Accuracy of a global positioning system (GPS) for weed mapping

Experiments were conducted to test the accuracy of a global positioning system (GPS) in measuring the area of simulated weed patches nf varying size and to determine the accuracy in navigating back to particular points in a field. Circular areas of 5, 50, and 500 m(2) were established and measured using point and polygon features of a GPS. The GPS estimations of the area of those patches had errors ranging from 7 to 45%, 6 to 15%, and 3 to 6%, respectively, when compared to actual measurements. As patch size increased, errors decreased. A curve describing the relationship between GPS error and patch size had an excellent fit (r(2) = 0.92). The error remained the same in all measurements across all patch sizes, but composed a smaller percentage of large patches. The GPS had submeter accuracy in navigation to the correct quadrat 73% of the time, located the correct quadrat 27% of the time, and invariably navigated to within 1.58 m of the correct quadrat. The relationship between patch size and measurement error was applied to natural infestations of hemp dogbane.