基于三维建模与组分发射辐射分离的异质性场景像元尺度表面温度模拟研究

Many in-situ measurements for validating the remotely sensed land surface temperature (LST) are currently conducted at homogeneous surfaces, and they are far from the requirements of the applications of the LST products for heterogeneous surfaces. Validations of LSTs of heterogeneous surfaces are still limited due to the mismatches between the in-situ measurements and thermal remote sensing pixels, such as the mismatches of spatial scales and the sun-surface-sensor geometrical relations. This proposal aims to simulate the surface temperature at the pixel scale for heterogeneous surfaces based 3D modeling and component emissions separation. By selecting typical vegetation-background surfaces as the object of study, this research will conduct the following investigations. First, the research will generate the 3D model for typical vegetation-background surfaces and simulate the sun-surface-sensor geometrical relations, and then component fractions in the field-of-view (FOV) of the pyrgeometer located at the ground level and in the pixel of the thermal remote sensing image will be calculated. Second, a component emissions separation model will be developed to estimate the emissions from the components located in the FOV of the pyrgeometer. With the emissions, the component temperatures will be calculated. Third, with the component fractions and component temperatures within the pixel, the surface temperature at the pixel scale will be calculated and it will be used as the "quasi-ground truth" for validating the remotely sensed LSTs; the "quasi-ground truth" will be evaluated with simulation datasets and the models reported by literatures. Finally, the proposed methods will be applied to the LSTs retrieved from the extensively used satellite thermal sensors, including Landsat-7 ETM+, Terra ASTER, Terra/Aqua MODIS, and HJ-1B IRS. In addition, the contributions from the errors contained by the in-situ measurements and the spatial and geometrical mismatches to the uncertainties of LST validating will investigated. In this research, the in-situ measurements, models, and applications to remotely sensed data will be integrated. We believe that the findings and achievements of this research will contribute to the validations of remote sensing products as well as thermal remote sensing.

地表异质性场景实测数据与卫星遥感像元存在的空间尺度匹配、太阳-地面场景-传感器几何关系匹配等难点，使异质性场景遥感地表温度验证受到严重制约。以异质性场景三维建模-地面实测数据与遥感像元转换-遥感地表温度验证为主线，开展以下研究：1）对典型植被-背景场景开展三维建模，模拟太阳-地面场景-传感器的几何关系，获得地面长波辐射传感器与星载热红外传感器视场内的组分覆盖比例；2）研究场景发射辐射的昼夜周期特征与时序演进特征，实现地面长波辐射传感器视场内组分发射辐射分离，获取稳定且有代表性的组分温度；3）实现植被-背景场景像元尺度表面温度模拟，并对模拟结果进行验证与评价；4）面向常用的热红外传感器开展应用分析，厘定实测数据误差、空间尺度误差与几何关系匹配误差等因素对遥感地表温度验证不确定性的影响。 研究将注重地面实测、模型模拟与遥感应用的充分结合，研究成果将有望提升卫星遥感产品真实性检验的定量化水准。

地表异质性场景实测数据与卫星遥感像元存在的空间尺度匹配、太阳-地面场景-传感器几何关系匹配等难点，使异质性场景遥感地表温度验证受到严重制约。自2014年以来，本项目以异质性场景建模-地面实测数据与遥感像元转换-遥感地表温度验证为主线，开展了以下研究：.1）野外观测试验与实测数据收集：在“黑河流域生态-水文过程综合遥感观测联合试验”等大型综合性遥感观测试验的支持下，获得了野外观测数据，包括水文气象数据和卫星/航空遥感数据等；开展了大量的地表温度多尺度观测试验，为地表温度的模拟、尺度转换和真实性检验提供了第一手的宝贵数据。.2）植被-背景场景三维建模与组分覆盖比例确定：基于地面人工量测与地面三维激光扫描等方式获得的场景结构参数，结合3D Max与Visual Studio+OpenGL对植被进行建模，确定场景组分比例。.3）基于昼夜周期特征与时序演进特征的发射辐射分离：基于异质性场景的长波辐射观测与场景中各组分的长波辐射观测与热红外辐射观测，分析了不同组分发射辐射的昼夜周期特征。.4）异质性场景表面温度模拟与评价：根据真实三维场景模型提取的各组分比例，结合野外试验观测的热像仪组分温度数据、热红外辐射组分温度数据，分别模拟异质性场景的表面温度。.5）遥感地表温度验证：针对ASTER、MODIS、AATSR等地表温度数据，在像元尺度开展了遥感地表温度的验证。.在本项目的支持下，项目组在IEEE Transactions on Geoscience and Remote Sensing、International Journal of Applied Earth Observation and Geoinformation等学术期刊上共发表学术论文9篇，其中SCI论文6篇，EI期刊论文1篇，其余核心期刊论文2篇。完成会议论文8篇，学位论文4篇。申报发明专利4项。培养硕士、博士研究生多人，其中4人已顺利毕业。.在本项目的支持下，先后共有4名青年教师受益。其中，项目负责人入选“四川省学术与技术带头人（后备人选）”和科技部国家遥感中心“遥感青年科技人才创新资助计划”，晋升为教授、博士生导师，所带领的课题组在地表温度的多源遥感协同反演、尺度转换与真实性检验方面已形成了较为鲜明的特色。