Modelling global Gross Primary Production using the correlation between key leaf traits

Abstract: Sophisticated ecosystem models make it possible to evaluate the potential future changes of the carbon sequestration capacity of the terrestrial biosphere, as a response to the rapid environmental and climatic changes. Accuracy of model estimates is however strongly dependent on the parametrisation of driving parameters. A previous study of Wang et al. (2012) suggests, that the knowledge of the relationship between key leaf traits may be used to constrain modelled global terrestrial GPP ranges. Access to extensive leaf trait databases (such as GLOPNET and TRY) open possibilities to develop a more mechanistic rather than empirical based methods of representing vegetation in ecosystem models. Prentice et al. (2015) suggests that a stochastic parametrisation approach – like the one applied here – should be considered as a future improvement in ecosystem model development. This thesis discusses the effect of varying key leaf attribute values on derived GPP estimates. Leaf parameters – specifically leaf longevity, leaf nitrogen content and leaf mass per area – are varied within their potential ranges, either individually one-at-a-time or leaf longevity and leaf N traits simultaneously. The methods are applied for LPJ-GUESS DGVM and a simple idealised model (LEIA), that accounts for GPP’s dependency on leaf traits. According to the results, adjusting leaf lifespan values for evergreen and summergreen groups, as well as leaf N yielded a substantial reduction in global annual GPP variance, along with a decrease in mean global estimates. Findings suggest that using the correlation between leaf attributes may significantly improve LPJ-GUESS’s performance.