面向低发射率物体的高光谱热红外数据表面温度与发射率反演方法研究

Surface temperature and emissivity in thermal infrared spectrum of objects have the very important significance in plant species classification, mineral exploration and recognition, military target detection and recognition field, etc. Compared with common natural objects, many artificial materials and target have lower thermal infrared emissivity. Thus, under the same atmospheric conditions, more serious land-atmosphere coupling problem can be found for low-emissivity object, which will increase greatly the retrieval difficulty. With the rapid development of high-resolution infrared earth observation techniques, the conventional methods are difficult to obtain high-precision temperature and emissivity of the low-emissivity objects. It is obvious that the hyperspectral data contain detailed information on the characteristic of emission and absorption of the atmosphere and land surface. In another hand, the large number of the channels helps us to stabilize the equations in the retrieval, and provides more constraints in describing the physical properties of the surface material to improve the separation of the LST and LSE. The project intends to take advantage of hyperspectral thermal infrared spectral and spatial information to develop a atmospheric correction method. Exploring the emissivity spectral characteristics for the low-emissivity object to search the physical emissivity constraints condition, a temperature / emissivity inversion model for low-emissivity object will be proposed with the aid of cost function and optimization algorithms for low emissivity objects. Finally, the field experiment will be carried out to verify the accuracies of the proposed models. The research is a deep exploration of new technology in thermal infrared remote sensing, which has a very important significance in improving the application of infrared earth observation technology.

物体表面温度及其在热红外谱段的发射率光谱在地物分类、矿物探测与识别、军事目标探测与识别等领域具有非常重要的意义。相比常见的自然地物，很多人工材料和目标都具有较低的热红外发射率，导致目标与大气间的耦合效应更加强烈，大大增加了温度与发射率分离的难度。随着高空间/高光谱分辨率红外对地观测技术的发展，现有热红外数据温度/发射率反演方法往往难以对低发射率物体进行高精度的温度/发射率反演。因此，本项目拟利用高光谱热红外遥感通道数量多、波谱连续细腻等特点，探索不依赖辅助信息的高光谱热红外大气校正算法；探索和挖掘符合实际物理现象的低发射率物体的波谱特征约束条件，构建适用于低发射率物体的高光谱热红外温度/发射率反演方法；结合野外试验，验证温度/发射率反演方法的精度。本研究成果是红外遥感新技术领域的深层探索，对于提升我国红外对地观测的应用潜能方面具有重要的价值。

地表温度和地表发射率是表征地表状态的重要物理参数，在农业生产、自然灾害防御、气象预报和地表能量平衡研究等领域有着广泛的应用。本项目主要利用高光谱热红外细腻的光谱信息和空间信息，围绕低发射率地物的温度/发射率反演实质上是从强背景中提取弱信号（温度和发射率）的问题开展研究，主要包括高光谱热红外数据的大气校正方法研究，基于大气吸收特征的高光谱热红外数据地表温度/发射率快速估计方法，基于大气下行辐射谱整体偏移特性的低发射率目标温度/发射率反演算法及实验验证研究等，达到了预期目标，具体包括：1）基于模拟数据，深入分析了大气强/弱吸收带情况下的大气透过率、大气上行辐射、大气下行辐射的内在关联性，构建了基于影像光谱和空间信息联合的高光谱热红外数据大气校正方法；2）开展了反演最优通道选择方法研究，提出了基于大气吸收线特征的高光谱热红外数据地表温度/发射率初值快速估计方法；3）提出了基于大气下行辐射谱整体偏移特性的低发射率目标表面温度/发射率反演方法，并利用实验获取的观测数据对算法进行了验证。通过本项目的研究，建立了一套较完善的高光谱热红外数据低发射率目标的表面地表温度和发射率反演算法。共发表论文10篇，其中SCI文章7篇。