型的冠层气孔导度模型，提高了冠层气孔导度模型估算精度，为冠层气孔导度遥感产品实现以及地表蒸散、能量平衡及陆面过程模型研究提供支持；2）以中高分辨率卫星数据为目标，在较精细的尺度上，以在特定尺度上具有一定意义的对象作为研究主体，建立基于对象的地表水热通量遥感估算方法。研究结果表明，引入分割对象的方法能明显改善地表水热通量反演结果的质量（RMSE由61w/m2 下降至48w/m2,MBE由39w/m2下降至为19w/m2）；3）针对非均匀下垫面水热通量估算提出不同分辨率数据的不同解决方案，基于Sentinel-2和Sentinel-3数据发展了低空间分辨率（1 km）和高空间分辨率（10m）水热通量遥感估算方法，从而得到更加精确的通量计算结果。