In this study, we aimed to determine the ammonia emission characteristics through analysis of ammonia concentration, ventilation rate, temperature, and relative humidity pattern in a mechanically ventilated swine finishing facility in Korea. Three pig rooms with similar environmental conditions were selected for repeated experimentation (Rooms A–C). Ammonia concentrations were measured using a photoacoustic gas monitor, and ventilation volume was estimated by applying the least error statistical model to supplement the missing data after measurement at several operation rates using a wind tunnel-based method. The mean ammonia concentrations were 4.19 ppm, and the ventilation rates were 24.9 m³ h⁻¹ pig⁻¹. Ammonia emissions were calculated within the range of 0.40–5.01, 0.25–4.16, and 0.37–5.68 g d⁻¹ pig⁻¹ for Room A, Room B, and Room C, respectively. Ammonia concentration and ventilation rate showed a weak negative correlation (r = −0.13). Ammonia emissions were more markedly affected by ammonia concentration (r = 0.88) than ventilation rate (r = 0.31). This indicates that ammonia concentration reduction can be effective in reducing ammonia emissions. The mean daily ammonia emissions, which increased exponentially over the finishing periods, were calculated as 1.78, 1.57, and 1.70 g d⁻¹ pig⁻¹ for Room A, Room B, and Room C, respectively (average 1.68 g d⁻¹ pig⁻¹).

Conventional methods for monitoring ammonia (NH₃) emissions from livestock farms have several challenges, such as a poor environment for measurement, difficulty in accessing livestock, and problems with long-term measurement. To address these issues, we applied various neural network models for the long-term prediction of NH₃ concentrations from sow farms in this study. Environmental parameters, including temperature, humidity, ventilation rate, and past records of NH₃ concentrations, were given as inputs to the models. These neural network models took the encoder or the feature extracting parts from the representative deep learning models, including Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and Transformer, to encode temporal patterns of time series. However, all of these models adopted dense layers for the decoder to format the task of long-term prediction as a regression problem. Due to their regression nature, all models showed a robust performance in predicting long-term NH₃ concentrations at a scale of weeks or even months despite there being a relatively short period of input signals (a few days to a week). Given one week of input, LSTM showed the minimum mean absolute errors (MAE) of 1.83, 1.78, and 1.87 ppm for the prediction of one, two, and three weeks, respectively, whereas Transformer performed best with a MAE of 1.73 ppm for a four-week prediction. In the long-term estimation of spanning months, LSTM showed the minimum MAEs of 1.95 and 1.90 ppm when trained on predicting two and three weeks of windows. At the same condition, Transformer gave the minimum MAEs of 1.87 and 1.83 when trained on predicting one and four weeks of windows. Overall, the neural network models can facilitate the prediction of national-level NH₃ emissions, the development of mitigation strategies for NH₃-derived air pollutants, odor management, and the monitoring of animal-rearing environments. Further, their integration of real-time measurement devices can significantly prolong device longevity and offer substantial cost savings.

This study was conducted in three similar-structured pig rooms for approximately 160 days from March 25 to August 31, 2021 to investigate greenhouse gas emissions and characteristics emitted from sow housing with forced ventilation systems. Greenhouse gas emissions and coefficients were calculated by real-time measurement of greenhouse gas (Nitrous Oxide, Methane) concentrations, ventilation rate, and inside temperature and humidity in a pig house based on the mega chamber method, which assumes a closed pig room in the form of a chamber. The average concentration of Nitrous Oxide (N₂O) in each room was in the range of 0.5 to 0.6 ppm, and Methane (CH₄) concentration was in the range of 28.9 to 33.5 ppm. The average N₂O emissions from Rooms 1 to 3 were 4.7 ± 1.4, 4.5 ± 1.3, and 5.2 ± 1.6 mg/day·m²·pig. In the case of CH₄, average emissions were 86.0 ± 15.6, 100.2 ± 31.2, and 103.3 ± 26.7 mg/day·m² ·pig. Finally, greenhouse gas emissions from sows calculated in this study were 1.87 ± 0.6 g/year·m²·pig (N₂O) and 35.4 ± 8.8 g/year·m²·pig (CH₄). Greenhouse gas emission coefficients were calculated as carbon dioxide equivalent (COSUB2-eq/SUB) by applying Global Warming Potential (GWP) and were 1.38 g COSUB2-eq/SUB and 2.64 g COSUB2-eq/SUB for N₂O and CH₄, respectively. This suggests that the contribution of CH₄ is higher than that of N₂O emitted by sows. The results of this study present useful data for domestic emission coefficient research.