Elsevier

Ecological Informatics

Volume 38, March 2017, Pages 12-25
Ecological Informatics

Individual-tree- and stand-based development following natural disturbance in a heterogeneously structured forest: A LiDAR-based approach

https://doi.org/10.1016/j.ecoinf.2016.12.004Get rights and content

Highlights

  • Forest development after natural disturbance was analyzed using LiDAR.

  • Individual trees were detected using LiDAR data supported by CIR aerial imagery.

  • Individual-tree-based growth of future forest development was simulated.

  • Naturally regenerating forests exhibit structural heterogeneity in the early-seral stages.

  • The feasibility of LiDAR data for modeling post-disturbed stands is shown.

Abstract

Large-scale severe natural disturbance events drive spatial and temporal patterns of forests by altering forest structure, composition, and functions. In the Bavarian Forest National Park in Germany, windthrow events led to large disturbances caused by the European bark beetle (Ips typographus L.). Until recently, it was assumed that at the initial stage of regeneration, trees tend to form a homogeneous stand structure, whereas structural heterogeneity is an attribute of later developmental stages. Yet recent studies provide evidence that under certain conditions structural heterogeneity can arise much earlier in stand development. Here we combined LiDAR data and forest growth modeling based on individual trees to develop a workflow for studying forest development in post-disturbed areas in the upper montane regions of the national park. The current forest structure was derived from LiDAR data of individually detected trees and a set of forest structural attributes were derived. The results served as input to simulate tree development spatio-temporally for a period of 80 years. Several spatial statistics, including landscape and spatial point pattern metrics, were calculated to assess the structural heterogeneity. The results showed that naturally regenerating forests on post-disturbed sites reveal structural heterogeneity already at the early-seral stage. Moreover, a significant portion of the eventual old-growth structural heterogeneity might already be determined in the early successional stages. Our workflow highlights the use of multi-sensor aerial remote sensing to provide detailed structural information useful for the investigation of early-phase forest dynamics.

Introduction

Forests are complex dynamic systems that contribute several crucial ecosystem services. Future forest management and conservation require the development of appropriate target objectives and the evaluation of different management strategies. The scientific knowledge needed for this includes a holistic description of forest status and development (Koch et al., 2009).

The structural composition of forests in particular has become an important factor in the analysis and management of forest ecosystems (Franklin et al., 2002, McElhinny et al., 2005, Pommerening and Stoyan, 2006). Forest structure is basically defined as “the physical and temporal distribution of trees in a forest stand” (Oliver and Larson, 1996) and is the result of natural processes as well as human intervention (Gadow et al., 2012). Forest structure is both a product and driver of ecosystem processes (Spies, 1998). Spatial structure is an important factor affecting forest stand dynamics, growth, and yield, and it also controls a range of forest functions, including soil protection and recreation (Pretzsch, 2009). In addition, other important ecosystem values, such as habitat and species diversity, are also related to spatial stand structure (Bergen et al., 2009, MacArthur and MacArthur, 1961, Pommerening, 2002, Vierling et al., 2008). Several studies have revealed that changes in the forest structure also affect species diversity (Lehnert et al., 2013, Müller et al., 2008) and composition (Bässler et al., 2010b, Müller et al., 2010).

Natural disturbance events, e.g., fires, windthrow, and insect outbreaks, are among the most crucial drivers that alter the structure of forest stands (Franklin et al., 2002, Turner, 2010, Swanson et al., 2011). Throughout the 20th century, the number of such disturbance events in Europe increased (Schelhaas et al., 2003, Seidl et al., 2014). These events are particularly important for forest evolution as they alter forest landscape and enable regeneration. However, natural forest development in post-disturbed areas in Central Europe is only insufficiently documented; almost all forest areas are influenced by human intervention and the available long-term documentations often do not cover total regeneration cycles (Fischer, 2003).

Conventional forest succession models usually treat early-seral stages simply as a period of re-organization and re-establishment (Donato et al., 2012). These models assume that regenerating trees on post-disturbed sites tend to initially form a homogeneous stand structure, and that structural heterogeneity evolves in the later developmental stages after a phase of self-thinning. The increasing competition among the trees culminates in density-dependent tree mortality across the areas commonly dominated by pioneer species. Gaps and dead wood change the microclimatic conditions and enable the development of complex forest structures, including understory vegetation (Bormann and Likens, 1979, Franklin et al., 2002, Oliver and Larson, 1996, Spies and Franklin, 1996). However, under certain initial conditions, post-disturbed forest sites might be spatially heterogeneous, similar to old-growth forests, even in early-seral stages (Donato et al., 2012). Because traditional management measures primarily place emphasis on fast timber production and the development of late-seral heterogeneity, the ecological importance of early-seral stages has been mostly underappreciated (Swanson et al., 2011). Simulations of such immediate forest recoveries should be captured by numerous factors. For example, it has been stated that the changed biogeochemical fluxes and their relationship with the vegetation are of high importance as they highly affect the long-term successional trajectories by altering species assemblages, biomass, leaf area index (LAI) recovery, chemical properties, stand and flow of water, biological legacies, soil carbon, and nutrient sources (Scheller and Swanson, 2015).

Traditionally, forest stand information is collected during expensive and time-consuming field surveys (Hyyppä et al., 2000), where tree locations and relevant structural attributes are often not measured. Nowadays, remote sensing is being applied more often. The available methods based on airborne, active and passive data represent powerful tools for efficiently measuring environmental variables over a large spatial extent, either for direct use or complementary to ground observations. The 3-D light detection and ranging (LiDAR) measurements have garnered a great deal of scientific and operational attention. LiDAR uses a narrow beam of visible or near-infrared light to measure the distance to a target object by measuring the elapsed time between the sent and returned laser signals (Wehr and Lohr, 1999) with discrete-return and waveform-recording devices (Lefsky et al., 2002). Because of the high sampling density of the point clouds generated by post-processing of waveform data, a portion of laser pulses in closed-canopy forests reflect off the crown surface and the remainder penetrate through the canopy to the ground. LiDAR data provide a direct 3-D representation of terrain, surface, and vegetation, in contrast to 2-D spectral information, which can only be used to indirectly infer vertical forest information. LiDAR data have been proven to accurately capture forest structural information, such as height, basal area, and mean diameter of stands (Coops et al., 2007, Holmgren and Jonsson, 2004, Latifi et al., 2012, Næsset, 2002). LiDAR has also been used for obtaining information on wood volume (Næsset, 2002, Latifi et al., 2010), overstory and understory vegetation cover (Latifi et al., 2016), above- and below-ground biomass (González-Ferreiro et al., 2012, Næsset, 2004), carbon stocks (Stephens et al., 2012), successional stages (Falkowski et al., 2009), and habitat characteristics (Goetz et al., 2010, Müller and Brandl, 2009, Latifi et al., 2016). Terrestrial LiDAR applications are complementary to airborne LiDAR platforms and have shown great potential for the investigation of forest understory structure. Several studies have already investigated the use of terrestrial LiDAR for detailed reconstruction of tree information (Dassot et al., 2011, Bayer et al., 2013, Liang et al., 2016). The application of LiDAR data in forest inventory has been comprehensively reviewed by Hyyppä et al. (2008), Latifi (2012), and Wulder et al. (2012), to which the reader is referred for further reading.

Individual-tree growth models offer the possibility of modeling forest growth by incorporating individual trees and their spatial arrangement in the prediction of post-disturbance stand development and by using different combinations of species assemblage and stand structures, management regimes, and regeneration methods (Pretzsch, 1997, Pretzsch, 2009). Therefore, such models provide higher flexibility and are particularly suitable to respond to new management goals (Pretzsch et al., 2002). The present study was designed to combine aerial remote sensing and growth modeling based on individual trees to study the development of horizontal and vertical structural complexities on post-disturbed forest sites. In this study, we captured the current spatial patterns of rejuvenation using LiDAR data of individual trees, simulated post-disturbance forest development using a forest growth simulator based on individual trees, and analyzed both the initial and simulated tree arrangements based on a set of calculated structural metrics. We also applied point pattern analysis in combination with the individual tree locations to examine and compare the spatial tree arrangement formed by various ecological processes.

Section snippets

Study area

The study area is located in the Bavarian Forest National Park (BFNP) in Germany. The park is part of the Bohemian Forest ecosystem, a low mountain range located in southeastern Germany and southwestern Czech Republic. Together with the adjacent Czech National Park of Šumava, the two parks form one of the most extensive and contiguous forest landscapes in Central Europe. The BFNP (approximately 13,500 ha) was established in 1970, and was expanded in 1997 to its current extent of 24,218 ha. The

Individual-tree information extraction

We summarized the results of the individual-tree extraction by calculating the 2-D distance between each pair of tree locations, which yielded seven distance classes ranging between 0 and 7 m (Table 2). The sample-based validation of the matching algorithm (Section 2.4) showed that the majority of trees extracted from the CHM were matched to the nearest trees in the image within a range of 0–1 m distance. The highest rate of such close matching was observed for site 2 (89.31%), and the values

Discussion

The forest sites analyzed in this study represent a near-natural mountain forest ecosystem. A major concern in such forests is their resistance and resilience to windthrow events and insect outbreaks. Changes in disturbance dynamics and climate conditions require an improved understanding of forest dynamics for the development of better and more flexible management practices. Therefore, it is crucial to gain a better understanding of both the impacts of such disturbances and the subsequent

Conclusions

Our results confirmed the usefulness of airborne LiDAR supported by CIR aerial imagery and field data to derive multi-scale forest structural attributes. Especially in natural primeval forests with insufficient accessibility, traditional approaches based on sample plots are often limited, and field surveys are expensive, laborious, and logistically difficult. LiDAR offers a relatively cost-effective and barrier-free means of measuring the horizontal and vertical forest structure across broad

Acknowledgements

The authors are grateful to the BFNP administration for providing the necessary remote sensing, photogrammetry, and field data, as well as for providing the first author with technical infrastructure and field material during his research stay in the BFNP. We thank Dr. Martin Isenburg (RapidLasso GmbH) for his support with LiDAR data processing and Dr. Peter Biber (Technische Universität München) for providing us with the SILVA 2.2. growth simulator.

References (89)

  • C. McElhinny et al.

    Forest and woodland stand structural complexity: its definition and measurement

    For. Ecol. Manag.

    (2005)
  • J. Müller et al.

    Learning from a “benign neglect strategy” in a national park: response of saproxylic beetles to dead wood accumulation

    Biol. Conserv.

    (2010)
  • E. Næsset

    Predicting forest stand characteristics with airborne scanning laser using a practical two-stage procedure and field data

    Remote Sens. Environ.

    (2002)
  • H. Pretzsch

    Analysis and modeling of spatial stand structures. Methodological considerations based on mixed beech-larch stands in lower Saxony

    For. Ecol. Manag.

    (1997)
  • H. Pretzsch et al.

    The single tree-based stand simulator SILVA: construction, application and evaluation

    For. Ecol. Manag.

    (2002)
  • P.R. Stephens et al.

    Airborne scanning LiDAR in a double sampling forest carbon inventory

    Remote Sens. Environ.

    (2012)
  • M. Svoboda et al.

    Natural development and regeneration of a central European montane spruce forest

    For. Ecol. Manag.

    (2010)
  • A. Wehr et al.

    Airborne laser scanning — an introduction and overview

    ISPRS J. Photogramm. Remote Sens.

    (1999)
  • M.A. Wulder et al.

    LiDAR sampling for large-area forest characterization: a review

    Remote Sens. Environ.

    (2012)
  • H. Arefi et al.

    A morphological reconstruction algorithm for separating off-terrain points from terrain points in laser scanning data

    Int. Arch. Photogramm. Remote. Sens. Spat. Inf. Sci.

    (2005)
  • P. Axelsson

    DEM generation from laser scanner data using adaptive TIN models

    Int. Arch. Photogramm. Remote Sens.

    (2000)
  • A. Baddeley et al.

    Spatial Point Patterns: Methodology and Applications with R

    (2015)
  • C. Bässler et al.

    The BIOKLIM Project: Biodiversity Research between Climate Change and Wilding in a Temperate Montane forest : The Conceptual Framework

    (2009)
  • C. Bässler et al.

    Detection of Climate-Sensitive Zones and Identification of Climate Change Indicators: A Case Study from the Bavarian Forest National Park

    (2010)
  • C. Bässler et al.

    Effects of resource availability and climate on the diversity of wood-decaying fungi

    J. Ecol.

    (2010)
  • D. Bayer et al.

    Structural crown properties of Norway spruce (Picea abies [L.] karst.) and European beech (Fagus sylvatica [L.]) in mixed versus pure stands revealed by terrestrial laser scanning

    Trees

    (2013)
  • K.M. Bergen et al.

    Remote sensing of vegetation 3-D structure for biodiversity and habitat-review and implications for LiDAR and radar spaceborne missions

    J. Geophys. Res. G: Biogeosci.

    (2009)
  • J. Besag

    Discussion of Dr Ripley's paper

    J. R. Stat. Soc.

    (1977)
  • P. Biber et al.

    SILVA 2.2: Benutzerhandbuch

    (2000)
  • F.H. Bormann et al.

    Pattern and Process in a Forested Ecosystem

    (1979)
  • L. Breiman

    Random forests

    Mach. Learn.

    (2001)
  • P. Čížková et al.

    Natural regeneration of acidophilous spruce mountain forests in non-intervention management areas of the Šumava National Park–the first results of the biomonitoring project

    Silva Gabreta

    (2011)
  • P.J. Clark et al.

    Distance to nearest neighbor as a measure of spatial relationships in populations

    Ecology

    (1954)
  • N. Coops et al.

    Estimating canopy structure of Douglas-fir forest stands from discrete-return LiDAR

    Trees

    (2007)
  • M. Dassot et al.

    The use of terrestrial LiDAR technology in forest science: application fields, benefits and challenges

    Ann. For. Sci.

    (2011)
  • D.C. Donato et al.

    Multiple successional pathways and precocity in forest development: can some forests be born complex?

    J. Veg. Sci.

    (2012)
  • J.S. Evans et al.

    A multiscale curvature algorithm for classifying discrete return LiDAR in forested environments

    Geosci. Remote Sens. IEEE Trans. On

    (2007)
  • A. Fischer

    Forstliche Vegetationskunde

    (2003)
  • K.v. Gadow et al.

    Forest Structure and Diversity

  • S.J. Goetz et al.

    LiDAR remote sensing variables predict breeding habitat of a Neotropical migrant bird

    Ecology

    (2010)
  • E. González-Ferreiro et al.

    Estimation of Stand Variables in Pinus radiata D. Don Plantations Using Different LiDAR Pulse Densities. Forestry

    (2012)
  • S. Gupta et al.

    Tree species detection using full waveform LiDAR data in a complex forest

  • M. Heurich

    Nationalpark Bayerischer Wald: Waldentwicklung im Bergwald nach Windwurf und Borkenkäferbefall

    (2001)
  • M. Heurich

    Progress of forest regeneration after a large-scale Ips typographus outbreak in the subalpine Picea abies forests of the Bavarian Forest National Park

    Silva Gabreta

    (2009)
  • Cited by (15)

    • Characterizing forest disturbance and recovery with thermal trajectories derived from Landsat time series data

      2022, Remote Sensing of Environment
      Citation Excerpt :

      However, the spatial variability in the thermal data increased significantly after the onset of disturbance and it was much larger than prior to the disturbance. This effect is caused by the high spatial canopy heterogeneity that characterizes natural recovery, as also documented by Svoboda et al. (2010) and Hill et al. (2017). By contrast, spectral Landsat time series do not show an increase in data variability during forest recovery in the same research area (Hais et al., 2009).

    • Extraction of individual trees based on Canopy Height Model to monitor the state of the forest

      2022, Trees, Forests and People
      Citation Excerpt :

      However, Ayrey et al. pointed out that the results of the algorithm depended significantly on the type of forest. Hill et al. (2017) concluded that the tree density parameters do not significantly affect the estimation of the tree attributes. On the other hand, various wireless data exchange protocols, such as Bluetooth, Bluetooth Low Energy (BLE), and ANT wireless, can also be used as positioning systems.

    • Metabarcoding reveals landscape drivers of beetle community composition approximately 50 years after timber harvesting

      2021, Forest Ecology and Management
      Citation Excerpt :

      Previous studies of ground-active beetles in managed landscapes in Tasmania and elsewhere have found high site-to-site turnover in assemblage composition, and that site effects are usually stronger than other ecological gradients like edge and riparian effects (Baker et al. 2006; Baker et al. 2007; Bitencourt and Da Silva 2016; da Silva et al. 2018). In light of the importance of fine-scale spatial position on beetle community composition, maintaining the spatial structural heterogeneity of regeneration forests with a range of age classes and reserved mature forests in a landscape is valuable for conserving beetle biodiversity in managed forests (Baker et al. 2016; Hill et al. 2017; Tinya et al. 2019). Consistent with the idea of niche characteristics contributing to spatial patterns of beetle communities, topographic position index (TPI), the precipitation of driest month and elevation had high relative importance among the variables considered in GDM.

    • Joint estimation of leaf area density and leaf angle distribution using TLS point cloud for forest stands

      2021, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
    View all citing articles on Scopus
    View full text