Comparing predictions of forest aboveground biomass of LINKAGES v2.2, PnET-II, and ED2 with long-term field data in temperate forests of the United States

Background/Question/Methods

Aboveground forest biomass (AFB) is an important source of forest products and plays a significant role in global carbon cycles.  Various forest models have been used to predict AFB dynamics.  These models generally fall into three groups: empirical gap models, ecosystem and/or biogeochemical process models, and hybrid models integrating both modeling approaches.  It was expected that the hybrid models would potentially predict more accurately for AFB than either gap or process models.  Comparisons among predictions of conceptually different models could help identify limitations and strengths in prediction accuracy and model applicability.  Thus, we compared AFB predictions of a gap model (LINKAGES v2.2), an ecosystem process model (PnET-II), and a hybrid model (ED2) with mid- and long-term field data at two temperate forest sites: Sinkin Experimental Forest in Missouri (approx. 30 years), and Kaskaskia Experimental Forest in Illinois (77 years).  The AFB predictions were compared among three models and to the field data in different time frames.

Results/Conclusions

In general, the AFB predictions by LINKAGES v2.2, PnET-II, and ED2 models are different at each site in short-term (< 20 years), mid-term (20-50 years) and long-term (> 50 years) time frames.  At each site, the AFB predictions of each model showed discrepancies from the field data, and these discrepancies generally increase from short-term to long-term time frames.  Despite the fact that the three temperate forest sites span over a wide environment gradient and forest compositions are different, ED2 simulations of AFB showed the best concordance with the field data at three sites, followed by the LINKAGES v2.2 simulations, and PnET-II simulations showed the least concordance.  This supports the expectation that models employing both empirical relationships and ecological processes could predict more accurately for AFB.  Results from this study revealed limitations and strengths of each model in predicting long-term AFB dynamics in temperate forests.