Evaluation of 4 carbon dioxide mass balance models for estimating the ventilation rate of naturally ventilated dairy barns.

Quantification of gaseous emissions from dairy cows requires a ventilation rate estimation which is in naturally ventilated barns often deduced from the modeled CO2 production rate. To date, the so-called CIGR CO2 production model has been widely used for this purpose. However, the cow-based parameters of the model were formulated based on animals whose characteristics might be outdated. In addition, the contribution of the indoor slurry storage to the total CO2 production at barn level is only roughly accounted for by the model (fixed relative contribution of 10% of the total). Therefore, the objective of the present study was to evaluate the prediction accuracy of the CIGR model for dairy cows and to compare it to those of 3 newer (animal-based) models recently published in 2024 by Kjeldsen et al. Additionally, the contribution of slurry to total CO2 production was assessed. In this study, the daily mean CO2 production rate (g/d per cow) of 2 mechanically ventilated compartments in an experimental barn were monitored during 3 periods (spring, summer, autumn) of 3 weeks. Each compartment housed 16 lactating Holstein-Friesian cows. Cows were intermittently taken out of the compartments for 7 d to differentiate CO2 production emitting from slurry from that of animals. During each period, the slurry volume was altered to investigate its effect on CO2 production. Predicted CO2 production rates by the different models were evaluated against the measured CO2 production rates. All models were evaluated on their animal (n = 12) prediction. Predicted CO2 production rates from slurry (n = 12) and at house level (n = 12) were evaluated only for the CIGR model. Results showed that the CIGR model underestimated animal-based CO2 production rate by 20 to 22% depending on the chosen productivity parameter (energy corrected milk yield or uncorrected milk yield). Animal-based production of CO2 as predicted by the 3 Kjeldsen models resulted in an 8-9% underestimation. The Kjeldsen models after mean bias correction performed even better (misestimation of 0 to 5%), with their model 2 after mean bias correction being the most accurate. At compartment level, slurry contributed on average 6% to the total CO2 production. This is below the 10% assumed in the CIGR model. However, more in-depth research is needed to evaluate slurry contribution to total barn emissions in response to varying climate conditions and slurry characteristics.