Reduction of location error in GPS collar tracking data of bovine cattle by using data screening
Loading...
Date
2012
Journal Title
Journal ISSN
Volume Title
Publisher
University of Namibia
Abstract
Global Positioning System (GPS) telemetry enables tracking of an individual animal over a long period of time and can provide researches with accurate information on animal movement. However, various environmental factors influence the satellite signals received by a GPS collar, potentially deteriorating accuracy of position. We tracked Caprivi Sanga cattle using Televilt Tellus Basic GPS collars in North-East Namibia in 2006 and 2007. Locations obtained during the night, when the cattle are gathered inside livestock enclosures, revealed that a significant proportion of the locations are inaccurate. We used data of seven GPS collars for testing different data screening options as a way to reduce location error. Basic analysis showed that simple measures of accuracy like dilution of precision (DOP) and figure of merit (FOM) are not sufficient to remove erroneous locations from the data. We removed the inaccurate locations with the following condition: 2D location with DOP >6 or 0< altitude <850 m or altitude >1050 m or DOP ≥10 or FOM ≥10 or walking speed of the animal over 4.5 km/h. This data screening option eliminated 75% of the most erroneous locations (>300 m from the livestock enclosures) retaining 97.2% of
the locations correctly located inside the livestock enclosures. Before data screening, 95% (1372) of the night-time locations that were located outside the livestock enclosures were located 71-406 m from the enclosures. The maximum error was over 10 km. After data screening, 95% (485) of the locations were 54-298 m from the enclosures; the maximum error was 4.4 km.
Description
Keywords
GPS, Bovine cattle, Data screening
Citation
Polojarvi, K., Colpaert, A. & Matengu, K. (2012). Reduction of location error in GPS collar tracking data of bovine cattle by using data screening. Journal for Studies in Humanities and Social Sciences, 1(2): 1-18.