The accuracy of ground based estimates using the six-class dwarf mistletoe rating system was evaluated in an old-growth Douglas-fir/western hemlock forest by comparing ground ratings by five different observers to an examination of tree crowns from a construction crane at the Wind River Canopy Crane Research Facility, Washington. A total of 139 dominant, codominant, and intermediate western hemlock were evaluated. No consistent pattern emerged to indicate where overall error was made. All but

more »

... e observer was relatively accurate at identifying noninfected trees (73 to 95% noninfected trees accurately rated by four observers). However, the observers were less accurate at estimating the DMR class of the infected trees (11% to 3 7% of the infected trees accurately rated), including a number of trees incorrectly rated by two or more DMR classes. One observer rated 98% of the trees as having infections, while the crane survey estimated only 53% of the trees as having infections. Each observer divided the individual tree crowns into thirds and estimated infections based on summing lower, middle, and upper canopy levels. There was no pattern to the errors associated with estimates by canopy level. One observer significantly overestimated all canopy levels, one observer estimated all accurately, one observer significantly underestimated all canopy levels, one observer underestimated the mid and upper canopies, and one observer underestimated the lower and upper canopies. The principal reasons for inaccurate dwarf mistletoe ratings were assumed to be difficulty in accurately estimating crown thirds, misidentification of infections because of various stem deformities or accumulation of organic debris resembling infections, and the difficulty in observing infections high above the ground and through dense vegetation. Only one of the five observers accurately represented the spatial pattern of the infection center. The implications of this research vary depending on whether the results are to be used for timber management or for research and modeling applications. West. J. Appl. For. 15(1):8-14.