Operator errors in tree crown transparency (CT) estimates using the semi-automatic image analysis system CROCO were examined. One experienced operator serving as a control and seven inexperienced operators analysed the same 73 digitised images obtained in Switzerland, including four European species, Norway spruce (Picea abies (L.) Karst.), Scots pine (Pinus sylvestris L.), Beech (Fagus sylvatica L.), Oak (Quercus robur L.). Before the processing, all operators were instructed in how to select a rectangular region of interest (ROI) to be analysed (that is, the part of the tree crown exposed to sunlight). Illustrative example images of tree crowns were used for this purpose. For Norway spruce and Scots pine, no significant differences in CT were found between any two operators. For Oak, two operators tended to underestimate CT as compared to the control and other inexperienced operators. For Beech, one operator tended to overestimate CT as compared to the others. The operator bias was mainly due to different delineation of the lower boundary of the sun-exposed crown. The existence of operator bias in selecting ROI suggests that the illustrative manual alone is not sufficient. Especially for broad-leaved trees, additional operator training is needed to obtain the most accurate estimates of CT using CROCO.
All Science Journal Classification (ASJC) codes
- Agronomy and Crop Science
- Computer Science Applications