Analysis of Airborne Optical and Thermal Imagery for Detection of Water Stress Symptoms

Analysis of Airborne Optical and Thermal Imagery for Detection of Water Stress Symptoms
复制标题

DOI:
10.3390/rs10071139
复制
发表时间:
2018-07-01
期刊:
影响因子:
5
通讯作者:
Inoue, Yoshio
Inoue, Yoshio
中科院分区:
工程技术2区
文献类型:
--
作者:
Gerhards, Max;Schlerf, Martin;Inoue, Yoshio

文献摘要

被引文献

相似文献

高分辨率的机载热红外(TIR)与太阳诱导荧光(SIF)和高光谱光学图像(可见光,近,短波红外; VNIR/SWIR)联合收购超过一个实验场地。本研究的目的是评估这些国家的最先进的遥感技术的潜力,用于检测类似的症状发生在水胁迫(以下简称水胁迫)在空气中的水平。2014年6月11日和12日,在意大利的Latisana,在一个商业草地(高羊茅和早熟禾)农场上使用两个摄像机系统(Telops Hyper-Cam LW,Specim HyPlant)进行了飞行,该农场的地块用抗蒸腾剂(Vapor Gard(R); VG)和高反射粉末(高岭土; KA)进行了处理。这两种试剂通过减少蒸腾作用并因此减少潜热耗散(VG)和通过增加蒸腾作用来影响植被的能量平衡,降低能量吸收(KA)。同时从现场气象站,表面温度和腔室通量测量现场气象数据。对图像数据进行处理,得到TIR指数(地表温度(Ts)、作物水分胁迫指数(CWSI))、SIF指数(F-687、F-780)和VNIR/SWIR指数(光化学反射指数(PRI)、归一化差异植被指数(NDVI)、水分胁迫指数(MSI)等)的正射校正图。一个线性混合效应模型,尊重嵌套结构的实验装置,分析处理效果的遥感参数。与现场T-s测量值相比,空中T-s具有良好的一致性(Δ T < 0.35 K)。基于TIR的指数的地图和箱形图显示了日变化:Ts在清晨最低,由于净辐射和空气温度(T-air)的增加而增加了6 K,直到上午晚些时候,由于植物蒸腾增加的补偿冷却效应,中午保持稳定;这也得到了室测量的证实。在清晨,VG处理的地块显示出显着更高的T-s相比,对照(CR)地块(p = 0.01),而SIF指数没有显着差异(p = 1.00),在任何立交桥。关于其水胁迫检测能力的谱域的比较评估是有限的,这是由于:(i)两个机载传感器的同步超越是不可行的,以及(ii)而不是一个真实的水胁迫发生只有水胁迫症状模拟的化学试剂。然而,研究结果表明,聚合物二-1-对-薄荷烯对植物具有抗蒸腾作用,而光反应的光合效率不受影响。VNIR/SWIR指数以及SIF指数对KA高度敏感,因为光谱反射率的整体增加,从而减少了吸收能量。与此相反,TIR域是高度敏感的微妙的变化,在温度制度引起的VG和KA,而VNIR/SWIR和SIF域的VG处理的影响较小。多传感器方法的好处不仅在于提供关于实际植物状况的有用信息,而且还在于提供关于生物物理、生理和光化学变化原因的有用信息。
High-resolution airborne thermal infrared (TIR) together with sun-induced fluorescence (SIF) and hyperspectral optical images (visible, near- and shortwave infrared; VNIR/SWIR) were jointly acquired over an experimental site. The objective of this study was to evaluate the potential of these state-of-the-art remote sensing techniques for detecting symptoms similar to those occurring during water stress (hereinafter referred to as water stress symptoms') at airborne level. Flights with two camera systems (Telops Hyper-Cam LW, Specim HyPlant) took place during 11th and 12th June 2014 in Latisana, Italy over a commercial grass (Festuca arundinacea and Poa pratense) farm with plots that were treated with an anti-transpirant agent (Vapor Gard (R); VG) and a highly reflective powder (kaolin; KA). Both agents affect energy balance of the vegetation by reducing transpiration and thus reducing latent heat dissipation (VG) and by increasing albedo, i.e., decreasing energy absorption (KA). Concurrent in situ meteorological data from an on-site weather station, surface temperature and chamber flux measurements were obtained. Image data were processed to orthorectified maps of TIR indices (surface temperature (T-s), Crop Water Stress Index (CWSI)), SIF indices (F-687, F-780) and VNIR/SWIR indices (photochemical reflectance index (PRI), normalised difference vegetation index (NDVI), moisture stress index (MSI), etc.). A linear mixed effects model that respects the nested structure of the experimental setup was employed to analyse treatment effects on the remote sensing parameters. Airborne T-s were in good agreement (Delta T < 0.35 K) compared to in situ T-s measurements. Maps and boxplots of TIR-based indices show diurnal changes: T-s was lowest in the early morning, increased by 6 K up to late morning as a consequence of increasing net radiation and air temperature (T-air) and remained stable towards noon due to the compensatory cooling effect of increased plant transpiration; this was also confirmed by the chamber measurements. In the early morning, VG treated plots revealed significantly higher T-s compared to control (CR) plots (p = 0.01), while SIF indices showed no significant difference (p = 1.00) at any of the overpasses. A comparative assessment of the spectral domains regarding their capabilities for water stress detection was limited due to: (i) synchronously overpasses of the two airborne sensors were not feasible, and (ii) instead of a real water stress occurrence only water stress symptoms were simulated by the chemical agents. Nevertheless, the results of the study show that the polymer di-1-p-menthene had an anti-transpiring effect on the plant while photosynthetic efficiency of light reactions remained unaffected. VNIR/SWIR indices as well as SIF indices were highly sensitive to KA, because of an overall increase in spectral reflectance and thus a reduced absorbed energy. On the contrary, the TIR domain was highly sensitive to subtle changes in the temperature regime as induced by VG and KA, whereas VNIR/SWIR and SIF domain were less affected by VG treatment. The benefit of a multi-sensor approach is not only to provide useful information about actual plant status but also on the causes of biophysical, physiological and photochemical changes.