12,000

We have over 12,000 students, from over 100 countries, within one of the safest campuses in the UK

93%

93% of Lancaster students go into work or further study within six months of graduating

Home > Research > Publications & Outputs > Uncertainty and sensitivity in global carbon cy...
View graph of relations

« Back

Uncertainty and sensitivity in global carbon cycle modelling.

Research output: Contribution to journalJournal article

Published

Journal publication date27/02/1998
JournalClimate Research
Journal number3
Volume9
Number of pages18
Pages157-174
Original languageEnglish

Abstract

This paper summarises the results obtained from a stochastic sensitivity study in the area of global carbon cycle modelling. In particular, it outlines how Monte Carlo Simulation (MCS) techniques and associated Generalised Sensitivity Analysis (GSA) have been applied to a modified box-diffusion model. Using parametric and input uncertainty measures derived from independent sources, the analysis shows that the model produces an ensemble response for atmospheric CO2 whose mean evolution is reasonably consistent with that observed over the industrial period (1765 to 1990). However, the situation is more ambiguous in relation to the amplitude distribution of the stochastic realisations: here, there is a larger than expected variance arising from the assumed uncertainty, although this may be due to unavoidable limitations in the assumptions about parameteric uncertainty used in the MCS analysis. The analysis also shows that the a priori model response is less consistent with 13C measurements and not at all consistent with the 14C observations, while GSA suggests that, from over 20 model parameters, only a small number have a statistically significant effect on the model response in 1990. In addition, the results from MCS, using the IPCC (Intergovernmental Panel on Climate Change) Scenario IS92a inputs, demonstrate that the uncertainties in the projections of the CO2 concentrations using this model are significantly higher than those suggested so far using deterministic methods. Finally, the advantages and disadvantages of the stochastic methodology are discussed.

Bibliographic note

Copyright © 1998 Inter-Research