Evaluating GHG simulation performance of DNDC in a boreal grassland setting

Food production in boreal regions is tending to expand into areas previously considered unproductive as a result of climate change. With increasing human populations and uncertainty regarding existing supply chains as well as the long-term viability of traditional bread-basket regions, northern countries are increasingly looking to their own under-exploited resources to increase food security. This is not without its risks and mismanagement can cause stable carbon sinks to become sources due to land-use change. Agroecological models are an important tool for assessing the long-term effects of management, yet in boreal regions their use has so far been limited and model evaluations have only been carried out in a handful of instances. One reason for this is the limited availability of measured data against which to effectively compare simulated data. To overcome these obstacles, we used existing eddy-covariance measurements to evaluate the ability of the process based DNDC model to simulate GHG emissions from a timothy/red clover grassland in eastern Finland over a 4-year period. These initial results suggest that DNDC is able to simulate gross primary production (GPP) R2 = 0.68, MAE = 14.4, RMSE = 26.6 and rBIAS = -30.8%, net ecosystem exchange (NEE) R2 = 0.54, MAE = 12.7, RMSE = 20.2 and rBIAS = 16.4%, and ecosystem respiration (Reco) R2 = 0.74, MAE = 11.35, RMSE = 15.43 and rBIAS = -41.2%. We then used an assessment criteria scoring method, where “poor” = 1, “fair” = 2, “good” = 3, and “excellent” = 4 for each evaluation method. Using this method, we determined that GPP and NEE simulations were “fair” whereas Reco was “good”. These results indicate that DNDC can satisfactorily simulate GHG fluxes in a boreal grassland setting but further work is needed to refine model calibration and evaluation methods.