Sensitivity analysis of a carbon simulation model and its application in a montane forest environment

Abstract

Accurate estimation of Net Primary Productivity (NPP), which is a key component of the terrestrial carbon cycle, is very important in studies of global climate. Ecosystem models have been used for NPP estimates. Determining how much each source of uncertainty contributes to modeled NPP is veiy important before ecosystem models can be used with confidence over larger areas and time periods. This research has systematically evaluated the boreal ecosystem productivity simulator (BEPS) carbon model in mountainous terrain, Kananaskis, Alberta. After parameterization of the model, sensitivity analysis was conducted as a controlled series of experiments involving sensitivity simulations with BEPS by changing a model input value in separate model runs. The results showed that NPP was sensitive to most model inputs measured in the study area, but that the most important input variables for BEPS were LAI and forest species. In addition, the NPP uncertainty resulting from topographic influence was approximately 3.5 %, which is equivalent to 140 kg C ha"1 yr"1. This suggested that topographic correction for the model inputs was also important for accurate NPP estimation. Using the topographically corrected data, the carbon dynamics were simulated, and average annual NPP production by forests in Kananaskis was estimated at 4.01 T ha"1 in 2003.