Forest biomass and volume estimation using airborne LiDAR in a cool-temperate forest of northern Hokkaido, Japan

https://doi.org/10.1016/j.ecoinf.2015.01.005Get rights and content

Highlights

  • We conducted large-scale (225 km2) airborne LiDAR survey in northernmost Japan.

  • Timber volume and biomass was linearly correlated with the LiDAR mean canopy height.

  • Regression including all forest types was adequate for large-scale biomass estimation.

Abstract

Trees are recognized as a carbon reservoir, and precise and convenient methods for forest biomass estimation are required for adequate carbon management. Airborne light detection and ranging (LiDAR) is considered to be one of the solutions for large-scale forest biomass evaluation. To clarify the relationship between mean canopy height determined by airborne LiDAR and forest timber volume and biomass of cool-temperate forests in northern Hokkaido, Japan, we conducted LiDAR observations covering the total area of the Teshio Experimental Forest (225 km2) of Hokkaido University and compared the results with ground surveys and previous studies. Timber volume and aboveground tree carbon content of the studied forest stands ranged from 101.43 to 480.40 m3 ha–1 and from 30.78 to 180.54 MgC ha–1, respectively. The LiDAR mean canopy height explained the variation among stands well (volume: r2 = 0.80, RMSE = 55.04 m3 ha–1; aboveground tree carbon content: r2 = 0.78, RMSE = 19.10 MgC ha–1) when one simple linear regression equation was used for all types (hardwood, coniferous, and mixed) of forest stands. The determination of a regression equation for each forest type did not improve the prediction power for hardwood (volume: r2 = 0.84, RMSE = 62.66 m3 ha–1; aboveground tree carbon content: r2 = 0.76, RMSE = 27.05 MgC ha–1) or coniferous forests (volume: r2 = 0.75, RMSE = 51.07 m3 ha–1; aboveground tree carbon content: r2 = 0.58, RMSE = 19.00 MgC ha–1). Thus, the combined regression equation that includes three forest types appears to be adequate for practical application to large-scale forest biomass estimation.

Introduction

Because trees are recognized as a carbon reservoir, precise and convenient methods for forest biomass estimation are required for adequate carbon management under the increasing CO2 concentration in the atmosphere. Traditionally, carbon densities have been assessed using field-based inventory plots, which are valuable but also expensive, time consuming and inherently limited in geographic representativeness (Asner, 2009). Remote sensing is considered to be a promising solution for large-scale forest biomass evaluation, and an airborne light detection and ranging (LiDAR) technique, in particular, has been developed for this purpose (Naesset, 1997). This technique can provide 3D information on land surfaces by measuring the distance between the laser ranging sensor installed on an airplane and points on the land surface and can yield aboveground carbon maps over thousands of hectares per day of flying, when combined with field calibration plots. The LiDAR-based maps can be used in a baseline analysis to initiate a long-term monitoring program that will subsequently rely primarily on low-cost satellite data and analysis methods (Asner, 2009).

Many LiDAR studies have evaluated the timber volume or biomass of forests with a relatively sparse canopy or plantation with simple age and size structures with considerable success (Boudreau et al., 2008, Hayashi et al., 2013, Tsuzuki et al., 2006, Yone et al., 2002, Yone et al., 2005, Zhao et al., 2009). On the other hand, quantification has been less successful in the case of natural forests with high aboveground biomass because of the difficulty in predicting the state of the forest below the canopy layer (Yone, 2008). However, several studies have succeeded in estimating forest biomass by simply relating the LiDAR canopy height to ground observations on forest biomass or volume (Asner, 2009, Asner et al., 2009, Lefsky et al., 2002). This approach has been recommended for dense natural forests to reduce the effort and expense required to develop global biomass estimates from satellite LiDAR data (Asner, 2009, Lefsky et al., 2002).

The aim of this study was to clarify the relationship between LiDAR canopy height and the timber volume/biomass of cool-temperate forests in northern Hokkaido, Japan. For this purpose, we conducted LiDAR observations covering the total area of the Teshio Experimental Forest (225 km2) of Hokkaido University and compared the results with ground surveys and previous studies.

Section snippets

Study site and ground surveys

The study site was located in the Teshio Experimental Forest, Hokkaido University (44°54′–45°06′N, 141°56′–142°10′E) in northernmost Hokkaido, Japan. The site has the characteristics of a mid-latitude cool-temperate climate, and it lies in the transition zone between temperate and sub-boreal forest ecosystems. About 90% of the studied forest is naturally regenerated forest and the other 10% is conifer plantation. The dominant tree species are Abies sachalinensis (F. Schmidt) Mast., Picea

Results and discussion

DSM or DTM showed good agreement between the two LiDAR observation in 2002 and 2004 in spite of the different sensor specifications and phenological stage of the forest trees and the understory (r2 = 0.99 for DSM and DTM), although the DTM in 2004 tended to be slightly higher than that in 2002 compared with the relationship between DSMs of the two years (slope of the regression line = 1.0058 for DSM and 1.0119 for DTM) (Fig. 2). Change in DSM by tree growth is considered to be little and within the

Acknowledgements

This research was part of the “CC-LaG experiment” project, which was carried out through the collaboration of Hokkaido University, the National Institute for Environmental Studies, and Hokkaido Electric Power Co., Inc. This work was partly supported by JSPS KAKENHI Grant Numbers 23255009, 25241002, 26292076 and 21114008. This study would not have been possible without the assistance of graduate students of Agriculture, Engineering, and Environment Science of Hokkaido University, and the

References (27)

  • A. Banskota et al.

    Synergistic use of very high-frequency radar and discrete-return lidar for estimating biomass in temperate hardwood and mixed forests

    Ann. For. Sci.

    (2011)
  • R.O. Dubayah et al.

    Estimation of tropical forest height and biomass dynamics using lidar remote sensing at La Selva, Costa Rica

    J. Geophys. Res.

    (2010)
  • M. Hayashi et al.

    Forest biomass mapping with airborne LiDAR in Yokohama City

    J. Jpn. Soc. Photogramm.

    (2013)
  • Cited by (28)

    • Hierarchical Bayesian geostatistics for C stock prediction in disturbed plantation forest in Zimbabwe

      2023, Ecological Informatics
      Citation Excerpt :

      Such information provides timely decision making and management by forest practitioners. With accuracies for both Landsat-8 and Sentinel-2 based models exceeding accuracies established in other similar environments (Ahmed et al., 2022; Fan et al., 2022; Jiang et al., 2021; Takagi et al., 2015; Xiong and Wang, 2022), the application of results of the present study can therefore be utilized for comprehensive and location-based ecological interventions. The Sentinel-2 based C stock predictive model yielded the shortest CIWs, implying more precise C stock predictions than the Landsat-8 based C stock estimates.

    • Combining hyperspectral imagery and LiDAR pseudo-waveform for predicting crop LAI, canopy height and above-ground biomass

      2019, Ecological Indicators
      Citation Excerpt :

      Airborne LiDAR systems can rapidly and accurately capture three-dimensional data (Qin et al., 2015). Previous studies showed that LiDAR data can reduce the saturation problem often encountered in optical remotely sensed data (Riaño et al., 2004; Zhao and Popescu, 2009), which have been broadly applied to predict canopy height (Gao et al., 2015; Glenn et al., 2011; Mielcarek et al., 2018; Pang et al., 2011), gap fraction and cover (Anderson et al., 2018; Hopkinson and Chasmer, 2009; Zheng et al., 2017), LAI (Heiskanen et al., 2015; Ma et al., 2014; Richardson et al., 2009) and biomass (Chen, 2015; Dalponte et al., 2018; Li et al., 2015; Ma et al., 2018; Nie et al., 2017; Takagi et al., 2015; Xi et al., 2016). Previous these studies showed that LiDAR data are able to reliably predict vegetation structural parameter and no saturation problem is found.

    • The use of fixed–wing UAV photogrammetry with LiDAR DTM to estimate merchantable volume and carbon stock in living biomass over a mixed conifer–broadleaf forest

      2018, International Journal of Applied Earth Observation and Geoinformation
      Citation Excerpt :

      Airborne light detection and ranging (LiDAR) data further improved the estimation accuracy of forest structural attributes such as volume, biomass, and carbon stock because of its ability to penetrate into the forest canopy, and reconstruct 3D forest structure. Further, the robustness of LiDAR data for forest resource assessment has been researched intensively over various forests types, e.g., boreal (Yoga et al., 2017; Næsset, 2002), tropical (Ioki et al., 2014; Asner et al., 2012), and temperate (Spriggs et al., 2017; Takagi et al., 2015) forests. However, satellite sensors may have limited applicability in operational forest management because they often require trade–offs between resolution, scale, and frequency whereas the application of LiDAR data is hindered by high acquisition costs (Ota et al., 2015; Puliti et al., 2017, 2015; White et al., 2015).

    • Aboveground biomass estimation using multi-sensor data synergy and machine learning algorithms in a dense tropical forest

      2018, Applied Geography
      Citation Excerpt :

      Remote sensing systems, both optical and active sensors, has been proved to be an effective alternative for measurement and monitoring of forest biomass at various scales and landscapes (Ali, Greifeneder, Stamenkovic, Neumann, & Notarnicola, 2015; Babcock et al., 2015; Kaasalainen et al., 2015; Su et al., 2016). Active remote sensing methods like Radio Detection and Ranging (RaDAR) and Light Detection and Ranging (LiDAR) have been regularly used in recent aboveground biomass estimation studies (Carreiras, Melo, & Vasconcelos, 2013; Carreiras, Vasconcelos, & Lucas, 2012; Cartus, Santoro, & Kellndorfer, 2012; Chen, 2015; Deng, Katoh, Guan, Yin, & Li, 2014; Li, Glenn, Olsoy, Mitchell, & Shrestha, 2015; Puliti et al., 2017; Takagi et al., 2015; Wang et al., 2016; Zhao et al., 2016). While RaDAR beams directly interacts with trunk and other parts containing biomass (Joshi et al., 2017), LiDAR measures the height of vegetation (Ghosh & Behera, 2017; Ho Tong Minh et al., 2016) which is a proxy to vegetation biomass (Erten, Lopez-Sanchez, Yuzugullu, & Hajnsek, 2016).

    View all citing articles on Scopus
    View full text