A revised above-ground maximum biomass layer for the Australian continent

Standard

A revised above-ground maximum biomass layer for the Australian continent. / Roxburgh, Stephen H.; Karunaratne, Senani B.; Paul, Keryn I.; Lucas, Richard; Armston, John A.; Sun, Jingyi.

In: Forest Ecology and Management, Vol. 432, 15.01.2019, p. 264-275.

Research output: Contribution to journalArticle

Harvard

Roxburgh, SH, Karunaratne, SB, Paul, KI, Lucas, R, Armston, JA & Sun, J 2019, 'A revised above-ground maximum biomass layer for the Australian continent' Forest Ecology and Management, vol. 432, pp. 264-275. https://doi.org/10.1016/j.foreco.2018.09.011

APA

Roxburgh, S. H., Karunaratne, S. B., Paul, K. I., Lucas, R., Armston, J. A., & Sun, J. (2019). A revised above-ground maximum biomass layer for the Australian continent. Forest Ecology and Management, 432, 264-275. https://doi.org/10.1016/j.foreco.2018.09.011

Vancouver

Roxburgh SH, Karunaratne SB, Paul KI, Lucas R, Armston JA, Sun J. A revised above-ground maximum biomass layer for the Australian continent. Forest Ecology and Management. 2019 Jan 15;432:264-275. https://doi.org/10.1016/j.foreco.2018.09.011

Author

Roxburgh, Stephen H. ; Karunaratne, Senani B. ; Paul, Keryn I. ; Lucas, Richard ; Armston, John A. ; Sun, Jingyi. / A revised above-ground maximum biomass layer for the Australian continent. In: Forest Ecology and Management. 2019 ; Vol. 432. pp. 264-275.

Bibtex - Download

@article{b5c846b98bc840c3a2515d1832dd2d11,
title = "A revised above-ground maximum biomass layer for the Australian continent",
abstract = "The carbon accounting model FullCAM is used in Australia’s National Greenhous Gas Inventory to provide estimates of carbon stock changes and emissions in response to deforestation and afforestation/reforestation. FullCAM-predicted above-ground woody biomass is heavily influenced by the parameter M, which defines the maximum upper limit to biomass accumulation for any location within the Australian continent. In this study we update FullCAM’s M spatial input layer through combining an extensive database of 5739 site-based records of above-ground biomass (AGB) with the Random Forest ensemble machine learning algorithm, with model predictions of AGB based on 23 environmental predictor covariates. A Monte-Carlo approach was used, allowing estimates of uncertainty to be calculated. Overall, the new biomass predictions for woodlands, with 20–50{\%} canopy cover, were on average 49.5 ± 1.3 (s.d.) t DM ha−1, and very similar to existing model predictions of 48.5 t DM ha−1. This validates the original FullCAM model calibrations, which had a particular focus on accounting for greenhouse gas emissions in Australian woodlands. In contrast, the prediction of biomass of forests with a canopy cover >50{\%} increased significantly, from 172.1 t DM ha−1, to 234.4 ± 5.1 t DM ha−1. The change in forest biomass was most pronounced at sub-continental scales, with the largest increases in the states of Tasmania (166 to 351 ± 22 t DM ha−1), Victoria (201 to 333 ± 14 t DM ha−1), New South Wales (210 to 287 ± 9 t DM ha−1), and Western Australia (103 to 264 ± 14 s.d. t DM ha−1). Testing of model predictions against independent data from the savanna woodlands of northern Australia, and from the high biomass Eucalyptus regnans forests of Victoria, provided confidence in the predictions across a wide range of forest types and standing biomass. When applied to the Australian Government’s National Inventory land clearing accounts there was an overall increase of 6{\%} in continental emissions over the period 1970–2016. Greater changes were seen at sub-continental scales calculated within 6° × 4° analysis tiles, with differences in emissions varying from −21{\%} to +35{\%}. Further testing is required to assess the impacts on other land management activities covered by the National Inventory, such as reforestation; and at more local scales for sequestration projects that utilise FullCAM for determining abatement credits",
keywords = "forest biomass, random forest, carbon accounting, national greenhouse gas inventory",
author = "Roxburgh, {Stephen H.} and Karunaratne, {Senani B.} and Paul, {Keryn I.} and Richard Lucas and Armston, {John A.} and Jingyi Sun",
year = "2019",
month = "1",
day = "15",
doi = "10.1016/j.foreco.2018.09.011",
language = "English",
volume = "432",
pages = "264--275",
journal = "Forest Ecology and Management",
issn = "0378-1127",
publisher = "Elsevier",

}

RIS (suitable for import to EndNote) - Download

TY - JOUR

T1 - A revised above-ground maximum biomass layer for the Australian continent

AU - Roxburgh, Stephen H.

AU - Karunaratne, Senani B.

AU - Paul, Keryn I.

AU - Lucas, Richard

AU - Armston, John A.

AU - Sun, Jingyi

PY - 2019/1/15

Y1 - 2019/1/15

N2 - The carbon accounting model FullCAM is used in Australia’s National Greenhous Gas Inventory to provide estimates of carbon stock changes and emissions in response to deforestation and afforestation/reforestation. FullCAM-predicted above-ground woody biomass is heavily influenced by the parameter M, which defines the maximum upper limit to biomass accumulation for any location within the Australian continent. In this study we update FullCAM’s M spatial input layer through combining an extensive database of 5739 site-based records of above-ground biomass (AGB) with the Random Forest ensemble machine learning algorithm, with model predictions of AGB based on 23 environmental predictor covariates. A Monte-Carlo approach was used, allowing estimates of uncertainty to be calculated. Overall, the new biomass predictions for woodlands, with 20–50% canopy cover, were on average 49.5 ± 1.3 (s.d.) t DM ha−1, and very similar to existing model predictions of 48.5 t DM ha−1. This validates the original FullCAM model calibrations, which had a particular focus on accounting for greenhouse gas emissions in Australian woodlands. In contrast, the prediction of biomass of forests with a canopy cover >50% increased significantly, from 172.1 t DM ha−1, to 234.4 ± 5.1 t DM ha−1. The change in forest biomass was most pronounced at sub-continental scales, with the largest increases in the states of Tasmania (166 to 351 ± 22 t DM ha−1), Victoria (201 to 333 ± 14 t DM ha−1), New South Wales (210 to 287 ± 9 t DM ha−1), and Western Australia (103 to 264 ± 14 s.d. t DM ha−1). Testing of model predictions against independent data from the savanna woodlands of northern Australia, and from the high biomass Eucalyptus regnans forests of Victoria, provided confidence in the predictions across a wide range of forest types and standing biomass. When applied to the Australian Government’s National Inventory land clearing accounts there was an overall increase of 6% in continental emissions over the period 1970–2016. Greater changes were seen at sub-continental scales calculated within 6° × 4° analysis tiles, with differences in emissions varying from −21% to +35%. Further testing is required to assess the impacts on other land management activities covered by the National Inventory, such as reforestation; and at more local scales for sequestration projects that utilise FullCAM for determining abatement credits

AB - The carbon accounting model FullCAM is used in Australia’s National Greenhous Gas Inventory to provide estimates of carbon stock changes and emissions in response to deforestation and afforestation/reforestation. FullCAM-predicted above-ground woody biomass is heavily influenced by the parameter M, which defines the maximum upper limit to biomass accumulation for any location within the Australian continent. In this study we update FullCAM’s M spatial input layer through combining an extensive database of 5739 site-based records of above-ground biomass (AGB) with the Random Forest ensemble machine learning algorithm, with model predictions of AGB based on 23 environmental predictor covariates. A Monte-Carlo approach was used, allowing estimates of uncertainty to be calculated. Overall, the new biomass predictions for woodlands, with 20–50% canopy cover, were on average 49.5 ± 1.3 (s.d.) t DM ha−1, and very similar to existing model predictions of 48.5 t DM ha−1. This validates the original FullCAM model calibrations, which had a particular focus on accounting for greenhouse gas emissions in Australian woodlands. In contrast, the prediction of biomass of forests with a canopy cover >50% increased significantly, from 172.1 t DM ha−1, to 234.4 ± 5.1 t DM ha−1. The change in forest biomass was most pronounced at sub-continental scales, with the largest increases in the states of Tasmania (166 to 351 ± 22 t DM ha−1), Victoria (201 to 333 ± 14 t DM ha−1), New South Wales (210 to 287 ± 9 t DM ha−1), and Western Australia (103 to 264 ± 14 s.d. t DM ha−1). Testing of model predictions against independent data from the savanna woodlands of northern Australia, and from the high biomass Eucalyptus regnans forests of Victoria, provided confidence in the predictions across a wide range of forest types and standing biomass. When applied to the Australian Government’s National Inventory land clearing accounts there was an overall increase of 6% in continental emissions over the period 1970–2016. Greater changes were seen at sub-continental scales calculated within 6° × 4° analysis tiles, with differences in emissions varying from −21% to +35%. Further testing is required to assess the impacts on other land management activities covered by the National Inventory, such as reforestation; and at more local scales for sequestration projects that utilise FullCAM for determining abatement credits

KW - forest biomass

KW - random forest

KW - carbon accounting

KW - national greenhouse gas inventory

U2 - 10.1016/j.foreco.2018.09.011

DO - 10.1016/j.foreco.2018.09.011

M3 - Article

VL - 432

SP - 264

EP - 275

JO - Forest Ecology and Management

JF - Forest Ecology and Management

SN - 0378-1127

ER -

Show download statistics
View graph of relations
Citation formats