Current Forestry Reports (2019) 5:169–183 https://doi.org/10.1007/s40725-019-00096-1 REMOTE SENSING (J SUAREZ, SECTION EDITOR) Early Diagnosis of Vegetation Health From High-Resolution Hyperspectral and Thermal Imagery: Lessons Learned From Empirical Relationships and Radiative Transfer Modelling R. Hernández-Clemente 1 & A. Hornero 1 & M. Mottus 2 & J. Penuelas 3,4 & V. González-Dugo 5 & J. C. Jiménez 6 & L. Suárez 7 & L. Alonso 8 & P. J. Zarco-Tejada 5,7 Published online: 20 August 2019 # The Author(s) 2019 Abstract Purpose of Review We provide a comprehensive review of the empirical and modelling approaches used to quantify the radiation– vegetation interactions related to vegetation temperature, leaf optical properties linked to pigment absorption and chlorophyll fluores- cence emission, and of their capability to monitor vegetation health. Part 1 provides an overview of the main physiological indicators (PIs) applied in remote sensing to detect alterations in plant functioning linked to vegetation diseases and decline processes. Part 2 reviews the recent advances in the development of quantitative methods to assess PI through hyperspectral and thermal images. Recent Findings In recent years, the availability of high-resolution hyperspectral and thermal images has increased due to the extraordinary progress made in sensor technology, including the miniaturization of advanced cameras designed for unmanned aerial vehicle (UAV) systems and lightweight aircrafts. This technological revolution has contributed to the wider use of hyperspectral imaging sensors by the scientific community and industry; it has led to better modelling and understanding of the sensitivity of different ranges of the electromagnetic spectrum to detect biophysical alterations used as early warning indicators of vegetation health. Summary The review deals with the capability of PIs such as vegetation temperature, chlorophyll fluorescence, photosynthetic energy downregulation and photosynthetic pigments detected through remote sensing to monitor the early responses of plants to different stressors. Various methods for the detection of PI alterations have recently been proposed and validated to monitor vegetation health. The greatest challenges for the remote sensing community today are (i) the availability of high spatial, spectral and temporal resolution image data; (ii) the empirical validation of radiation–vegetation interactions; (iii) the upscaling of physiological alterations from the leaf to the canopy, mainly in complex heterogeneous vegetation landscapes; and (iv) the temporal dynamics of the PIs and the interaction between physiological changes. Keywords Vegetation health . Hyperspectral and thermal data . Physiological indicators . Radiative transfer models . Vegetation indices This article is part of the Topical Collection on Remote Sensing * R. Hernández-Clemente 5 Instituto de Agricultura Sostenible (IAS), Consejo Superior de r.hernandez-clemente@swansea.ac.uk Investigaciones Científicas (CSIC), s/n 14004 Alameda del Obispo, Cordoba, Spain 1 Department of Geography, Swansea University (SU), Swansea SA2 6 8PP, UK University of Valencia (GCU/IPL), Catedrático José Beltrán 2, 46980 Paterna, Valencia, Spain 2 Technical Research Centre of Finland (VTT), P.O. Box 1000, FI-02044 Espoo, Finland 7 School of Agriculture and Food, Faculty of Veterinary and 3 Centro de Investigación Ecológica y Aplicaciones Forestales Agricultural Sciences (FVAS) & Department of Infrastructure (CREAF), 08193 Cerdanyola del Valles, Catalonia, Spain Engineering, Melbourne School of Engineering (MSE), University 4 of Melbourne, Parkville, Victoria 3010, Australia Consejo Superior de Investigaciones Científicas (CSIC), Global Ecology Unit CREAF-CSIC-UAB, 08193 Bellaterra, Catalonia, 8 Spain University of Valencia, Valencia, Spain 170 Curr Forestry Rep (2019) 5:169–183 Introduction Part 1: Physiological Indicators of Vegetation Health This review focuses on the recent progress in remote sens- ing detection of early responses of plants to biotic and New strategies using the radiation pathways of the photon– abiotic stresses. One of the main challenges for the re- vegetation interaction have recently been proposed as a way to search community is to develop viable detection methods detect plant physiological status through remote sensing. to anticipate long-term irreversible damage to stressed veg- Light energy absorbed by vegetation can be used for photo- etation before symptoms are visible so that action can be synthesis, re-emitted as sun-induced fluorescence (SIF) or dis- taken [1]. At an early stage of stress, alterations in the sipated as heat. Any alteration in the efficiency of one of the photosynthetic rate and stomatal conductance affect sun components will affect the other two. Light re-emitted as SIF radiation–vegetation interactions, including the energy dis- is produced by the pigment–protein complex photosystems I sipated as heat or emitted at longer wavelengths as chloro- (PSI) and II (PSII) involved in photosynthesis. Although SIF phyll fluorescence, and the gradual effects on the photo- is a small emission (2–3% of the incoming radiation), it is a synthetic pigment concentration in plant leaves. Recent strong proxy for plant photosynthesis. Plant responses to en- studies have demonstrated the capability of spectral data vironmental stress (e.g. light, nutrients, water availability, ex- to detect physiological alterations and anticipate vegetation treme temperatures) affect the photosynthetic rate and there- diseases such as Phytophthora [2], Xylella fastidiosa [3••], fore fluorescence emission. For this reason, SIF has been used Verticillium wilt [4] and almond red leaf blotch [5], as well for decades as a powerful non-invasive marker to track the as environmental stressors such as water stress [6]. These status, resilience and recovery of vegetation [7]. achievements represent a major step forward in the moni- Additionally, vegetation temperature is related to stomatal toring of vegetation health from airborne remote sensing conductance and transpiration-associated evaporative cooling. imagery and its potential application to satellite scales. Specifically, when stomatal conductance decreases due to sto- Therefore, this review focuses on recent developments in ma closure under water stress conditions, vegetation temper- the early detection of biotic- and abiotic-induced symp- ature increases and the transpiration rate decreases according- toms using hyperspectral and thermal images. It describes ly. Consequently, vegetation temperature has been used to four groups of physiological indicators (PIs) of critical im- derive information about transpiration and as an indicator of portance in plant functioning for improving vegetation vegetation health. The first study using this approach was monitoring: (i) vegetation temperature, (ii) chlorophyll conducted in 1964 on Quercus macrocarpa and demonstrated fluorescence, (iii) photoprotective pigments or the xantho- the suitability of temperature to monitor transpiration [8]. phyll cycle and (iv) photosynthetic pigments (Fig. 1). Since this pioneering research, vegetation temperature has Fig. 1 Physiological indicators used to quantify early-stage vegetation (WDI); (ii) the Fraunhofer line depth principle (FLD); (iii) the photo- health from hyperspectral and thermal data. Retrieval methods for four chemical reflectance index (PRI); and (iv) the chlorophyll (Cab), carot- physiological-based groups used as indicators in remote sensing include enoid (Cxc) and anthocyanin (Anth) vegetation indices (VI), and radiative (i) the difference between crown temperature and air temperature (Tc – transfer models (RTM) Ta), the crop water stress index (CWSI) and the water deficit index Curr Forestry Rep (2019) 5:169–183 171 been used to determine water status [9, 10], irrigation require- soluble flavonoids involved in plant protection against ments in agriculture [11] and pest and disease incidence [3••, drought and stress [21] and photooxidative processes [2, 12••, 13] and forest decline [14]. Compared to point ground- 22] with an exponential decay with increases in Chlab based measurements, the ability to remotely detect temporal content. Their assessment has typically been conducted and spatial variations in vegetation temperature as an indicator in the context of senescence and fruit-ripening processes of water stress is considered a major advantage. Moreover, the [23–25]. relation between transpiration and vegetation temperature is underpinned by the energy balance equation [15], which im- plies the balance between incoming energy and energy loss Part 2: Retrieval Methods and Applications and provides a theoretical basis for modelling methods and to Vegetation Health applications. Various mechanisms overcome the excessive reduction of Given the importance that vegetation functioning plays in the the light energy transferred when photosynthesis is reduced monitoring of vegetation health, great efforts have been direct- by stress or limited resources. These mechanisms downregulate ed towards non-destructive retrieval methods at different spa- the rate of electron transport in plant cells and dissipate excess tial scales using various sensors and analytical methods. energy controlling the facultative pool of xanthophyll pigments Figure 2 shows an example of the main PIs used for monitor- in plants known as facultative changes. The dissipation of ex- ing vegetation health that can be quantified with hyperspectral cess energy by facultative non-photochemical quenching and thermal images using a broad range of approaches de- (NPQ) through the xanthophyll cycle is one of these mecha- scribed below. nisms. NPQ is particularly interesting because it can provide an optical signal of excess of reducing power, increased NADPH/ Vegetation Temperature NADP ratios and reduced light-use efficiency (LUE). Therefore, NPQ is linked to stress, caused by limited resources, Vegetation temperature (VT) is a key indicator for vegetation and to the plant photosynthetic activity through associated health monitoring and several related applications have been changes in reflectance in the blue and green regions of the identified [27]. However, accurate retrieval of VT from remote spectrum [16]. When excess excitation energy accumulates in sensing data requires an accurate atmospheric correction. The the leaf, the chloroplast thylakoid pH gradient builds up and the main problem in obtaining accurate VT retrieval is the coupling xanthophyll cycle is activated: violaxanthin is de-epoxidized to between land surface temperature (LST) and land surface emis- zeaxanthin via antheraxanthin. These chemical changes affect sivity (LSE). Retrieval of vegetation emissivity is also of inter- the spectrally selective absorption of light, the relative absorp- est because the emissivity spectrum of vegetation provides in- tion by each pigment and hence the leaf spectral reflectance and formation about leaf composition and health status [28]. transmittance factors. These changes in reflectance result from Several LST (and LSE) retrieval methods have been pub- changes in the de-epoxidation state of the xanthophylls, with lished in recent decades (see a review by Li et al. [29]). These increases in the concentration of zeaxanthin and consequent methods can be classified according to the spectral configura- decreases in reflectance around 531 nm. tion of the thermal infrared (TIR) sensor. Hence, single- Increasing levels of vegetation stress over time can channel (SC) algorithms have been developed for LST retriev- ultimately cause variations among leaf photosynthetic pig- al from a single TIR band, such as the previous Landsat series ment pools known as constitutive changes. The [30]. However, these methods are typically very sensitive to discolouration produced by the variation in pigment con- uncertainties in the atmospheric correction (e.g. in humid at- tent has also been used as an indicator of vegetation mospheres). This problem of atmospheric correction in SC health [17•]. The three main pigment groups are com- algorithms can partly be solved when two TIR bands are posed of chlorophylls, carotenoids and anthocyanins. available. This is the case of the two-channel (TC) or split- Chlorophyll occurs in the form of chlorophyll a (Chla) window (SW) algorithms. These algorithms perform the at- as the major pigment and chlorophyll b (Chlb) as the mospheric correction based on the differential absorption be- accessory pigment. Both are genuine components of the tween two TIR bands located in atmospheric windows and photosynthetic membrane and occur in a ratio a/b value usually provide robust retrievals in all types of conditions. between 1 and 4 reflecting growth and stress [18, 19]. An example is LST products generated from TIR data ac- Drought stress led to a decrease in Chla and Chlb and quired by polar-orbiting or geostationary meteorological sat- increase in the Chl a/b ratio due to greater reduction in ellites (e.g. MetOp, Meteosat, GOES) and the recent Sentinel- Chlb compared to Chla. Carotenoids provide structure to 3. Both SC and TC algorithms require a priori knowledge of the photosynthetic membrane, and play a role in light surface emissivity, which can be retrieved from vegetation harvesting processes, energy transfer and NPQ through indices and/or data measured in other parts of the electromag- the xanthophyll cycle [20]. Anthocyanins are water- netic spectrum [31]. 172 Curr Forestry Rep (2019) 5:169–183 Fig. 2 Pre-visual physiological indicators based on hyperspectral and on vegetation [26]; c relationship between the crop water stress index thermal data: a leaf reflectance alterations in the visible and infrared (CWSI) measured with infrared thermal sensors (IRT) sensors and the ranges produced by the variation in chlorophyll content (10–60 μg/ ratio between daily values of actual to potential transpiration (E/Ep) in cm−2) [6]; b reflectance difference after light excitation showing the mandarin trees in deficit-irrigated (DI) and well-irrigated (control trees) non-photochemical quenching and chlorophyll fluorescence emission conditions; d high-resolution hyperspectral and thermal images The innovation of multispectral TIR sensors (with three or In the 1980s, Idso and Jackson’s group demonstrated the more TIR bands in atmospheric windows) has encouraged the need to normalize vegetation temperature according to envi- development of temperature and emissivity separation (TES) ronmental conditions. Air temperature, relative humidity and methods for simultaneous retrieval of LST and band emissiv- solar radiation strongly influence the absolute value of vege- ities. This is the case of the TES algorithm developed for tation temperature and should, therefore, be considered in de- ASTER [32] and also adapted to other satellite sensors such riving any information about vegetation health. The final for- as MSG/SEVIRI [33] and Terra&Aqua/MODIS [34]. In prin- mulation developed by Idso et al. [38] to account for the in- ciple, TES methods are preferred for emissivity retrieval be- fluence of environmental conditions on vegetation tempera- cause they are extracted from TIR data, thus reflecting the ture was termed crop water stress index (CWSI). This index is actual conditions of the surface [35]. TES methods have also derived from the energy balance and relies on normalization been applied to hyperspectral TIR data, usually acquired from according to air temperature and two bounding limits (Eq. 1): airborne sensors or ground-based instruments based on Fourier transform infrared (FTIR) spectroscopy. Some of ðTc−TaÞ−ðTc−TaÞLL these techniques are based on spectral smoothness [36]. CWSI ¼ ð1Þ ðTc−TaÞUL −ðTc−TaÞLL Emissivity spectrum retrieval from hyperspectral TIR imaging or FTIR laboratory spectroscopy has led to the use of emis- where (Tc − Ta)LL is the lower limit of the differential be- sivity spectral features for identification of plant species [28] tween vegetation and air temperature and corresponds to and changes in canopy leaf area index [37]. the Tc−Ta value of the vegetation transpiring at the Curr Forestry Rep (2019) 5:169–183 173 potential rate. (Tc − Ta)UL is the vegetation differential limitation on use of this approach. Compared with the empir- when transpiration is completely halted. ical approach, it can be applied under a wider range of weather The CWSI can be determined empirically by assessing the conditions, as the net radiation is incorporated into the formu- value of the lower and upper limits or by modelling the energy lation. It is especially relevant for humid environments. balance at the canopy level [39]. In the empirical approach, the In open canopies, the surface temperature is a composite of lower limit is computed using the non-water stress baseline soil and vegetation, which makes the CWSI not reliable, ham- (NWSB), which accounts for the relationship between the pering its application on local and regional scales. This prob- vapour pressure deficit (VPD) and the Tc−Ta value of the lem was explored by Moran et al. [45], who introduced the vegetation transpiring at the potential rate. It is species- water deficit index (WDI). The WDI is based on the vegeta- specific and is affected by climatic conditions; thus, the ex- tion index/temperature trapezoid. It combines spectral vegeta- trapolation is restricted by the environment. For some species, tion indices with surface temperature measurements to allow it also varies according to plant phenology [40]. It can be application of the CWSI theory to partially vegetated fields determined by the continuous monitoring of temperature in without data on foliage temperature. Nevertheless, this ap- plants that are not limited by water availability. The upper proach does not always work well when the goal is to detect limit can be calculated by solving the NWSB equation for subtle temperature changes associated with vegetation stress. VPD = 0, and subsequently correcting for the difference in vapour pressure induced by the difference in temperature Chlorophyll Fluorescence Emission Tc – Ta [38]. The required inputs are limited to vegetation and air temperature and the VPD. However, this method has Research carried out over the last 20 years has provided the- several limitations. First, the NWSB should be determined oretical and experimental evidences that SIF can be retrieved before the practical application of the CWSI is implemented. from imaging spectroscopy and used as an indicator of plant When the NWSB is not available for a specific species/ photosynthetic activity. The SIF emission spectrum is maxi- environment combination, the Tc value of plants transpiring mum at 682 nm for the PSII and 700 to 750 nm for PSI. There at the potential rate must be known, ideally by determining the are many different parameters retrieved from SIF. For in- NWSB during the season. Moreover, as the CWSI is not nor- stance, the integral of the spectral fluorescence radiance in malized by the solar radiation, the measurement must be ac- the observation direction (F) over the full retrieval range quired under cloudless conditions and at midday to minimize (e.g. 670–780 nm, 650–850 nm) (FINT), the maximum value this effect and obtain more robust estimates [41]. This can be a of F in the red region (maxFR), the maximum value of F in the significant drawback in humid or temperate regions [39]. far-red region (maxFFR), F value at 687 nm (F687), F value at In its simplest form, the CWSI can also be determined 740 nm (F740) or F value at 760 nm (F760). However, the directly by measuring the temperature of a wet (Twet) or dry retrieval of SIF from images is extremely challenging due to surface [42]. These surfaces are included in the image and can the small contribution of the emission to total radiance. be leaves either sprayed with water (Twet) or covered with Several methods have been developed to quantify SIF at the petroleum jelly (Tdry), or an artificial surface with a thermal leaf and canopy levels. The most common approaches to re- behaviour similar to that of the canopy. Although this method trieve SIF from reflected radiance under field conditions are (i) is very appealing because it does not require meteorological the Fraunhofer line depth principle or discriminator (FLD) measurements, its accuracy is strongly affected by the refer- [46] using radiance spectra from the O2 absorption bands; ences used, as Twet is often lower than potential T [39]. (ii) radiative transfer modelling (RTM); (iii) spectral fitting Moreover, there is a scale issue, as the surface references usu- methods (SFM) and singular vector decomposition (SVD). ally have the size of a leaf and are thus used at the plant level, Initial attempts to track the effects of fluorescence emission making the upscaling to airborne imagery difficult [43]. on reflectance signatures were based on computing a combi- The CWSI can also be determined theoretically by deriving nation of spectral reflectance bands: a band near the two max- it from the energy balance (Eq. (2)), as proposed by Jackson imum fluorescence emissions (about 685 and 740 nm) and a et al. [44]: reference band that is less affected. Some of the formulations have been validated in vegetation at the canopy level [26]: γ  ð1 þ rc =ra Þ−γ * R680/R630, R685/R630, R686/R630, R690/R630, D730/D706, DλP/ CWSI ¼ 1−E=Ep ¼ ð2Þ D703, R6832/(R675 × R691) and R685/R655, D705/D722 and D730/ Δ þ γ  ð1 þ rc =ra Þ D706, R690/R600 and R740/R800. where E and Ep are actual and potential transpiration, respec- The FLD method initially proposed for the H absorption tively, Δ is the slope of the saturated vapour pressure– calculates SIF using the sun’s irradiance and canopy radiance temperature relationship, γ is the psychrometric constant, ra inside and outside the O2 absorption bands (such as the O2-A is the aerodynamic resistance and rc is the canopy resistance to band at 760.5 nm). The main limitation of this formulation is vapour transport. The complex data requirement is the main to assume that reflectance and fluorescence inside and outside 174 Curr Forestry Rep (2019) 5:169–183 the O2-A band are the same. Several methods have been used fluorescence emission in the radiative transfer of 3-D scenes. to improve the formulation and linearly correlate reflectance DART is able to simulate more complex 3-D scenes with with fluorescence. Some examples using airborne image data different types of elements and distributions. Yet, models are 3FLD [47], corrected FLD (cFLD) [48] and improved using a large set of input parameters could also increase the FLD (iFLD) [49] reviewed by [50]. By contrast, SFM and errors, so that further of this complex model validation over SVD use all available hyperspectral bands to quantify the different types of ecosystems is still a delicate task. contribution of fluorescence and reflectance [51]. One of the first attempts to model the effects of fluores- cence emission on the radiance and reflectance signatures was Energy Dissipated During Photosynthetic made by the European Space Agency (ESA). ESA developed Downregulation the FluorMOD leaf [52] and the canopy FluorSAIL [53] A common way to quantify stress-related spectral changes is models, including the fluorescence excitation–emission ma- to use the photochemical reflectance index (PRI) calculated trix to simulate the fluorescence effects at the canopy level. from the leaf spectral reflectance factor R(λ), where λ is the FluorMODleaf was based on the widely used and validated wavelength given in nanometers, as indicated in Eq. 3: PROSPECT leaf optical properties model. It required the in- put of the σII/σI ratio, referring to the relative absorption Rð531Þ−Rð570Þ cross-section of PSI and PSII, and the fluorescence quantum PRI ¼ ð3Þ Rð531Þ þ Rð570Þ efficiency of PSI and PSII, τI and τII. This model was recently computationally improved with FLUSPECT [54]. The canopy The two wavelengths are located on different sides of the model was based on the turbid medium SAIL model green absorption minimum (λ = 550 nm), where λ = 531 nm (FluorSAIL), modified to internally compute the illumination is the wavelength that is most affected by reversible changes levels through the canopy. [3••] and λ = 570 nm is a reference wavelength where the Extensive work has been conducted with FluorSAIL in ho- typical absorption of other pigments of a green leaf is the same mogeneous canopies. Yet, progress has been made in the de- as at 531 nm. The diurnal variation of the PRI is mainly driven velopment of modelling methods to retrieve pure vegetation by the mentioned changes in the xanthophyll cycle [58, 59] fluorescence from aggregated pixels with FluorFLIM using which are considered facultative changes in pigments. The the linked FluorMODleaf-FluorSAIL with the geometric forest PRI is also proportional to the chlorophyll and carotenoid ratio light interaction model (FLIM) [55]. Further strategies to quan- relative to the energy status and photosynthetic activities as- tify SIF from heterogeneous natural ecosystems have been lim- sociated with changes in chlorophyll/carotenoid ratios ited by difficulties in simulating canopy structure, crown over- throughout foliar development, ageing or stress [60]. In sea- lapping, mutual shading and multiple scattering between sonal time scales (i.e. over weeks and months), the PRI vari- crowns. The development of the model SCOPE (Soil Canopy ation thus combines changes in the xanthophyll cycle and Observation, Photochemistry and Energy fluxes) model [56] changes in the pools of carotenoids and chlorophylls improved the effect of the environment (e.g. sun position and [60–62], which are considered constitutive changes in pig- biochemical composition of the vegetation). Nevertheless, ments. Recent studies have shown that these constitutive SCOPE is still a 1-D model and therefore ignores the vertical changes in pigment pool size have a dominant influence on and horizontal variations in surface characteristics. This limits the chlorophyll/carotenoid index (CCI, a modified PRI calcu- its range of applications in heterogeneous ecosystems. To fill lated using the chlorophyll absorption band as reference) over this gap, a 3-D canopy FluorFLIGHT RTM was proposed by long periods [63, 64]. Hernández-Clemente et al. [12••] to properly parameterize the The tracking of LUE and photosynthetic performance via heterogeneous canopy structure. This model proposes a novel the PRI has been extensively explored over the last two de- approach by linking the FLUSPECT model to the 3-D ray- cades. The PRI is a good predictor of photosynthetic efficien- tracing model (FLIGHT) to estimate F signals in heterogeneous cy and the connection between this index and dynamic forest canopies. Forest canopies are characterized by high hor- photoprotection is well established at leaf level across species, izontal and vertical heterogeneities. functional types and nutrient treatments. The PRI accounted Consequently, the radiative transfer within a forest canopy for about 60% of the variability of LUE at the leaf, canopy and depends on the spatial distribution of the canopy elements ecosystem levels in unique exponential relationships for all relative to each other and the subsequent complex radiative the vegetation types studied [65, 66]. It tracks the rapid phys- processes such as multiple scattering, mutual shading of the iological changes that are usually difficult to follow in ever- crowns and shading of the background. The model has been green species using indices of greenness and canopy structure, validated for detecting oak trees infected by Phytophthora in such as the normalized difference vegetation index (NDVI) [12••] and olives trees infected by Xylella fastidiosa [3••]. [58, 59, 65–67]. Furthermore, the PRI is a good indicator of Additionally, the DART RTM [57] has recently incorporated the photosynthetic apparatus across functional types and Curr Forestry Rep (2019) 5:169–183 175 spatiotemporal scales [65, 68••] and has detected the reactiva- photoprotective mechanisms apparent as changes in leaf pig- tion of photosynthesis from winter stress in evergreen species ment pools, which cannot be neglected since they affect the [63, 64, 69]. The effects of seasonal drought on the photosyn- reference band at 570 nm. To correct for this and detect thetic apparatus have also been detected by satellite-based PRI photoprotection, the unstressed (i.e. predawn) PRI value [70, 71]. Maximum CO2 assimilation has been efficiently es- needs to be subtracted from leaf measurements, which can timated by the PRI under severe drought conditions [72]. In be further elaborated to deconvolving daily and seasonal addition, Rossini et al. [73] demonstrated that changes in the changes [79]. Unfortunately, this task requires multiple suc- PRI were correlated with water stress in maize. Photosynthetic cessful image acquisitions in a single day, which may not variability induced by heat and drought is simultaneously ac- always be feasible. The consistency of the relationships be- companied by complex physiological and biochemical pro- tween PRI, LUE and ecosystemic CO2 uptake [66] suggests a cesses that could, however, constrain the PRI-based estima- surprising degree of ‘functional convergence’ of biochemical, tion of the photosynthetic apparatus. physiological and structural components that affect plant and However, scaling of the PRI from leaf to canopy level carbon fluxes. As an alternative to the PRI, it has been pro- requires considering numerous factors other than leaf bio- posed to integrate the whole spectral area affected by the chemistry. While generally referred to as canopy structural downregulation (the 507–556-nm region) [80]. The xantho- effects, they also include the influence of measurement direc- phyll cycle status can also be retrieved by inverting the recent tionality, mixing of the spectra of the various canopy compo- modifications of the PROSPECT leaf optical model nents and complex illumination conditions. This refers to ef- (PROSPECT-D [81] and Fluspect-CX [82]. These methods, fects caused by the shift from measuring a well-characterized, which are computationally more complex, have shown prom- uniform target to working under natural illumination with a ising results, but are nevertheless hindered by the canopy target composed of a large number of individual and direc- structural effects. tional scatterers. Indeed, multiple scattering in vegetation can be calculated using basic canopy structural information and Spectral Alterations Associated with Pigment Content the harmonic mean of leaf spectral reflectance at the two Variation wavelengths used in the index. Unfortunately, the other factors are more difficult to quantify, may co-vary with changes in The main spectral regions sensitive to pigment absorption are view or sun angle (e.g. visibility of soil background, a fraction located between 400 and 700 nm where the light spectrum of shaded leaves in sensor field of view and blue sky irradi- useful for plant is largest [18, 83]. This determines the absorp- ance on the visible leaves) and are difficult to separate in tion of natural light by vegetation. The red edge region has empirical data. also been demonstrated to change with chlorophyll concentra- To address this major research gap, various studies have tion and quick photosynthetic rate changes [84, 85]. Hence, proposed different methods for minimizing background and the main difficulty of remote pigment assessment is to disen- illumination effects on the PRI. An option is to use the tangle the signals related to individual pigment and structural illumination-corrected PRI (DeltaPRI) proposed by Takala changes in plant canopies. Retrieval methods need to account and Mõttus [74] using the relationship between the PRI and for this overlap, especially when differentiating between chlo- the shadow fraction. According to this study, illumination- rophyll and other pigment groups [86, 87]. induced effects are nearly half of the observed dynamic range Traditional remote sensing approaches to assess pigment of canopy PRI calculated using high spatial resolution air- concentration in heterogeneous forest areas have mainly fo- borne imaging spectroscopy data (less than 2-m resolution). cused on establishing empirical relationships between pig- Some authors have demonstrated the need to consider modi- ment concentration values measured in the field and vegeta- fied PRI formulations (such as the PRI512 index) that are less tion index results of the combination of spectral bands. The affected by structural components to detect vegetation stress use of vegetation indices is a popular choice to reduce unwant- in forest canopies [14]. Others have focused on improving the ed structural, atmospheric and background effects at the can- PRI to decrease the influence of these physiological and phys- opy level [59, 88]. The development of indices for quantifying ical factors that hamper the detection of photosynthetic activ- pigment concentration is based on the known location of spec- ity [68••]. The differential PRI is one of these corrections tral features related to specific absorptions [89]. While these obtained by subtracting dark-state PRI from light-exposed indices are based on the leaf spectral signal as a result of PRI [75, 76], by subtracting predawn PRI from midday PRI pigment concentration, their performance can decrease when [77], or by subtracting the actual PRI from the theoretical PRI applied to canopy image data where the mixed-pixel signal is calculated by RTM inversion techniques [78]. The DeltaPRI affected by woody elements, canopy structure, crown archi- can eliminate the impacts of canopy structure and foliar pig- tecture, the understory and soil background [90]. Croft et al. ments on PRI interpretations. In addition to dynamic photo- [88] reviewed the performance of indices used at both the leaf synthetic downregulation, the PRI also reflects long-term and canopy scales for both conifer and broadleaf species. 176 Curr Forestry Rep (2019) 5:169–183 Several studies have attempted to estimate pigment con- ground and their spectral signal need to be considered [14, centration through empirical approaches correlating broad- 92, 93, 102]. Models simulate canopy reflectance with vary- band vegetation indices calculated from multispectral satellite ing levels of detail. They range from 1-D models where the imagery with field data [88]. Although these methods are ap- canopy layer is considered a continuous medium with a ho- propriate for assessing large areas, the use of narrowband mogeneous composition to three-dimensional models that al- spectral information, typically provided by hyperspectral sen- low the representation of heterogenous discontinuous cano- sors, is suggested as the best procedure for pigment estimation pies with a very high level of detail [103]. The selection of where spectral responses are provided as narrow signal fea- the model to use is a compromise among the level of detail tures [91]. Another limitation of traditional empirical methods needed to properly describe the study area, input data avail- based on vegetation indices is their poor performance in forest ability and computational requirements needed to conduct the areas with high structural complexity. Croft et al. [88] demon- simulations or the model inversion. Homogeneous and con- strated that the performance of broad and narrowband indices tinuous canopies composed of a single species can be assessed varies significantly for different forest structures. The effect of using 1-D models [93, 104, 105]. Discontinuous canopies and canopy structure and soil background on the performance of areas where specific vegetative components impart an impor- methods relying on vegetation indices has also been demon- tant effect on overall reflectance (e.g. woody material) require strated through RTM [92, 93]. To minimize the level of un- accounting for the contribution of the canopy structure effects certainty in the predictions based on only a few spectral bands, on radiative transfer schemes to obtain accurate assessments methods using the full information acquired by hyperspectral [87, 106, 107]. systems in numerous continuous narrow bands have been de- veloped [94]. They make it possible to conduct an extended analysis over multiple spectral regions and to quantify spectral Conclusion, Gaps and Future Prospects peaks affected by both biochemistry and structure through continuum removal analysis and band depth normalizations Data Acquisition and Quality [95–97]. Some statistical approaches that have been applied to There is consensus in the remote sensing community on hyperspectral imagery do not require previous knowledge of the lack of availability of high spatial, spectral and tempo- the location of spectral features. They range in complexity level ral resolution VIS, NIR and TIR data [108•]. The highest from simple models such as linear discriminant analysis to spectral resolution in satellites with regular acquisition complex machine learning techniques and neural networks. schedule is currently provided by MODIS while the Although statistical models are very accurate locally, they can highest spatial resolution is provided by Sentinel and the only be applied to the areas they were developed for and are not highest temporal range by Landsat. The spatial, spectral readily applicable to other sites and vegetation types [88, 98]. and temporal resolutions of satellite sensors constrain the In search of more synoptic approaches, several authors reliability of monitoring plant biochemical variables such have used RTM with various degrees of success. Model de- as pigment content or LAI at different spatial and temporal velopments since the early 1990s opened the possibility of scales [109]. Several TIR sensors have been launched for integrating structural and biophysical data into the radiative Earth Observation (EO) purposes since the development of transfer scheme, thereby providing the means to understand the first experimental HCMM TIR mission in the late the effects of the different components on the overall reflec- 1970s. However, most of these sensors provide TIR data tance signal. The development of leaf models, simulating the at low spectral and spatial resolutions. Applications of TIR leaf spectral response as a function of its main structural com- data for studies on natural resources were boosted by the ponents including chlorophyll content and later carotenoids, launch of Landsat-4 in the early 1980s, which included the provided a tool to explore more universally applicable first TIR sensor at a high spatial resolution. The launch of methods [99, 100]. Leaf spectrum simulations can be used ASTER in December 1999 was a further step towards TIR as input in canopy models that consider the vegetation struc- applications because of its unique multispectral TIR char- ture and background properties at different degrees of com- acteristics (five TIR bands at high spatial resolution). The plexity [87]. Models can be either inverted to derive input ECOsystem Spaceborne Thermal Radiometer Experiment values from image spectra [101] or used as a tool to develop on Space Station (ECOSTRESS) was recently mounted on robust algorithms based on vegetation indices. Although there the International Space Station (ISS) to monitor plant pro- is extensive literature documenting pigment absorption spec- cesses from space and is providing multispectral TIR im- tra and indices to use with leaf spectral data, methods devel- agery in five spectral bands at a spatial resolution below oped at the leaf level are not ready to be used with canopy 100 m (ecostress.jpl.nasa.gov). Other future TIR missions reflectance [6]. The effect of viewing and illumination geom- such as Copernicus’ Sentinel-8 or NASA’s HyspIRI may etry and the arrangement of the different elements on the also include multispectral capabilities at high spatial Curr Forestry Rep (2019) 5:169–183 177 resolutions. In spite of these EO advances, significant gaps studying the physiological condition of forest canopies in remain which clearly need addressing. higher detail, as current available satellite sensors are limited The aforementioned limitations of EO satellites for TIR by their spatial and spectral resolutions for SIF retrieval applications requiring high spatial/spectral/temporal resolu- purposes. tions have typically been solved by using aerial remote sens- To address the current gap in SIF collected at the global ing technologies. The procedure has been to use broadband, scale, the European Space Agency’s Earth Explorer Mission multispectral and hyperspectral TIR sensors mounted on of the ‘Fluorescence Explorer’ (FLEX) [117], the first mission planes under well-controlled conditions. This makes it possi- designed to observe the photosynthetic activity of the vegeta- ble to organize dedicated field campaigns for the acquisition tion layer on a global level, has recently been approved, with of high-quality thermal imagery. Since the development of the 2022 as the tentative launch date. For the first time, this mis- thermal infrared multispectral scanner (TIMS) in the early sion will enable the assessment of the dynamics of photosyn- 1980s [110], various airborne multi-/hyperspectral TIR sen- thesis in forest canopies through sun-induced fluorescence at sors have operated in dedicated field campaigns. Some exam- 300-m spatial resolutions; in addition, it will be able to distin- ples are the MODIS/ASTER (MASTER) simulator, the guish different fluorescence signals from PSI and PSII [118]. Spatially Enhanced Broadband Array Spectrograph System This offers a great advantage over current techniques used for (SEBASS), the Airborne Hyperspectral Scanner (AHS), the photosynthesis monitoring based on green biomass estima- Thermal Airborne Spectrographic Imager (TASI) and the tions rather than on physiological processes acquired from Hyperspectral Thermal Emission Spectrometer (HyTES). conventional Earth-resource satellites. Detection of photosyn- SEBASS and HyTES have been used to identify plant species thetic processes at the global level will change the scope of [28], AHS has been used to detect water stress in an olive environmental and ecosystem management approaches and orchard [111•] and MASTER has been used to assess LST opens up a new line of research for ecological management variability and crop stress [112]. and conservation. One of the major uncertainties to get there is Given that dedicated field campaigns using high-quality understanding the relationship between SIF, PRI and LST airborne sensors incur high costs, proximal remote sensing towards estimating photosynthetic rates under different stress using unmanned aerial vehicles (UAVs) has rapidly developed conditions. A recent experiment with passive proximal remote in recent years. However, due to their low load capacity, sensing at the leaf level demonstrates that it is feasible to UAVs do not allow configurations using heavy instruments. estimate NPQ based on PRI and APAR, but under stress con- In practice, small-sized and uncooled TIR cameras are cur- ditions, the relationship is strongly alternated [119]. The val- rently used for thermal monitoring with UAVs. These un- idation of these results at the canopy level will imply having cooled microbolometer TIR cameras (FLIR Tau 2 or into account the additional effect of other geometrical and Thermoteknix MicroCAM 3) operate without an active tem- structural components of the canopy and the background. perature stabilization component, which can lead to LST re- trievals whose precision is too low for some agricultural ap- From the Leaf to the Canopy: Upscaling Limitations plications [113, 114]. This problem may be partly solved by applying dedicated corrections to the data acquired with the Many constraints arise from the accurate scaling up and re- uncooled TIR instrument [114]. However, the calibration of trieval of PI from the leaf to the canopy. Such constraints are uncooled TIR cameras to absolute temperature measurement mainly related to the variations in plant functioning, canopy is a time-consuming and complex process [115]. structure, atmosphere, pigments and soil background at differ- The rapid development of new hyperspectral sensors to be ent scales and have different impacts on the PI. carried onboard manned airborne and UAV platforms has giv- Atmospheric eddies induce variability in the thermal range en rise to the retrieval of high spatial resolution SIF at local that can limit the precision of LST retrievals [27, 120]. The scales, which is becoming a novel area of research [12••, 54, contribution of pixels outside the field of view of the sensor 114, 115]. Some of the in situ-based SIF measuring systems (i.e. neighbour pixels) to the TIR radiance of the target pixel currently tested are FloX, JB Hyperspectral, PhotoSpec, (i.e. adjacency effect) also induces uncertainty into the LST Piccolo and HyScreen. Fluorescence data previously amassed retrieval. Yet, the impacts of this effect on TIR measurements from space were collected on missions not purposely designed have yet to be adequately studied. Overall, precise characteri- for SIF but employing hyperspectral sensors such as OCO-2, zation of surface emissivity is the limiting factor for accurate GOME or TANSO/GOSAT [116]. However, it remains very LST retrievals (emissivity uncertainties of 1% lead to LST un- challenging to cover the large areas required for forest moni- certainties between 0.4 and 0.6 K; Gomis-Cebolla et al. [121]). toring analysis at a very high resolution. The measurement of Fortunately, most of these impacts are somehow minimized SIF requires highly accurate radiance and retrieval methods over green vegetation because of its thermal homogeneity and reliant on instrument characterization and measurement setups high emissivity. However, emissivity characterization over dry and protocols. This has hitherto been the main limitation for (i.e. senescent) vegetation is more problematic, with a high 178 Curr Forestry Rep (2019) 5:169–183 impact on LST [122]. The scaling issue is still problematic in measurements of other leaf optical properties, such as reflec- some applications because spectral features observed at the leaf tance, transmittance and, indirectly, absorbance and absorbed level from in situ or laboratory measurements disappear at the photosynthetically active radiation (APAR), as well as fluo- metre scale when using airborne data [123]. rescence yield. These parameters have made it possible to The greatest challenges with the routine use of the PRI in validate retrieval algorithms, helped to fine-tune leaf RTMs vegetation health monitoring at the global scales lie in the that include fluorescence emission and been used as a basis for difficulty of separating the long- and short-term dynamic com- upscaling from leaf to canopy. Similar challenges have been ponents in the PRI and the difficulty in scaling up the index found for scaling up leaf pigment content to the canopy. from canopy to leaf level. Solving these issues may require Pigment leaf data has been successfully retrieved using de- evaluating the full spectral information available in the region structive biochemical analysis or indirect measurements mea- affected by the dynamic photoprotection processes (505– sured with a SPAD chlorophyll meter [125]. However, when 560 nm) instead of focusing on a single index, or at the very scaling pigment content data from leaf to canopy level, the least reformulating the PRI. We also need more research on heterogeneity in pigment composition and canopy structure scaling with a full description of leaf optics as well as com- contribute to a higher variation of pigments in the canopy. prehensive and temporally extensive calibrated leaf-level On balance, the uncertainty of scaling up PI from the leaf to measurements to accompany remote spectral imagery. the canopy requires special attention [126]. Special care needs Further research is warranted to better understand the spectral to be applied to standardized retrieval methods for different response produced by the temporal and spatial dynamics of leaf types (i.e. broad leaves and needles) and spatial sampling the photosynthetic activity so that it can be generalized to the strategies that can be combined with airborne or satellite im- ecosystemic and biospheric scales. In short, these problems agery. Other important challenges are related to assessments at are related to the structural differences of canopies, the vary- the canopy scale and over larger areas using imagery. There is ing ‘background effects’ (e.g. soil colour, moisture, shadows, still a limited number of studies that have attempted to assess the presence of other non-green landscape components), the mixed forest areas with high species diversity with varying effects of seasonality or the signals derived from variations in structural and composition properties. Previous studies have illumination and viewing angles. demonstrated that the main spectral differences found in One of the main challenges for scaling up chlorophyll fluo- mixed forests were the result of species composition and not rescence is the establishment of a relationship between leaf pigment concentration [127]. This is especially relevant in chlorophyll fluorescence and satellite-based measurements. high-production tropical forests considered to be the ‘lungs In recent years, passive methods for remotely measuring chlo- of the Earth’ and highly significant for carbon balance ac- rophyll fluorescence have been developed and applied. Field- counting at the global scale. portable pulse amplitude modulated (PAM) devices have been Future direction to solve these limitations is the improve- widely applied to individual leaves. They provide an approach ment of RTM, incorporating new modules accounting for can- for bottom–up validation of steady-state fluorescence yield opy structural effects and additional biochemical components (Fs) for canopy photosynthesis models. Alternatively, Fs can of the leaf. At the canopy level, 3D models and data derived also be measured with a FluorPen FP100 (Photon Systems from active sensor imagery (e.g. LiDAR) may overcome some Instruments, Brno, Czech Republic). Although measurements of the aforementioned limitations [12••, 57]. At the leaf level, taken with both fluorescence instruments differ from airborne recent improvements of PROSPECT, including anthocyanin SIF retrievals, leaf data have been successfully used as a field- content [81], and Fluspect, including xanthophyll content level assessment in vegetation health analysis [12••, 121]. [82], may provide a better understanding of physiological However, fluorescence is a low-intensity signal that overlaps processes and vegetation dynamics. with the much brighter red and near-infrared reflected radia- Nevertheless, RTM modelling approaches are powerful but tion. These methods are designed to disentangle both signals, remain computationally demanding for complex environ- providing an indirect measure of fluorescence. Early on, it ments, so they are limited to smaller areas. One of the most became clear that it was necessary to have a way of validating recent solutions proposed to overcome this limitation is the the retrievals or, at least, to have an estimate of the order of combined use of RTM with a surrogate machine learning or magnitude of the expected fluorescence signal. This need led emulator reducing the processing time while preserving accu- to the design of a clip that allowed an attached leaf sample to racy [128]. be fully illuminated by natural sunlight, inducing photosyn- It can therefore be concluded that remote sensing enables thesis; yet, at the same time it blocked the overlapping light the assessment of physiological indicators to detect vegetation fluorescence, providing a direct measurement of the emitted health over large areas. The close link between vegetation fluorescence from 650 to 850 nm and discriminating between temperature, leaf optical properties linked to pigment absorp- the contribution of the red and far-red emission peaks. The tion and chlorophyll fluorescence emission with plant func- FluoWat leaf clip was proposed by [124] and provided tioning opens a wide range of applications and methods to Curr Forestry Rep (2019) 5:169–183 179 monitor vegetation health remotely. Current and future re- blotch using high-resolution hyperspectral and thermal imagery. Remote Sens. 2016;8:276. https://doi.org/10.3390/rs8040276. search should address the main challenges, focusing on the 6. Hernández-Clemente R, Navarro-Cerrillo RM, Zarco-Tejada PJ. improvement of data quality and successful upscaling to con- Carotenoid content estimation in a heterogeneous conifer forest sider the main drivers of plant physiology. using narrow-band indices and PROSPECT+DART simulations. Remote Sens Environ. 2012;127:298–315. https://doi.org/10. Funding Information The authors received funding provided by the 1016/j.rse.2012.09.014. FluorFLIGHT (GGR801) Marie Curie Fellowship, the QUERCUSAT 7. Porcar-Castell A, Tyystjärvi E, Atherton J, van der Tol C, Flexas J, and ESPECTRAMED projects (Spanish Ministry of Economy and Pfündel EE, et al. Linking chlorophyll a fluorescence to photosyn- Competitiveness), the Academy of Finland (grants 266152, 317387) thesis for remote sensing applications: mechanisms and chal- and the European Research Council Synergy grant ERC-2013-SyG- lenges. J Exp Bot. 2014;65:4065–95. https://doi.org/10.1093/jxb/ 610028 IMBALANCE-P. eru191. 8. Gates DM. Leaf temperature and transpiration 1. Agron J. 1964;56: 273–7. https://doi.org/10.2134/agronj1964. Compliance with Ethical Standards 00021962005600030007x. 9. Möller M, Alchanatis V, Cohen Y, Meron M, Tsipris J, Naor A, Conflict of Interest Rocio Hernandez-Clemente, Alberto Hornero, Matti et al. Use of thermal and visible imagery for estimating crop water Mottus, Josep Penuelas, Victoria González-Dugo, Juan Carlos Jiménez, status of irrigated grapevine. J Exp Bot. 2007;58:827–38. https:// Lola Suárez, Luis Alonso and Pablo J. Zarco-Tejada state that there are no doi.org/10.1093/jxb/erl115. conflicts of interests to declare. 10. Reinert S, Bögelein R, Thomas FM. Use of thermal imaging to determine leaf conductance along a canopy gradient in European Human and Animal Rights This article does not contain any studies beech (Fagus sylvatica). Tree Physiol. 2012;32:294–302. https:// with human or animal subjects performed by any of the authors. doi.org/10.1093/treephys/tps017. 11. Irmak S, Haman DZ, Bastug R. Determination of crop water stress Open Access This article is distributed under the terms of the Creative index for irrigation timing and yield estimation of corn. Agron J. Commons Attribution 4.0 International License (http:// 2000;92:1221–7. https://doi.org/10.2134/agronj2000.9261221x. creativecommons.org/licenses/by/4.0/), which permits unrestricted use, 12.•• Hernández-Clemente R, North PRJ, Hornero A, Zarco-Tejada PJ. distribution, and reproduction in any medium, provided you give appro- Assessing the effects of forest health on sun-induced chlorophyll priate credit to the original author(s) and the source, provide a link to the fluorescence using the FluorFLIGHT 3-D radiative transfer model Creative Commons license, and indicate if changes were made. to account for forest structure. Remote Sens Environ. 2017;193: 165–79. https://doi.org/10.1016/j.rse.2017.02.012 Suggests a novel approach for assessing forest health using sun-induced chlorophyll fluorescence and 3-D radiative transfer modelling. 13. Smigaj M, Gaulton R, Suárez JC, Barr SL. Canopy temperature References from an unmanned aerial vehicle as an indicator of tree stress associated with red band needle blight severity. For Ecol Manag. 2019;433:699–708. https://doi.org/10.1016/j.foreco.2018.11.032. Papers of particular interest, published recently, have been 14. Hernández-Clemente R, Navarro-Cerrillo RM, Suárez L, Morales highlighted as: F, Zarco-Tejada PJ. Assessing structural effects on PRI for stress detection in conifer forests. Remote Sens Environ. 2011;115: • Of importance 2360–75. https://doi.org/10.1016/j.rse.2011.04.036. •• Of major importance 15. Lambers H, Chapin FS, Pons TL. Leaf energy budgets: effects of radiation and temperature. In: Lambers H, Chapin FS, Pons TL, 1. Martinelli F, Scalenghe R, Davino S, Panno S, Scuderi G, Ruisi P, editors. Plant physiological ecology. New York: Springer New et al. Advanced methods of plant disease detection. A review. Agron York; 1998. p. 210–29. Sustain Dev. 2015;35:1–25. https://doi.org/10.1007/s13593-014- 16. Peñuelas J, Bartrons M, Llusia J, Filella I. Sensing the energetic 0246-1. status of plants and ecosystems. Trends Plant Sci. 2015;20:528– 2. Steyn WJ, Wand SJE, Holcroft DM, Jacobs G. Anthocyanins in 30. https://doi.org/10.1016/j.tplants.2015.07.002. vegetative tissues: a proposed unified function in photoprotection. 17.• Tuominen J, Lipping T, Kuosmanen V, Haapanen R. Remote sens- New Phytol. 2002;155:349–61. https://doi.org/10.1046/j.1469- ing of forest health. Geosci Remote Sens. 2009. https://doi.org/10. 8137.2002.00482.x. 5772/8283 This review summarizes forest health definitions and 3.•• Zarco-Tejada PJ, Camino C, Beck PSA, Calderon R, Hornero A, the use of vegetation indices extracted from remote sensing Hernández-Clemente R, et al. Previsual symptoms of Xylella data. fastidiosa infection revealed in spectral plant-trait alterations. Nat 18. Lichtenthaler HK. Vegetation stress: an introduction to the stress Plants. 2018. https://doi.org/10.1038/s41477-018-0189-7 This concept in plants. J Plant Physiol. 1996;148:4–14. https://doi.org/ study provides a practical framework for evaluating the 10.1016/S0176-1617(96)80287-2. effectiveness of detecting pre-visual symptoms of Xylella 19. Lichtenthaler, H.K.: [34] Chlorophylls and carotenoids: pigments fastidiosa infection using high-resolution hyperspectral im- of photosynthetic biomembranes. In: Methods in enzymology. pp. ages, thermal data and radiative transfer modelling. 350–382. Academic Press (1987). 4. Calderón R, Navas-Cortés JA, Zarco-Tejada PJ. Early detection 20. Demmig-Adams B, Gilmore AM, Adams WW. Carotenoids 3: and quantification of Verticillium wilt in olive using hyperspectral in vivo function of carotenoids in higher plants. FASEB J Off and thermal imagery over large areas. Remote Sens. 2015;7:5584– Publ Fed Am Soc Exp Biol. 1996;10:403–12. 610. https://doi.org/10.3390/rs70505584. 21. Chalker-Scott L. Environmental significance of anthocyanins in 5. López-López M, Calderón R, González-Dugo V, Zarco-Tejada PJ, plant stress responses. Photochem Photobiol. 1999;70:1–9. https:// Fereres E. Early detection and quantification of almond red leaf doi.org/10.1111/j.1751-1097.1999.tb01944.x. 180 Curr Forestry Rep (2019) 5:169–183 22. Barker DH, Seaton GGR, Robinson SA. Internal and external Obs Geoinf. 2016;53:40–7. https://doi.org/10.1016/j.jag.2016.08. photoprotection in developing leaves of the CAM plant 002. Cotyledon orbiculata. Plant Cell Environ. 1997;20:617–24. 38. Idso SB, Jackson RD, Pinter PJ, Reginato RJ, Hatfield JL. https://doi.org/10.1111/j.1365-3040.1997.00078.x. Normalizing the stress-degree-day parameter for environmental 23. Herold B, Truppel I, Zude M, Geyer M. Spectral measurements on variability. Agric Meteorol. 1981;24:45–55. https://doi.org/10. ‘Elstar’ apples during fruit development on the tree. Biosyst Eng. 1016/0002-1571(81)90032-7. 2005;2:173–82. https://doi.org/10.1016/j.biosystemseng.2005.03. 39. Maes WH, Steppe K. Estimating evapotranspiration and drought 005. stress with ground-based thermal remote sensing in agriculture: a 24. Robinson R. Formation of anthocyanins in plants. Nature. review. J Exp Bot. 2012;63:4671–712. https://doi.org/10.1093/ 1936;137:172–3. https://doi.org/10.1038/137172a0. jxb/ers165. 25. Schaberg PG, van den Berg AK, Murakami PF, Shane JB, 40. Bellvert J, Marsal J, Girona J, Zarco-Tejada PJ. Seasonal evolution Donnelly JR. Factors influencing red expression in autumn foliage of crop water stress index in grapevine varieties determined with of sugar maple trees. Tree Physiol. 2003;23:325–33. https://doi. high-resolution remote sensing thermal imagery. Irrig Sci. org/10.1093/treephys/23.5.325. 2015;33:81–93. https://doi.org/10.1007/s00271-014-0456-y. 26. Zarco-Tejada PJ, Miller JR, Mohammed GH, Noland TL. 41. Gardner, B.R. (BP R., Nielsen, D.C., Shock, C.C.: Infrared ther- Chlorophyll fluorescence effects on vegetation apparent reflec- mometry and the crop water stress index. I. History, theory, and tance: I. Leaf-level measurements and model simulation. baselines. J Prod Agric USA (1992). Remote Sens Environ. 2000;74:582–95. https://doi.org/10.1016/ 42. Jones PD, Briffa KR. Global surface air temperature variations S0034-4257(00)00148-6. during the twentieth century: part 1, spatial, temporal and seasonal 27. Sobrino JA, Frate FD, Drusch M, Jiménez-Muñoz JC, Manunta P, details. The Holocene. 1992;2:165–79. Regan A. Review of thermal infrared applications and require- 43. Alchanatis V, Cohen Y, Cohen S, Moller M, Sprinstin M, Meron ments for future high-resolution sensors. IEEE Trans Geosci M, et al. Evaluation of different approaches for estimating and Remote Sens. 2016;54:2963–72. https://doi.org/10.1109/TGRS. mapping crop water status in cotton with thermal imaging. 2015.2509179. Precis Agric. 2010;11:27–41. https://doi.org/10.1007/s11119- 28. Ribeiro da Luz B, Crowley JK. Spectral reflectance and emissivity 009-9111-7. features of broad leaf plants: prospects for remote sensing in the 44. Jackson RD, Idso SB, Reginato RJ, Pinter PJ. Canopy temperature thermal infrared (8.0–14.0 μm). Remote Sens Environ. 2007;109: as a crop water stress indicator. Water Resour Res. 1981;17:1133– 393–405. https://doi.org/10.1016/j.rse.2007.01.008. 8. https://doi.org/10.1029/WR017i004p01133. 45. Moran MS, Clarke TR, Inoue Y, Vidal A. Estimating crop water 29. Li Z-L, Tang B-H, Wu H, Ren H, Yan G, Wan Z, et al. Satellite- deficit using the relation between surface-air temperature and derived land surface temperature: current status and perspectives. spectral vegetation index. Remote Sens Environ. 1994;49:246– Remote Sens Environ. 2013;131:14–37. https://doi.org/10.1016/j. 63. https://doi.org/10.1016/0034-4257(94)90020-5. rse.2012.12.008. 46. Plascyk JA. The MK II Fraunhofer line discriminator (FLD-II) for 30. Jimenez-Munoz JC, Cristobal J, Sobrino JA, Soria G, Ninyerola airborne and orbital remote sensing of solar-stimulated lumines- M, Pons X, et al. Revision of the single-channel algorithm for land cence. Opt Eng. 1975;14:144339. https://doi.org/10.1117/12. surface temperature retrieval from Landsat thermal-infrared data. 7971842. IEEE Trans Geosci Remote Sens. 2009;47:339–49. https://doi. 47. Maier SW, Günther KP, Stellmes M. Sun-induced fluorescence: a org/10.1109/TGRS.2008.2007125. new tool for precision farming. Digit Imaging Spectr Tech Appl 31. Sobrino JA, Jimenez-Munoz JC, Soria G, Romaguera M, Guanter Precis Agric Crop Physiol. asaspecialpubli. 2003:209–22. https:// L, Moreno J, et al. Land surface emissivity retrieval from different doi.org/10.2134/asaspecpub66.c16. VNIR and TIR sensors. IEEE Trans Geosci Remote Sens. 48. Moya I, Daumard F, Moise N, Ounis A, Goulas Y. First airborne 2008;46:316–27. https://doi.org/10.1109/TGRS.2007.904834. multiwavelength passive chlorophyll fluorescence measurements 32. Gillespie A, Rokugawa S, Matsunaga T, Cothern JS, Hook S, over La Mancha (Spain) fields. Second Recent Adv Quant Kahle AB. A temperature and emissivity separation algorithm Remote Sens. 2006:820–5. for advanced spaceborne thermal emission and reflection radiom- 49. Alonso L, Gomez-Chova L, Vila-Frances J, Amoros-Lopez J, eter (ASTER) images. IEEE Trans Geosci Remote Sens. 1998;36: Guanter L, Calpe J, et al. Improved Fraunhofer line discrimination 1113–26. https://doi.org/10.1109/36.700995. method for vegetation fluorescence quantification. IEEE Geosci 33. Jiménez-Muñoz JC, Sobrino JA, Mattar C, Hulley G, Göttsche F. Remote Sens Lett. 2008;5:620–4. https://doi.org/10.1109/LGRS. Temperature and emissivity separation from MSG/SEVIRI data. 2008.2001180. IEEE Trans Geosci Remote Sens. 2014;52:5937–51. https://doi. 50. Meroni M, Rossini M, Guanter L, Alonso L, Rascher U, Colombo org/10.1109/TGRS.2013.2293791. R, et al. Remote sensing of solar-induced chlorophyll fluores- 34. Hulley GC, Hook SJ, Baldridge AM. Investigating the effects of cence: review of methods and applications. Remote Sens soil moisture on thermal infrared land surface temperature and Environ. 2009;113:2037–51. https://doi.org/10.1016/j.rse.2009. emissivity using satellite retrievals and laboratory measurements. 05.003. Remote Sens Environ. 2010;114:1480–93. https://doi.org/10. 51. Guanter L, Rossini M, Colombo R, Meroni M, Frankenberg C, 1016/j.rse.2010.02.002. Lee J-E, et al. Using field spectroscopy to assess the potential of 35. French AN, Inamdar A. Land cover characterization for hydrolog- statistical approaches for the retrieval of sun-induced chlorophyll ical modelling using thermal infrared emissivities. Int J Remote fluorescence from ground and space. Remote Sens Environ. Sens. 2010;31:3867–83. https://doi.org/10.1080/01431161.2010. 2013;133:52–61. https://doi.org/10.1016/j.rse.2013.01.017. 483491. 52. Miller, J.R., Berger, M., Alonso, L., Cerovic, Z., Goulas, Y., 36. Borel C. Error analysis for a temperature and emissivity retrieval Jacquemoud, S., Louis, J., Mohammed, G., Moya, I., Pedros, R., algorithm for hyperspectral imaging data. Int J Remote Sens. Moreno, J.F., Verhoef, W., Zarco-Tejada, P.J.: Progress on the 2008;29:5029–45. https://doi.org/10.1080/01431160802036540. development of an integrated canopy fluorescence model. In: 37. Neinavaz E, Darvishzadeh R, Skidmore AK, Groen TA. IGARSS 2003. 2003 IEEE International Geoscience and Measuring the response of canopy emissivity spectra to leaf area Remote Sensing Symposium. Proceedings (IEEE Cat. index variation using thermal hyperspectral data. Int J Appl Earth No.03CH37477). pp. 601–603 vol.1 (2003). Curr Forestry Rep (2019) 5:169–183 181 53. Verhoef, W.: Extension of SAIL to model solar - induced canopy (PRI) for remotely sensing foliar, canopy and ecosystemic fluorescence spectra. Proc. 2nd Int. Workshop Remote Sens. Veg. radiation-use efficiencies. Remote Sens. 2016;8:677. https://doi. Fluoresc. 17–19 Nov 2004 Montr. Can. Paris ESA 2004 ESA org/10.3390/rs8090677 A practical guide detecting the main WPP 242 18 P. (2004). factors affecting the relationship between the PRI and the 54. Vilfan N, van der Tol C, Muller O, Rascher U, Verhoef W. terrestrial gross primary productivity. Fluspect-B: a model for leaf fluorescence, reflectance and trans- 69. Wong CYS, Gamon JA. The photochemical reflectance index mittance spectra. Remote Sens Environ. 2016;186:596–615. provides an optical indicator of spring photosynthetic activation https://doi.org/10.1016/j.rse.2016.09.017. in evergreen conifers. New Phytol. 2015;206:196–208. https://doi. 55. Zarco-Tejada PJ, Morales A, Testi L, Villalobos FJ. Spatio- org/10.1111/nph.13251. temporal patterns of chlorophyll fluorescence and physiological 70. Goerner A, Reichstein M, Rambal S. Tracking seasonal drought and structural indices acquired from hyperspectral imagery as effects on ecosystem light use efficiency with satellite-based PRI compared with carbon fluxes measured with eddy covariance. in a Mediterranean forest. Remote Sens Environ. 2009;113:1101– Remote Sens Environ. 2013;133:102–15. https://doi.org/10. 11. https://doi.org/10.1016/j.rse.2009.02.001. 1016/j.rse.2013.02.003. 71. Vicca S, Balzarolo M, Filella I, Granier A, Herbst M, Knohl A, 56. van der Tol C, Verhoef W, Timmermans J, Verhoef A, Su Z. An et al. Remotely-sensed detection of effects of extreme droughts on integrated model of soil-canopy spectral radiances, photosynthe- gross primary production. Sci Rep. 2016;6:1–13. https://doi.org/ sis, fluorescence, temper ature and en ergy balanc e. 10.1038/srep28269. Biogeosciences. 2009;6:3109–29. https://doi.org/10.5194/bg-6- 72. Ripullone F, Rivelli AR, Baraldi R, Guarini R, Guerrieri R, 3109-2009. Magnani F, et al. Effectiveness of the photochemical reflectance 57. Gastellu-Etchegorry J, Lauret N, Yin T, Landier L, Kallel A, index to track photosynthetic activity over a range of forest tree Malenovský Z, et al. DART: recent advances in remote sensing species and plant water statuses. Funct Plant Biol. 2011;38:177– data modeling with atmosphere, polarization, and chlorophyll 86. https://doi.org/10.1071/FP10078. fluorescence. IEEE J Sel Top Appl Earth Obs Remote Sens. 73. Rossini M, Fava F, Cogliati S, Meroni M, Marchesi A, Panigada 2017;10:2640–9. https://doi.org/10.1109/JSTARS.2017.2685528. C, et al. Assessing canopy PRI from airborne imagery to map 58. Gamon, J. A., Peñuelas, J., Field, C.B.: A narrow-waveband spec- water stress in maize. ISPRS J Photogramm Remote Sens. tral index that tracks diurnal changes in photosynthetic efficiency, 2013;86:168–77. https://doi.org/10.1016/j.isprsjprs.2013.10.002. (1992). 74. Takala TLH, Mõttus M. Spatial variation of canopy PRI with 59. Penuelas, J., Filella, I., Gamon, J.: Assessment of photosynthetic shadow fraction caused by leaf-level irradiation conditions. radiation use efficiency with spectral reflectance., (1995). Remote Sens Environ. 2016;182:99–112. https://doi.org/10. 60. Filella I, Porcar-Castell A, Munné-Bosch S, Bäck J, Garbulsky 1016/j.rse.2016.04.028. MF, Peñuelas J. PRI assessment of long-term changes in 75. Gamon JA, Berry JA. Facultative and constitutive pigment effects carotenoids/chlorophyll ratio and short-term changes in de- on the photochemical reflectance index (PRI) in sun and shade epoxidation state of the xanthophyll cycle. Int J Remote Sens. conifer needles. Isr J Plant Sci. 2012;60:85–95. https://doi.org/ 2009;30:4443–55. https://doi.org/10.1080/01431160802575661. 10.1560/ijps.60.1-2.85. 61. Porcar-Castell A, Garcia-Plazaola JI, Nichol CJ, Kolari P, 76. Gamon JA, Surfus JS. Assessing leaf pigment content and activity Olascoaga B, Kuusinen N, et al. Physiology of the seasonal rela- with a reflectometer. New Phytol. 1999;143:105–17. https://doi. tionship between the photochemical reflectance index and photo- org/10.1046/j.1469-8137.1999.00424.x. synthetic light use efficiency. Oecologia. 2012;170:313–23. https://doi.org/10.1007/s00442-012-2317-9. 77. Magney TS, Huggins DR, Eitel JUH, Vierling LA, Garrity SR. 62. Sims DA, Gamon JA. Relationships between leaf pigment content Response of high frequency photochemical reflectance index and spectral reflectance across a wide range of species, leaf struc- (PRI) measurements to environmental conditions in wheat. tures and developmental stages. Remote Sens Environ. 2002;81: Remote Sens Environ. 2016. 337–54. https://doi.org/10.1016/S0034-4257(02)00010-X. 78. Suárez L, Zarco-Tejada PJ, Berni JAJ, González-Dugo V, Fereres 63. Gamon JA, Huemmrich KF, Wong CYS, Ensminger I, Garrity S, E. Modelling PRI for water stress detection using radiative transfer Hollinger DY, et al. A remotely sensed pigment index reveals models. Remote Sens Environ. 2009;113:730–44. https://doi.org/ photosynthetic phenology in evergreen conifers. Proc Natl Acad 10.1016/j.rse.2008.12.001. Sci. 2016;113:13087–92. https://doi.org/10.1073/pnas. 79. Mõttus M, Aragão L, Bäck J, Hernández-Clemente R, Maeda EE, 1606162113. Markiet V, et al. Diurnal changes in leaf photochemical reflectance 64. Wong CYS, Gamon JA. Three causes of variation in the photo- index in two evergreen forest canopies. IEEE J Sel Top Appl Earth chemical reflectance index (PRI) in evergreen conifers. New Obs Remote Sens. 2019:1–8. https://doi.org/10.1109/JSTARS. Phytol. 2015;206:187–95. https://doi.org/10.1111/nph.13159. 2019.2891789. 65. Garbulsky MF, Peñuelas J, Gamon J, Inoue Y, Filella I. The pho- 80. Kovac D, Navratil M, Malenovsky Z, Stroch M, Spunda V, Urban tochemical reflectance index (PRI) and the remote sensing of leaf, O. Reflectance continuum removal spectral index tracking the canopy and ecosystem radiation use efficiencies. A review and xanthophyll cycle photoprotective reactions in Norway spruce meta-analysis. Remote Sens Environ. 2011;115:281–97. https:// needles. Fac Sci Med Health - Pap Part A. 2012:987–98. https:// doi.org/10.1016/j.rse.2010.08.023. doi.org/10.1071/FP12107. 66. Penuelas J, Garbulsky MF, Filella I. Photochemical reflectance 81. Feret J-B, Gitelson A, Noble SD, Jacquemoud S. PROSPECT-D: index (PRI) and remote sensing of plant CO2 uptake. New towards modeling leaf optical properties through a complete Phytol. 2011;191:596–9. https://doi.org/10.1111/j.1469-8137. lifecycle. Remote Sens Environ. 2017;193:204–15. https://doi. 2011.03791.x. org/10.1016/j.rse.2017.03.004. 67. Stylinski CD, Gamon JA, Oechel WC. Seasonal patterns of reflec- 82. Vilfan N, Van der Tol C, Yang P, Wyber R, Malenovský Z, tance indices, carotenoid pigments and photosynthesis of ever- Robinson SA, et al. Extending Fluspect to simulate xanthophyll green chaparral species. Oecologia. 2002;131:366–74. https:// driven leaf reflectance dynamics. Remote Sens Environ. doi.org/10.1007/s00442-002-0905-9. 2018;211:345–56. https://doi.org/10.1016/j.rse.2018.04.012. 68.•• Zhang C, Filella I, Garbulsky MF, Peñuelas J. Affecting factors 83. Gitelson AA, Merzlyak MN. Signature analysis of leaf reflectance and recent improvements of the photochemical reflectance index spectra: algorithm development for remote sensing of chlorophyll. 182 Curr Forestry Rep (2019) 5:169–183 J Plant Physiol. 1996;148:494–500. https://doi.org/10.1016/ Remote Sens Environ. 2008;112:3234–47. https://doi.org/10. S0176-1617(96)80284-7. 1016/j.rse.2008.04.005. 84. Carter GA, Knapp AK. Leaf optical properties in higher plants: 99. Jacquemoud S, Baret F. PROSPECT: a model of leaf optical prop- linking spectral characteristics to stress and chlorophyll concen- erties spectra. Remote Sens Environ. 1990;34:75–91. https://doi. tration. Am J Bot. 2001;88:677–84. org/10.1016/0034-4257(90)90100-Z. 85. Hernández-Clemente R, Kolari P, Porcar-Castell A, Korhonen L, 100. Hernandez-Clemente R, Navarro-Cerrillo RM, Zarco-Tejada PJ. Mõttus M. Tracking the seasonal dynamics of boreal forest pho- Deriving predictive relationships of carotenoid content at the can- tosynthesis using EO-1 Hyperion reflectance: sensitivity to struc- opy level in a conifer forest using hyperspectral imagery and mod- tural and illumination effects. IEEE Trans Geosci Remote Sens. el simulation. IEEE Trans Geosci Remote Sens. 2014;52:5206– 2016;54:5105–16. https://doi.org/10.1109/TGRS.2016.2554466. 17. https://doi.org/10.1109/TGRS.2013.2287304. 86. Gitelson AA, Yacobi YZ, Schalles JE, Rundquist DC, Han L, 101. Sampson PH, Zarco-Tejada PJ, Mohammed GH, Miller JR, Stark R, et al. Remote estimation of phytoplankton density in Noland TL. Hyperspectral remote sensing of forest condition: productive waters. Arch HydrobioI Spec Issues Adv Limnol estimating chlorophyll content in tolerant hardwoods. For Sci. Limnol Lake Manag. 2000;55(2000):121–36. 2003;49:381–91. https://doi.org/10.1093/forestscience/49.3.381. 87. Hernández-Clemente R, Navarro-Cerrillo RM, Ramírez FJR, 102. Suárez L, Zarco-Tejada PJ, Sepulcre-Cantó G, Pérez-Priego O, Miller Hornero A, Zarco-Tejada PJ. A novel methodology to estimate JR, Jiménez-Muñoz JC, et al. Assessing canopy PRI for water stress single-tree biophysical parameters from 3D digital imagery com- detection with diurnal airborne imagery. Remote Sens Environ. pared to aerial laser scanner data. Remote Sens. 2014;6:11627–48. 2008;112:560–75. https://doi.org/10.1016/j.rse.2007.05.009. https://doi.org/10.3390/rs61111627. 103. Widlowski J-L, Pinty B, Lopatka M, Atzberger C, Buzica D, 88. Croft H, Chen JM, Zhang Y. The applicability of empirical vege- Chelle M, et al. The fourth radiation transfer model intercompar- tation indices for determining leaf chlorophyll content over differ- ison (RAMI-IV): proficiency testing of canopy reflectance models ent leaf and canopy structures. Ecol Complex. 2014;17:119–30. with ISO-13528. J Geophys Res-Atmos. 2013;118:6869–90. https://doi.org/10.1016/j.ecocom.2013.11.005. https://doi.org/10.1002/jgrd.50497. 89. Blackburn GA. Wavelet decomposition of hyperspectral data: a 104. Zarco-Tejada PJ, Miller JR, Harron J, Hu B, Noland TL, Goel N, novel approach to quantifying pigment concentrations in vegeta- et al. Needle chlorophyll content estimation through model inver- tion. Int J Remote Sens. 2007;28:2831–55. https://doi.org/10. sion using hyperspectral data from boreal conifer forest canopies. 1080/01431160600928625. Remote Sens Environ. 2004;89:189–99. https://doi.org/10.1016/j. rse.2002.06.002. 90. Kuusk A. Monitoring of vegetation parameters on large areas by 105. Zarco-Tejada PJ, Miller JR, Mohammed GH, Noland TL, the inversion of a canopy reflectance model. Int J Remote Sens. Sampson PH. Estimation of chlorophyll fluorescence under natu- 1998;19:2893–905. https://doi.org/10.1080/014311698214334. ral illumination from hyperspectral data. Int J Appl Earth Obs 91. Broge NH, Mortensen JV. Deriving green crop area index and Geoinf. 2001;3:321–7. https://doi.org/10.1016/S0303-2434(01) canopy chlorophyll density of winter wheat from spectral reflec- 85039-X. tance data. Remote Sens Environ. 2002;81:45–57. https://doi.org/ 106. Demarez V, Gastellu-Etchegorry JP. A modeling approach for study- 10.1016/S0034-4257(01)00332-7. ing forest chlorophyll content. Remote Sens Environ. 2000;71:226– 92. Barton CVM, North PRJ. Remote sensing of canopy light use 38. https://doi.org/10.1016/S0034-4257(99)00089-9. efficiency using the photochemical reflectance index: model and 107. Yáñez-Rausell L, Malenovský Z, Rautiainen M, Clevers JGPW, sensitivity analysis. Remote Sens Environ. 2001;78:264–73. Lukeš P, Hanuš J, et al. Estimation of spruce needle-leaf chloro- https://doi.org/10.1016/S0034-4257(01)00224-3. phyll content based on DART and PARAS canopy reflectance 93. Haboudane D, Miller JR, Tremblay N, Zarco-Tejada PJ, Dextraze models. IEEE J Sel Top Appl Earth Obs Remote Sens. 2015;8: L. Integrated narrow-band vegetation indices for prediction of 1534–44. https://doi.org/10.1109/JSTARS.2015.2400418. crop chlorophyll content for application to precision agriculture. 108.• Sobrino JA, Frate FD, Drusch M, Jiménez-Muñoz JC, Manunta P, Remote Sens Environ. 2002;81:416–26. https://doi.org/10.1016/ Regan A. Review of thermal infrared applications and require- S0034-4257(02)00018-4. ments for future high-resolution sensors. IEEE Trans Geosci 94. Kokaly RF, Clark RN. Spectroscopic determination of leaf Remote Sens. 2016). doi:https://doi.org/10.1109/TGRS.2015. biochemistry using band-depth analysis of absorption fea- 2509179;54:2963–72 This review neatly summarizes the tures and stepwise multiple linear regression. Remote Sens applicability of thermal infrared data for different Environ. 1999;67:267–87. https://doi.org/10.1016/S0034- applications, including vegetation condition. 4257(98)00084-4. 109. Navarro-Cerrillo RM, Trujillo J, Sanchez de la Orden M, 95. Curran PJ, Dungan JL, Peterson DL. Estimating the foliar bio- Hernández-Clemente R. Hyperspectral and multispectral satellite chemical concentration of leaves with reflectance spectrometry: sensors for mapping chlorophyll content in a Mediterranean Pinus testing the Kokaly and Clark methodologies. Remote Sens sylvestris L. plantation. Int J Appl Earth Obs Geoinf. 2014;26:88– Environ. 2001;76:349–59. https://doi.org/10.1016/S0034- 96. https://doi.org/10.1016/j.jag.2013.06.001. 4257(01)00182-1. 110. Kahle, A.B.: The new airborne thermal infrared multispectral 96. Malenovský Z, Homolová L, Zurita-Milla R, Lukeš P, Kaplan V, scanner (TIMS). Presented at the 1983 International Geoscience Hanuš J, et al. Retrieval of spruce leaf chlorophyll content from and Remote Sensing Symposium , August 31–September 2, 1983 airborne image data using continuum removal and radiative trans- January 1 (1983). fer. Remote Sens Environ. 2013;131:85–102. https://doi.org/10. 111.• Sepulcre-Cantó G, Zarco-Tejada PJ, Jiménez-Muñoz JC, Sobrino 1016/j.rse.2012.12.015. JA, de Miguel E, Villalobos FJ. Detection of water stress in an 97. Schlerf M, Atzberger C, Hill J, Buddenbaum H, Werner W, olive orchard with thermal remote sensing imagery. Agric For Schüler G. Retrieval of chlorophyll and nitrogen in Norway Meteorol. 2006;136:31–44. https://doi.org/10.1016/j.agrformet. spruce (Picea abies L. Karst.) using imaging spectroscopy. Int J 2006.01.008 This study provides a practical example of how Appl Earth Obs Geoinf. 2010;12:17–26. https://doi.org/10.1016/j. thermal data can be used to monitor water stress in plants. jag.2009.08.006. 112. Shivers SW, Roberts DA, McFadden JP. Using paired thermal and 98. Zhang Y, Chen JM, Miller JR, Noland TL. Leaf chlorophyll con- hyperspectral aerial imagery to quantify land surface temperature tent retrieval from airborne hyperspectral remote sensing imagery. variability and assess crop stress within California orchards. Curr Forestry Rep (2019) 5:169–183 183 Remote Sens Environ. 2019;222:215–31. https://doi.org/10.1016/ 122. Olioso A, Soria G, Sobrino J, Duchemin B. Evidence of low land j.rse.2018.12.030. surface thermal infrared emissivity in the presence of dry vegeta- 113. Kusnierek K, Korsaeth A. Challenges in using an analog uncooled tion. IEEE Geosci Remote Sens Lett. 2007;4:112–6. https://doi. microbolometer thermal camera to measure crop temperature. Int J org/10.1109/LGRS.2006.885857. Agric Biol Eng. 2014;7:60–74. https://doi.org/10.25165/ijabe. 123. Meerdink SK, Roberts DA, King JY, Roth KL, Dennison PE, v7i4.1041. Amaral CH, et al. Linking seasonal foliar traits to VSWIR-TIR 114. Ribeiro-Gomes K, Hernández-López D, Ortega JF, Ballesteros R, spectroscopy across California ecosystems. Remote Sens Environ. Poblete T, Moreno MA. Uncooled thermal camera calibration and 2016;186:322–38. https://doi.org/10.1016/j.rse.2016.08.003. optimization of the photogrammetry process for UAV applications 124. Alonso, L., Gomez-Chova, L., Vila-Frances, J., Amoros-Lopez, J., in agriculture. Sensors. 2017;17:2173. https://doi.org/10.3390/ Guanter, L., Calpe, J., Moreno, J.: Sensitivity analysis of the s17102173. Fraunhofer line discrimination method for the measurement of 115. Budzier H, Gerlach G. Calibration of uncooled thermal infrared chlorophyll fluorescence using a field spectroradiometer. In: cameras. J Sens Sens Syst. 2015;4:187–97. https://doi.org/10. 2007 IEEE International Geoscience and Remote Sensing 5194/jsss-4-187-2015. Symposium. pp. 3756–3759 (2007). 116. Guanter L, Frankenberg C, Dudhia A, Lewis PE, Gómez-Dans J, 125. Zarco-Tejada PJ, González-Dugo V, Williams LE, Suárez L, Berni Kuze A, et al. Retrieval and global assessment of terrestrial chloro- JAJ, Goldhamer D, et al. A PRI-based water stress index combin- phyll fluorescence from GOSAT space measurements. Remote Sens ing structural and chlorophyll effects: assessment using diurnal Environ. 2012;121:236–51. https://doi.org/10.1016/j.rse.2012.02.006. narrow-band airborne imagery and the CWSI thermal index. 117. Kraft, S., Bello, U.D., Bouvet, M., Drusch, M., Moreno, J.: FLEX: Remote Sens Environ. 2013;138:38–50. https://doi.org/10.1016/ ESA’s Earth Explorer 8 candidate mission. In: 2012 IEEE j.rse.2013.07.024. International Geoscience and Remote Sensing Symposium. pp. 126. Woodgate W, Jones SD, Suarez L, Hill MJ, Armston JD, Wilkes P, 7125–7128 (2012). et al. Understanding the variability in ground-based methods for 118. Rossini M, Panigada C, Cilia C, Meroni M, Busetto L, Cogliati S, retrieving canopy openness, gap fraction, and leaf area index in et al. Discriminating irrigated and rainfed maize with diurnal fluo- diverse forest systems. Agric For Meteorol. 2015;205:83–95. rescence and canopy temperature airborne maps. ISPRS Int J Geo- https://doi.org/10.1016/j.agrformet.2015.02.012. Inf. 2015;4:626–46. https://doi.org/10.3390/ijgi4020626. 127. Asner GP, Martin RE, Ford AJ, Metcalfe DJ, Liddell MJ. Leaf 119. Alonso L, Wittenberghe SV, Amorós-López J, Vila-Francés J, chemical and spectral diversity in Australian tropical forests. Ecol Gómez-Chova L, Moreno JF. Diurnal cycle relationships between Appl. 2009;19:236–53. https://doi.org/10.1890/08-0023.1. passive fluorescence, PRI and NPQ of vegetation in a controlled stress experiment. Remote Sens. 2017;9:770. https://doi.org/10. 128. Verrelst J, Rivera Caicedo JP, Muñoz-Marí J, Camps-Valls G, 3390/rs9080770. Moreno J. SCOPE-based emulators for fast generation of synthet- 120. Balick, L.K., Jeffery, C.A., Henderson, B.G.: Turbulence-induced ic canopy reflectance and sun-induced fluorescence spectra. spatial variation of surface temperature in high-resolution thermal Remote Sens. 2017;9:927. https://doi.org/10.3390/rs9090927. IR satellite imagery. In: Remote sensing for agriculture, ecosys- tems, and hydrology IV. pp. 221–231. International Society for Publisher’s Note Springer Nature remains neutral with regard to jurisdic- Optics and Photonics (2003). tional claims in published maps and institutional affiliations. 121. Gomis-Cebolla, J., Jiménez-Muñoz, J.C., Sobrino, J.A.: Extreme thermal episodes analyzed with MODIS products during the bo- real winter (2000-2016). Rev Teledetec 0, 19–30 (2016). doi: https://doi.org/10.4995/raet.2016.5709.