离散森林场景组分温度模拟及热辐射方向性建模

Temperatures within a forest ecosystem are essential inputs for applications such as fire danger prediction and productivity estimation. The thermal radiation directionality affects the inversion of forest temperatures from thermal infrared observations, and its modeling is the theoretical basis to improve accuracy of inverted results. This project intends to develop a suitable model for studying thermal radiation directionality in discrete forest scenes. Complex structural characteristics of forests, including leave clumping phenomenon caused by crown shape and vertical stratification, are not fully considered in existing directional thermal radiation models. Therefore, based on hybrid geometric optical and radiative transfer models in the visible/near infrared domain, a directional thermal radiation model is proposed to explain the relationship between component temperatures and thermal infrared observations. Because surface temperature may vary significantly as solar radiation and meteorological conditions, a combined model between the new hybrid model and energy balance method is developed, in which component temperatures and directional thermal radiation for discrete forest scenes can be synchronously simulated. The combined model can be used for multi-angle simulations during a continuous period by means of the iteration between radiative transfer and energy balance processes. The combined model can be validated based on simulated and measured data sets, and its applications will be studied such as analyzing diurnal and seasonal variations of thermal radiation directionality, angular normalization of temperatures and component temperature inversion for forest scenes.

温度是森林生态系统火灾预警和生产力估算等研究的重要输入数据。遥感方法获取森林温度还面临着热辐射方向性问题，对其建模是提高温度产品精度的理论基础。本项目拟发展一套适用于离散森林场景的热辐射方向性模型。针对森林复杂的结构特征，包括树冠形状引起的叶片聚集现象和垂直分层现象，借鉴可见光/近红外波段辐射传输-几何光学混合模型的建模思路，构建组分温度与热红外观测的关系，解决现有模型刻画森林结构特征不充分的问题。由于温度随太阳辐射和气象状况变化显著，因此将新提出的混合模型与能量平衡方法耦合，通过二者的迭代实现离散森林场景组分温度和方向热辐射同步模拟。最后，基于模拟和实测数据集对新模型进行验证，并研究其应用方案，如分析离散森林场景热辐射方向性日变化和季节变化规律，支撑地表温度角度归一化算法和组分温度反演算法研究。

森林具有巨大的经济和社会效益，也兼具涵养水源、保育土壤和固氮制氧等生态功能。温度是影响森林生态系统光合、呼吸作用的关键因素，遥感的方法可以大尺度、连续地获取森林的温度信息，在森林的生产力估算及火灾预警等研究中扮演着重要角色。现有遥感温度产品存在方向性问题和混合像元问题，其精度难以满足研究和应用需求，在森林区域尤其明显，其结构复杂、多样，且树冠有明确的几何形态。本研究针对森林热辐射方向性问题开展了建模研究，基于几何光学和光谱不变理论，构建了针对森林结构特征的热辐射方向性模型，满足了对秋冬季节北方落叶林和复合分层森林的热辐射模拟需要；然后开展了地表温度模拟研究，发展了基于深度学习方法的地表温度模拟经验模型，其展示出了与物理模型相当的模拟能力，但是仍旧存在着由于训练样本不足导致的精度问题；最后开展了遥感实验数据处理，对模型和方法进行了验证。本项目可以提供森林温度反演研究的理论依据，有望进一步提升森林温度的精度。