1 INTRODUCTION

Temperature is a fundamental component of climate and a key driver of vegetation distribution and functionality. Chemical and biological reactions are regulated by temperature, which influences all levels of organization in ecological systems. At the cellular level, high temperature extremes can cause irreversible damage to photosynthetic apparatus (Berry & Bjorkman, 1980; Teskey et al., 2015). Temperature, along with other biotic and abiotic factors such as nutrient and water availability, regulates biogeochemical cycles at ecosystem scales and influences the biogeographical distribution of plant and animal species at regional and global scales (Figure 1; Jeffree & Jeffree, 1994). Furthermore, anthropogenic climate change is expected to alter temperature regimes, with impacts on plants from leaf to global scales. Despite the importance of temperature across scales, evaluating trends across space and time is difficult. This is not a unique issue—scaling challenges are persistent and fundamental in all fields of ecology (Wiens, 1989). How do we understand the influence of photosynthetic kinetics on global vegetation patterns, or assess the impacts of periodic temperature extremes on future species ranges? One way to address scaling issues is to explore phenomena concurrently at multiple spatial and temporal scales. Thermal remote sensing is a key tool that can be used to investigate the relationships between temperature and vegetation at multiple levels of organization. Furthermore, thermal remote sensing is more accessible to plant ecologists than ever before—new sensors, increased data availability and emerging technologies could facilitate novel, cross-scale approaches to advance our understanding of temperature impacts on vegetation in the terrestrial biosphere.

Thermal remote sensing data can be obtained at various spatial, temporal and spectral resolutions coming from a range of platforms including satellites, aircraft, uncrewed aerial vehicles (UAVs), mounted on towers and ground-based systems including handheld cameras (Figure 2). Most thermal infrared (TIR) sensors used for vegetation analyses measure emitted radiation in the mid-range IR (3–8 μm) and, more commonly, the long-range IR (8–14 μm) regions (Kuenzer & Dech, 2013; Zhu et al., 2018). In remote sensing, there are inherent trade-offs among spatial, temporal and spectral resolutions of thermal instruments that must be considered when assessing a given science question. For example, UAV-based sensors can address questions at the leaf and canopy scale with single thermal pixels covering 40 cm or less (Figure 2), allowing differentiation between individual plants or even individual leaves. Conversely, spaceborne sensors provide global thermal coverage of the earth's surface, but the spatial resolution of pixels has been typically greater than 60 m, limiting application to canopy scales and larger (Kuenzer et al., 2013). Spaceborne satellites in geostationary orbits can obtain thermal imagery every 15 min, but the spatial resolution is typically too coarse for canopy scale measurements (>2 km); higher spatial resolution sensors typically sacrifice high temporal resolution (Fisher et al., 2017). Clouds are problematic for both air- and space-borne sensors as they can scatter, absorb and re-emit long and short-wave radiation, confounding temperature estimates obtained from thermal retrievals. With low-altitude UAVs, clouds are less of an issue but there are logistical challenges associated with dedicated long-term sampling to observe slowly changing ecological phenomena at canopy scales (Figure 2). Despite the trade-offs, with careful consideration of confounding factors, scaling issues and the selection of appropriate data sources, thermal remote sensing data can help address important questions in plant biology across spatial scales from leaf to globe (Figure 1).

One of the most ecologically useful products derived from thermal remote sensing data is surface temperature (T_surf) estimates. Surface temperatures both impact and respond to a variety of chemical, biological and physical properties, providing valuable insight into plant patterns and underlying plant controls in ecosystems (Still, Powell, et al., 2019). For example, when plants experience stressful heat and moisture conditions, they respond by closing stomata, which causes reduced transpiration and associated increases in leaf T_surf. Remote sensing of T_surf can help identify plants experiencing heat and water stress (Buitrago et al., 2016; Coates et al., 2015). Across larger scales, thermal data can improve plant land cover classifications and can be leveraged to evaluate spatial distributions of certain species, communities or management regimes (Barnes et al., 2021). Overall, remote sensing of T_surf provides valuable information into the water-energy exchange between plants, soil and the atmosphere.

The thermal radiation emitted from an object depends on both its temperature and emissivity; as such, emissivity is required to retrieve T_surf from TIR measurements. Emissivity is the effectiveness of a material to emit and absorb thermal radiation relative to a blackbody, and is generally close to 1 (0.96–0.99) for healthy vegetation, but can be as low as 0.88 to 0.94 for dry leaves (Kuenzer & Dech, 2013). Accurate T_surf measurements from thermal remote sensing require calibrations that account for the impact of emissivity on radiometric temperatures (Fuchs & Tanner, 1966).

Surface temperatures can deviate substantially from air temperatures (T_air), yet T_air are commonly used to describe plant temperature in plant ecological studies (Körner & Hiltbrunner, 2018). Near-surface T_air (2 m above the land surface) are used to assess vegetation–climate interactions at regional and global scales in models, projections and gridded climatologies. Gridded T_air estimates typically rely on data from weather stations, with uneven spatial coverage (Menne et al., 2012). In contrast, spaceborne TIR measurements, used to compute T_surf (also called land surface temperature; LST), are available at high spatial resolutions and global coverage (Duan et al., 2017; Lawrimore et al., 2011). Plant biological and biophysical properties can influence decoupling between T_air and T_surf. In densely vegetated forests, deviations between T_air and T_surf (ΔT) are generally small compared to ΔT differences observed in grasslands, savannas, heath and tundra systems with smaller stature vegetation (Mildrexler et al., 2011). As such, T_surf estimates more accurately reflect the extreme and dynamic thermal environment experienced by plants and other organisms living on the land surface. Furthermore, T_surf provides insight into facets of plant physiology including growth, stomatal conductance, stress and transpiration. Despite its more direct relevance to plant biological processes, regional and global-scale studies of terrestrial vegetation seldom examine T_surf explicitly, presenting an exciting opportunity for the use of thermal remote sensing for studies of plant physiology, ecology and biogeography at leaf, canopy and broader scales.

Here, we consider three ecological scales (Figure 1) where thermal remote sensing has strong potential to advance understanding of plant physiology and ecology. Each level of organization—leaf, canopy and beyond canopy (e.g. regional, global) scale—has its own theoretical underpinnings, methodological and scaling considerations, and scientifically motivating future directions. We also highlight specific opportunities where thermal remote sensing could address open questions in plant ecology. We conclude by delving into principles of thermal remote sensing and outline current and future technologies that could be most useful to ecologists. We hope this review will serve as both a bridge between thermal remote sensing and plant ecology and as motivation for ecologists to consider incorporating thermal remote sensing into studies of plant dynamics.

中文翻译：

---

从叶到地球的植物生态热遥感

1 简介

温度是气候的基本组成部分，也是植被分布和功能的关键驱动因素。化学和生物反应受温度调节，影响生态系统中各个层次的组织。在细胞水平上，极端高温会对光合装置造成不可逆转的损害（Berry & Bjorkman,  1980 ; Teskey et al.,  2015）。温度以及其他生物和非生物因素，如养分和水的可用性，在生态系统尺度上调节生物地球化学循环，并影响区域和全球尺度上动植物物种的生物地理分布（图 1；Jeffree 和 Jeffree，  1994 年））。此外，人为气候变化预计会改变温度状况，对植物产生从叶片到全球范围的影响。尽管跨尺度的温度很重要，但评估跨空间和时间的趋势是困难的。这不是一个独特的问题——规模化挑战在生态学的所有领域都是持久的和根本的（Wiens，  1989）。我们如何理解光合动力学对全球植被模式的影响，或评估周期性极端温度对未来物种范围的影响？解决尺度问题的一种方法是在多个空间和时间尺度上同时探索现象。热遥感是一种关键工具，可用于在多个组织层次上研究温度与植被之间的关系。此外，植物生态学家比以往任何时候都更容易获得热遥感——新的传感器、增加的数据可用性和新兴技术可以促进新的跨尺度方法，以促进我们对温度对陆地生物圈植被影响的理解。

热遥感数据可以在各种空间、时间和光谱分辨率下从一系列平台获得，包括卫星、飞机、无人驾驶飞行器 (UAV)、安装在塔上和地面系统，包括手持摄像机（图 2）。大多数用于植被分析的热红外 (TIR) 传感器测量中程 IR (3–8 μm) 以及更常见的远程 IR (8–14 μm) 区域的发射辐射（Kuenzer & Dech，  2013 年；朱等，  2018）。在遥感中，热仪器的空间、时间和光谱分辨率之间存在固有的权衡，在评估给定的科学问题时必须考虑这些因素。例如，基于无人机的传感器可以解决叶子和冠层尺度的问题，单个热像素覆盖 40 厘米或更小（图 2），从而可以区分单个植物甚至单个叶子。相反，星载传感器提供地球表面的全球热覆盖，但像素的空间分辨率通常大于 60 m，限制了对冠层尺度和更大尺度的应用（Kuenzer et al.,  2013）。地球静止轨道上的星载卫星可以每 15 分钟获取一次热图像，但空间分辨率通常太粗糙，无法进行冠层尺度测量（>2 公里）；更高的空间分辨率传感器通常会牺牲高时间分辨率（Fisher et al.,  2017）。云对于机载和星载传感器都是有问题的，因为它们可以散射、吸收和重新发射长波和短波辐射，从而混淆从热反演中获得的温度估计。使用低空无人机，云不是一个问题，但与专门的长期采样相关的后勤挑战，以观察树冠尺度上缓慢变化的生态现象（图 2）。尽管权衡取舍，但在仔细考虑混杂因素、缩放问题和选择适当的数据源后，热遥感数据可以帮助解决从叶子到地球的空间尺度上植物生物学中的重要问题（图 1）。

从热遥感数据中获得的对生态最有用的产品之一是地表温度 ( T _surf ) 估计值。地表温度会影响和响应各种化学、生物和物理特性，从而为了解生态系统中的植物模式和潜在植物控制提供有价值的见解（Still, Powell 等人，  2019 年）。例如，当植物经历高温和潮湿的压力条件时，它们会通过关闭气孔做出反应，这会导致蒸腾减少和叶片T _surf的相关增加。T _surf的遥感可以帮助识别遭受热和水分胁迫的植物（Buitrago 等人，  2016 年；Coates 等人， 2015 年）。在更大范围内，热数据可以改进植物土地覆盖分类，并可用于评估某些物种、群落或管理制度的空间分布（Barnes 等人，  2021 年）。总体而言， T _surf的遥感为植物、土壤和大气之间的水能交换提供了有价值的信息。

物体发出的热辐射取决于它的温度和发射率。因此，需要发射率才能从 TIR 测量中检索T _{surf 。}发射率是一种材料相对于黑体发射和吸收热辐射的有效性，对于健康的植被，通常接近 1 (0.96–0.99)，但对于干树叶，可以低至 0.88 到 0.94 (Kuenzer & Dech,  2013 年）。来自热遥感的 准确T _{surf测量需要进行校准，以考虑发射率对辐射温度的影响（Fuchs & Tanner，} 1966 年）。

地表温度可能与气温 ( T _air ) 有很大差异，但T _air通常用于描述植物生态学研究中的植物温度 (Körner & Hiltbrunner,  2018 )。在模型、预测和网格化气候学中，近地表T_空气（高于地表 2 m）用于评估区域和全球尺度的植被-气候相互作用。网格T_空气估计通常依赖于来自气象站的数据，空间覆盖不均匀（Menne 等人，  2012年）。相比之下，用于计算T _{surf的星载 TIR 测量}（也称为地表温度；LST）在高空间分辨率和全球覆盖范围内可用（Duan 等人，  2017 年；Lawrimore 等人，  2011 年）。植物生物学和生物物理特性会影响T _air和T _surf之间的解耦。在植被茂密的森林中，与在植被较小的草原、稀树草原、荒地和苔原系统中观察到的ΔT差异相比， T _air和T _surf ( ΔT ) 之间的偏差通常较小（Mildrexler 等人，  2011 年）。因此，T _surf估计更准确地反映了生活在陆地表面的植物和其他生物所经历的极端和动态的热环境。此外，T _surf可以深入了解植物生理学的各个方面，包括生长、气孔导度、压力和蒸腾作用。尽管它与植物生物过程更直接相关，但对陆地植被的区域和全球尺度研究很少明确地检查T _surf，这为利用热遥感研究植物生理学、生态学和叶、冠层和生物地理学提供了一个令人兴奋的机会。更广泛的尺度。

在这里，我们考虑了三个生态尺度（图 1），其中热遥感具有促进对植物生理学和生态学理解的强大潜力。每个级别的组织——叶子、树冠和树冠之外（例如区域、全球）规模——都有自己的理论基础、方法和规模考虑，以及科学地激励未来的方向。我们还强调了热遥感可以解决植物生态学中未解决问题的具体机会。最后，我们深入研究了热遥感原理，并概述了可能对生态学家最有用的当前和未来技术。我们希望这篇综述既能成为热遥感与植物生态学之间的桥梁，也能成为生态学家考虑将热遥感纳入植物动力学研究的动力。