Sensitivity analysis of 3-PG model for Pinus taeda (Loblolly Pine)

Background/Question/Methods

The 3-PG (Physiological Principles Predicting Growth) model is the most widely used model world-wide to simulate tree growth, stand productivity and carbon sequestration. It requires relatively few (42) physiological input parameters and when accurately parameterized, it can produce accurate predictions of growth. While widely used, there has been little attention given to parameter sensitivity.  Sensitivity analysis, which is the response of the model outputs to the changing values of input parameters, can provide valuable information, guiding researchers to the most important parameters to specify accurately. It can also give us a deeper understanding of the model, needed to accurately broaden its application to different site, weather and soil conditions. The goal of our study is to investigate 3-PG model parameter sensitivity with parameter values developed for loblolly pine. The model was calibrated using growth data from a loblolly pine stand in the coastal plain of Georgia and validated using data from three additional sites in Georgia, one in the coastal plain and two in the piedmont region.  Model sensitivity analysis was conducted by calculating the percentage change in all major outputs based on modifying the value of each input parameter by +10% and +20%, respectively.

Results/Conclusions

Our results indicate that the model is especially sensitive to the values of 8 input parameters. All of the outputs are highly affected by changes in the values of canopy quantum efficiency (α), ratio of net primary productivity to gross primary productivity (NPP/GPP), fertility modifier (fN0), extinction coefficient for PAR absorption by canopy (k), production days lost by frost (kF) and specific leaf area (SLA1), and optimum and maximum growth temperature. Our results also indicate that stand based mean tree diameter (avDBH) is highly sensitive to the stem taper function (as high as 300% increase with a 20%change in avDBH). All biomass outputs are sensitive to the turnover rate, maximum litterfall rate (gammaFx) and foliage to stem partitioning ratio (pFS20).  To our knowledge, this study is the first thorough sensitivity analysis for 3-PG parameterized for loblolly pine.  Our study can aid the future applications of the 3-PG model in this species and other species by clearly determining the importance of the imprecision or uncertainty in model inputs.