Short communicationEstimation of leaf area index in understory deciduous trees using digital photography
Introduction
Accurate estimates of leaf area index (L) are essential to the ecological characterization of forest ecosystems (Chen et al., 1997) and for modeling stand structure and dynamics (Xue et al., 2011). Because direct measurements of L in forests are destructive, time-consuming and often impractical, indirect optical methods have been widely used to indirectly estimate L from measurements of radiation transmittance through the canopy (for a review, see Bréda, 2003, Jonckheere et al., 2004, Welles and Norman, 1991). Beer–Lambert's law has often been used to model canopy transmittance (Eq. (1), based on Nilson, 1971):where P(θ) is the canopy gap fraction, G(θ) is the foliage projection coefficient and Ω(θ) is the foliage clumping index at zenith angle θ. Lt is the plant area index, including foliar and woody materials. Over the last few decades, several comparisons have been made between direct and indirect methods to estimate the overstory leaf area index in forest ecosystems, as demonstrated in the previously cited reviews. However, very few attempts have been made to estimate the leaf area index of forest understory. Some studies showed that the understory leaf area may exceed that of the overstory (Law et al., 2001, Macfarlane et al., 2010). Accurate estimates of the understory leaf area index are also required for processed-based canopy photosynthesis models (Beaudet et al., 2002, Jolly et al., 2004), for designing silvicultural systems aimed at promoting natural tree regeneration (Caccia and Ballaré, 1998) and for understanding energy and mass exchange processes (Xue et al., 2011). As a consequence, rapid, non-destructive and reliable methods are strongly needed to estimate the understory leaf area index.
Thanks to recent technological development, digital cameras with high spatial and radiometric resolutions are becoming increasingly affordable and promote the use of digital photographic methods to indirectly estimate canopy structural variables such as gap fraction, foliage cover, leaf angle distribution and leaf area index. In addition, the digital image format is well suited to process photographs taken from above the canopy looking downward. For example, the use of vegetation indices has long been explored in crops and weed plants for indirectly estimating L from downward-looking cameras (e.g., Liu et al., 2013, Meyer and Neto, 2008). Basically, vegetation indices involve the transformation of the digital number (DN) of image pixels from each channel to generate features that are able to separate green vegetation (foreground) from non-vegetation elements (background). The use of vegetation indices derived from downward-looking photography has rarely been explored in forest ecosystems (e.g., Graham et al., 2009). Recently, Macfarlane and Ogden (2012) tested nadir photography in forest stands and proposed robust and affordable image classification methods to estimate the understory foliage cover (i.e., the portion of the ground area covered by the vertical projection of the foliage; Walker and Tunstall, 1981). Unfortunately, the authors did not test the accuracy of their photographic methods to estimate L because they had no estimates of the foliage projection coefficient or foliage clumping available to invert leaf area from the foliage cover at the time. In addition, they had no direct measurements available to verify the performance of their method. This calls for a validation of their method using direct reference measurements.
To infer leaf area from foliage cover ff and its complement, vertical gap fraction P(0), information of the foliage leaf angle distribution f(θL), which is related to the foliage projection coefficient G(θ), is required. These estimates can be obtained from digital photography using a leveled camera approach, a method recently proposed by Ryu et al. (2010) that has also been validated with direct leaf angle measurements in tall trees by Pisek et al. (2011). The method is potentially suitable for estimating the leaf angle distribution in understory owing to the accessible height of leaves in short canopies. For example, Zou et al. (2014) tested and validated this method in field crops.
Another variable needed for the indirect estimation of L is foliage clumping; by knowing the foliage cover, the vertical clumping index Ω(0) can be estimated from the vertical gap fraction using theoretical gap fraction formulas.
In this paper, we tested whether digital photography can be used to estimate the leaf area index in understory deciduous trees. Different photographic methods were combined for this purpose. Nadir photography was used to estimate foliage cover and its complementary vertical gap fraction. Leveled digital photographs were used for estimating leaf angle and the foliage projection coefficient G(0). Finally, we estimated the foliage clumping index Ω(0) from a gap size distribution approach (Chen and Cihlar, 1995, Leblanc, 2002). Leaf area index derived from digital photography was compared with destructive L measurements obtained from harvesting.
Section snippets
Study site and experimental design
The study was performed in May 2014 at the Forestry Research Centre, Arezzo, Italy (43.48°N; 11.88°E), in understory plots established within a 0.25 ha, 30-year old, 15-m tall Turkey oak (Quercus cerris L.) forest. Three deciduous forest understory species were examined: hornbeam (Carpinus betulus L.), Turkey oak and beech (Fagus sylvatica L.). For each species, a rectangular 60 × 40 cm understory plot was established in which 40 three-year old saplings that were about 0.5 m tall were planted with
Results
Leaf area index measured by harvesting (L) ranged from 0.5 to 2.4 m2 m−2. Specifically, L ranged from 1.0 to 2.4 in hornbeam (average ± standard error 1.5 ± 0.1), from 0.5 to 1.4 in Turkey oak (average 0.9 ± 0.1) and from 0.7 to 2.2 in beech (average 1.4 ± 0.2).
The foliage cover estimates (ff) obtained from the LAB2 method ranged from 0.18 to 0.84 in hornbeam (average 0.58 ± 0.07), from 0.11 to 0.70 in Turkey oak (average 0.43 ± 0.06) and from 0.11 to 0.85 in beech (average 0.53 ± 0.08). The foliage cover
Discussion and conclusions
In this study, we have demonstrated that digital photography can be used to obtain indirect estimates of leaf area index of understory tree species. Both of the tested image classification methods yielded good estimates of ff, although we observed that LAB2 was the most accurate method overall. In contrast, the accuracy of Rosin's method was sensitive to larger ff and thus to L values. Macfarlane and Ogden (2012) first indicated that the LAB2 method should be used to estimate ff in nadir images
Acknowledgments
The research was supported by the research grant “Relationships between stand structure and biodiversity in forest ecosystems—ForBIO”. We thank Giulio Puletti for providing the extendable pole for the nadir photograph acquisition. We thank the Forest Service staff of Pieve Santo Stefano, AR, particularly the chief officer Alberto Veracini, for providing vegetation materials. We thank two anonymous reviewers for helping to greatly improve the manuscript with their comments.
References (35)
- et al.
Predictions of understorey light conditions in northern hardwood forests following parameterization, sensitivity analysis, and tests of the SORTIE light model
For. Ecol. Manage.
(2002) - et al.
Estimation of canopy properties in deciduous forests with digital hemispherical and cover photography
Agric. For. Meteorol.
(2013) - et al.
Budburst and leaf area expansion measured with a novel mobile camera system and simple color thresholding
Environ. Exp. Bot.
(2009) - et al.
Review of methods for in situ leaf area index determination: Part I, Theories, sensors and hemispherical photography
Agric. For. Meteorol.
(2004) - et al.
Estimation of leaf area index in open-canopy ponderosa pine forests at different successional stages and management regimes in Oregon
Agric. For. Meteorol.
(2001) - et al.
Hemispherical photography simulations with an architectural model to assess retrieval of leaf area index
Agric. For. Meteorol.
(2014) - et al.
Assessment of in situ crop LAI measurement using unidirectional view photography
Agric. For. Meteorol.
(2013) - et al.
Estimation of leaf area index in eucalypt forest using digital photography
Agric. For. Meteorol.
(2007) - et al.
Verification of color vegetation indices for automated crop imaging applications
Comput. Electron. Agric.
(2008) A theoretical analysis of the frequency of gaps in plant stands
Agric. For. Meteorol.
(1971)
Is the spherical leaf inclination angle distribution a valid assumption for temperate and boreal broadleaf tree species?
Agric. For. Meteorol.
Unimodal thresholding
Pattern Recognit.
How to quantify tree leaf area index in an open savanna ecosystem: a multi-instrument and multi-model approach
Agric. For. Meteorol.
Comparison of leaf angle distribution functions: effects on extinction coefficient and fraction of sunlit foliage
Agric. For. Meteorol.
Influences of canopy structure and physiological traits on flux partitioning between understory and overstory in an eastern Siberian boreal larch forest
Ecol. Modell.
Photographic measurement of leaf angles in field crops
Agric. For. Meteorol.
Image analysis compared with other methods for measuring ground cover
Arid Land Res. Manage.
Cited by (40)
Estimating the leaf area index of urban individual trees based on actual path length
2023, Building and EnvironmenthemispheR: an R package for fisheye canopy image analysis
2023, Agricultural and Forest MeteorologyComprehensive evaluation of global CI, FVC, and LAI products and their relationships using high-resolution reference data
2022, Science of Remote SensingUsing fractal dimension to correct clumping effect in leaf area index measurement by digital cover photography
2021, Agricultural and Forest MeteorologyCitation Excerpt :Another method is to remove the large between-crown gaps that are considered the main factor of the clumping effect. The between-crown gaps are selected from the DCP images according to a threshold (e.g., larger than 2.1% of the total image area; Chianucci et al. 2014). Then crown porosity (i.e., one minus the ratio of foliage cover to crown cover) can be calculated, and the LAI can be estimated using a modified Beer-Lambert law.
Testing an expanded set of sustainable forest management indicators in Mediterranean coppice area
2021, Ecological Indicators