Quantifying Current and Future CO2 Fluxes Using Daycent and PEST Models from a Marginal Land Seeded to Switchgrass Production in South Dakota.

More accurate measurements and prediction of CO₂ fluxes are important factors for greenhouse gas (GHG) mitigation strategies. However, it is practically difficult to monitor CO₂ fluxes from different environmental conditions. Therefore, process-based models have been used to estimate GHG fluxes, and for developing mitigation strategies. In this study, DAYCENT, a process-based ecosystem model, was used for simulating CO₂ from switchgrass (Panicum virgatum L.) seeded to a marginal land previously used for cropland in South Dakota. The parameter estimation (PEST), an automatic calibration method, was used to calibrate the DAYCENT model. Data show that PEST significantly improved DAYCENT model estimations of CO₂ fluxes with a modeling efficiency (ME) of 0.66. The PEST-DAYCENT modeled CO₂ fluxes matched measured data reasonably well with a coefficient of determination (R²) value of 0.67, RSR (ratio of RMSE to SD) value of 0.58, and percent bias (PBIAS) value of 1.32%. Further, the calibrated model was used to predict CO₂ fluxes for future 2011 through 2095 based on changing temperature values from 0.1 to 3°C and precipitation changes from ±30%. Furthermore, we compared future CO₂ fluxes based on temperature changes under wet to drought conditions and precipitation changes under increase of 3°C temperature to normal current weather condition. This study concluded that (i) PEST can effectively improve DAYCENT model’s calibration using measured CO₂ fluxes; (ii) soil surface CO₂ fluxes showed an increased trend from 2011 to 2095 with mean of 3426.6 kg ha^-1 yr^-1; (iii) temperature changes under normal and wet conditions showed a positive linear impact on CO₂ fluxes, whereas, precipitation changes could slightly and non-monotonically impacted increased soil CO₂ fluxes under normal temperature.