Usability of time-lapse digital camera images to detect characteristics of tree phenology in a tropical rainforest
Introduction
To evaluate the interaction between vegetation and atmospheric processes under rapid climatic change it is necessary to accurately detect spatial and temporal variations in plant phenology, such as the timing of flowering, leaf expansion, and leaf-fall (Peñuelas et al., 2009, Polgar and Primack, 2011, Richardson et al., 2010, Richardson et al., 2013, Schwartz et al., 2013). Such variations have important consequences for carbon cycling and sequestration. Tropical forests in particular account for about 20% of terrestrial carbon stock and about 30% of net primary productivity worldwide (Prentice et al., 2001). The ecosystem functions in tropical rainforests, such as photosynthesis and evapotranspiration, strongly affect global heat, water, and carbon-budget cycles (Kumagai and Kume, 2012, Kumagai et al., 2004, Kumagai et al., 2005, Kumagai et al., 2006). Kumagai et al. (2013) reported that the deforestation and degradation of tropical rainforests induced changes in atmospheric circulation, leading to a gradual long-term decrease in precipitation. Despite humid climate conditions throughout the year, the spatial distributions of vegetation and ecosystem functions in Borneo are strongly affected by meteorological changes caused by changes in the frequency and strength of the El Niño Southern Oscillation (ENSO) and the summer and winter Asian monsoons (Erasmi et al., 2009, Kumagai and Porporato, 2012, Nagai et al., 2007). These observations suggest the importance of accurately detecting the spatial and temporal variability of tree phenology in tropical rainforests under rapid meteorological and climatic change (Huete et al., 2008).
In contrast with the temporal variability of tree phenology in deciduous forests in the temperate zone, which is reasonably well understood, that in tropical rainforests is unclear. Leaf-fall and leaf flushing usually occur throughout the year (Harrison, 2001, Kishimoto-Yamada et al., 2010); however, occasional severe drought caused by ENSO and other phenomena can trigger canopy-scale mass leaf shedding, leaf flushing, and flowering (Harrison, 2001, Ichie et al., 2004, Itioka and Yamauti, 2004, Kobayashi et al., 2013, Sakai et al., 2006). These temporal variations are complicated by the fact that tropical rainforests have such high biodiversity, which creates a wide range of species-specific phenology. For instance, the 52-ha permanent-plot study site in the Lambir Hills National Park in Borneo harbors 1192 known plant species (Lee et al., 2002). Reich et al. (2004) described differences in the characteristics of leafing phenology among tree species in Amazonian rainforests. In addition, the interaction between leafing phenology and fluctuations of insect populations in response to changes in the intensity of ENSO was evident in a lowland mixed dipterocarp forest (Itioka and Yamauti, 2004). These facts suggest the importance of long-term continuous phenological observations and analyses of both individual tree species and the whole canopy to accurately detect tree phenology in tropical rainforests.
Although satellite images permit relatively inexpensive, high-frequency phenological monitoring over wide areas, the low resolution of the images prevent researchers from achieving the goal of monitoring individual trees or species; in addition, the lack of daily-resolution data can prevent researchers from capturing important short-term changes such as leaf flushing. Long-term continuous phenological observations using a time-lapse digital camera at a daily time-step can solve these problems (Graham et al., 2010, Ide and Oguma, 2010, Richardson et al., 2007, Richardson et al., 2009). Previous studies have indicated that the temporal patterns of red, green, and blue digital numbers (denoted as DNRGB) extracted from phenological images can reveal the patterns and timing of leaf expansion, leaf coloring, and leaf-fall of different tree species (Ahrends et al., 2008, Nagai et al., 2011). The data can also reveal the year-to-year variability in the timing of flowering, leaf expansion, and leaf-fall of the whole canopy and of individual trees (Inoue et al., 2014, Nagai et al., 2013b, Nagai et al., 2014c). Time-lapse digital cameras have the advantages of low operating cost and low labor input (Ahrends et al., 2008, Inoue et al., 2014, Richardson et al., 2007), and because they permit daily image capture, they can therefore collect high-frequency phenological information on shoots, individual trees, and the overall canopy in tropical rainforests at multiple locations.
Previous studies have obtained phenological observations in deciduous broad-leaved forests (Ahrends et al., 2008, Inoue et al., 2014, Mizunuma et al., 2013, Nagai et al., 2011, Nagai et al., 2013b, Richardson et al., 2007, Richardson et al., 2009, Sonnentag et al., 2012), evergreen coniferous forests (Bater et al., 2011, Nagai et al., 2013a, Richardson et al., 2009, Sonnentag et al., 2012), grassland (Ide and Oguma, 2013, Inoue et al., 2015, Migliavacca et al., 2011), cropland (Meyer and Neto, 2008, Sakamoto et al., 2012, Zhou et al., 2013), and Cerrado savanna (Alberton et al., 2014, Almeida et al., 2014). In contrast, phenological observations and analysis have been limited in evergreen broad-leaved forests, and particularly in tropical rainforests (Zhao et al., 2012). Nakaji et al. (2014) and Nagai et al. (2014b) collected continuous phenological observations by using a time-lapse digital camera in an evergreen broad-leaved forest on the Malay Peninsula and in Borneo, respectively. However, these authors did not analyze the temporal patterns of DNRGB values extracted from such images.
Accordingly, we examined similarities and differences in the phenological characteristics among tree species and of the whole canopy by collecting daily field observations with a time-lapse digital camera and with quantum sensors in a tropical rainforest in Borneo over 2 years. Our aim was to validate the usefulness of long-term, continuous, time-lapse digital photos for detecting differences in tree phenology in tropical rainforests.
Section snippets
Study site and period
Our study site was located in a primary lowland mixed dipterocarp forest in Lambir Hills National Park (LHNP site; 4°11′43.71″N, 114°2′25.55″E, 150 to 200 m above sea level) in Malaysian Borneo. The mean annual temperature was about 27 °C, and the mean annual rainfall at the site from 2000 to 2006 was around 2600 mm (Kishimoto-Yamada et al., 2010). There is no clear seasonality in solar radiation, temperature, vapor-pressure deficit, or precipitation at the site. There is occasional severe
Temporal patterns in camera-based %RGB, saturation, lightness, hue, and GEI
Fig. 2, Fig. 3 show, respectively, typical canopy surface images and the spatial distribution of camera-based %RGB, saturation (S), lightness (L), hue (H), and GEI values from 2013. During late July 2013 and early September 2013, many individual trees in the study forest flowered (Fig. 2d), particularly trees Sb3, Sb5, Da1, Sa1, Sw1, Sw2, and Sov1 (T. Itioka et al., unpublished data; Yayoi Takeuchi, National Institute for Environmental Studies [NIES], unpublished data). The color of the canopy
Usability of phenological images for detecting phenological characteristics
We did not detect any clear temporal patterns in tree phenology in a tropical rainforest by analyzing %RGB values and HSL values of the whole canopy based on data extracted from phenological images (Figs. 4a, 5a, 6a, 7a) or by measuring spectral reflectance of the whole canopy (Fig. 9). In contrast, the analysis of %RGB values and saturation values of individual trees could detect characteristics of tree phenology based on changes in the color of the canopy surface due to flowering, leaf
Conclusions
In this study, we show the utility of analyzing temporal patterns of DNRGB values of individual trees based on data extracted from phenological images for detecting the characteristics of tree phenology in a tropical rainforest. Baldocchi et al. (2005) encouraged the installation of time-lapse digital cameras at tower-flux observation sites to improve our understanding of the relationships between the temporal patterns of CO2 flux and plant phenology at the canopy scale. Phenological images
Acknowledgments
We thank the Forest Department Sarawak and all members of the Lambir Hills National Park site community, and particularly Dr. Masayuki Matsuoka (Kochi University), Dr. Eri Yamasaki (Zürich University), and Dr. Kazuho Matsumoto (Ryukyu University) for their assistance in the field. We are grateful to Dr. Keiko Kishimoto-Yamada (Niigata University) and Dr. Yayoi Takeuchi (National Institute for Environmental Studies) for their valuable discussion and for providing phenological information. We
References (65)
- et al.
Using phenological cameras to track the green up in a Cerrado savanna and its on-the-ground validation
Ecol. Inform.
(2014) - et al.
Applying machine learning based on multiscale classifiers to detect remote phenology patterns in Cerrado savanna trees
Ecol. Inform.
(2014) - et al.
Spectroscopy of canopy chemicals in humid tropical forests
Remote Sens. Environ.
(2011) - et al.
Overview of the radiometric and biophysical performance of the MODIS vegetation indices
Remote Sens. Environ.
(2002) - et al.
Multiple site tower flux and remote sensing comparisons of tropical forest dynamics in monsoon Asia
Agric. For. Meteorol.
(2008) - et al.
Use of digital cameras for phenological observations
Ecol. Inform.
(2010) - et al.
A cost-effective monitoring method using digital time-lapse cameras for detecting temporal and spatial variations of snowmelt and vegetation phenology in alpine ecosystems
Ecol. Inform.
(2013) - et al.
Detection of the different characteristics of year-to-year variation in foliage phenology among deciduous broad-leaved tree species by using daily continuous canopy surface images
Ecol. Inform.
(2014) - et al.
Utilization of ground-based digital photography for the evaluation of seasonal changes in the aboveground green biomass and foliage phenology in a grassland ecosystem
Ecol. Inform.
(2015) - et al.
Influences of diurnal rainfall cycle on CO2 exchange over Bornean tropical rainforests
Ecol. Model.
(2012)
Annual water balance and seasonality of evapotranspiration in a Bornean tropical rainforest
Agric. For. Meteorol.
Verification of color vegetation indices for automated crop imaging applications
Comput. Electron. Agric.
Using digital repeat photography and eddy covariance data to model grassland phenology and photosynthetic CO2 uptake
Agric. For. Meteorol.
Photosynthetic and structural characteristics of canopy and shrub trees in a cool-temperate deciduous broadleaved forest: implication to the ecosystem carbon gain
Agric. For. Meteorol.
Seasonal changes in camera-based indices from an open canopy black spruce forest in Alaska, and comparison with indices from a closed canopy evergreen coniferous forest in Japan
Policy. Sci.
Estimation of light-use efficiency through a combinational use of the photochemical reflectance index and vapor pressure deficit in an evergreen tropical rainforest at Pasoh, Peninsular Malaysia
Remote Sens. Environ.
Climate change, phenology, and phenological control of vegetation feedbacks to the climate system
Agric. For. Meteorol.
Assessing the use of camera-based indices for characterizing canopy phenology in relation to gross primary production in a deciduous broad-leaved and an evergreen coniferous forest in Japan
Ecol. Informat.
An alternative method using digital cameras for continuous monitoring of crop status
Agric. For. Meteorol.
Comparing carbon flux and high-resolution spring phenological measurements in a northern mixed forest
Agric. For. Meteorol.
Relationships between leaf pigment content and spectral reflectance across a wide range of species, leaf structures and developmental stages
Remote Sens. Environ.
Digital repeat photography for phenological research in forest ecosystems
Agric. For. Meteorol.
Applicability of time-lapse digital cameras for ground-truth measurements of satellite indices in the boreal forests of Alaska
Policy. Sci.
Red and photographic infrared linear combinations for monitoring vegetation
Remote Sens. Environ.
Using digital cameras for comparative phenological monitoring in an evergreen broad-leaved forest and a seasonal rain forest
Ecol. Inform.
Modeling winter wheat phenology and carbon dioxide fluxes at the ecosystem scale based on digital photography and eddy covariance data
Ecol. Inform.
Quantitative phenological observations of a mixed beech forest in northern Switzerland with digital photography
J. Geophys. Res. Biogeosci.
Predicting the onset of net carbon uptake by deciduous forests with soil temperature and climate data: a synthesis of FLUXNET data
Int. J. Biometeorol.
Using digital time-lapse cameras to monitor species-specific understorey and overstorey phenology in support of wildlife habitat assessment
Environ. Monit. Assess.
Spatial patterns of NDVI variation over Indonesia and their relationship to ENSO warm events during the period 1982–2006
J. Clim.
Public Internet-connected cameras used as a cross-continental ground-based plant phenology monitoring system
Glob. Chang. Biol.
Drought and the consequences of El Niño in Borneo: a case study of figs
Popul. Ecol.
Cited by (25)
Phenology of fine root and shoot using high frequency temporal resolution images in a temperate larch forest
2022, RhizosphereCitation Excerpt :The RR peak indicates advances in leaf senescence, which will subsequently be followed by leaf out. Seasonal changes in canopy color indices, extracted from repeat canopy photography, are closely related to seasonal changes in the phenology of deciduous forests, such as bud-break, leaf expansion, and the start/end of the photosynthetic period (Sakamoto et al., 2012; Ide and Oguma, 2013; Keenan et al., 2014; Toomey et al., 2015; Wingate et al., 2015; Moore et al., 2016; Nagai et al., 2016). In other words, by combining time series data of PAI and color indices, it is possible to more accurately identify fluctuations in shoot biomass and the period of photosynthesis and estimate the temperature effects of phenological fluctuations on C budgets in deciduous trees.
Monitoring leaf phenology in moist tropical forests by applying a superpixel-based deep learning method to time-series images of tree canopies
2022, ISPRS Journal of Photogrammetry and Remote SensingEvaluating land surface phenology from the Advanced Himawari Imager using observations from MODIS and the Phenological Eyes Network
2019, International Journal of Applied Earth Observation and GeoinformationCitation Excerpt :It employs cameras mounted at different positions to provide a comprehensive picture of phenological changes in the ecosystem (Nagai et al., 2018). The time-lapse images collected by PEN have been used in a wide range of scientific studies including: 1) exploring the phenological changes at ground level in understudied ecosystems such as evergreen forests and tropical rainforests (Nagai et al., 2013, 2016a; Kobayashi et al., 2018), 2) evaluating the quality of satellite-derived LSP (Motohka et al., 2009), and 3) modeling ecosystem productivity (Nagai et al., 2010). The recently launched next generation geostationary satellite, Himawari-8, provides an opportunity to quantitatively evaluate LSP detections from geostationary satellites in the Asia-Pacific region.
Climate and nutrient effects on Arctic wetland plant phenology observed from phenocams
2018, Remote Sensing of EnvironmentGenomics meets remote sensing in global change studies: monitoring and predicting phenology, evolution and biodiversity
2017, Current Opinion in Environmental SustainabilityCitation Excerpt :They demonstrated that drought is highly related to the intensity of GF, and that photosynthetically active radiation and low temperature are also relevant. Nagai et al. [37] analyzed images obtained from interval cameras installed on a canopy observation crane and found that the ratio of the reflectance in the red, green and blue wavelength range can detect characteristics of phenology of individual trees. Future climate scenarios for the tropics suggest that the frequency of extreme events such as severe droughts will increase [38].
Introducing digital cameras to monitor plant phenology in the tropics: applications for conservation
2017, Perspectives in Ecology and ConservationCitation Excerpt :However, the application of repeated digital photographs is also efficient for the phenology monitoring of temperate grasslands (Inoue et al., 2015; Julitta et al., 2014), peatland (Peichl et al., 2014), and evergreen forest (Toomey et al., 2015). Its reliability for tropical vegetation was recently validated for woody cerrado savanna (Alberton et al., 2014) and applied for tropical forest (Nagai et al., 2016; Lopes et al., 2016). The use of camera-derived vegetation indices in association with leaf demography-ontogeny models has been recently applied in the Amazon forest to investigate ecosystem-scale photosynthetic seasonality (Wu et al., 2016).