Modelling methane emissions from pig manure using statistical and machine learning methods

Abstract

Manure production and its management in the livestock sector have been increasingly receiving global attention due to its contribution to generating greenhouse gases, especially methane (CH4). This study was conducted to quantify and characterize daily manure including its moisture, dry matter (DM), ash, volatile solid (VS) contents, and model CH4 production rate as a function of feed intake and mass of pigs. Two statistical (multiple linear regression and polynomial regression) and three machine learning algorithms (ridge regression, random forest regression, and artificial neural network) were employed to predict CH4 emission. The result showed body mass ranged from 60 to 90 kg pig produced around 4.78 kg of manure per day consisting of 67% moisture content and 33% DM. The manure's ash content was 28% DM (0.45 kg pig(-1) day(-1)), while the VS was 72% DM (1.21 kg pig(-1) day(-1)). Moreover, the average CH4 production rate was estimated as 0.018 kg pig(-1) day(-1) which was lower than IPCC's (2006) recommended value for Oceania, Western Europe, and even North America regions. The current study found that the performance of the ridge regression was comparatively better, where the model with a coefficient of determination (R-2) greater than 90% was suitable for describing the relationship between the explanatory (feed intake and mass of pigs) and the response (CH4 emissions) variables. Further research may be conducted to improve this model's prediction accuracy, providing a wider range of diets and management conditions.