Dairy manure temperature dynamics during storage

Abstract. One of the main factors that influence nitrogen (N) loss from manure storage is temperature. Increased microbial activities, rates of chemical reactions and processes, and higher N loss rates occur at higher manure temperatures. This study focuses on establishing the relationship between measured manure temperature and the ambient air conditions during the storage period at the farm-scale level. Manure storage structures, a clay pit, and a concrete tank used in the study had temperature sensors mounted on poles at different depths. The poles, placed at various locations in the storage structure. At each farm, a weather monitoring station (adjacent to the storage structure) collects ambient air temperature, rainfall, solar radiation, windspeeds, wind direction, and relative humidity. Data analysis included visualization, statistical analysis, and machine learning to discern relationships between manure temperature and weather parameters. In general, the average manure temperature was higher than the ambient air temperature by 3.5 and 6°C in the clay pit and the concrete tank, respectively. Manure temperatures were different by depth; the manure closer to the bottom of the storage structure was generally higher than those at depths above it during cold periods and vice versa. Another observation was that the temperature of the ambient air closer to the manure surface was higher than the general ambient air temperature at the farm. Of the different machine learning algorithms used to explore potential to predict the manure temperature from the ambient weather conditions, the Ensembles Bagged Trees and Time-Series Neural Networks presented the best (lowest RMSE and highest R2). The model predictions were better in warm seasons (lowest RMSE). Improving manure temperature estimation not only make models for estimating emissions more accurately but also informs the understanding of microbial and biogeochemical reactions in manure during storage.