Rapid Quantitative Analysis of Nitrogen and Phosphorus through the Whole Chain of Manure Management in Dairy Farms by Fusion Model

Abstract Field monitoring technology plays a vital role for returning the animal manure back to the cropland with high-efficiency and accuracy, particular in the complex rotation system of manure management in Chinese intensive farms. The comprehensive quantitative analysis models were proposed and built for determining the content of the nitrogen (N) and the phosphorus (P) through the whole chain of manure management in different dairy farms under multiple conditions. 249 manure samples were collected from 31 intensive dairy farms in Tianjin both in summer and autumn. The effect of seasons on the distribution characteristics of the N and P in the manure was analyzed. Near infrared spectra were collected and principal component analysis (PCA) was performed. Partial least squares (PLS) was used to establish the intra-season and inter-season models. It was found that the contents of the N and P in the manure varied with seasons. The prediction performance of intra-season models was better than that of inter-season models. Fusion model of two seasons were also established. The coefficient of determination of external validation (R2pred) for the N and P were 0.972 and 0.901, respectively. The residual predictive deviations (RPD) were 5.98 and 3.18, respectively. The results showed that the fusion model could enhance the universality and stability for predicting the N and P contents through the whole chain of manure management under the influence of various factors. The study not only supports for the development of on-spot detecting instrument but also guides for the rational recycling of manure in practice as well.