Epsilon Archive for Student Projects

Abstract

Information regarding cows grazing behaviour and time spent grazing can be of great interest to dairy farmers, since this information can indicate how well suited the pasture is to the cows. Feed intake at pasture is however challenging to measure. Automated systems for monitoring the behaviour of cows within dairy production have become increasingly important and relatively common.

The aim of this study was to validate an activity measuring device (HOBO® G Logger Pendant Data Logger, USA), a triaxial accelerometer that registers the cow’s head positions during grazing, in order to see if cows’ grazing behaviour on pasture could be distinguished from their non grazing behaviour. Another three axis accelerometer (IceTag3D™, IceRobotics Ltd, Scotland) was included and registered if the cows were in a standing or lying position. The validation was performed by comparing data from behavioural observations to information about head- and leg positions. By continuous sampling, the behaviour of 20 cows was studied during 5 hours each, divided into periods of 30 min.

Discriminant analysis showed that the HOBO® logger had an accuracy of 83.6% when estimating grazing behaviour. With addidtion of the IceTag™ logger the accuracy was 75.6%. When also non grazing behaviour was included the estimated accuracy for the HOBO® logger alone was 81.4%, which was lower compared to when the IceTag™ logger was included (84.2%). The HOBO® logger overestimated grazing behaviour with 5.1%, while the IceTag™ logger underestimated grazing behaviour with 4.3%.

The present study showed that the HOBO® logger is a relatively accurate and useful tool to automatically determine cows’ grazing behaviour at pasture. To ensure the result, further validation for a longer period of time, with a larger variation in behaviours is needed. In order to know how large the actual feed intake is, additional development of the logger is required.