Abstract
Abstract In 2018 Sterman et al (2018a) published a simple dynamic lifecycle analysis (DLCA) model for forest-sourced bioenergy. The model has been widely cited since its publication, including widespread reporting of the model’s headline results within the media. In adapting a successful replication of the Sterman et al (2018a) model with open-source software, we identified a number of changes to input parameters which improved the fit of the model’s forest site growth function with its training data. These relatively small changes to the input parameters result in relatively large changes to the model predictions of forest site carbon uptake: up to 92 tC.ha −1 or 18% of total site carbon at year 500. This change in estimated site carbon resulted in calculated payback periods (carbon sequestration parity) which differed by up to 54 years in a clear-fell scenario when compared with results obtained using previously published parameters. Notably, this uncertainty was confined to forests which were slower growing and where the model’s training dataset was not sufficiently long for forests to reach maturity. We provide improved parameterisations for all forest types used within the original Sterman et al (2018a) paper, and propose that these provide better fits to the underlying data. We also provide margins of error for the generated growth curves to indicate the wide range of possible results possible with the model for some forest types. We conclude that, while the revised model is able to reproduce the earlier Sterman et al (2018a) results, the headline figures from that paper depend heavily on how the forest growth curve is fitted to the training data. The resulting uncertainty in payback periods could be reduced by either obtaining more extensive training data (including mature forests of all types) or by modification of the forest growth function.
Abstract from the paper’s OpenAlex record; open copy licensed cc-by. Read at the source ↗
The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.
Claims 11
Claim · model-written summary
Changes to model inputs improved fit of the forest site growth function to its training data.
Verbatim
e0“changes to input parameters which improved the fit of the model’s forest site growth function with its training data”
Claim · model-written summary
Small parameter changes altered predicted forest site carbon uptake by up to 92 tC.ha −1, or 18% at year 500.
Verbatim
e0“These relatively small changes to the input parameters result in relatively large changes to the model predictions of forest site carbon uptake: up to 92 tC.ha −1 or 18% of total site carbon at year 500”
Claim · model-written summary
Changed forest site carbon estimates altered calculated payback periods by up to 54 years in a clear-fell scenario.
Verbatim
e0“This change in estimated site carbon resulted in calculated payback periods (carbon sequestration parity) which differed by up to 54 years in a clear-fell scenario when compared with results obtained using previously published parameters”
t0“forest”
Claim · model-written summary
The uncertainty was confined to slower growing forests whose training data did not extend to maturity.
Verbatim
e0“this uncertainty was confined to forests which were slower growing and where the model’s training dataset was not sufficiently long for forests to reach maturity”
Claim · model-written summary
The authors provide improved parameterisations for all forest types of the original model.
Verbatim
e0“We provide improved parameterisations for all forest types used within the original Sterman et al (2018a) paper”
Claim · model-written summary
The revised forest parameterisations provide better fits to the underlying data.
Verbatim
e0“propose that these provide better fits to the underlying data”
t0“forest”
Claim · model-written summary
The authors provide margins of error for generated forest growth curves.
Verbatim
e0“We also provide margins of error for the generated growth curves to indicate the wide range of possible results possible with the model for some forest types”
Claim · model-written summary
The revised model can reproduce the earlier forest biomass model results.
Verbatim
e0“the revised model is able to reproduce the earlier Sterman et al (2018a) results”
t0“forest”
Claim · model-written summary
The earlier headline figures depend heavily on how the forest growth curve is fitted to training data.
Not reportedDirection
Verbatim
e0“the headline figures from that paper depend heavily on how the forest growth curve is fitted to the training data”
Claim · model-written summary
The authors recommend obtaining more extensive training data, including mature forests, to reduce payback-period uncertainty.
Verbatim
e0“The resulting uncertainty in payback periods could be reduced by either obtaining more extensive training data (including mature forests of all types)”
Claim · model-written summary
The authors recommend modifying the forest growth function to reduce payback-period uncertainty.
Verbatim
e0“or by modification of the forest growth function”