Detection of observer bias in ongoing forest health monitoring programmes

This article describes a method for detecting bias between observers in monitoring programmes where crown density is used as the indicator of forest health. The method concentrates on trees where the surveyor team has changed from year to year and compares the changes reported by each separate pair of observers. It shows that observer bias is present in the Norwegian monitoring programme. Observer bias can give an artificial impression of geographical patterns if surveyors are allowed to operate regionally and may also result in inconsistent or even false reports about change in crown density. Monitoring programmes cannot differentiate between effects of observer bias and real environmental changes unless the observers are assigned to monitoring sites on a permanent basis.