Grazemore DSS för att prediktera beteskvalitet för mjölkkor

Abstract: The aim of this study was to examine if the predictions of the herbage quality in the software Grazemore Decision Support System (DSS) gives a reliable ground for milk production in the north of Scandinavia. Pasture samples from one research farm (Umeå) and one organic farm (Nordingrå) was analysed on crude protein and organic matter digestibility. The results were statistically compared to the predicted values. Measured and predicted herbage mass was compared and a control if the predictions of milk production improved if the predicted input were replaced by the values from the analysis, was made. The concentration of crude protein was underestimated by the model on both farms and the relationship between actual and predicted values was poor. Mean Prediction Error (MPE) was 24% and 31% respectively. The organic matter digestibility was slightly overestimated, but there were a significant relationship between the analysed and the predicted values and both farms had a MPE at 7%. Herbage mass was measured during two summers in Umeå and one summer in Nordingrå. The model gave underestimations of the values in Umeå both years, while the mass was overestimated for Nordingrå. The relationship was statistically significant (p<0,05) on both farms 2004, but no significance was found for Umeå 2005. The relationship between measured milk yield and milk yield predicted with input from HGM was stronger then when the actual inputs were used. Though, the MPE was relatively low, 6%, when using predicted input and MSPE was mostly due to line (83%). When using actual values as input to HIM, the MPE was 16% and MSPE was mostly due to bias. The model has a good potential to predict the organic matter digestibility and the herbage mass, but it takes further development to make the predictions of crude protein more reliable.