Understanding 3-PG using a sensitivity analysis

Abstract

A detailed parameter sensitivity analysis of the 3-PG model [J.J. Landsberg and R.H. Waring, A generalised model of forest productivity using simplified concepts of radiation-use efficiency, carbon balance and partitioning, For. Ecol. Manage. 95 (1997), 209–228] was conducted as an aid to understanding the model, and to identify model parameters and site factors that need to be accurately determined for reliable applications of 3-PG to plantation-grown Eucalyptus grandis. The analysis examined the sensitivity of 3-PG outputs to parameters that characterise allometric relationships and biomass partitioning, canopy structure, branch and bark fractions, litterfall, canopy conductance and the effects of temperature, soil water and site nutrition on canopy quantum efficiency, and to variations in the site and climatic factors needed to run 3-PG. The analysis was based on data from 31 sites with widely differing climatic and site conditions, and the environmental variation of sensitivity was examined. Stand volume is highly sensitive to the ratio of net to gross primary production, canopy quantum efficiency, canopy conductance and basic wood density, and moderately sensitive to root partitioning, effects of fertility on canopy quantum efficiency and the light extinction coefficient. The sensitivities of stand volume to parameters characterising the foliage:stem partitioning ratio and the specific leaf area of seedlings vary markedly across sites. Leaf area index is highly sensitive to the ratio of net to gross primary production, canopy conductance, canopy quantum efficiency, and to the litterfall rate, foliage:stem partitioning ratio and specific leaf area of mature trees. Both stand volume and leaf area index are moderately sensitive to, and non-linear functions of, the optimum temperature for growth. On the sites studied, stand volume and root biomass were sensitive to site fertility, rainfall and mean temperature, but not to the maximum available soil water. Leaf area index was highly sensitive to the power in the allometric relationship between stem mass and mean stem diameter, and moderately sensitive to the constant in that relationship. On the other hand, stem volume was far less sensitive to these two parameters. These effects of stem allometry on stand volume and canopy leaf area were explained by constructing a causal loop diagram to visualise the feedbacks and dynamics inherent in the relevant parts of 3-PG. The work presented here is the first comprehensive sensitivity analysis of 3-PG, and the results will prove invaluable in subsequent rigorous parameterisations of the model for E. grandis and for other species. It also illustrates how a sensitivity analysis can lead to greater understanding of the manner in which complex models function.