Airborne laser scanner (LiDAR) proxies for understory light conditions
Introduction
Understory light conditions within a forest are determined by a number of factors, with the structure of the tree canopy being particularly important (Jennings et al., 1999). The quantity, quality, and spatio-temporal distribution of light near the forest floor are primarily controlled by the structure of the canopy, which influence a broad range of biological components in forest ecosystems, from the demography and population dynamics of individual species (Svenning, 2002, Svenning and Magård, 1999) over species distributions (Svenning, 2000) to community structure (Frelich et al., 2003). Plants differ in their requirement for light, with some plants preferring to grow in shade, others in full sunlight, and yet others in intermediate conditions (Ellenberg, 1988). The forest canopy is often manipulated for creating conditions favourable for the survival and growth of certain tree species of commercial or, more rarely, of conservation value (Frelich et al., 2003). Both ecologists and foresters are therefore interested in mapping forest understory light conditions (Jennings et al., 1999).
Canopy cover and canopy closure are two closely related measures which are useful for estimating the microclimate and light conditions at the forest floor. These measures are also useful for assessing habitat suitability for different plants and animals, and for estimating functional variables such as Leaf Area Index (LAI). Canopy cover is the proportion of the forest floor covered by the vertical projection of tree crowns, and is often estimated in the field using transects and vertical sighting tubes. Canopy closure is the proportion of the sky obscured by vegetation when viewed from a single point, and is estimated from hemispherical photographs or using spherical densiometer (Jennings et al., 1999, Korhonen et al., 2006, Korhonen et al., 2011).
Canopy closure should provide a better description of the light conditions under a canopy than canopy cover as all the directions in which light reaches a point below the canopy are taken into consideration (Jennings et al., 1999). Hemispherical photographs are often used for estimating canopy gap fraction, the fraction of the sky visible from a point below the canopy, which is a complementary measure to canopy closure. Canopy gap fraction can be used to determine leaf area index (LAI), defined as the one-sided green leaf area per unit ground area in broadleaved canopies, or as the projected needle-leaved area per unit ground area in needle canopies (Law et al., 2008). The sky (light) and canopy (dark) pixels in a grey-scale hemispherical image are classified based on a threshold, which is dependent on sky conditions. Hemispherical photographs should be taken under uniform sky lighting either early or late in the day, or under overcast conditions, as sun glare may lead to overestimation of canopy gaps (Steele-Feldman et al., 2006).
Terrestrial laser scanners (TLS) have been used to estimate canopy gap fraction, providing results comparable to those from hemispherical photographs (Danson et al., 2007, Seidel et al., 2012). TLS have the advantage that they also provide additional information about the vertical structure of vegetation. Estimates of canopy gap fraction using a TLS are usually obtained from two orthogonal vertical scans (Danson et al., 2007), and the precision is high, especially in the areas where the scans overlap. However, field-based methods for the estimation of canopy cover or closure are time-consuming. Remote sensing could provide a method for large-scale mapping of canopy cover and closure, if these estimates could be validated by comparing them to field-based measurements (Fiala et al., 2006, Korhonen et al., 2011).
Airborne laser scanners (ALS) use light detection and ranging (LiDAR) to obtain geo-referenced points on and above the terrain. A few studies have explored the use of ALS in estimating canopy gap fraction and LAI. Riaño et al. (2004) found that the percentage of ALS points from the canopy had a high correlation with canopy gap fraction estimated from hemispherical photographs, and consequently also with LAI. The percentage of canopy hits were calculated within different radial distances from a point, and their results showed that the radius yielding the best estimates was influenced by canopy height. They additionally found differences between the estimates for deciduous and evergreen trees. Morsdorf et al. (2006) similarly considered different radial distances to compute canopy cover and LAI showing that radii of 2 m and 15 m provided the best results for canopy cover and LAI, respectively.
Most of the estimates of canopy cover from airborne laser scanner data are based on the proportion of echoes from vegetation above a certain height (Korhonen et al., 2011, Morsdorf et al., 2006), often referred to as fractional cover (Morsdorf et al., 2006). A method based on the number of points could overestimate canopy cover if laser scanning systems that collect multiple echoes are used in open forests, or there are variations in point density within an area due to overlapping flight strips. This problem could be addressed if surface area, rather than number of points, is used to estimate canopy cover. Areas of Thiessen polygons, also known as Voronoi or Dirichlet tessellations, generated from ALS points could be used to estimate canopy cover based on surface area.
There are inherent differences between estimates of the vertical distribution of foliage by airborne and terrestrial measurement techniques. The viewing directions of the two methods differ, with ALS looking downwards from above the canopy and terrestrial instruments looking upwards from the forest floor. Notably, the presence of tree trunks and branches in images affects estimates of LAI from hemispherical photographs, spherical densiometer and TLS (Seidel et al., 2012). For field-based instruments with an angular field of view, the observed volume is an upward facing cone. However, in previous studies concerning canopy closure the observed volume has been cylindrical (Solberg et al., 2009). The penetration of the canopy by ALS is dependent on sensor properties and flight settings, influencing point density, as well as the density of foliage. Even with these differences, if it is possible to derive an angular measure of canopy closure from ALS, it could provide a measure that is not influenced by sky conditions or the threshold used for binary classification, as is the case with hemispherical photography. It would be more representative of the upper part of the canopy than estimates from terrestrial instruments, and could cover large areas in much less time than field-based methods.
The extent to which ALS-derived canopy structure variables are good estimators of understory light availability may be evaluated by examining their correlation with the average Ellenberg indicator values for light (EIVlight) for plant species within understory vegetation plots (Diekmann, 1995, Dzwonko, 2001). Indicator values have been developed as estimates of the preferred light conditions for plants, first for Central Europe (Ellenberg, 1988) and later expanded to other regions, e.g., the British Isles (Hill et al., 1999). The Ellenberg indicator values range from 1 to 9, with 1 denoting species preferring deep shade and 9 denoting species preferring full sunlight (Ellenberg, 1988).
The overall aim of this study was to develop proxies for canopy cover and canopy closure based on discrete-return ALS data, and to evaluate whether these ALS-based canopy variables were correlated with understory light conditions as estimated from the understory vegetation by EIVlight. Our specific objectives were (1) to develop proxies for canopy cover and canopy closure based on ALS data, and (2) to determine whether there is a relationship between these ALS-based canopy structure proxies and EIVlight, predicting a stronger relationship for angular canopy closure than vertical canopy cover.
Section snippets
Study Area
A number of sites in Denmark are monitored through the National Monitoring and Assessment Programme for the Aquatic and Terrestrial Environment (NOVANA), designed to monitor Danish habitats within the European NATURA 2000 network established under the Habitats Directive of the European Union, 1992. The study area includes seven NOVANA sites, which constitute three groups of sites (Fig. 1), each including both semi-open habitat and forest. The sites included the following NATURA 2000 habitat
Field Datasets
Species occurrence data for vascular plants, bryophytes and lichens in the study area were collected under the NOVANA programme within 5-m radius plots in the open sites, and in 5- and 15-m radius plots in the forest sites. We used the species occurrence data collected in 2006, 2007 and 2009 (Table 1) for this study. The EIVlight for each of these species was selected as the modified values for Britain (Hill et al., 1999), whenever they were available, as these values are considered to be more
Processing of ALS Data
ALS data were collected in April and May 2006 by COWI (2007), an international engineering and planning company, using an Optech ALTM 3100 airborne laser scanner. The data were recorded at an altitude of 1.6 km with a footprint diameter of 50 cm at nadir (directly below the aircraft), with a 24° maximum off-nadir angle. The point density was approximately 1.5 m− 2, including the points from overlapping flight strips. Both the first and the last echoes were used in the reported analyses. Aerial
Canopy Cover from ALS Data
Thiessen polygon-estimated ALS-based canopy cover visually corresponded well with the aerial images for each site. It was highly correlated (r = 0.99) with the canopy cover estimate based on number of echoes. The estimate based on Thiessen polygons was used for all further analyses, since it has a stronger theoretical relation with the surface area of canopy than the number of points. The plant-based EIVlight generally was higher for the open sites than for the forest sites, and in most cases
Discussion
Estimation of understory light conditions using ALS data could supplement, or even replace, field-based methods. ALS-based canopy closure was shown to be a better predictor of understory light conditions than ALS-based canopy cover. However, as with field-based estimates of canopy closure using densiometer, hemispherical photographs or TLS, the estimates are scale-dependent. Zhao and Popescu (2009) noted that the estimation of LAI from ALS data, as in optical remote sensing, is also
Conclusion
ALS-derived canopy variables has the big advantage that they can be rapidly estimated for extensive areas, and hence ALS may be better suited for large-area mapping of canopy variables than more accurate, but also more laborious, time-consuming and, sometimes, subjective field-based methods. Angular canopy closure estimates from ALS have been compared with estimates from hemispherical photographs (Korhonen et al., 2011). All the previous studies compared fractional cover, or laser penetration
Acknowledgement
We gratefully acknowledge funding from the Aarhus University Research Foundation through the Center for Interdisciplinary Geospatial Informatics Research (CIGIR), and the Danish National Research Foundation through the Center for Massive Data Algorithmics (MADALGO). We also thank COWI, Denmark for the LiDAR point cloud data and the aerial images of the study area, and the reviewers for their comments and suggestions.
References (34)
- et al.
Comparison of five canopy cover estimation techniques in the western Oregon Cascades
Forest Ecology and Management
(2006) - et al.
Assessment of defoliation during a pine sawfly outbreak: calibration of airborne laser scanning data with hemispherical photography
Forest Ecology and Management
(2007) - et al.
Airborne discrete-return LIDAR data in the estimation of vertical canopy cover, angular canopy closure and leaf area index
Remote Sensing of Environment
(2011) - et al.
Estimation of 3D vegetation structure from waveform and discrete return airborne laser scanning data
Remote Sensing of Environment
(2012) - et al.
Estimation of LAI and fractional cover from small footprint airborne laser scanning data based on gap fraction
Remote Sensing of Environment
(2006) - et al.
Estimation of leaf area index and covered ground from airborne laser scanner (Lidar) in two contrasting forests
Agricultural and Forest Meteorology
(2004) - et al.
Analyzing forest canopies with ground-based laser scanning: a comparison with hemispherical photography
Agricultural and Forest Meteorology
(2012) - et al.
Mapping LAI in a Norway spruce forest using airborne laser scanning
Remote Sensing of Environment
(2009) - et al.
Population ecology and conservation status of the last natural population of English yew Taxus baccata in Denmark
Biological Conservation
(1999) - et al.
Lidar-based mapping of leaf area index and its use for validating GLOBCARBON satellite LAI product in a temperate forest of the southern USA
Remote Sensing of Environment
(2009)
DEM generation from laser scanner data using adaptive TIN models
International Archives of Photogrammetry and Remote Sensing
Comparison of methods for estimating forest overstory cover: differences among techniques
Canadian Journal of Forest Research
Principles of Geographical Information Systems
Investigating laser pulse penetration through a conifer canopy by integrating airborne and terrestrial lidar
Canadian Journal of Remote Sensing
Kvalitetsbeskrivelse af laserscanningsdata
Forest canopy gap fraction from terrestrial laser scanning
IEEE Geoscience and Remote Sensing Letters
Use and improvement of Ellenberg's indicator values in deciduous forests of the Boreo-Nemoral Zone in Sweden
Ecography
Cited by (39)
Mapping subcanopy light regimes in temperate mountain forests from Airborne Laser Scanning, Sentinel-1 and Sentinel-2
2023, Science of Remote SensingPredictive mapping of bryophyte diversity associated with mature forests using LiDAR-derived indices in a strongly managed landscape
2022, Ecological IndicatorsCitation Excerpt :These indices can be used, therefore, in biodiversity assessment (Listopad et al. 2015). The different indices that are derived can permit, among other things, the mapping of soil moisture accumulation patterns (Murphy et al. 2007), light levels (Alexander et al. 2013, Tanhuanpää et al. 2015), deadwood availability (Pesonen et al. 2008, Tanhuanpää et al. 2015), microtopography (Grohmann et al., 2011), and estimate the organic layer thickness in peatland boreal forest (Laamrani et al. 2014b). The value of using LiDAR for predictive mapping of bryophyte diversity lies in the fact that the technology permits the analyst to describe habitat characteristics at scales down to the meter.
Enhancing airborne LiDAR data for improved forest structure representation in shortwave transmission models
2020, Remote Sensing of EnvironmentModeling and simulation of the spectral reflectance for the natural environment: Area pester plateau
2020, Computers and Electronics in AgricultureLiDAR data as a proxy for light availability improve distribution modelling of woody species
2020, Forest Ecology and ManagementCitation Excerpt :However, non-woody species inhabiting the understorey strongly depend on light conditions. Indeed, recent studies show that LiDAR-derived data are well suited to approximate light-demands of herbaceous understory communities (Alexander et al., 2013). Remote sensing products already have proven their value in the management of invasive species (e.g., Hantson et al., 2012; Müllerová et al., 2017), and we expect that LiDAR-derived predictors that inform on light availability also have the potential to improve maps of the potential distributions of invasives.
Estimation of sub-canopy solar radiation from LiDAR discrete returns in mixed temporal forest of Białowieża, Poland
2019, International Journal of Applied Earth Observation and GeoinformationCitation Excerpt :The canopy cover computation designed by McGaughey (2018) is one of the ratio metrics with the strongest link to under canopy sunlight in this study (R2 = 0.71 (average from 3 sunlight components)). By definition, canopy cover is the area of the ground covered by the vertical projection of the canopy (Jennings, 1999) and has been used to estimate penetration of light to the understory (e.g. Alexander et al., 2013; Canham et al., 1990). It is not surprising therefore that it is one of the best estimators of sunlight whether field-estimated or ALS-based.