Advertisement

Remote Sensing Measurements of Forest Structure Types for Ecosystem Service Mapping

  • Rico FischerEmail author
  • Nikolai Knapp
  • Friedrich Bohn
  • Andreas Huth
Chapter

Abstract

Forests represent an important pool in the global carbon cycle. However, biomass stocks and carbon fluxes are variable due to the fact that forest dynamics are driven by processes that act on different spatial and temporal scales. Estimating forest biomass and productivity for larger regions is therefore a major challenge. In this study, horizontal and vertical forest structure is used to improve forest ecosystem service mapping by remote sensing. By linking remote sensing techniques with vegetation modelling (here FORMIND) and forest inventories, forest structure maps were derived for Germany (resolution 4 km). Using these maps, the role of forest structure for selected ecosystem services of forests has been investigated. For forest state estimations (like biomass) horizontal forest structure plays a key role while for productivity estimations both horizontal and vertical structures are relevant. This concept of forest structure classification in combination with forest modelling and remote sensing has high potential for applications at continental scales as future remote sensing missions will provide information on forest structure.

Keywords

Forest structure Biomass Productivity Remote sensing Forest model Germany 
  • Which ecosystem services are addressed? Ecosystem services provided by forests: forest biomass and forest productivity. Knowledge about these ecosystem services can be used to deliver further forest ecosystem services related to climate regulation, soil protection, biodiversity protection, water regulation, disturbance regulation, and bioenergy

  • What is the research question addressed? How can we estimate forest structure from remote sensing, and what is the role of forest structure for forest biomass and productivity estimations?

  • Which method has been applied? Linking forest inventory data, forest model simulations, and remote sensing

  • What is the main result? Forest structure can be estimated from remote sensing by using structural indices. Additionally, we show that over a broad range of forest stands, forest structure is the important driver for estimating forest biomass and forest productivity

  • What is concluded, recommended? Future remote sensing missions will provide information on forest structure with a high degree of detail. This will lead to more accurate estimations of forest biomass and productivity

11.1 Introduction

Twenty-five percent of Earth’s land surface is covered by forests, and they are habitat for more than 70% of all terrestrial species [1, 2, 3]. Forests represent an important pool in the global carbon cycle as they bind huge amounts of carbon in their living biomass [4, 5, 6]. They are also able to regulate the water cycle through processes of evapotranspiration, which is important for stabilizing the global climate [7]. Additionally, forest management is an important economic sector in Europe.

Global forests are characterized by complex patterns and structures. Forest dynamics are driven by processes that act on different spatial and temporal scales. Consequently, biomass stocks and carbon fluxes are variable in space and time. Therefore, estimating forest properties such as biomass or productivity for larger regions is a major challenge. The IPCC (Intergovernmental Panel on Climate Change) reported that missing knowledge on biomass distribution is one large source for uncertainty in the global carbon cycle [6, 8].

Forest canopy height, derived from active remote sensing systems such as lidar or radar, is often used as predictor for forest biomass. However, a significant amount of variance remains unexplained. One approach to improving height-to-biomass relationships is to consider horizontal and vertical forest structure , as forest structure is a key element for forest properties. Ground-based tests have shown that classifying stands according to structure indices, e.g., the stand density index [9] and modified species profile index [10], can lead to more accurate height-to-biomass relationships within each structure class. Hence, an important goal is to classify forest stands into structure types (horizontal and vertical structure) based on remote sensing measurements. For each forest structure type, biomass and productivity of forests could then be estimated more accurately compared to a general estimation (see Fig. 11.1).
Fig. 11.1

Workflow to classify forest stands into 16 structure types and predict forest attributes (i.e., biomass and productivity)

11.2 Methods

11.2.1 Study Site

For this study we used ground-inventory data from the new forest megaplot Traunstein (https://forestgeo.si.edu/sites/europe/traunstein). It is a large permanent research plot, established as a new super test site (25 ha, 30,000 measured trees) in a highly diverse structured forest district in the German alpine upland—including even- and uneven-aged forest stands and 26 tree species. In 2016, a full tree survey was performed (stem diameter and position for each tree) and an airborne lidar campaign was conducted by the German Aerospace Center (DLR). Since 2017 this unique research plot is part of the global Smithsonian Tropical Forest Institute Network (ForestGEO).

To apply our local findings from Traunstein to Germany, we used the German national forest inventory data (‘BWI’; [11]), which consists of 48,562 field plots distributed across Germany.

11.2.2 Describing Forest Structure by Structural Indices

Different structural indices can be used to characterize the horizontal and vertical structure of forests. For each forest stand we calculated a horizontal and vertical structure index and related this stand to a forest structure type (Fig. 11.1). Horizontal structure can by described, for example, by stand basal area BA [m2], which is the sum of all tree basal area values:
$$ BA=\sum \limits_{trees}\frac{\uppi}{4}{d}^2, $$
where d [m] is the stem diameter of a tree.
Vertical structure can be quantified by tree height heterogeneity σheight [m]:
$$ {\upsigma}_{height}=\sqrt{\frac{1}{n-1}\sum \limits_{trees}{\left(h-\overline{h}\right)}^2}, $$
where h [m] is the height of a tree and \( \overline{h} \) [m] the mean tree height of a stand.
We focus in this study on these two indices (basal area and tree height heterogeneity), but other indices can also be applied. Another possible index for horizontal structure is, for example, stand density index SDI [−]:
$$ \mathrm{SDI}=\mathrm{N}\cdotp {\left(\frac{25}{\overline{\mathrm{d}}}\right)}^{-1.605}, $$
where N [1/ha] is the number of trees and \( \overline{d} \) [cm] the quadratic mean stem diameter (i.e., square root of the mean of the squares) of all trees of a stand. For the vertical structure, the modified species profile index S [−] can also be used, describing the basal area distribution in different height layers:
$$ S=-\frac{1}{\ln\;3}\;\sum \limits_1^3{p}_i\cdotp \ln {p}_i,\kern0.98em {p}_i=\frac{B{A}_i}{B{A}_{tot}}, $$
where BAi [m2/ha] is the basal area in height layer i and BAtot [m2/ha] is the total basal area of a stand. Three height layers were used , which were equally spaced between the ground and the maximum height.

11.2.3 Estimating Structural Indices from Lidar Remote Sensing

To find relationships between field-based forest structure and lidar, we used ground-inventory data from the Traunstein megaplot (see Sect. 11.2.1). Additionally, we analysed the airborne lidar campaign conducted for this forest (see Sect. 11.2.1). We explored relationships between field-based forest structure (here, basal area and tree height heterogeneity) and lidar on different scales (e.g., 20 m, 100 m). Depending on the spatial scale, we found good relationships between field-based and lidar-based structure index for horizontal forest structure (e.g., r2 = 0.77 for field basal area vs. lidar top-of-canopy height at the scale of 20 m). Relations for the vertical index are more challenging (e.g., r2 = 0.41 for tree height heterogeneity vs. lidar 90% height quantile at the scale of 20 m).

11.2.4 Classifying Forest Stands into Structure Types

Basal area and tree height heterogeneity can be used to classify forest stands into different structure types. For this we divided both indices into four classes: basal area (m2/ha) as horizontal structural descriptor: H1: 0–15, H2: 15–25, H3: 25–35, H4: >35, and tree height heterogeneity (m) as vertical structural descriptor: V1: 0–1, V2: 1–2, V3: 2–3, V4: >3. With this classification scheme we assign each forest stand to a vertical and horizontal structure class. Both indices can be estimated from lidar due to the derived relationships between field-based metrics and lidar.

This classification scheme was applied to the German national forest inventory (BWI) data [11]. As no wall-to-wall lidar data for Germany was available, we generated lidar data for the BWI plots using a lidar simulation model [12]. At the end we developed a Germany-wide forest structure map estimated from lidar remote sensing. The same type of maps can be derived also from radar (e.g., L-Band) as radar measurements can be also used to quantify forest structure. Similar maps can be generated using other structural indices (like SDI, not shown).

11.2.5 Forest Biomass and Productivity

In a second step, 300,000 virtual forest stands were analysed to identify the importance of forest structure for biomass and productivity estimations (“forest factory approach,” [13]). The virtual stands were generated with the individual-based forest model FORMIND [14]. We calculate for each forest stand (400 m2) biomass and productivity (here, aboveground woody productivity AWP) and relate them to the structural indices: here, basal area BA and standard deviation of tree heights σheight.

11.3 Results and Discussion

11.3.1 Classifying Forest Stands into Structural Classes Using Field Data and Lidar

A first application of the workflow shows how forest structure types can be derived from remote sensing (here, lidar). The maps for vertical and horizontal structure types in Germany estimated from lidar are shown in Fig. 11.2, the frequency distribution of the structure type classes in Fig. 11.3.
Fig. 11.2

Map of horizontal (H) and vertical (V) forest structure types in Germany estimated from simulated lidar data based on the German national forest inventory [11]. As horizontal index we used basal area (H1: 0–15, H2: 15–25, H3: 25–35, H4: >35 [m2]), as vertical index the tree height heterogeneity (V1: 0–1, V2: 1–2, V3: 2–3, V4: >3 [m]). Class 1 stands for a homogenous structure, class 4 for a heterogeneous structure

Fig. 11.3

Amount of forest area with a specific forest structure in Germany: horizontal (a) and vertical (b) forest structure types, estimated from simulated lidar data based on the German national forest inventory [11]. As horizontal index we used basal area, as vertical index tree height heterogeneity

According to this analysis, most forest stands in Germany have a high basal area >35 m2/ha (see Fig. 11.3). The amount of forest area with heterogeneous vertical structure (σheight > 2 m) is as high as the forest area with homogenous vertical structure (σheight < 2 m).

11.3.2 The Importance of Forest Structure for Biomass and Productivity Estimates

Analysing all 300,000 virtual forest stands (using FORMIND), we find that forest productivity (AWP) is hardly affected by species diversity. Instead, forest structure emerges as the key variable [12]. Here, we group the forest stands into sixteen forest structure classes, four horizontal and four vertical structure classes like in the presented forest structure maps (Fig. 11.2). We find an increase in biomass with basal area (basal area as proxy for horizontal structure), whereby biomass in forests with low basal area is much more influenced by vertical heterogeneity than forests with large basal area (Fig. 11.4a).
Fig. 11.4

Mean above-ground biomass (a) and mean aboveground wood productivity AWP (b) vs. forest structure classes. As horizontal structure index we used basal area (m2 / ha), as vertical structure index we used tree height heterogeneity (m; in colours). Error bars indicate the standard deviation

Forest productivity increases with basal area for stands with a low vertical heterogeneity (Fig. 11.4b). However, with increasing vertical heterogeneity, the positive effect of basal area on productivity diminishes. The reason is that large trees shade smaller trees, which reduces the productivity of smaller trees. To sum up, for forest state estimations (like biomass) the horizontal forest structure plays a key role (here, basal area; Fig. 11.4). For productivity estimations, however, the horizontal and vertical structures are relevant.

Using the derived forest structure maps for Germany, we can estimate forest biomass and forest productivity distributions for forests and explore relationships between forest structure and other forest properties . We show that over a broad range of forest stands, forest structures are the important drivers for estimating forest biomass and forest productivity. Knowledge about ecosystem functions like biomass and productivity is crucial for evaluating timber volume for forestry, estimating disturbance state of forests, and understanding the effects of climate change. The concept of forest structure types in combination with forest modelling and remote sensing has high potential for applications at larger scales . Future remote sensing missions (like BIOMASS, GEDI, Tandem-L) will provide information on forest structure with a high degree of detail.

Notes

Acknowledgements

We thank the Thünen Institute for providing the German national forest inventory (BWI) data. We also want to thank Hans Pretzsch, Peter Biber, and Michael Heym (TUM) for their input on forest structure and structure metrics. Kostas Papathanassiou, Victor Cazcarra-Bes, Matteo Pardini and Marivi Tello Alonso (DLR) gave useful insights into linking forest structure and remote sensing. This study was part of the Helmholtz-Alliance Remote Sensing and Earth System Dynamics. NK was funded by the German Federal Ministry for Economic Affairs and Energy (BMWi) under the funding reference 50EE1416.

References

  1. 1.
    Gibson L, Lee TM, Koh LP, Brook BW, Gardner TA, Barlow J, et al. Primary forests are irreplaceable for sustaining tropical biodiversity. Nature. 2011;478:378–81.CrossRefGoogle Scholar
  2. 2.
    Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J. Biodiversity hotspots for conservation priorities. Nature. 2000;403:853–8.CrossRefGoogle Scholar
  3. 3.
    Pimm SL, Jenkins CN, Abell R, Brooks TM, Gittleman JL, Joppa LN, et al. The biodiversity of species and their rates of extinction, distribution, and protection. Science. 2014;344:1246752.CrossRefGoogle Scholar
  4. 4.
    Bonan GB. Forests and climate change: forcings, feedbacks, and the climate benefits of forests. Science. 2008;320:1444–9.CrossRefGoogle Scholar
  5. 5.
    Grace J, Mitchard E, Gloor E. Perturbations in the carbon budget of the tropics. Glob Chang Biol. 2014;20:3238–55.CrossRefGoogle Scholar
  6. 6.
    Pan Y, Birdsey RA, Fang J, Houghton R, Kauppi PE, Kurz WA, et al. A large and persistent carbon sink in the world’s forests. Science. 2011;333:988–93.CrossRefGoogle Scholar
  7. 7.
    Reifsnyder WE. The role of forests in the global and regional water and energy balances. In: CAgM, editor. CAgM report, no 8. Geneva: World Meteorological Organization; 1982.Google Scholar
  8. 8.
    Intergovernmental Panel on Climate Change. Climate change 2014: impacts, adaptation, and vulnerability. Cambridge: Cambridge University Press; 2015.Google Scholar
  9. 9.
    Reineke LH. Perfecting a stand-density index for even-aged forests. J Agric Res. 1933;46:627–38.Google Scholar
  10. 10.
    Pretzsch H. Forest dynamics, growth and yield. Berlin: Springer Verlag; 2009. p. 281.Google Scholar
  11. 11.
    Dritte Bundeswaldinventur. Thuenen-Institut – Basisdaten (Stand 20.03.2015). 2015. https://bwi.info/Download/de/BWI-Basisdaten/ACCESS2003/. Accessed 7 Oct 2017.
  12. 12.
    Knapp N, Fischer R, Huth A. Linking lidar and forest modeling to assess biomass estimation across scales and disturbance states. Remote Sens Environ. 2018;205:199–209.CrossRefGoogle Scholar
  13. 13.
    Bohn FJ, Huth A. The importance of forest structure to biodiversity–productivity relationships. R Soc Open Sci. 2017;4:160521.CrossRefGoogle Scholar
  14. 14.
    Fischer R, Bohn F, Dantas de Paula M, Dislich C, Groeneveld J, Gutiérrez AG, et al. Lessons learned from applying a forest gap model to understand ecosystem and carbon dynamics of complex tropical forests. Ecol Model. 2016;326:124–33.CrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2019

Authors and Affiliations

  • Rico Fischer
    • 1
    Email author
  • Nikolai Knapp
    • 1
  • Friedrich Bohn
    • 1
  • Andreas Huth
    • 1
    • 2
    • 3
  1. 1.Department of Ecological ModellingHelmholtz Centre for Environmental Research–UFZLeipzigGermany
  2. 2.German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-LeipzigLeipzigGermany
  3. 3.Institute of Environmental Systems Research, University of OsnabrückOsnabrückGermany

Personalised recommendations