Evaluating the ability of LPJ-GUESS to simulate the tree size structures of tropical forests

Sammanfattning: Tropical forests are of great importance to all living-beings due to their high biodiversity and the valu-able resources, such as food and fuel, they provide. In addition, tropical trees sequester a high amount of carbon and consequently over half of the global forest carbon stock can be found in the tropics. Climate change, however, might weaken this carbon sink and possibly result in tropical forests turning into net carbon sources. The tree size structures of forests and growth and mortality processes are strongly related to each other and determine the accumulation of biomass. LPJ-GUESS is a Dynamic Global Vegetation Model (DGVM), which can be used to simulate tropical forest dynamics. However, it has not been thoroughly evaluated against field data of tropical tree size distributions. The aim of this study was therefore to further enhance our understanding of the behaviour of LPJ-GUESS by means of comparing simulated distributions of tree size versus tree density and tree size versus biomass to a validation dataset. Six locations within the tropical rainforest ecozone were chosen across Africa, South America and Southeast-Asia. It was found that LPJ-GUESS generally underestimates the total amount of biomass. A total of eleven parameters were adjusted according to the one-at-a-time principle, which showed large variability between the six sites. The results of the parameter changes are described and are consistent with expectations from the literature. An increase in the disturbance interval, for example, stimulated the accumulation of biomass whilst reducing the tree density. This can be explained by the fact that fewer disturbances cause trees to grow older and consequently store more biomass, whilst their bigger crowns reduce the amount of light passing through the canopy. Recommendations for further studies are presented and stress the importance of exploring parameters interactions. Based on the results from this study it is advised to explore the re-sponse of LPJ-GUESS when altering the disturbance interval, crown area and kallom1 (an allometric constant) simultaneously.