Assessing the generality and accuracy of the TRIPLEX model using in situ data of boreal forests in central Canada

Abstract

TRIPLEX1.0 is a hybrid model that integrates three well-established process models including 3-PG, TREEDYN3.0 and CENTURY4.0. We have conducted calibrations using eight sites to determine and generalize parameters of the TRIPLEX. We also performed model validation using 66 independent data sets to examine the model accuracy and the generality of its application. Simulations were conducted for plots with large sample size from the boreal ecosystem atmosphere study (BOREAS) program, including the northern study area (NSA) near Thompson, Manitoba (55.7° N, 97.8° W) and the southern study area (SSA) near Prince Albert, Saskatchewan (53.7° N, 105.1° W). The calibrations and simulations emphasized on generating average parameters and initial statuses for applying a complex model in a broad region where site detailed information such as photosynthetic capacity, soil carbon, nutrient, soil water, and tree growth is not always available. A suggestion was presented regarding adjusting the sensitive parameter by estimating tree growth rate corresponding to different site conditions. The study actually presented a reasonable and balanced parameter generalization procedure that did not lead to a significant reduction of model accuracy, but did increase the model practicability. The comparison of observations and simulations produced a good agreement for tree density, mean tree height, DBH, soil carbon, above-ground and total biomass, net primary productivity (above-ground) and soil nitrogen in both short- and long-term simulation. Results presented here imply that the set of parameters generalized and suggested in this study can be used as basic referenced values, in which TRIPLEX can be applied to simulate the general site conditions of boreal forest ecosystems.