No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Sensitivity of colour indices for discriminating leaf colours from digital photographs
Methods in Ecology and Evolution
Abstract
Digital images of tree canopies have been analysed to understand how forest phenology responds to climate change. Researchers have used different colour indices to carry out quantitative analyses, but uncertainties over the performance of the various indices are hampering progress in their use.
To compare the various indices under controlled conditions, we carried out experiments using a low‐cost off‐the‐shelf digital camera with a set of standard colour charts as model leaves for different stages: emerging leaves, yellowish green; newly expanded leaves, green; fully mature leaves, dark green; senescent leaves, yellow. Two models of cameras, a compact digital camera and a surveillance ‘live image’ camera were used, and photographs were taken by two cameras for each model under clear or overcast sky conditions with two colour balance settings. The indices were also compared with those derived from spectral reflectance.
Colour indices based on hue distinguished leaf colour samples with only a small influence of camera models, balance setting and sky conditions, while indices based on green were strongly influenced by camera models and were relatively insensitive to leaf colours. The strength of the green channel relative to the total of digital numbers took similar values for the mature and senescent replica leaves, highlighting its poor ability to identify the change of colour in autumn. Spectral‐based hue was also sensitive to the gradation of leaf colours and showed a good correlation with the digital representation of hue regardless of camera models and balance setting. Remarkably, the primitive digital number of red, N red , also discriminated leaf colours well, with a small influence of the factors investigated here, showing a good correlation with the reflectance of the red band, except from images taken by the surveillance cameras with auto balance.
Hue was a robust index across the image set, while the green‐based indices often used to quantify canopy phenology in previous studies performed poorly. Hue was well correlated with spectral reflectance indices and worked better than all other indices to discriminate leaf colours. We recommend using hue as a colour index for tracking different stages of leaf development.
... On the other hand, digital photography is a fast, noninvasive alternative which has become the method of choice for measuring color both in animals and in plants (Bergman & Beehner, 2008;Garcia, Greentree, Shrestha, Dorin, & Dyer, 2014;Kendal et al., 2013;Mizunuma et al., 2014;Stevens, Lown, & Wood, 2014). With relatively simple camera settings, a few precautions before taking the photograph, and easy image processing (Stevens, Párraga, Cuthill, Partridge, & Troscianko, 2007;Troscianko & Stevens, 2015;White et al., 2015), digital imaging is an efficient and reliable method to quantify color, even in the field (Bergman & Beehner, 2008;Macfarlane & Ogden, 2012;Stevens et al., 2014). ...
... In the literature, there are a myriad of indices that can be obtained from color spectra data (Endler, 1990;Gitelson et al., 2009;Gomez, 2006;Montgomerie, 2006) and digital image data (i.e. RGB values; Gillespie, Kahle, & Walker, 1987;Woebbecke, Meyer, Von Bargen, & Mortensen, 1995;Mizunuma et al., 2014). For spectral reflectance, we used indices related to the physical properties of light which are independent of the observer's visual system, except for segment analysis indices, chosen because it compares different regions of the wavelength spectra (Endler, 1990;Kemp et al., 2015). ...
... As anthocyanin pigments show absorption in the green region of the spectrum, and reflect red, blue, and purple wavelengths (Gitelson, Merzlyak, & Chivkunova, 2001;van der Kooi et al., 2016); it follows that variation in the G channel is particularly effective for estimating anthocyanin concentration using digital images. In this way, Mizunuma et al. (2014) showed the suitability of digital images to discriminate leaf colors and estimate chlorophyll concentration. Chlorophylls absorb in the red region of spectra, and accordingly, indices accounting for the R channel showed the best performance (Mizunuma et al., 2014). ...
Anthocyanin pigments have become a model trait for evolutionary ecology as they often provide adaptive benefits for plants. Anthocyanins have been traditionally quantified biochemically or more recently using spectral reflectance. However, both methods require destructive sampling and can be labor intensive and challenging with small samples. Recent advances in digital photography and image processing make it the method of choice for measuring color in the wild. Here, we use digital images as a quick, noninvasive method to estimate relative anthocyanin concentrations in species exhibiting color variation. Using a consumer-level digital camera and a free image processing toolbox, we extracted RGB values from digital images to generate color indices. We tested petals, stems, pedicels, and calyces of six species, which contain different types of anthocyanin pigments and exhibit different pigmentation patterns. Color indices were assessed by their correlation to biochemically determined anthocyanin concentrations. For comparison, we also calculated color indices from spectral reflectance and tested the correlation with anthocyanin concentration. Indices perform differently depending on the nature of the color variation. For both digital images and spectral reflectance, the most accurate estimates of anthocyanin concentration emerge from anthocyanin content-chroma ratio, anthocyanin content-chroma basic, and strength of green indices. Color indices derived from both digital images and spectral reflectance strongly correlate with biochemically determined anthocyanin concentration; however, the estimates from digital images performed better than spectral reflectance in terms of r2 and normalized root-mean-square error. This was particularly noticeable in a species with striped petals, but in the case of striped calyces, both methods showed a comparable relationship with anthocyanin concentration. Using digital images brings new opportunities to accurately quantify the anthocyanin concentrations in both floral and vegetative tissues. This method is efficient, completely noninvasive, applicable to both uniform and patterned color, and works with samples of any size.
... The archived images used in the present analysis are all taken between 11:00 and 13:00 local time (LT). The camera setup is specific to each camera type but a common requirement to observe the seasonal colour fraction time series is to set the colour balance to "fixed" mode (on the Stardot cameras) or the white balance to "manual" mode (for the Nikon Coolpix cameras) (Mizunuma et al., 2013(Mizunuma et al., , 2014. ...
... Most commercial cameras already contain UV filter protection and in addition they are protectively housed from the elements and sit behind a glass shield that protects the camera 6010 L. Wingate et al.: Linking ecosystem fluxes to canopy phenology in Europe from the damage by UV or other environmental conditions, such as water. Also, the studies of Sonnentag et al. (2012) and Mizunuma et al. (2013Mizunuma et al. ( , 2014 demonstrated that different camera brands and even different cameras of the same brand that produce different colour fraction time series reflecting differences in their spectral response still produce coherent phenological metrics. Nonetheless, it would be useful to develop a calibration scheme by digital photography of radiometrically characterised colour sheets such as those used to detect the health and nutritional status of plants (Mizunuma et al., 2014) whilst the camera is in the field and exposed to a variety of light conditions. ...
... Also, the studies of Sonnentag et al. (2012) and Mizunuma et al. (2013Mizunuma et al. ( , 2014 demonstrated that different camera brands and even different cameras of the same brand that produce different colour fraction time series reflecting differences in their spectral response still produce coherent phenological metrics. Nonetheless, it would be useful to develop a calibration scheme by digital photography of radiometrically characterised colour sheets such as those used to detect the health and nutritional status of plants (Mizunuma et al., 2014) whilst the camera is in the field and exposed to a variety of light conditions. However, the deployment of such a colour checker for long-term continuous monitoring is problematic as the spectral quality of the colour checker will alter over time as particles, such as dust and insects, accumulate on its surface. ...
Plant phenological development is orchestrated through subtle changes in
photoperiod, temperature, soil moisture and nutrient availability.
Presently, the exact timing of plant development stages and their response
to climate and management practices are crudely represented in land surface
models. As visual observations of phenology are laborious, there is a need
to supplement long-term observations with automated techniques such as those
provided by digital repeat photography at high temporal and spatial
resolution. We present the first synthesis from a growing observational
network of digital cameras installed on towers across Europe above deciduous
and evergreen forests, grasslands and croplands, where vegetation and
atmosphere CO2 fluxes are measured continuously. Using colour indices
from digital images and using piecewise regression analysis of time series,
we explored whether key changes in canopy phenology could be detected
automatically across different land use types in the network. The piecewise
regression approach could capture the start and end of the growing season,
in addition to identifying striking changes in colour signals caused by
flowering and management practices such as mowing. Exploring the dates of
green-up and senescence of deciduous forests extracted by the piecewise
regression approach against dates estimated from visual observations, we
found that these phenological events could be detected adequately (RMSE < 8 and 11 days for leaf out and leaf fall, respectively). We also
investigated whether the seasonal patterns of red, green and blue colour
fractions derived from digital images could be modelled mechanistically
using the PROSAIL model parameterised with information of seasonal changes
in canopy leaf area and leaf chlorophyll and carotenoid concentrations. From
a model sensitivity analysis we found that variations in colour fractions,
and in particular the late spring `green hump' observed repeatedly in
deciduous broadleaf canopies across the network, are essentially dominated
by changes in the respective pigment concentrations. Using the model we were
able to explain why this spring maximum in green signal is often observed
out of phase with the maximum period of canopy photosynthesis in ecosystems
across Europe. Coupling such quasi-continuous digital records of canopy
colours with co-located CO2 flux measurements will improve our
understanding of how changes in growing season length are likely to shape
the capacity of European ecosystems to sequester CO2 in the future.
... Nonetheless, in a comparative study by Sonnentag et al. (2012), it was shown that a camera model similar to ours yielded results equivalent to cameras with a fixed white balance. Also, despite their usefulness, greenness indices have not been derived from a calibrated scientific instrument although some guidelines, protocols, and recommendations have been proposed (Alberton et al., 2017;Mizunuma et al., 2014;Richardson et al., 2009Richardson et al., , 2018Sonnentag et al., 2012). As yet, there is no basis for an absolute scale (e.g., to compare Ig from arid and mesic sites), in contrast with common remote sensing products such as NDVI. ...
... However, the PhenoCam Network has specific guidelines and protocols that may help in standardizing how vegetation phenology is monitored with consumer-grade cameras (Richardson et al., 2018). It has been discussed that comparative laboratory tests may improve standardization (Mizunuma et al., 2014). ...
Although drylands cover >40% of the land surface, models of ecosystem gross primary productivity (GPP) generally have been designed for mesic temperate ecosystems. Arguably, GPP models often lack a good representation of vegetation phenology, particularly not estimating the ecosystem effects of the prolonged foliage senescence which may be common in drylands. To estimate daily GPP for a water‐limited Mediterranean shrubland, we propose a simple framework (GPPmod) using light use efficiency, a spectral vegetation index derived from digital cameras, and five meteorological variables, including an index of functional senescence of foliage (i.e., heat degree‐days). We tested the model with different combinations of meteorological variables but without senescence, using 1 year's data. The best formulation showed good agreement with GPP derived from eddy covariance (GPPEC; r² = 0.53, RMSE = 0.77). However, including the foliage senescence parameter significantly improved model performance (r² = 0.74, RMSE = 0.49), especially during the fall season. In the following year, we validated the parameters: The overall GPPmod and GPPEC comparison yielded an r² = 0.78. We postulate that models that mainly rely on meteorological variables or greenness indices could yield an overestimation of annual GPP between 24% and 90%, while models including the foliage senescence parameter reduced that bias by 10% to 34%. Our results highlight the importance of incorporating the phenology of foliage senescence in models regarding productivity in drylands or dry sclerophyll ecosystems.
... Several common spectral indices for 3 and 8 band images were calculated [43,44] and are listed in SI 1; for example, for WorldView 2 several measures of NDVI were calculated and for Nearmap images examples of calculated indices included GEI and GRVI [43,44]. The mean, median, and standard deviation of each of these indices were extracted for each tree crown polygon. ...
... Several common spectral indices for 3 and 8 band images were calculated [43,44] and are listed in SI 1; for example, for WorldView 2 several measures of NDVI were calculated and for Nearmap images examples of calculated indices included GEI and GRVI [43,44]. The mean, median, and standard deviation of each of these indices were extracted for each tree crown polygon. ...
Urban tree identification is often limited by the accessibility of remote sensing imagery but has not yet been attempted with the multi-temporal commercial aerial photography that is now widely available. In this study, trees in Detroit, Michigan, USA are identified using eight high resolution red, green, and blue (RGB) aerial images from a commercial vendor and publicly available LiDAR data. Classifications based on these data were compared with classifications based on World View 2 satellite imagery, which is commonly used for this task but also more expensive. An object-based classification approach was used whereby tree canopies were segmented using LiDAR, and a street tree database was used for generating training and testing datasets. Overall accuracy using multi-temporal aerial images and LiDAR was 70%, which was higher than the accuracy achieved with World View 2 imagery and LiDAR (63%). When all data were used, classification accuracy increased to 74%. Taxa identified with high accuracy included Acer platanoides and Gleditsia, and taxa that were identified with good accuracy included Acer, Platanus, Quercus, and Tilia. Our results show that this large catalogue of multi-temporal aerial images can be leveraged for urban tree identification. While classification accuracy rates vary between taxa, the approach demonstrated can have practical value for socially or ecologically important taxa.
... The NDVI is widely used as a descriptor of plant photosynthetic activity; thus, plant condition and the SRG are indices of anthocyanin content. For the RGB orthomosaics Green Excess Index (GEI; Eq. (5)) (Mizunuma et al., 2014) was calculated instead of the NDVI. A total of 75 predictors were extracted from all orthoimages for the years 2011, 2012 and 2014 used in the study. ...
Platanus sp. pl. (plane trees) are common ornamental tree in Poland that produces a large amount of wind-transported pollen, which contains proteins that induce allergy symptoms. Allergy sufferers can limit their contact with pollen by avoiding places with high pollen concentrations, which are restricted mainly to areas close to plane trees. Their location is thus important, but creating a detailed street tree inventory is expensive and time-consuming. However, high-resolution remote sensing data provide an opportunity to detect the location of specific plants. But acquiring high-resolution spatial data of good quality also incurs costs and requires regular updates. Therefore, this study explored the potential of using open access remote sensing data to detect plane trees in the highly urbanized environment of Poznań (western Poland). Airborne light detection and ranging (LiDAR) was used to detect training treetops, which were subsequently marked as young plane trees, mature plane trees, other trees or artefacts. Spectral and spatial variables were extracted from circular buffers (r = 1 m) around the treetops to minimize the influence of shadows and crown overlap. A random forest machine learning algorithm was applied to assess the importance of variables and classify the treetops within a radius of 6.2 km around the functioning pollen monitoring station. The model performed well during 10-fold cross-validation (overall accuracy ≈ 92%). The predicted Platanus sp. pl. locations, aggregated according to 16 wind directions, were significantly correlated with the hourly pollen concentrations. Based on the correlation values, we established a threshold of prediction confidence, which allowed us to reduce the fraction of false-positive predictions. We proposed the spatially continuous index of airborne pollen exposure probability, which can be useful for allergy sufferers. The results showed that open-access geodata in Poland can be applied to recognize major local sources of plane pollen.
... With ongoing leaf senescence, Sblue increased rapidly, causing a decrease in Sgreen. The comparison of several colour indices showed that the HUE value provided a more robust metric than the other indices (Mizunuma et al., 2011(Mizunuma et al., , 2014, while the redness index was better for estimating leaf senescence and the greenness index was better for estimating leaf development events (Zhao et al., 2012). In this study site, the Sred and HUE values did not show an evident pattern associated with the phenological events. ...
Vegetation phenology is an important indicator of environmental change and strongly connected to forest ecosystem productivity change. This study aimed to analyse the pattern of phenological variations derived from digital imagery for the interpretation of ecosystem productivity. For 2014, 2015 and 2016, the seasonal phenological development of savanna was analysed by using towerbased imagery from a digital camera. The green excess index (GEI) was the best at representing the phenological transition dates (PTDs) and useful for investigating the gross primary production (GPP) in the savanna ecosystem. There was a significant correlation between the monthly pattern of the strength of green (Sgreen), green excess index (GEI) and vegetation contrast index (VCI) and GPP throughout the year. Additionally, the annual pattern of colour indices had significant relationship (p < 0.05) with GPP but this was not seasonal. The air temperature (air T) and soil temperature (soil T) were strongly significantly correlated (p < 0.001) with the start of growing season (SGS) and caused the advance in green-up and the timing of the start of the growing season in 2014 and 2016. The short growing season length (GSL) had an impact on the productivity. The colour indices from the digital camera images not only provided the phenological pattern of a forest canopy but also revealed the forest ecosystem productivity by showing the response to environmental factors. Our results indicate that daily continuous digital camera images might be useful for ecologists to use as a tool for future prediction of the long-term phenological modelling.
... At present, most camera-based phenology studies only focus on a single region of interest (ROI) in the image to track vegetation phenology and consider this region as a reference for the average behavior of the entire ecosystem [19][20][21]. Few studies have used multiple ROIs to evaluate phenological differences between different species or even different individuals in the same canopy image [22]. Besides, there is no unified conclusion on the optimal color index for monitoring the phenological change of canopy [23]. ...
Background
Forest canopies are highly sensitive to their growth, health, and climate change. The study aims to obtain time sequence images in mix foresters using a near-earth remote sensing method to track the seasonal variation in the color index and select the optimal color index. Three different regions of interest (RIOs) were defined and six color indexes (GRVI, HUE, GGR, RCC, GCC, and GEI) were calculated to analyze the microenvironment difference. The key phenological phase was identified using the double logistic model and the derivative method, and the phenology forecast of color indexes was performed based on the long short-term memory (LSTM) model.
Results
The results showed that the same color index in different RIOs and different color indexes in the same RIO present a slight difference in the days of growth and the days corresponding to the peak value, exhibiting different phenological phases; the mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) of the LSTM model was 0.0016, 0.0405, 0.0334, and 12.55%, respectively, indicating that this model has a good forecast effect.
Conclusions
In different areas of the same forest, differences in the micro-ecological environment in the canopies were prevalent, with their internal growth mechanism being affected by different cultivation ways and the external environment. Besides, the optimal color index also varies with species in phenological response, that is, different color indexes are used for different forests. With the data of color indexes as the training set and forecast set, the feasibility of the LSTM model in phenology forecast is verified.
... The authors preliminarily concluded that the RGB digital camera-derived Vegetation Indices are phenotypically associated to the NAPP. In the same way, Mizunuma et al. (2014) recommend using hue as a colour index for tracking different stages of leaf development. However, it should be noted that Stevens et al. (2007) do not recommend utilizing indices derived from the conversion to other values of color in the space as HSB (hue, saturation and brightness) because of their high inaccuracy. ...
The present study is aimed at quantifying and comparing the net aerial primary productivity (NAPP) of two alfalfa varieties (Medical sativa L.) by determining the Radiation Use Efficiency (Ɛ) for each variety, estimating the NAPP though the Red Vegetation Index and relating it to the quantified NAPP. Significant differences between the individual NAPP of each variety were not found: G969 = 1564 g dm m-2 and M901 = 1636 g dm m-2 (T = 0.92; p>0.05). The Ɛ of the G969 was 0.56 g Mj-1 while that of M901 was 0.58 g Mj-1. Significant direct relationships between the quantified NAPP and that calculated using the Red Vegetation Index were found. The models obtained were: NAPP G969 = 506.06x-343.25 (R 2 = 0.88; p<0.001) and NAPP M901 = 420.28x + 37.82 (R 2 = 0.98; p<0.001). The Ɛ values of the alfalfa varieties under study, determined at local level, reduce uncertainty when generating predictive models of productivity. The NAPP of alfalfa varieties can be non-destructively predicted using the Red Vegetation Index obtained by a reflex RGB digital camera.
... The authors preliminarily concluded that the RGB digital camera-derived Vegetation Indices are phenotypically associated to the NAPP. In the same way, Mizunuma et al. (2014) recommend using hue as a colour index for tracking different stages of leaf development. However, it should be noted that Stevens et al. (2007) do not recommend utilizing indices derived from the conversion to other values of color in the space as HSB (hue, saturation and brightness) because of their high inaccuracy. ...
The present study is aimed at quantifying and comparing the net aerial primary productivity (NAPP) of two alfalfa varieties (Medical sativa L.) by determining the Radiation Use Efficiency (Ɛ) for each variety, estimating the NAPP though the Red Vegetation Index and relating it to the quantified NAPP. Significant differences between the individual NAPP of each variety were not found: G969 = 1564 g dm m-2 and M901 = 1636 g dm m-2 (T = 0.92; p>0.05). The Ɛ of the G969 was 0.56 g Mj-1 while that of M901 was 0.58 g Mj-1. Significant direct relationships between the quantified NAPP and that calculated using the Red Vegetation Index were found. The models obtained were: NAPPG969 = 506.06x – 343.25 (R2 = 0.88; p<0.001) and NAPPM901 = 420.28x + 37.82 (R2 = 0.98; p<0.001). The Ɛ values of the alfalfa varieties under study, determined at local level, reduce uncertainty when generating predictive models of productivity. The NAPP of alfalfa varieties can be non-destructively predicted using the Red Vegetation Index obtained by a reflex RGB digital camera. Keywords: radiation use efficiency, digital camera, canopy reflectance, RGB indices
... Color information detected by the DSLR was in RGB color space, which was not intuitive, and its process of distinguishing color aberrations was nonlinear (Chien and Tsai, 2014), so RGB color space was not suitable for color recognition. HSL color space, however, was more intuitive and consistent with human visual characteristics (Mizunuma et al., 2014;Lin et al., 2015;Qiao et al., 2016) and is typically used in color recognition applications. In the HSL color mode, H [hue, H∈ (0 • , 360 • )] was defined to characterize the type of color, and the value of hue could be used to represent the color warmth-the hue value looped from 0 • to 360 • and defined 0 • and 180 • as the y axis and 90 • and 270 • as the x axis; the chromophore was divided into four quadrants. ...
Understanding phenotypic responses is crucial for predicting and managing the effects of environmental change on native species. Color and display size are typically used to evaluate the utilization value of ornamental plants, which are also important ornamental characters of Lonicera nervosa Maxim. (L. nervosa). However, there is limited documentation of its floral environmental adaptation. The environmental conditions for the development of an organism changes with altitudinal variation. The aim of this research was to find flower trait variability maintenance and the tradeoff among the organs in five different populations of L. nervosa growing at distinct altitudes. We investigated the distribution patterns of floral color, floral display, and biomass tradeoff along a 700-m altitude gradient from 2,950 to 3,650 m. One-way ANOVA analysis was performed to assess the variability of flower traits and floral color across different altitudes. Moreover, correlations and tradeoffs between flowers and vegetative organs were also observed at different altitude ranges. The results indicated that L. nervosa flowers had a strong adaptability along the elevation and divergent altitude-range-specific patterns, which was divided by an altitude breakpoint at around 3,300 m. Below 3,300 m, petal lightness (petal L) decreased, but total floral display area (TFDA), individual floral dry mass (IFDM), and total floral dry mass (TFDM) increased with an increase in altitude. Whereas, above 3,300 m no significant difference was observed in petal L, TFDA, IFDM, and TFDM decreased slightly with an increase in altitude. The responsibility for the selection on floral color at a lower altitude was stronger than that at a higher altitude, while the selection agents on floral biomass had significant effects within the entire altitude range. However, the effects on floral biomass were opposite on both sides of 3,300 m. Thus, floral trait and floral color can be useful indicators for the domestication of horticultural plants and help to evaluate and initiate management and conservation actions.
... Because the RGB DN values stored across the multiple scenes can be strongly affected by the varying illumination conditions (Woebbecke et al., 1995;Richardson et al., 2007), colour indices are calculated in order to diminish this effect and allow consistent time series to be generated . The Green Chromatic Coordinate (GCC, Eq. (2-2)) has been successfully used to track green leaf phenology Klosterman et al., 2014;Moore et al., 2016), but a variety of other indices can potentially be formulated (Mizunuma et al., 2014), dependent upon the objective of the study (Zhao et al., 2012). ...
Vegetation phenology is the study of plant natural life cycle stages. Plant phenological events are related to carbon, energy and water cycles within terrestrial ecosystems, operating from local to global scales. As plant phenology events are highly sensitive to climate fluctuations, the timing of these events has been used as an independent indicator of climate change. The monitoring of forest phenology in a cost-effective manner, at a fine spatial scale and over relatively large areas remains a significant challenge. To address this issue, unmanned aerial vehicles (UAVs) appear to be a potential new platform for forest phenology monitoring. The aim of this research is to assess the potential of UAV data to track the temporal dynamics of spring phenology, from the individual tree to woodland scale, and to cross-compare UAV results against ground and satellite observations, in order to better understand characteristics of UAV data and assess potential for use in validation of satellite-derived phenology. A time series of UAV data were acquired in tandem with an intensive ground campaign during the spring season of 2015, over Hanging Leaves Wood, Northumberland, UK. The radiometric quality of the UAV imagery acquired by two consumer-grade cameras was assessed, in terms of the ability to retrieve reflectance and Normalised Difference Vegetation Index (NDVI), and successfully validated against ground (0.84≤R2≥0.96) and Landsat (0.73≤R2≥0.89) measurements, but only NDVI resulted in stable time series. The start (SOS), middle (MOS) and end (EOS) of spring season dates were estimated at an individual tree-level using UAV time series of NDVI and Green Chromatic Coordinate (GCC), with GCC resulting in a clearer and stronger seasonal signal at a tree crown scale. UAV-derived SOS could be predicted more accurately than MOS and EOS, with an accuracy of less than 1 week for deciduous woodland and within 2 weeks for evergreen. The UAV data were used to map phenological events for individual trees across the whole woodland, demonstrating that contrasting canopy phenological events can occur within the extent of a single Landsat pixel. This accounted for the poor relationships found between UAV- and Landsat-derived phenometrics (R2<0.45) in this study. An opportunity is now available to track very fine scale land surface changes over contiguous vegetation communities, information which could improve characterization of vegetation phenology at multiple scales.
(Permalink to thesis@ http://hdl.handle.net/10443/4131)
... Consumer-grade cameras are usually employed as imaging sensors and the aim is to monitor changes in canopy colour associated with the different plant development stages and/or photosynthetic activity (Toomey et al., 2015). A variety of colour indices can potentially be formulated from the RGB DN values (please see Mizunuma et al. (2014)), including the commonly used GCC (Table 3). ...
Vegetation phenology is the study of recurring plant life cycle stages, seasonality which is linked to many ecosystem processes and is an important proxy of climate and environmental change. Remote sensing has been playing an important and increasing role in the monitoring and assessment of vegetation phenology. The aim of this review is to critically examine key studies related to remote sensing of vegetation phenology, with a special focus on temperate and boreal forests. Specifically, we focus on how the latest ground, near-surface and aerial data have been used to assess the satellite-derived Land Surface Phenology (LSP) metrics and the agreements that has been achieved in the last 15 years. Results demonstrated that the timing of satellite-derived LSP events can be detected, in the best-case scenarios, with a certainty of around half-week for spring metrics (e.g. Day of Year -DOY- of start of growing season) and around one week for autumn metrics (e.g. DOY of end of growing season). With expected shifts in plant phenology averaging <1 day per decade, such LSP uncertainties (in terms of absolute phenological dates) could greatly over- or under-estimate these species-level shifts; but the spatial variation in phenology can be consistently monitored. An increasing number of studies have investigated autumn phenology in the last decade, but autumn phenological dates continue to be more challenging to retrieve and interpret than spring dates. Emerging opportunities to further advance remote sensing of forest phenology is presented that includes synergetic use of multiple orbital sensors and its LSP evaluation with data from new sensors at a ground, near-surface and airborne level; yet traditional ground-based observations will continue to be highly useful to accurately record the timing of species-specific phenological events. This review might provide a guide for planning and managing remote sensing of forest phenology.
... In order to track the seasonality ("phenology") of vegetation activity in different ecosystems, digital cameras have been deployed to record highfrequency images of the canopy at hundreds of research sites around the world (Richardson, 2018). From each image, color-channel information (e.g., RGB [red-green-blue] values of each pixel) are extracted and converted to a suite of "vegetation indices" derived from linear or nonlinear transformations of the RGB or other color spaces (Sonnentag et al., 2012;Mizunuma et al., 2014;Toomey et al., 2015;Nguy-Robertson et al., 2016). These indices have been used to identify the timing of seasonal phenomena such as leaf-out, senescence, and abscission, and to monitor how these phenomena are changing in response to ongoing climatic change (Sonnentag et al., 2012). ...
In this work we define a spatial concordance coefficient for second-order stationary processes. This problem has been widely addressed in a non-spatial context, but here we consider a coefficient that for a fixed spatial lag allows one to compare two spatial sequences along a 45-degree line. The proposed coefficient was explored for the bivariate Mat\'ern and Wendland covariance functions. The asymptotic normality of a sample version of the spatial concordance coefficient for an increasing domain sampling framework was established for the Wendland covariance function. To work with large digital images, we developed a local approach for estimating the concordance that uses local spatial models on non-overlapping windows. Monte Carlo simulations were used to gain additional insights into the asymptotic properties for finite sample sizes. As an illustrative example, we applied this methodology to two similar images of a deciduous forest canopy. The images were recorded with different cameras but similar fields-of-view and within minutes of each other. Our analysis showed that the local approach helped to explain a percentage of the non-spatial concordance and to provided additional information about its decay as a function of the spatial lag.
... All images were collected around noon (11:30-12:00 local time) on sunny or cloudy days. The white balance of camera was set to the sunny mode [30,42,43]. The optical sensor in the camera was a CCD image sensor set with the auto mode of exposure; and all images were stored in JPEG format with a resolution of 4320 × 3240. ...
Near-surface remote sensing is an effective tool for in situ monitoring of canopy phenology, but the uncertainties involved in sensor-types and their deployments are rarely explored. We comprehensively compared three types of sensor (i.e., digital camera, spectroradiometer, and routine radiometer) at different inclination-and azimuth-angles in monitoring canopy phenology of a temperate deciduous forest in Northeast China for three years. The results showed that the greater contribution of understory advanced the middle of spring (MOS) for large inclination-angle of camera and spectroradiometer. The length of growing season estimated by camera from the east direction extended 11 d than that from the north direction in 2015 due to the spatial heterogeneity, but there was no significant difference in 2016 and 2018.The difference infield of view of sensors caused the MOS and the middle of fall, estimated by camera, to lag a week behind those by spectroradiometer and routine radiometer. Overall, the effect of azimuth-angle was greater than that of inclination-angle or sensor-type. Our assessments of the sensor types and their deployments are critical for the long-term accurate monitoring of phenology at the site scale and the regional/global-integration of canopy phenology data.
... An understanding of canopy physiology and its diurnal dynamics is important for evaluating both whole tree performance and ecosystem carbon and water fluxes (Flanagan et al., 2012;Mizunuma et al., 2014;Lin et al., 2015;Bussotti et al., 2018;Grossiord et al., 2018). Daily maximum values carbon assimilation (A) and stomatal conductance (g s ) are generally observed in the morning, followed by midday declines (Koch et al., 1994;Brodribb et al., 2002;Kosugi et al., 2008;Tucci et al., 2010;Yang et al., 2012;Resco de Dios et al., 2017). ...
Tree trunks not only provide physical support for canopy leaves but also supply and store water for transpiration. However, the relationships between trunk hydraulic properties and canopy leaf physiology in tropical trees are not well-understood. In this study we concurrently measured morning and midday canopy leaf photosynthesis (A), stomatal conductance (g s ), and leaf water potentials (Ψ L ) in 40 tropical trees representing 14 species at the beginning of the rainy season in Xishuangbanna, Southwest China. We also measured trunk sapwood capacitance (C), wood density, and sap flux density to assess their association with canopy leaf physiology. Among the 14 studied species, only three and four species did not show a significant midday reduction in A and g s respectively. The diurnally maximum A and g s were significantly positively related to sapwood hydraulic capacitance, maximum sap flux density (midday), and sap flux density at 11:00. Those species with lower wood density and higher C showed a lower reduction in Ψ L at midday, whereas, species with high C, and large values of maximum sap flux density also showed high carbon assimilation at midday. Our results provide new insights into the close coordination between canopy physiology and trunk sapwood hydraulic properties in tropical trees.
... Changes in canopy greenness can be quantified using a variety of colour indices such as the Gcc index, green excess indices (GEI), green red vegetation index (GRVI) or colour space transformation such as hue saturation value (HSV) colour space (Andresen et al., 2018;Nasahara and Nagai, 2015;Mizunuma et al., 2014;Richardson et al., 2009;Woebbecke et al., 1995). Time series of canopy greenness provide information about key processes of the vegetation at the canopy level. ...
The presence or absence of leaves within plant canopies exert a strong influence on the carbon, water and energy balance of ecosystems. Identifying key changes in the timing of leaf elongation and senescence during the year can help to understand the sensitivity of different plant functional types to changes in temperature. When recorded over many years these data can provide information on the response of ecosystems to long-term changes in climate. The installation of digital cameras that take images at regular intervals of plant canopies across the Integrated Carbon Observation System ecosystem stations will provide a reliable and important record of variations in canopy state, colour and the timing of key phenological events. Here, we detail the procedure for the implementation of cameras on Integrated Carbon Observation System flux towers and how these images will help us understand the impact of leaf phenology and ecosystem function , distinguish changes in canopy structure from leaf physiology and at larger scales will assist in the validation of (future) remote sensing products. These data will help us improve the representation of phenological responses to climatic variability across Integrated Carbon Observation System stations and the terrestrial biosphere through the improvement of model algorithms and the provision of validation datasets.
... Seasonal changes in canopy color indices (such as green chromatic coordinate, G cc ), extracted from the region of interest (ROI) in repeat canopy photography, were closely related to seasonal changes in phenology of deciduous forests, such as bud-break, leaf expansion, and leaf abscission (Woebbecke et al. 1995, Richardson et al. 2009, Sakamoto et al. 2012, Ide and Oguma 2013, Keenan et al. 2014, Filippa et al. 2015, Moore et al. 2016, Nagai et al. 2016). In addition, G cc can track plant physiological characters, such as leaf area index (LAI; Ryu et al. 2012, Liu et al. 2015, leaf chlorophyll concentration (Keenan et al. 2014, Yang 2014, and gross primary production (GPP; Ahrends et al. 2009, Mizunuma et al. 2014, Wingate et al. 2015, even though mismatches between G cc and chlorophyll concentration may exist (Yang et al. 2017). Based on the widespread application of canopy repeat imagery, regional camera-based phenological networks were initiated and extended, such as the PhenoCam Network (Richardson et al. 2009(Richardson et al. , 2018, the National Ecological Observatory Network (Kampe et al. 2010), EUROPheno Network (Wingate et al. 2015), Phenological Eyes Network (PEN; Nasahara , and Australian Phenocam Network (Moore et al. 2016). ...
Camera‐based observation of forest canopies allows for low‐cost, continuous, high temporal‐spatial resolutions of plant phenology and seasonality of functional traits. In this study, we extracted canopy color index (green chromatic coordinate, Gcc) from the time‐series canopy images provided by a digital camera in a deciduous forest in Massachusetts, USA. We also measured leaf‐level photosynthetic activities and leaf area index (LAI) development in the field during the growing season, and corresponding leaf chlorophyll concentrations in the laboratory. We used the Bayesian change point (BCP) approach to analyze Gcc. Our results showed that (1) the date of starting decline of LAI (DOY 263), defined as the start of senescence, could be mathematically identified from the autumn Gcc pattern by analyzing change points of the Gcc curve, and Gcc is highly correlated with LAI after the first change point when LAI was decreasing (R2 = 0.88, LAI < 2.5 m2/m2); (2) the second change point of Gcc (DOY 289) started a more rapid decline of Gcc when chlorophyll concentration and photosynthesis rates were relatively low (13.4 ± 10.0% and 23.7 ± 13.4% of their maximum values, respectively) and continuously reducing; and (3) the third change point of Gcc (DOY 295) marked the end of growing season, defined by the termination of photosynthetic activities, two weeks earlier than the end of Gcc curve decline. Our results suggested that with the change point analysis, camera‐based phenology observation can effectively quantify the dynamic pattern of the start of senescence (with declining LAI) and the end of senescence (when photosynthetic activities terminated) in the deciduous forest.
... L 为土壤调节因子 L indicated the soil adjusted factor. [44][45] 和欧洲 [46] 碳水通量网中广泛应用，并且提高了公用陆面模式物候模块的模拟精度 [47] 。但该技术在我国还处于起步阶段，而且大多应用于农作物和草地中，在森林中应用非常少 [48][49] 。 用于监测冠层物候变化的最优颜色指数尚无统一的定论，以相对绿度指数(GCC)和绝对绿度指数 (GEI)应用最多。GCC 对光照条件不敏感，有效降低了相机视角和感光度对图像的影响 [50] ，而且 GCC 和 GEI 与 GPP 的关系紧密程度接近 [51] ；但周磊等 [49] 认为，GEI 比 GCC 更能准确表征草地季相变化。 Mizunuma 等 [52] 推荐色相为最优颜色指数；然而 Keenan 等 [53] 发现色相与 GPP 并不相关，且更容易受白 平衡影响。相对红度指数更适合估算常绿森林落叶期 [54] 。值得注意的是，红外相机由于存在红外波段 而可以计算基于红外相机的 NDVI，能够与 GCC 相互补足 [20] 。不同研究的差异可能是植被类型、相机型 号和设置(如白平衡)以及安装方式的差异等多种因素造成的。 ...
Near-surface remote sensing technique is an important way to in-situ monitor forest phenology and a robust tool for scaling of the phenology with a high temporal resolution and moderate spatial coverage. In this review, we first reviewed the methods of near-surface remote sensing with three major optical sensors (i.e., radiometer, spectrometer, and digital camera) for observing forest phenology. Second, we analyzed sources of uncertainty from distinguishing the phenophases by using the data obtained at the Maoershan flux site in the Chinese temperate forest as a case study, and found that the error was mainly attributed to the extracting method. Third, we analyzed the linkage of near-surface remote sensing with other methods and its intrinsic issues. Finally, we proposed four research priorities in this field: 1) linking optical (or canopy structural) phenology with functional phenology (physiological and ecological processes); 2) integrating the regional networks of canopy phenology to achieve global networking observation and data sharing of canopy phenology; 3) integrating multi-source and multi-scale phenological data with the help of near-surface remote sensing; 4) developing phenology models based on near-surface remote sensing in order to improve the phenology simulation in the dynamic global vegetation models.
... B+G+R 3 [35,40,84] Can describe prominent visual changes in foliage (e.g., white flowering of Prunus dulcis). RGB conversion to Hue, Saturation and Value; HSV See [85] [85][86][87][88] Alternative colour space for describing canopy changes. Compared to RGB-derived indices, can be more effective and robust as a proxy for leaf development. ...
Most recent studies relating to the classification of vegetation species on the individual level use cutting-edge sensors and follow a data-driven approach, aimed at maximizing classification accuracy within a relatively small allocated area of optimal conditions. However, this approach does not incorporate cost-benefit considerations or the ability of applying the chosen methodology for applied mapping over larger areas with higher natural heterogeneity. In this study, we present a phenology-based cost-effective approach for optimizing the number and timing of unmanned aerial vehicle (UAV) imagery acquisition, based on a priori near-surface observations. A ground-placed camera was used in order to generate annual time series of nine spectral indices and three color conversions (red, green and blue to hue, saturation and value) in four different East Mediterranean sites that represent different environmental conditions. After outliers’ removal, the time series dataset represented 1852 individuals of 12 common vegetation species and annual herbaceous patches. A feature selection process was used for identifying the optimal dates for species classification in every site. The feature selection can be designed for various objectives, e.g., optimization of overall classification, discrimination between two species, or discrimination of one species from all others. In order to evaluate the a priori findings, a UAV was flown for acquiring five overhead multiband orthomosaics (five bands in the visible-near infrared range based on the five optimal dates identified in the feature selection of the near-surface time series of the previous year. An object-based classification methodology was used for the discrimination of 976 individuals of nine species and annual herbaceous patches in the UAV imagery, and resulted in an average overall accuracy of 85% and an average Kappa coefficient of 0.82. This cost-effective approach has high potential for detailed vegetation mapping, regarding the accessibility of UAV-produced time series, compared to hyper-spectral imagery with high spatial resolution which is more expensive and involves great difficulties in implementation over large areas.
... Several vegetation indices were designed to minimize these influences (Gitelson et al. 2002;Woebbecke et al. 1995). Studies by Mizunuma et al. (2014), 2013) have shown that in the absence of a calibration procedure, the green-based indices provided a poor estimate of canopy phenology when comparing camera models. These studies suggest that alternative parameters such as hue may provide a better estimate, as they remove differences in the optical properties between camera models. ...
Digital cameras can collect quantitative leaf data, such as chlorophyll content and leaf area index (LAI), because they act as a simple broadband radiometer. However, a cross-calibration between cameras is needed for the purpose of extracting vegetation information from various image repositories. The objective of this study was to examine the variation between multiple consumer-grade camera types – single reflex lens (SLR), point-and-shoot, and cellphone cameras – for the purpose of collecting reliable quantitative data when monitoring vegetation. The specific objectives were to: 1) identify the optimal light conditions for the calibration procedure, 2) determine the variability of exposure value (EV)-corrected calibrated digital numbers (cDNev) values among eight consumer-grade digital cameras, and 3) compare the cDNev values with the raw digital numbers (DN), exposure-adjusted digital numbers (DNev), and calibrated digital numbers (cDN) as these latter three components are easier to compute. This study demonstrated that light intensity was important for calibrating cameras to ensure sensor saturation, and that an improper white-balance setting can negatively impact data collection. In one experiment, the coefficient of variation (CV) between the eight cameras examined in the study was reduced from 29% using raw DN to 16% using cDNev values. Likewise, the root mean square error in estimating leaf chlorophyll-a using a common vegetation index for digital camera, excess green index (EGI), was reduced from 131 to 96 mg g−2. However, for both experiments, there was only a weak statistical difference between cDNev and DNev, indicating that exposure information was the most useful in minimizing the differences between cameras. Although digital cameras are not nearly as accurate as specialized remote-sensing equipment, they do offer the potential for greater collection opportunities. This study demonstrates the potential of using consumer-grade digital cameras to derive quantitative information from citizen science projects.
Background and Aims. Rapid, large-scale monitoring is critical to understanding spatiotemporal plant stress dynamics, but current physiological stress markers are costly, destructive, and time-consuming. This study aimed to evaluate the potential of machine learning to non-destructively predict leaf betalains-yellow to reddish pigments unique to Caryophyllales species-for the first time, and to explore betalains' intra-individual variation on a clonal species and its role to respond to stressful periods.
Methods. We characterized the betalainic profile of an invasive clonal plant for the first time, Carpobrotus edulis (L.) NE Br. (the hottentot-cape cape fig), via HPLC. We measured multiple stress markers over a year, including betalain content using our optimized method, where the species is spreading. Additionally, 3,735 digital images at the leaf level were taken. Machine learning regression algorithms were trained to predict betalain accumulation from digital images, outperforming classic spectroradiometer measurements.
Key Results. Betalain content increased sharply in non-reproductive ramets during extreme abiotic conditions in summer and during senescence in reproductive ramets. The stress markers revealed a strong intra-individual functional mosaic, underscoring the importance of spatiotemporal dimensions in stress tolerance.
Conclusions. We developed a scalable, non-destructive tool for betalain research that integrates digital imaging with machine learning. This approach opens new possibilities for understanding spatiotemporal stress responses, particularly in clonal plant systems, using artificial intelligence.
Anthocyanins are widely found in plants and have significant functions. The accurate detection and quantitative assessment of anthocyanin content are essential to assess its functions. The anthocyanin content in plant tissues is typically quantified by wet chemistry and spectroscopic techniques. However, these methods are time-consuming, labor-intensive, tedious, expensive, destructive, or require expensive equipment. Digital photography is a fast, economical, efficient, reliable, and non-invasive method for estimating plant pigment content. This study examined the anthocyanin content of Rosa chinensis petals using digital images, a back-propagation neural network (BPNN), and the random forest (RF) algorithm. The objective was to determine whether using RGB indices and BPNN and RF algorithms to accurately predict the anthocyanin content of R. chinensis petals is feasible. The anthocyanin content ranged from 0.832 to 4.549 µmol g⁻¹ for 168 samples. Most RGB indices were strongly correlated with the anthocyanin content. The coefficient of determination (R²) and the ratio of performance to deviation (RPD) of the BPNN and RF models exceeded 0.75 and 2.00, respectively, indicating the high accuracy of both models in predicting the anthocyanin content of R. chinensis petals using RGB indices. The RF model had higher R² and RPD values, and lower root mean square error (RMSE) and mean absolute error (MAE) values than the BPNN, indicating that it outperformed the BPNN model. This study provides an alternative method for determining the anthocyanin content of flowers.
Anthocyanins are precious industrial raw materials. Purple corn is rich in anthocyanins, with large variation in their content between organs. It is imperative to find a rapid and non-destructive method to determine the anthocyanin content in purple corn. To this end, a field experiment with ten purple corn hybrids was conducted, collecting plant images using a digital camera and determining the anthocyanin content of different organ types. The average values of red (R), green (G) and blue (B) in the images were extracted. The color indices derived from RGB arithmetic operations were applied in establishing a model for estimation of the anthocyanin content. The results showed that the specific color index varied with the organ type in purple corn, i.e., ACCR for the grains, BRT for the cobs, ACCB for the husks, R for the stems, ACCB for the sheaths and BRT for the laminae, respectively. Linear models of the relationship between the color indices and anthocyanin content for different organs were established with R² falling in the range of 0.64–0.94. The predictive accuracy of the linear models, assessed according to the NRMSE, was validated using a sample size of 2:1. The average NRMSE value was 11.68% in the grains, 13.66% in the cobs, 8.90% in the husks, 27.20% in the stems, 7.90% in the sheaths and 15.83% in the laminae, respectively, all less than 30%, indicating that the accuracy and stability of the model was trustworthy and reliable. In conclusion, this study provided a new method for rapid, non-destructive prediction of anthocyanin-rich organs in purple corn.
Anthocyanins are a group of polyphenolic pigments that are ubiquitously found in the plant kingdom. These compounds have attracted attention in recent years due to their beneficial effects.
The color of plant leaves can be assessed qualitatively by color charts or after processing of digital images. This pilot study employed a novel pocket-sized sensor to obtain the color of plant leaves. In order to assess its performance, a color-dependent parameter (SPAD index) was used as the dependent variable, since there is a strong correlation between SPAD index and greenness of plant leaves. A total of 1,872 fresh and intact leaves from 13 crops were analyzed using a SPAD-502 m and scanned using the Nix™ Pro color sensor. The color was assessed via RGB and CIELab systems. The full dataset was divided into calibration (70% of data) and validation (30% of data). For each crop and color pattern, multiple linear regression (MLR) analysis and multivariate modeling [least absolute shrinkage and selection operator (LASSO), and elastic net (ENET) regression] were employed and compared. The obtained MLR equations and multivariate models were then tested using the validation dataset based on r, R², root mean squared error (RMSE), and mean absolute error (MAE). In both RGB and CIELab color systems, the Nix™ Pro color sensor was able to differentiate crops, and the SPAD indices were successfully predicted, mainly for mango, quinoa, peach, pear, and rice crops. Validation results indicated that ENET performed best in most crops (e.g., coffee, corn, mango, pear, rice, and soy) and very close to MLR in bean, grape, peach, and quinoa. The correlation between SPAD and greenness is crop-dependent. Overall, the Nix™ Pro color sensor was a fast, sensible and an easy way to obtain leaf color directly in the field, constituting a reliable alternative to digital camera imagery and associated image processing.
The segmentation of digital images in red, green and blue (RGB) components is a low-cost method for monitoring leaf chlorophyll concentrations and seedling quality. The two congeneric species, Cariniana legalis and C. estrellensis, are distinguished based on differences in bark texture and the colour of their new leaves. We compared indices based on leaf colour segmentation in RGB to predict total chlorophyll concentrations (Chlt) in the leaves of seedlings of these two species. Mature leaves were digitalised in a flatbed scanner and segmented in red (R), green (G) and blue (B). The relationships between the three RGB indices and Chlt were tested. Additionally, we calculated the anthocyanin content-chroma basic (ACcb). The mean value of ACcb was significantly higher in C. legalis than in C. estrellensis, demonstrating a higher anthocyanin concentration in C. legalis leaves. Based on the highest coefficients of determination (R²) and lowest prediction errors (PE), for all indices, the best results were obtained for C. estrellensis. The presence of anthocyanins in the leaves of C. legalis and the limitation of the RGB colour segmentation indices for separating all leaf pigments might be the main causes of the differences in Chlt prediction in the leaves of these two congeneric tree species.
In this work we define a spatial concordance coefficient for second-order stationary processes. This problem has been widely addressed in a non-spatial context, but here we consider a coefficient that for a fixed spatial lag allows one to compare two spatial sequences along a 45°line. The proposed coefficient was explored for the bivariate Matérn and Wendland covariance functions. The asymptotic normality of a sample version of the spatial concordance coefficient for an increasing domain sampling framework was established for the Wendland covariance function. To work with large digital images, we developed a local approach for estimating the concordance that uses local spatial models on non-overlapping windows. Monte Carlo simulations were used to gain additional insights into the asymptotic properties for finite sample sizes. As an illustrative example, we applied this methodology to two similar images of a deciduous forest canopy. The images were recorded with different cameras but similar fields-of-view and within minutes of each other. Our analysis showed that the local approach helped to explain a percentage of the non-spatial concordance and provided additional information about its decay as a function of the spatial lag.
Autumn leaf phenology and its color brightness provide valuable information for managing forest carbon cycles and cultural ecosystem services. Digital repeat photography has provided standard phenological data, but the methodologies for detecting autumn leaf coloring have various strengths and weaknesses. We assessed the accuracy, sensitivity, and uncertainty of various model and color index combinations for detecting autumn leaf coloring. Then we identified the most robust and sensitive methods, using digital repeat photography data from Japanese alpine vegetation. For determining autumn leaf color duration, quadratic or multinomial discriminant analysis using RGB digital numbers had the highest accuracy (hit ratio > 0.7). For determining the peak day of autumn leaf color and its color brightness, we compared uncertainty of methodologies by randomly resampling 80% of the data 20 times to mimic observation errors (e.g., due to heavy rain). The spline-fitted red/green reflectance ratio (RGR) and visible atmospherically resistant index (VARI) proved robust for detecting the peak day (median SD = 1.25). Uncertainty of color brightness was also low when using VARI fitted by a double logistic model for both red and yellow leaves (median coefficient of variation = 1.03). These two indexes are stable despite atmospheric effects, which may result in robustness to daily variation in conditions (e.g., fog). We compared sensitivity of leaf color brightness: RGR and excess red (ExR) fitted by a double logistic model had the highest sensitivity to red and yellow leaves exceeding the average of other combinations by 26% and 88% in median values, respectively. The small denominator or lack of a denominator of these indexes increases the sensitivity to red or yellow. Our results demonstrate the averaged accuracy, sensitivity, and robustness of each methodology among our research sites with different camera observations. These methods should help in utilizing hidden big data from web cameras or past photos that were not intended for scientific research to properly assess autumn leaf phenology and its color brightness.
We evaluated the usability of the red (R), green (G), and blue (B) digital numbers (DNRGB) extracted from daily phenological images of a tropical rainforest in Malaysian Borneo. We examined temporal patterns in the proportions of DNR, DNG, and DNB as percentages of total DN (denoted as %R, %G and %B), in the hue, saturation, and lightness values in the HSL color model, and in a green excess index (GEI) of the whole canopy and of individual trees for 2 years. We also examined temporal patterns in the proportions of the red, green, and blue reflectance of the whole canopy surface as percentages of total reflectance (denoted as %ref_R, %ref_G, and %ref_B), and vegetation indices (the normalized-difference vegetation index, enhanced vegetation index, and green–red vegetation index) of the whole canopy by using daily measurements from quantum sensors. The temporal patterns of %RGB and saturation of individual trees revealed the characteristics of tree phenology caused by flowering, coloring, and leaf flushing. In contrast, those of the whole canopy did not, nor did those of %ref_R, %ref_G, or %ref_B, or the vegetation indices. The temporal patterns of GEI, however, could detect differences among individual trees caused by leaf flushing and coloring. Our results show the importance of installing multiple time-lapse digital cameras in tropical rainforests to accurately evaluate the sensitivity of tree phenology to meteorological and climatic changes. However, more work needs to be done to adequately describe whole-canopy changes.
Endemic species are often replaced by plantings from non-local areas in new suburbs in the developed world. Does this lead to colour changes? This paper compares colour in the leaves of exotic trees planted in suburbs to that of endemics in the Southwest Australian Floristic Region. Colour in plant parts was assessed by the Natural Colour System of Sweden, which enabled quantitative comparison between species. Hue, chromaticness, percentage yellow, blackness, whiteness, luminescence, and visual lightness were determined. The leaves of Australian trees were less chromatic and darker than exotic trees, suggesting that colour changes are occurring with suburbanisation in this region.
Plant phenological development is orchestrated through subtle changes in photoperiod, temperature, soil moisture and nutrient availability. Presently, the exact timing of plant development stages and their response to climate and management practices are crudely represented in land surface models. As visual observations of phenology are laborious, there is a need to supplement long-term observations with automated techniques such as those provided by digital repeat photography at high temporal and spatial resolution. We present the first synthesis from a growing observational network of digital cameras installed on towers across Europe above deciduous and evergreen forests, grasslands and croplands, where vegetation and atmosphere CO2 fluxes are measured continuously. Using colour indices from digital images and using piecewise regression analysis of time-series, we explored whether key changes in canopy phenology could be detected automatically across different land use types in the network. The piecewise regression approach could capture the start and end of the growing season, in addition to identifying striking changes in colour signals caused by flowering and management practices such as mowing. Exploring the dates of green up and senescence of deciduous forests extracted by the piecewise regression approach against dates estimated from visual observations we found that these phenological events could be detected adequately (RMSE
the purpose of understanding pat-terns and processes controlling carbon budgets across a broad range of scales, explicit activities to assess the impact of phenol-ogy on ecosystem carbon bal-ance are still somewhat lacking within the carbon cycle commu-nity. The reasons are clear: long-term observations, otherwise called 'monitoring' are not popu-lar with those that sponsor re-search in this area; three or five year projects are the norm, when in practice much longer records are required to detect long-term trends and their rela-tionships to climatic drivers. There is however, evidence for a shift in attitudes. Keeling's meas-urements of atmospheric CO 2 concentrations, that began in 1958, are an outstanding exam-ple of the value long-term moni-toring represents in the context of a changing world (Nisbet, 2007). Moreover, continuous eddy covariance measurements of CO 2 fluxes began in the early 1990s at a handful of sites. Every year, more and more sites have been added to FLUXNET, and many of these are now providing useful long term data not only with regard to spatial patterns of carbon uptake and release, but also in relation to the influence of phenology on carbon seques-tration. One example of a synergy be-tween phenology and flux moni-toring networks in Europe has Why observe phenology within FLUXNET? Phenology is the study of the timing of lifecycle events, espe-cially as influenced by the sea-sons and by the changes in weather patterns from year to year. The oldest phenological records, observations of cherry flowering at the Royal Court in Kyoto date back to 705 AD, and are still maintained to this day across Japan where the Japanese Meteorological Agency use these data to provide weekly forecast maps of expected blooming dates (http: Marsham, the father of modern phenological recording, was a wealthy landowner and amateur naturalist who recorded "Indications of spring" in Nor-folk, England, beginning in 1736. His family maintained these re-cords until the 1950s. In the modern era, phenology has gained a new impetus, as people realize that such records, if sus-tained over many years, can reveal how plants and animals respond to climate change. Moreover, phenological events such as the spring leaf-out and the autumn fall exert a strong control on both spatial and tem-poral patterns of the carbon cycle. Phenology also influences hydrologic processes, as spring leaf-out is accompanied by a marked increase in evapotranspi-ration, and nutrient cycling as autumn senescence results in a flush of fresh litter (nutrient) input to the forest floor. Phenology is a robust integrator of the effects of climate change on natural systems (Schwartz et al., 2006; IPCC 2007), and it is recognized that improved moni-toring of phenology on local-to-continental scales is needed. Historically, phenological obser-vations were a pastime of ama-teur naturalists (e.g. the Mar-sham family) and reliable records were often dependent on the skills and effort of the observer. The increased demand for inter-national co-operation and stan-dardisation in this area led to the creation of many large-scale phenological monitoring net-works such as the International Phenology Garden (IPG) pro-gram (http: (established in 1998) as well as the recently-established USA-National Phenology Network (U S A -N P N) (http://www.usanpn.org) and associated regional networks (e.g., http://www.nerpn.org). These networks have focused on developing standardized proto-cols for phenological observa-tions, and ensuring overlap be-tween plant species found across locations. Although there are obvious advantages in creating explicit linkages between these
Phenological metrics are of potential value as direct indicators of climate change. Usually they are obtained via either satellite imaging or ground based manual measurements; both are bespoke and therefore costly and have problems associated with scale and quality. An increase in the use of camera networks for monitoring infrastructure offers a means of obtaining images for use in phenological studies, where the only necessary outlay would be for data transfer, storage, processing and display. Here a pilot study is described that uses image data from a traffic monitoring network to demonstrate that it is possible to obtain usable information from the data captured. There are several challenges in using this network of cameras for automatic extraction of phenological metrics, not least, the low quality of the images and frequent camera motion. Although questions remain to be answered concerning the optimal employment of these cameras, this work illustrates that, in principle, image data from camera networks such as these could be used as a means of tracking environmental change in a low cost, highly automated and scalable manner that would require little human involvement.
A two-band digital imaging system-one band for the visible red band (RED, 630-670 nm) and the other for the near infrared band (NIR, 820-900 nm)- was devised and positioned at a height of 12 m above a rice field of 300 m 2 in area during the 2007 growing season. The imaging system automatically logged bird'seye view images at 10-min intervals from 0800-1600 every day. Radiometric corrections for the pairs of two-band images were done using solar irradiance sensors and preceding calibrations to calculate daily band-reflectance and the normalized difference vegetation index (NDVI) values for 9 plots of rice plants, with 3 levels of planting density and basal fertilization. The daily-averaged reflectance values in the RED and the NIR bands showed different but smooth seasonal changing patterns according to the growth of plants. At the maximum tiller number and the panicle formation stages, the RED and NIR reflectance values had correlation coefficients (r) of 0.79 and 0.81 with above-ground nitrogen absorption per unit land area (NA, g m -2), respectively, whereas the NDVI using the two band reflectance values showed r-value of -0.13. An empirically derived equation for the NA using two band reflectance values showed r-value of 0.96 and a root mean square of error (RMSE) 0.5 g m -2 (10% of the mean observed NA) in the estimation for the original (not validated) data set acquired at the maximum tiller number and the panicle formation stages. The results indicated that reflectance observation in the RED and NIR bands using the digital imaging system was potentially effective for assessing rice growth.
Digital control of color television monitors—in particular, via frame buffers—has added precise control of a large subset of human colorspace to the capabilities of computer graphics. This subset is the gamut of colors spanned by the red, green, and blue (RGB) electron guns exciting their respective phosphors. It is called the RGB monitor gamut. Full-blown color theory is a quite complex subject involving physics, psychology, and physiology, but restriction to the RGB monitor gamut simplifies matters substantially. It is linear, for example, and admits to familiar spatial representations. This paper presents a set of alternative models of the RGB monitor gamut based on the perceptual variables hue (H), saturation (S), and value (V) or brightness (L). Algorithms for transforming between these models are derived. Particular emphasis is placed on an RGB to HSV non-trigonometric pair of transforms which have been used successfully for about four years in frame buffer painting programs. These are fast, accurate, and adequate in many applications. Computationally more difficult transform pairs are sometimes necessary, however. Guidelines for choosing among the models are provided. Psychophysical corrections are described within the context of the definitions established by the NTSC (National Television Standards Committee).
Background: To understand how forests and woodland respond to global climate change, phenological observations are being made at a number of sites worldwide. Recently, digital cameras have been deployed as part of the existing network of ecosystem CO2 flux towers to provide a time-series of canopy images, and various numerical indices have so far been used by different authors.Aims: To identify which are the most effective colour indices to calculate from the signals extracted from digital cameras, in order to provide recommendations to the scientific community.Methods: Sample images of a Japanese beech (Fagus crenata) forest on Mt. Tsukuba (Japan) were used to define and calculate 12 colour signals and vegetation indices.Results: Although the strength of green signal and green excess index were reliable indicators for estimating foliage growth period, the indices were susceptible to low-visibility weather conditions and distance from the camera. Hue provided a robust metric, showing much less scatter during the vegetative period and a good indication of spring bud break. The bud break dates derived from the indices were slightly earlier than those assessed by visual observation, while the abscission dates were later.Conclusions: We propose that of all the candidate colour indices, hue is the most promising for the detection of bud break as it was least affected by atmospheric conditions.
There are thousands of outdoor webcams which offer live images freely over the Internet. We report on methods for discovering and organizing this already existing and massively distributed global sensor, and argue that it provides an interesting alternative to satellite imagery for global-scale remote sensing applications. In particular, we characterize the live imaging capabilities that are freely available as of the summer of 2009 in terms of the spatial distribution of the cameras, their update rate, and characteristics of the scene in view. We offer algorithms that exploit the fact that webcams are typically static to simplify the tasks of inferring relevant environmental and weather variables directly from image data. Finally, we show that organizing and exploiting the large, ad-hoc, set of cameras attached to the web can dramatically increase the data available for studying particular problems in phenology.
The environmental constraints to agriculture imply that nitrogen (N) fertilizer management should be adjusted to crop N requirements determined by target yields. Nowadays for environmental and economical reasons target yield of farmers can be lower than the potential crop yields as permitted by soil and climatic conditions. So it is important to provide farmers crop N status diagnostic tools in order to decide the rate and the timing of N fertilizer applications. Theory on crop N uptake and allocation allows the determination of a diagnostic tool, the Nitrogen Nutrition Index, based on the determination of the critical N dilution curve for each crop species considered. During the vegetative growth period of all the crop species studied, including C3 and C4 species and monocots and dicots, plant N concentration decreases monotonically as crop grows because of (i) the ontogenetic decline in leaf area per unit of plant mass, and (ii) the remobilisation of N from shaded leaves at the bottom of the canopy to well illuminated leaves at the top. NNI appears then as an indicator well connected with the physiological regulation of N uptake at canopy level. So this indicator can be used as the basis for determination of crop N nutrition status, and then for decision making on the necessity of an N application for achieving target yield. Nevertheless despite its high physiological relevance, NNI cannot be used directly in farm conditions because its determination is very time consuming. So it is necessary to develop indirect methods for NNI estimation through more operational procedures. Several methods have been proposed in literature, such as nitrate concentration in sap or chlorophyll meter. But the calibration or validation of these methods with NNI have not been always made and, when they have been, they did not give univocal relationships, showing a strong dependence of the relationship with cultivar and environment, that limits considerably the relevance of such diagnostic tools in a large range of situations. Easier to use is the indirect estimation of crop NNI by remote sensing measurements. This method allows the estimation of both actual crop mass, through LAI estimation and crop N content, through crop chlorophyll content. The possibility to have repeated estimations of crop NNI during the period of vegetative growth would allow a dynamic diagnostic tool of crop N status. The coupling of indirect measurements of crop N status with dynamic models of crop growth and development should allow a very promising method for crop N diagnostics for decision tools in N fertilization.
The terrestrial carbon sink has been large in recent decades, but its size and location remain uncertain. Using forest inventory
data and long-term ecosystem carbon studies, we estimate a total forest sink of 2.4 ± 0.4 petagrams of carbon per year (Pg
C year–1) globally for 1990 to 2007. We also estimate a source of 1.3 ± 0.7 Pg C year–1 from tropical land-use change, consisting of a gross tropical deforestation emission of 2.9 ± 0.5 Pg C year–1 partially compensated by a carbon sink in tropical forest regrowth of 1.6 ± 0.5 Pg C year–1. Together, the fluxes comprise a net global forest sink of 1.1 ± 0.8 Pg C year–1, with tropical estimates having the largest uncertainties. Our total forest sink estimate is equivalent in magnitude to the
terrestrial sink deduced from fossil fuel emissions and land-use change sources minus ocean and atmospheric sinks.
Vegetation phenology is an important indicator of climate change and climate variability and it is strongly connected to biospheric-atmospheric gas exchange. We aimed to evaluate the applicability of phenological information derived from digital imagery for the interpretation of CO2 exchange measurements. For the years 2005-2007 we analyzed seasonal phenological development of 2 temperate mixed forests using tower-based imagery from standard RGB cameras. Phenological information was jointly analyzed with gross primary productivity (GPP) derived from net ecosystem exchange data. Automated image analysis provided reliable information on vegetation developmental stages of beech and ash trees covering all seasons. A phenological index derived from image color values was strongly correlated with GPP, with a significant mean time lag of several days for ash trees and several weeks for beech trees in early summer (May to mid-July). Leaf emergence dates for the dominant tree species partly explained temporal behaviour of spring GPP but were also masked by local meteorological conditions. We conclude that digital cameras at flux measurement sites not only provide an objective measure of the physiological state of a forest canopy at high temporal and spatial resolutions, but also complement CO2 and water exchange measurements, improving our knowledge of ecosystem processes.
Understanding relationships between canopy structure and the seasonal dynamics of photosynthetic uptake of CO(2) by forest canopies requires improved knowledge of canopy phenology at eddy covariance flux tower sites. We investigated whether digital webcam images could be used to monitor the trajectory of spring green-up in a deciduous northern hardwood forest. A standard, commercially available webcam was mounted at the top of the eddy covariance tower at the Bartlett AmeriFlux site. Images were collected each day around midday. Red, green, and blue color channel brightness data for a 640 x 100-pixel region-of-interest were extracted from each image. We evaluated the green-up signal extracted from webcam images against changes in the fraction of incident photosynthetically active radiation that is absorbed by the canopy (f (APAR)), a broadband normalized difference vegetation index (NDVI), and the light-saturated rate of canopy photosynthesis (A(max)), inferred from eddy flux measurements. The relative brightness of the green channel (green %) was relatively stable through the winter months. A steady rising trend in green % began around day 120 and continued through day 160, at which point a stable plateau was reached. The relative brightness of the blue channel (blue %) also responded to spring green-up, although there was more day-to-day variation in the signal because blue % was more sensitive to changes in the quality (spectral distribution) of incident radiation. Seasonal changes in blue % were most similar to those in f (APAR) and broadband NDVI, whereas changes in green % proceeded more slowly, and were drawn out over a longer period of time. Changes in A(max) lagged green-up by at least a week. We conclude that webcams offer an inexpensive means by which phenological changes in the canopy state can be quantified. A network of cameras could offer a novel opportunity to implement a regional or national phenology monitoring program.
The carbon balance of terrestrial ecosystems is particularly sensitive to climatic changes in autumn and spring, with spring and autumn temperatures over northern latitudes having risen by about 1.1 degrees C and 0.8 degrees C, respectively, over the past two decades. A simultaneous greening trend has also been observed, characterized by a longer growing season and greater photosynthetic activity. These observations have led to speculation that spring and autumn warming could enhance carbon sequestration and extend the period of net carbon uptake in the future. Here we analyse interannual variations in atmospheric carbon dioxide concentration data and ecosystem carbon dioxide fluxes. We find that atmospheric records from the past 20 years show a trend towards an earlier autumn-to-winter carbon dioxide build-up, suggesting a shorter net carbon uptake period. This trend cannot be explained by changes in atmospheric transport alone and, together with the ecosystem flux data, suggest increasing carbon losses in autumn. We use a process-based terrestrial biosphere model and satellite vegetation greenness index observations to investigate further the observed seasonal response of northern ecosystems to autumnal warming. We find that both photosynthesis and respiration increase during autumn warming, but the increase in respiration is greater. In contrast, warming increases photosynthesis more than respiration in spring. Our simulations and observations indicate that northern terrestrial ecosystems may currently lose carbon dioxide in response to autumn warming, with a sensitivity of about 0.2 PgC degrees C(-1), offsetting 90% of the increased carbon dioxide uptake during spring. If future autumn warming occurs at a faster rate than in spring, the ability of northern ecosystems to sequester carbon may be diminished earlier than previously suggested.
“Near-surface” remote sensing provides a novel approach to phenological monitoring. Optical sensors mounted in relatively close proximity (typically 50 m or less) to the land surface can be used to quantify, at high temporal frequency, changes in the spectral properties of the surface associated with vegetation development and senescence. The scale of these measurements—intermediate between individual organisms and satellite pixels—is unique and advantageous for a variety of applications. In this chapter, we review and discuss a variety of approaches to near-surface remote sensing of phenology, including methods based on broad- and narrow-band radiometric sensors, and using commercially available digital cameras as inexpensive imaging sensors.
Numerous studies have used the satellite-derived Normalized Difference Vegetation Index (NDVI) to estimate the phenology of vegetation cover. However, little is known about the effect of species difference on the susceptibility of NDVI-based estimation approaches, such as the threshold approach and the abrupt variation approach, for estimating the phenology of forest trees. In this study, to clarify the utility of NDVI in cool temperate deciduous forests, which consist of many tree species, we investigated the effect of the species difference on the estimation accuracy of two traditional approaches at the scale of the individual tree. We observed a canopy NDVI of 6 tree species by using a high resolution spectral camera, and compared the NDVI-based estimate of the phenological stages (green-up, green peak, senescence and leaf fall) and the ground truth data on the basis of foliar chlorophyll content.In the threshold approach, the optimal threshold value of NDVI was higher in the autumn leaf fall than the spring green-up. Species difference did not strongly affect the threshold of the green-up, but the threshold of the leaf fall was higher in tree species which flower in summer. The mean estimation error of the leafy period was +1.3 days in this approach when the simple threshold value was used for all species. In the abrupt variance approach the estimation error was larger and the leafy period was over estimated (mean: +26.1 days). The degree of overestimation in the leaf fall tended to be larger in species that flower and have a late abscission.These results suggest that the threshold approach is a better method than the abrupt variation approach if the optimal threshold value can be calculated by using a ground truth data set. Furthermore, species specific leaf senescence type and the existence of flowering affect the accuracy of NDVI-based estimates, indicating that we should confirm the composition of tree species when evaluating the NDVI-based phenology data of cool temperate deciduous forests.
Color slide images of weeds among various soils and residues were digitized and analyzed for red, green, and blue (RGB) color content. Red, green, and blue chromatic coordinates (rgb) of plants were very different from those of background soils and residue. To distinguish living plant material from a nonplant background, several indices of chromatic coordinates were studied, tested, and were successful in identifying weeds. The indices included r-g, g-b, (g-b)||r-g|, and 2g-r-b. A modified hue was also used to distinguish weeds from non-plant surfaces. The modified hue, 2g-r-b index, and the green chromatic coordinate distinguished weeds from a nonplant background (0.05 level of significance) better than other indices. However, the modified hue was the most computationally intense. These indices worked well for both nonshaded and shaded sunlit conditions. These indices could be used for sensor design for detecting weeds for spot spraying control.
Recent studies have reported that seasonal variation in camera-based indices that are calculated from the digital numbers of the red, green, and blue bands (RGB_DN) recorded by digital cameras agrees well with the seasonal change in gross primary production (GPP) observed by tower flux measurements. These findings suggest that it may be possible to use camera-based indices to estimate the temporal and spatial distributions of photosynthetic productivity from the relationship between RGB_DN and GPP. To examine this possibility, we need to investigate the characteristics of seasonal variation in three camera-based indices (green excess index [GE], green chromatic coordinate [rG], and HUE) and the robustness of the relationship between these indices and tower flux-based GPP and how it differs among ecosystems. Here, at a daily time step over multiple years in a deciduous broad-leaved and an evergreen coniferous forest, we examined the relationships between canopy phenology assessed by using the three indices and GPP determined from tower CO2 flux observations, and we compared the camera-based indices with the corresponding spectra-based indices estimated by a spectroradiometer system. We found that (1) the three camera-based indices and GPP showed clear seasonal patterns in both forests; (2) the amplitude of the seasonal variation in the three camera-based indices was smaller in the evergreen coniferous forest than in the deciduous broad-leaved forest; (3) the seasonal variation in the three camera-based indices corresponded well to seasonal changes in potential photosynthetic activity (GPP on sunny days); (4) the relationship between the three camera-based indices and GPP appeared to have different characteristics at different phenological stages; and (5) the camera-based and spectra-based HUE indices showed a clear relationship under sunny conditions in both forests. Our results suggest that it might be feasible for ecologists to establish comprehensive networks for long-term monitoring of potential photosynthetic capacity from regional to global scales by linking satellite-based, in situ spectra-based, and in situ camera-based indices.
Phenology is the study of reoccurring life-cycle events that are initiated and driven by environmental factors. It is an inter-disciplinary and integrative field that presents some key cross-cutting challenges. However, some recent advances in data and technology can help rise to these challenges. With these advances, phenology is poised to help address some critical ecological issues. This article presents an overview of phenology and some of its cross-cutting challenges. It then reviews some advances being made in the field and how these connect to land management, climate change, and climate modeling studies.
Carbon dioxide flux measurements using the eddy covariance (EC) methodology have helped researchers to develop models of ecosystem carbon balance. However, making reliable predictions of carbon fluxes is not straightforward due to phenological changes and possible abiotic/biotic stresses that profoundly influence tree functioning.
To assess the influence of canopy phenological state on CO 2 flux, we installed two different digital camera systems at different viewing angles (an outdoor webcam with a near‐horizontal view and a commercial ‘fisheye’ digital camera with a downward view) on a flux measurement tower in southern England and tracked the visual change of the canopy in this oak‐dominated ( Quercus robur L.) forest over two growing seasons.
Changes in the setting of the camera's white balance substantially affected the quality of the webcam images. However, the timing of the onset of greening and senescence was, nevertheless, detectable for the individual trees as well as the overall canopy for both years. The greening‐up date assessed from the downward images from a hemispherical lens was ∼5 days earlier than from the horizontal‐view images, because of ground vegetation development (not visible in the horizontal view).
The effects of a late air frost in 2010 were evident in the canopy greenness, and these led to reductions in daily gross primary productivity ( GPP ). The cameras recorded differences between individual tree crowns, showing their different responses to the late frost.
A major new finding from this work is the strong relationship between GPP and H ue, which was stronger than the relationship between GPP and NDVI .
Background: Recent studies have described a technique that incorporates a digital camera to observe aspects of tree phenology such as leaf expansion and leaf fall. This technique has shown that seasonal patterns of red, green, and blue digital numbers (RGB_DN) extracted from digital images differ between species.Aims: To identify the different characteristics of phenology between species by examining RGB_DN, the relationship between the seasonal patterns of RGB_DN and ecological characteristics for various species need to be evaluated throughout the year.Methods: The relationship between the normalised RGB_DN values extracted from digital images and in situ leaf area index (LAI) and leaf chlorophyll content (indicated by soil and plant analyser development; SPAD) was examined for three dominant species for multiple years in a cool-temperate, deciduous, broad-leaved forest in Japan.Results: The RGB_DN values in spring were not useful in detecting the different characteristics of leaf-flush patterns between species. In contrast, RGB_DN values in autumn showed differences in leaf-colouring as well as in leaf-fall patterns and timings between species.Conclusion: Differences in autumn phenology between tree species can be detected by using the normalised RGB_DN technique while the technique cannot be applied in spring.
Plant phenology is highly sensitive to changes in environmental conditions and can vary widely across landscapes. Current observation methods are either manual for small-scale, high precision measurements or by satellite remote sensing for large-scale, low spatial resolution measurement. The development of inexpensive approaches is necessary to advance large scale, high precision phenology monitoring. The use of publicly available, Internet-connected cameras, often associated with airports, national parks, and roadway conditions, for detecting and monitoring plant phenology at a continental scale can augment existing ground and satellite-based methodologies. We collected twice-daily images from over 1100 georeferenced public cameras across North America from February 2008 to 2009. Using a test subset of these cameras, we compared modeled spring ‘green-up’ with that from co-occurring remote sensing products. Although varying image exposure and color correction introduced noise to camera measurements, we were able to correlate spring green-up across North America with visual validation from images and detect a latitudinal trend. Public cameras had an equivalent or higher ability to detect spring compared with satellite-based data for corresponding locations, with fewer numbers of poor quality days, shorter continuous bad data days, and significantly lower errors of spring estimates. Manual image segmentation into deciduous, evergreen, and understory vegetation allowed detection of spring and fall onset for multiple vegetation types. Additional advantages of a public camera-based monitoring system include frequent image capture (subdaily) and the potential to detect quantitative responses to environmental changes in organisms, species, and communities. Public cameras represent a relatively untapped and freely available resource for supporting large-scale ecological and environmental monitoring.
Over the last decade, technological developments have made it possible to quickly and nondestructively assess, in situ , the chlorophyll (Chl) status of plants. We evaluated the performance of these optical methods, which are based on the absorbance or reflectance of certain wavelengths of light by intact leaves.
As our benchmark, we used standard extraction techniques to measure Chl a , Chl b , and total Chl content of paper birch ( Betula papyrifera ) leaves. These values were compared with the nominal Chl index values obtained with two hand‐held Chl absorbance meters and several reflectance indices correlated with foliar Chl.
The noninvasive optical methods all provided reliable estimates of relative leaf Chl. However, across the range of Chl contents studied (0.0004–0.0455 mg cm ⁻² ), some reflectance indices consistently out‐performed the hand‐held meters. Most importantly, the reflectance indices that performed best
were not those most commonly used in the literature.
We report equations to convert from index values to actual Chl content, but caution that differences in leaf structure may necessitate species‐specific calibration equations.
In the spring of 2010, temperatures averaged ~3 °C above the long-term mean (March–May) across the northeastern United States. However, in mid-to-late spring, much of this region experienced a severe frost event. The spring of 2010 therefore provides a case study on how future spring temperature extremes may affect northeastern forest ecosystems. We assessed the response of three northern hardwood tree species (sugar maple, American beech, yellow birch) to these anomalous temperature patterns using several different data sources and addressed four main questions: (1) Along an elevational gradient, how was each species affected by the late spring frost? (2) How did differences in phenological growth strategy influence their response? (3) How did the late spring frost affect ecosystem productivity within the study domain? (4) What are the potential long-term impacts of spring frost events on forest community ecology? Our results show that all species exhibited early leaf development triggered by the warm spring. However, yellow birch and American beech have more conservative growth strategies and were largely unaffected by the late spring frost. In contrast, sugar maples responded strongly to warmer temperatures and experi- enced widespread frost damage that resulted in leaf loss and delayed canopy development. Late spring frost events may therefore provide a competitive advantage for yellow birch and American beech at the expense of sugar maple. Results from satellite remote sensing confirm that frost damage was widespread throughout the region at higher elevations (>500 m). The frost event is estimated to have reduced gross ecosystem productivity by 70–153 g C m␣2, or 7–14% of the annual gross productivity (1061 ± 82 g C m␣2) across 8753 km2 of high-elevation forest. We conclude that frost events following leaf out, which are expected to become more common with climate change, may influence both forest composition and ecosystem productivity.
Conventional enhancements for the color display of multispectral images are usually based on independent contrast modifications (“stretches”) of three image channels. This approach generally does not produce colorful pictures if the image channels are strongly correlated. Procedures that selectively emphasize the weakly correlated component of the image data may better utilize the full range of colors. In Part I of this series of articles, two such methods were discussed that exaggerated color saturation without greatly modifying hue. These methods utilized principal-component analysis and HSI transformation of the image data. Herein we discuss two other methods, based on ratioing of data from different image channels. In the first approach, images of the ratioed data are “stretched” and assigned primary colors for display as color pictures. An alternative ratio method uses the difference, normalized to the intensity, instead of the ratio itself. Hues in pictures thus enhanced bear no resemblance to those in the unenhanced picture: hence, interpretation may be difficult. In the second approach, the intensity for each of three image channels is ratioed, pixel by pixel, to the sum of the intensities. Thus the data are transformed to image-domain “chromaticity” coordinates. These data are next “stretched” and then multiplied by the summed intensities to return them to the original coordinates for display. This technique is similar to those of Part I in that the hues need not be greatly altered.
Plant phenology relates strongly to primary productivity and the energy that enters into ecological food webs, and thus is vital in understanding ecosystem function and the effects of climate and climate change. The manual collection of phenological data is labor-intensive and not easily scalable, thus the ability to quantify leaf flush and other parameters at many locations requires innovative new methodologies such as the use of visible light digital cameras. Improved imaging performance was obtained by using a cabled, mobile camera system that allowed a repeated image census of branches of Rhododendron occidentale in the understory along a 30 m transect during leaf flush. Automatic division of acquired images into areas of interest (leaves) and background for calculating leaf area was accomplished by thresholding images in different color spaces. Transformation of the color space into the hue, saturation, and luminance (HSL) color space before thresholding resulted in a mean RMS error of 21.2 cm2 compared to hand-counts of leaf area. Thresholding in the native red, green, and blue (RGB) color space to isolate leaves resulted in a larger error, as did using algebraic combinations of the color components or color ratios. Relating physiological function to images, as for sap flow for branches of R. occidentale, indicates that the greening and calculated leaf area of a species as detected by imagers requires additional meteorological sensor data for interpretation.
Vegetation phenology such as the onset of green-up and senescence is strongly controlled by climate and other environmental factors, and in turn affects the terrestrial carbon balance. Therefore, phenological observation is important as an indicator of global warming and for estimation of the terrestrial carbon balance. Because phenological responses differ from species to species, precise monitoring from the species scale to the global scale is required. In this study, we analyzed images from digital cameras, which have proliferated in recent years, to investigate their utility as remote sensors. We collected daily images taken by digital cameras in national parks across Japan over 8years in wetland mixed deciduous forest, and evergreen broadleaved forest. Values of red, green, and blue (RGB) channels in each pixel within images were extracted, and a vegetation green excess index (2G-RBi) was calculated to detect phenology. The time series of 2G-RBi showed clear phenological patterns of each vegetation type in each year at the species or community scale. Even physiological damage due to a typhoon was detected. The dates of green-up were estimated easily and objectively from the second derivative of 2G-RBi, and a trend in yearly green-up dates of various types of vegetation was demonstrated. Furthermore, a strong correlation between interannual variations in green-up dates and local spring temperature was found, and the sensitivity of green-up date to temperature was revealed. The results suggest the utility of digital cameras for phenological observations at precise temporal and spatial resolutions, despite a year-to-year drift of color balance of camera as a technical device. As a form of near-surface remote sensing, digital cameras could obtain significant ecological information. Establishing camera networks could help us understand phenological responses at a wide range of scales.
Digital repeat photography has the potential to become an important long-term data source for phenological research given its advantages in terms of logistics, continuity, consistency and objectivity over traditional assessments of vegetation status by human observers. Red-green-blue (RGB) color channel information from digital images can be separately extracted as digital numbers, and subsequently summarized through color indices such as excess green (ExG = 2G − [R + B]) or through nonlinear transforms to chromatic coordinates or other color spaces. Previous studies have demonstrated the use of ExG and the green chromatic coordinate (gcc = G/[R + G + B]) from digital landscape image archives for tracking canopy development but several methodological questions remained unanswered. These include the effects of diurnal, seasonal and weather-related changes in scene illumination on ExG and gcc, and digital camera and image file format choice.
Crop physiological and phenological status is an important factor that characterizes crop yield as well as carbon exchange between the atmosphere and the terrestrial biosphere in agroecosystems. It is difficult to establish high frequency observations of crop status in multiple locations using conventional approaches such as agronomical sampling and also remote sensing techniques that use spectral radiometers because of the labor intensive work required for field surveys and the high cost of radiometers designed for scientific use. This study explored the potential utility of an inexpensive camera observation system called crop phenology recording system (CPRS) as an alternative approach for the observation of seasonal change in crop growth. The CPRS consisting of two compact digital cameras was used to capture visible and near infrared (NIR) images of maize in 2009 and soybean in 2010 for every hour both day and night continuously. In addition, a four channel sensor SKYE measured crop reflectance and Moderate Resolution Imaging Spectroradiometer (MODIS) satellite images were acquired over crop fields. The six different camera- radiometer- and MODIS-derived vegetation indices (VIs) were calculated and compared with the ground-measured crop biophysical parameters. In addition to VIs that use digital numbers, we proposed to use daytime exposure value-adjusted VIs. The camera-derived VIs were compared with the VIs calculated from spectral reflectance observations taken by SKYE and MODIS. It was found that new camera-derived VIs using daytime exposure values are closely related to VIs calculated using SKYE and MODIS reflectance and good proxies of crop biophysical parameters. Camera-derived green chlorophyll index, simple ratio and NDVI were found to be able to estimate the total leaf area index (LAI) of maize and soybean with high accuracy and were better than the widely used 2g-r-b. However, camera-derived 2g-r-b showed the best accuracy in estimating daily fAPAR in vegetative and reproductive stages of both crops. Visible atmospherically resistant vegetation index showed the highest accuracy in the estimation of the green LAI of maize. A unique VI, calculated from nighttime flash NIR images called the nighttime relative brightness index of NIR, showed a strong relationship with total aboveground biomass for both crops. The study concludes that the CPRS is a practical and cost-effective approach for monitoring temporal changes in crop growth, and it also provides an alternative source of ground truth data to validate time-series VIs derived from MODIS and other satellite systems.
Four rates of ammonium nitrate (NH4NO3) (0, 151, 454, and 908 g actual N/tree) were applied each spring for 6 years to 'Golden Delicious' (Malus domestica) apple trees. High rates of nitrogen (N) increased N concentration of Orchardgrass (Dactylis glomerata) blades and increased cover-grass growth whereas various legume species were prevalent at the low rates. Leaf N in spur or mid-terminal leaves increased yearly, and was related to leaf color by visual comparison and reflectance. Fruit from the higher N rates had greener peel and lower firmness, soluble solids content and titratable acidity. In vitro freeze tests indicated trees fertilized with lower rates of N were more cold hardy during the fall, winter and spring than those receiving the higher rates. In a similar long-term study on 'Delicious,' cold hardiness was related not only to seasonal temperature cycles and shoot dry matter, but to total sugars and sorbitol content in wood or sap.
Normal human color perception is a product of three independent sensory systems. By mirroring this mechanism, full-color display devices create colors as mixtures of three primaries. Any displayable color can be described by the corresponding values of these primaries. Frequently it is more convenient to define various other color spaces, or coordinate systems, for color representation or manipulation. Several such color spaces are presented which are suitable for applications involving user specification of color, along with the defining equations and illustrations. The use of special color spaces for particular kinds of color computations is discussed.
There is a need to document how plant phenology is responding to global change factors, particularly warming trends. "Near-surface" remote sensing, using radiometric instruments or imaging sensors, has great potential to improve phenological monitoring because automated observations can be made at high temporal frequency. Here we build on previous work and show how inexpensive, networked digital cameras ("webcams") can be used to document spatial and temporal variation in the spring and autumn phenology of forest canopies. We use two years of imagery from a deciduous, northern hardwood site, and one year of imagery from a coniferous, boreal transition site. A quantitative signal is obtained by splitting images into separate red, green, and blue color channels and calculating the relative brightness of each channel for "regions of interest" within each image. We put the observed phenological signal in context by relating it to seasonal patterns of gross primary productivity, inferred from eddy covariance measurements of surface-atmosphere CO2 exchange. We show that spring increases, and autumn decreases, in canopy greenness can be detected in both deciduous and coniferous stands. In deciduous stands, an autumn red peak is also observed. The timing and rate of spring development and autumn senescence varies across the canopy, with greater variability in autumn than spring. Interannual variation in phenology can be detected both visually and quantitatively; delayed spring onset in 2007 compared to 2006 is related to a prolonged cold spell from day 85 to day 110. This work lays the foundation for regional- to continental-scale camera-based monitoring of phenology at network observatory sites, e.g., National Ecological Observatory Network (NEON) or AmeriFlux.
Color gamet transformation pairs
- Smith
Smith, A.R. (1978) Color Gamet Transformation Pairs. SIGGRAPH 78, pp. 12-19.
The Global Network of Outdoor Webcams: Properties and Applications
- N Jacobs
- W Burgin
- N Fridrich
- A Abrams
- K Miskell
- B Braswell
- A Richardson
- R Pless
Jacobs, N., Burgin, W., Fridrich, N., Abrams, A., Miskell, K., Braswell, B., Richardson, A. &
Pless, R. (2009) The Global Network of Outdoor Webcams: Properties and
Applications. ACM SIGSPATIAL GIS 2009. Seattle, Washington, USA.
Accepted Article
This article is protected by copyright. All rights reserved.
The relationship between carbon dioxide uptake and canopy colour from two camera systems in a deciduous forest in southern England
- Mizunuma