Quantifying Australian forest floristics and structure using small footprint LiDAR and large scale aerial photography

doi:10.1016/j.foreco.2005.11.021

Forest Ecology and Management

Volume 223, Issues 1–3, 1 March 2006, Pages 379-394

https://doi.org/10.1016/j.foreco.2005.11.021 Get rights and content

Abstract

Light detection and ranging (LiDAR) data and large scale (1:4000) photography (LSP) were investigated for their potential to quantify the floristics and structure of mixed species forests near Injune, central east Queensland, and to scale these up to the region for purposes of baseline assessment and on-going monitoring. For a 220,000 hectare (ha) area, LiDAR and LSP were acquired over 150 500 m × 150 m (7.5 ha) primary sampling units (PSUs) located on a ∼4 km systematic grid. Based on LSP interpretation, 292 species combinations were observed, although forests were dominated or co-dominated primarily by Callitris glaucophylla, Eucalyptus melanaphloia, Eucalyptus populnea and Angophora Leiocarpa. Comparisons with species distributions mapped using LSP and in the field suggested a 79% correspondence for dominant species. Robust relationships were observed between LiDAR and field measurements of individual tree (r² = 0.91, S.E. = 1.34 m, n = 100) and stand (r² = 0.84, S.E. = 2.07 m, n = 32) height. LiDAR-derived estimates of plot level foliage/branch projected cover (FBPC), defined by the percentage of returns >2 m, compared well (r² of 0.74, S.E. = 8.1%, n = 29) with estimates based on field transects. When translated to foliage projected cover (FPC), a close correspondence with field measurements (r² = 0.62, S.E. = 6.2%, n = 29) was observed. Using these relationships, floristics and both height and FPC distributions were estimated for forests contained with the PSU grid and extrapolated to the study area. Comparisons with National Forest Inventory (NFI), National Vegetation Information System (NVIS) and Queensland Herbarium data suggested that sampling using LSP and LiDAR aggregated to the landscape provided similar estimates at the broad level but allowed access to a permanent and more detailed record. Observed differences were attributed to different scales of data acquisition and mapping. The cost of survey was also reduced compared to more traditional methods. The method outlined in the paper has relevance to national forest monitoring initiatives, such as the Continental Forest Monitoring Framework currently being evaluated in Australia.

Introduction

As a signatory to international agreements, including the United Nations Framework Convention on Climate Change (UNFCCC) and the Montreal Process for sustainable forest management, Australia is increasingly required to provide accurate and quantitative information on the species/community composition (herein referred to as floristics), structure and condition of it's forests through time (MPIG, 2001, Barrett et al., 2001). In addition, such information is required by governments, industry, private landholders and the public to detect trends in commercial, biodiversity and greenhouse values (NFI, 1998, NFI, 2003, AGO, 2000, Henry et al., 2002), assess the performance of management practices and public policies, guide sustainable development and forecast the future condition of these ecosystems (NFI, 2003). However, undertaking such assessments within Australia represents a significant challenge for two main reasons. First, Australia has an estimated 164 million hectares (ha) of native forests, which are distributed largely around the outer margins of the continent. Second, around 70% of these forests are under private management and less than 10% are in commercial public forest estates (NFI, 2003). In the areas under private management, the information available on structure and condition is especially limited (MPIG, 2001). The development of efficient and cost-effective methods for retrieving this essential information is therefore critical if international obligations are to be better fulfilled and the sustainable development and conservation of forest resources optimised.

The overall objective of this research, therefore, was to evaluate whether large scale (1:4000) stereo aerial photography (herein referred to as LSP) and/or small footprint light detection and ranging (LiDAR) data could be used as tools, either singularly or in combination, for routinely sampling, describing and quantitatively assessing the floristics and structure of these forests. Focusing on areas of agricultural land and mixed species forests in central Queensland, which were considered typical of those occurring across large areas of Australia, the study aimed specifically to evaluate whether: (a) floristics could be described through air photograph interpretation (API) of LSP, (b) measures of structure (e.g., height and canopy cover) could be estimated from LiDAR data, (c) the resulting quantitative estimates of each could be extrapolated to the landscape with levels of reliability comparable to or better than those currently available and (d) data from these sensors combined offered a viable and cost-effective alternative or supplement to methods used currently for on-going regional assessment and monitoring of forests.

Section snippets

Background

Although LSP has been used as a basic forest inventory tool for some time (e.g., Spencer, 1992), the integration of LSP and LiDAR data has only been possible in the past few years due to advances in sensor design and data acquisition and processing. The following sections therefore provide a brief overview of these two systems and their use in Australia.

Study area

To evaluate the use of both LSP and LiDAR for quantifying the floristics and structure of forests, an area of 37 km × 60 km (222,000 ha) of private and public land near Injune, central Queensland (Lat 25°32′S, Long 147°32′E), was selected (Fig. 1). The study area was chosen, as the forests are typical to those of much of Australia¹

Image and field data collection

The acquisition of image and field data was undertaken in four main stages (Table 1). In stage I, a systematic sampling scheme (Schreuder et al., 1993) was implemented to guide the acquisition of LSP (stage II) and LiDAR data (stage III). Following collection and initial interpretation of these data, forest inventory data were collected from selected areas (stage IV). The majority of the fieldwork was carried out during the period of LiDAR data acquisition and within 1-month of the LSP data

Post-processing of field and remote sensing

Following collection, the inventory data were analysed primarily to determine the species composition of the forests, so that the API could be better evaluated, and to generate tree and stand level estimates of height and cover that could be regressed against LiDAR data. For this purpose, further stages of LSP (stage V) and LiDAR data (stage VI) processing were necessary (Table 1).

Data analysis

To provide summary information on the forests, their floristic composition was described using LSP (stage VII) whilst estimates of tree and stand height and cover were retrieved from LiDAR data (stage VIII).

Results: tree and stand level estimates

Based on the analysis outlined above, the use of both LSP and LIDAR for tree and stand level assessment, in terms of floristics, tree height and canopy cover was evaluated.

Scaling up to the landscape

On the basis of the plot level relationships established with LSP (floristics) and LiDAR (height and canopy cover), predictions of mean attribute values and distributions at both the PSU (150 predictions) and SSU (4500 predictions) level for the entire 220,000 ha study area were generated. The following sections present a summary of the extrapolations and then compare the sampled distributions with the mapped distributions based on datasets currently used by both the Queensland and Federal

Discussion

The study has shown that LSP and LiDAR can provide estimates of stand level floristics and structure (e.g., canopy cover) which are more comprehensive, precise and of greater number compared to field measurements alone. Through API and the development of empirical relationships with LIDAR data, regional level estimates can be generated through simple extrapolation. This approach provides options for operational mapping of such attributes. These options are discussed in greater detail in the

Conclusions and recommendations

The research has demonstrated that sampling using LSP and/or LIDAR can provide quantitative assessments of floristics and key structural attributes (height, cover) which can be extrapolated across the landscape. These estimates are comparable to those generated using traditional wall-to-wall mapping approaches although absolute comparison is limited because of the coarser level of detail associated with many existing datasets. This feature highlights then the additional information that can be

Acknowledgements

The authors would like to acknowledge the support provided by the Australian Research Council, under their SPIRT program, the Cooperative Research Centre for Greenhouse Accounting, and Agriculture, Fisheries and Forestry Australia. We would also like to thank Kerstin Jones, Robert Denham, Norm Good, and the staff of UNSW, QDNR&M, BRS, QUT and QDPI Tropical Beef Centre.

References (49)

J.L. Lovell et al.
Simulation study for finding optimal LiDAR acquisition parameters for forest height retrieval
Forest Ecol. Manage.
(2005)
R.M. Lucas et al.
Empirical relationships between AIRSAR backscatter and forest biomass, Queensland, Australia
Remote Sens. Environ.
(2006)
J.E. Means et al.
Use of large footprint scanning airborne LIDAR to estimate forest stand characteristics in the western Cascades of Oregon
Remote Sens. Environ.
(1999)
T. Moffiet et al.
Airborne laser scanning: exploratory data analysis indicates potential variables for classification of individual trees or stands according to species
ISPRS J. Photogram. Remote Sens.
(2005)
E. Naesset
Determination of mean tree height of forest stands using airborne laser scanner data
ISPRS J. Photogram. Remote Sens.
(1997)
R. Nelson
Modelling forest canopy heights: the effects of canopy shape
Remote Sens. Environ.
(1997)
R.F. Nelson et al.
Determining forest canopy characteristics using airborne laser data
Remote Sens. Environ.
(1984)
R.F. Nelson et al.
Estimating forest biomass and volume using airborne laser data
Remote Sens. Environ.
(1988)
D. Riaño et al.
Estimation of leaf area index and covered ground from airborne laser scanner (LiDAR) in two contrasting forests
Agric. Forest Meterol.
(2004)
J.C. Suárez et al.
Use of airborne LiDAR and aerial photography in the estimation of individual tree heights in forestry
Comp. Geosci.
(2005)

A. Wehr et al.

Airborne laser scanning—an introduction and overview

Photogram. Eng. Remote Sens.

(1999)

A.H. Aldred et al.

Application of Airborne Laser to Forest Surveys, Information Report PI-X-51, Canadian Forestry Service

(1985)

Australian Greenhouse Office, 2000. National Greenhouse Gas Inventory: Land Use Change and Forestry Sector 1990–1999,...

D.J. Barrett et al.

Quantifying uncertainty in estimates of C-emissions from above-ground biomass due to historic land-use change to cropping in Australia

Global. Ch. Biol.

(2001)

Bureau of Meteorology, 2004....

Bureau of Rural Sciences, 2003. A Continental Forest Monitoring Framework for Australia – Background, Concept and...

W.H. Burrows et al.

Growth and carbon stock change in eucalypt woodlands in north-east Australia: ecological and greenhouse sink implications

Global. Ch. Biol.

(2002)

Carnahan, J.A., 1990. Vegetation. Atlas of Australian Resources, vol. 6, Third Series. Australian Surveying and Land...

Commonwealth of Australia, 2002. Implementation of Montreal Process Criteria and Indicator Reporting–An operational...

R. Dowling et al.

Vegetation classification of the riparian zone along the Brisbane River, Queensland, Australia using light detection and ranging (lidar) data and forward looking digital video

Can. J. Remote Sens.

(2003)

R.J. Fensham et al.

Estimating clearance of Acacia dominated ecosystems in central Queensland using Land-system mapping data

Aust. J. Bot.

(1998)

M.R. Harris et al.

An investigation into the possible inhibitory effects of white cypress pine (Callitris glaucophylla) litter on the germination and growth of associated ground cover species

Aust. J. Bot.

(2003)

B.K. Henry et al.

A review of the potential role of greenhouse gas abatement in native vegetation management in Queensland rangelands

Rangeland J.

(2002)

J. Hyyppa et al.

A segmentation-based method to retrieve stem volume estimates from 3-D tree height models produced by laser scanners

IEEE Trans. Geosci. Remote Sens.

(2001)

Cited by (64)

Modelling canopy gap probability, foliage projective cover and crown projective cover from airborne lidar metrics in Australian forests and woodlands
2020, Remote Sensing of Environment
Tree canopy density metrics (TCDM) derived from airborne lidar data are used in a range of crucial environmental monitoring, forestry and natural resource management applications. The derivation of spatially and temporally consistent TCDM, however, typically requires field calibration to account for differences in instrument/survey parameters. Lidar surveys with no coincident field measurements consequently will have an unknown error associated with TCDM limiting their application. In this study, we analysed an extensive set of lidar captures with coincident field data to determine the lidar TCDM that best match the canopy gap probability (P_gap), foliage projective cover (FPC) and crown projective cover (CPC). Furthermore, we developed and evaluated models designed to reduce the bias introduced by variations in lidar instrument and survey acquisition parameters. The dataset incorporated 148 field sites (100 m diameter circular plots) coincident with 13 different lidar surveys between 2008 and 2015, distributed across a range of Australian forests and woodlands. The best lidar metric for 1 − P_gap, achieving a root mean square error (RMSE) of 6.7% with 95% confidence intervals (CI) of 6.1–7.3%, was the proportion of all returns greater than a canopy height threshold (t_canopy) of 1.5 m above ground (d_all). The best metric for FPC (RMSE = 6.0%, CI = 5.3–6.7%) used the proportion of returns, weighted as the fraction of the number of returns recorded from each pulse (d_weighted), with t_canopy of 1.7 m. The best metric for CPC (RMSE = 7.0%, CI = 6.4–7.7%) was the proportion of 0.5 m pixels greater than 0.8 m above the ground, for an interpolated canopy height model (d_interp). Overall bias for these metrics was low (~1%), however, the bias for individual surveys varied significantly. For example, for one survey d_all consistently underestimated 1 − P_gap with a bias of −8.3%, while a different survey consistently overestimated 1 − P_gap with a bias of 3.8%. Elastic net regression models, using instrument, survey and plot parameters as predictor variables, were unable to consistently remove the bias. No relationships could be discerned between lidar parameters and the bias between lidar metrics and field measurements, potentially due to complex interactions between parameters, the spatial scale of the field plots, and uncertainties in field measurements and lidar attributes. Although the bias could not be modelled, the results provide metrics to derive P_gap, FPC and CPC with less than 10% error from lidar surveys captured with similar parameters across Australia (>600,000 km²).
Terrestrial laser scanner based 3D reconstruction of trees and retrieval of leaf area index in a forest environment
2019, Ecological Informatics
Three-dimensional reconstruction of trees and the estimation of biophysical parameters is significant for the management of forest resources, ecological studies carbon cycle and biodiversity. Terrestrial LiDAR data provides detailed, objective and three-dimensional measurement of forest structure and exact metrics of the tree canopies. Several methods for tree detection including canopy height models and raster interpolation models are based on commercial software and huge data processing. The objective of the given study is the three-dimensional reconstruction of trees by implementing segmentation algorithms and thereby estimating the Leaf Area Index of individual tree segments by terrestrial laser scanned data in the Mudumalai forests of Western Ghats, India. The hierarchical minimum cut segmentation method is used for the three-dimensional reconstruction of the individual trees by tracking cylinders along individual branches and trees in a hierarchical order. Super voxel clustering method is also implemented in the study for tree reconstruction and estimating the tree parameters. Leaf area index is calculated by applying a multivariate regression technique for the heights and the diameter obtained from both the segmentation methods. Results obtained indicated a strong correlation with the in-situ measurements which are obtained from the instruments. The approach addresses the applicability of segmentation algorithms which can be run fully automatically. The approach successfully reconstructed a high precision and realistic model of trees in the Western Ghats region which failed in the case of traditional tree modeling methods which requires multiple instruments operating simultaneously for extracting each parameter. The method proved that using TLS; multiple forest parameters can be estimated simultaneously.
Hierarchical integration of individual tree and area-based approaches for savanna biomass uncertainty estimation from airborne LiDAR
2018, Remote Sensing of Environment
Understanding the role that the vast north Australian savannas play in the continental carbon cycle requires reliable quantification of their carbon stock at landscape and regional scales. LiDAR remote sensing has proven efficient and accurate for the fine-scale estimation of above-ground tree biomass (AGB) and carbon stocks in many ecosystems, but tropical savanna remain under studied. We utilized a two-phase LiDAR analysis procedure which integrates both individual tree detection (ITC) and area-based approaches (ABA) to better understand how the uncertainty of biomass estimation varies with scale. We used estimations from individual tree LiDAR measurements as training/reference data, and then applied these data to develop allometric equations related to LIDAR metrics. We found that LiDAR individual tree heights were strongly correlated with field-estimated AGB (R² = 0.754, RMSE = 90 kg), and that 63% of individual trees crowns (ITC) could be accurately delineated with a canopy maxima approach. Area-based biomass estimation (ABA), which incorporated errors from the ITC steps, identified the quadratic mean of canopy height (QMCH) as the best single independent variable for different plot sample sizes (e.g. for 4 ha plots: R² = 0.86, RMSE = 3.4 Mg ha^− 1; and 1 ha plots: R² = 0.83, RMSE = 4.0 Mg ha^− 1). Our results show how ITC and ABA approached can be integrated to understand how biomass uncertainty varies with scale across broad landscapes. Understanding these scaling relationships is critical for operationalizing regional savanna inventories, monitoring and mapping.
Multi-resolution time series imagery for forest disturbance and regrowth monitoring in Queensland, Australia
2015, Remote Sensing of Environment
Citation Excerpt :
Both LiDAR datasets were processed to a 1 m spacing. These data were used to obtain estimates of height and cover for each PSU for the years 2000 and 2009 based on the methods outlined in Tickle et al. (2006). A comparison of these data allowed areas that had remained undisturbed or otherwise to be identified.
High spatio-temporal resolution optical remote sensing data provide unprecedented opportunities to monitor and detect forest disturbance and loss. To demonstrate this potential, a 12-year time series (2000 to 2011) with an 8-day interval of a 30 m spatial resolution data was generated by the use of the Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) with Landsat sensor observations and Moderate Resolution Imaging Spectroradiometer (MODIS) data as input. The time series showed a close relationship over homogeneous forested and grassland sites, with r² values of 0.99 between Landsat and the closest STARFM simulated data; and values of 0.84 and 0.94 between MODIS and STARFM. The time and magnitude of clearing and re-clearing events were estimated through a phenological breakpoint analysis, with 96.2% of the estimated breakpoints of the clearing event and 83.6% of the re-clearing event being within 40 days of the true clearing. The study highlights the benefits of using these moderate resolution data for quantifying and understanding land cover change in open forest environments.
Mapping forest growth and degradation stage in the Brigalow Belt Bioregion of Australia through integration of ALOS PALSAR and Landsat-derived foliage projective cover data
2014, Remote Sensing of Environment
Differentiation of forest growth stages through classification of single date or time-series of Landsat sensor data is limited because of insensitivity to their three-dimensional structure. This study therefore evaluated the benefits of integrating the Advanced Land Observing Satellite (ALOS) Phased Array L-band Synthetic Aperture Radar (PALSAR) L-band HH and HV polarisation response from the woody components of vegetation with Landsat-derived foliage projective cover (FPC). Focus was on 12 regional ecosystems (REs) distributed across the Brigalow Belt Bioregion (BRB) of Queensland, Australia, where different stages of growth dominated by brigalow (Acacia harpophylla) were widespread. From remnant areas of brigalow-dominated forests mapped previously for each RE by the Queensland Herbarium through field visits and interpretations of aerial imagery, frequency distributions of all three channels were extracted and compared to those of image segments generated using FPC and PALSAR data. For woody vegetation (with an FPC threshold of ≥ 9%) outside of the remnant areas, mature (non-remnant) forests were associated with segments where the HH and HV backscatter thresholds were within one standard deviation of the mean extracted for remnant forest. Early-stage regrowth was differentiated using an L-band HH threshold of <− 14 dB, common for all REs because of similarities in structure at this stage. The early-stage included forests regrowing over several decades and often occurred in areas recovering from recent clearing events. Objects falling between the early and mature stages were considered to be intermediate regrowth and/or degraded forest. All areas with an FPC < 9% were mapped as non-forest.
Within the BRB, the Queensland Herbarium established that forests with brigalow as a dominant or subdominant component originally occupied over 7.3 million ha but were reduced to 586,364 ha by 2009, with 460,499 ha (78.5%) having brigalow as the dominant component. Using the Landsat FPC and ALOS PALSAR data, an additional 722,686 ha of brigalow-dominated regrowth forest were identified giving a total forested area (brigalow-dominated remnant and secondary forest) of 1,183,185 ha or 17.2% of the area of the 12 REs. Within this area, the greater proportion of regrowth (368,473 ha or 31.1%) was mapped as early stage primarily because of recovery following recent clearance events. 230,551 (19.5%) ha and 123,662 ha (10.5%) were mapped as intermediate and mature (non-remnant) stages respectively and the remainder (38.9%) was remnant forest. Users’ and producers’ accuracies were, respectively, 81% and 69% for early regrowth and 71% and 89% for mature and intermediate stage forests combined. The mapping, which used Queensland Herbarium’s RE data to delineate brigalow extent, provided a structural, rather than age-based classification of growth stage, as is typically retrieved using time-series comparison of optical imagery. The regional estimates of growth/degradation stage generated for the BRB provide a basis for optimising the use and recovery of these threatened brigalow ecosystems with benefits for biodiversity and carbon sequestration.
Overstorey and topographic effects on understories: Evidence for linkage from cork oak (Quercus suber) forests in southern Spain
2014, Forest Ecology and Management
The understorey is a critical feature of forest ecosystems, affecting nutrient cycling, biodiversity, regeneration capacity and wildfire regimes. Understanding the interaction of environmental and canopy controls on understorey development is therefore important for forest management. We use airborne lidar to elucidate this interaction in a forest of cork oak (Quercus suber) mixed with the deciduous oak (Quercus canariensis) on complex topographic relief in southern Spain. Solar radiation (controlled by aspect) and topographic position were highly influential in the differential distribution of the two dominant canopy trees, and exerted some control on their canopy density and height. There was a detectable effect of both canopy characteristics and solar radiation on the amount of understorey shrub cover, indicating direct and indirect controls of the environment on this stratum. In cork oak forest, understorey shrubbery increased towards valley bottoms, at lower levels of solar radiation and under shorter canopies. Our results are relevant to the management of these forest systems in the face of future change. They suggest that maturation and closure of these even-aged stands, coupled with climate warming, may lead to impoverishment of the shrub layer, reducing the build-up of fuel but also the biodiversity value of these systems.

View all citing articles on Scopus

¹: Tel.: +61 2 62499111.

²: Tel.: +61 7 38969832.

View full text

Quantifying Australian forest floristics and structure using small footprint LiDAR and large scale aerial photography

Abstract

Introduction

Section snippets

Background

Study area

Image and field data collection

Post-processing of field and remote sensing

Data analysis

Results: tree and stand level estimates

Scaling up to the landscape

Discussion

Conclusions and recommendations

Acknowledgements

Forest Ecol. Manage.

Remote Sens. Environ.

Remote Sens. Environ.

ISPRS J. Photogram. Remote Sens.

ISPRS J. Photogram. Remote Sens.

Remote Sens. Environ.

Remote Sens. Environ.

Remote Sens. Environ.

Agric. Forest Meterol.

Comp. Geosci.

Photogram. Eng. Remote Sens.

Application of Airborne Laser to Forest Surveys, Information Report PI-X-51, Canadian Forestry Service

Quantifying uncertainty in estimates of C-emissions from above-ground biomass due to historic land-use change to cropping in Australia

Global. Ch. Biol.

Growth and carbon stock change in eucalypt woodlands in north-east Australia: ecological and greenhouse sink implications

Global. Ch. Biol.

Vegetation classification of the riparian zone along the Brisbane River, Queensland, Australia using light detection and ranging (lidar) data and forward looking digital video

Can. J. Remote Sens.

Estimating clearance of Acacia dominated ecosystems in central Queensland using Land-system mapping data

Aust. J. Bot.

An investigation into the possible inhibitory effects of white cypress pine (Callitris glaucophylla) litter on the germination and growth of associated ground cover species

Aust. J. Bot.

A review of the potential role of greenhouse gas abatement in native vegetation management in Queensland rangelands

Rangeland J.

A segmentation-based method to retrieve stem volume estimates from 3-D tree height models produced by laser scanners

IEEE Trans. Geosci. Remote Sens.