Observer Calibration Problem in Aerial Surveys of Pink Salmon

Aerial counts are the most practical means of monitoring escapement of pink salmon in Alaska, but these counts often have a large bias and variation due to observer-specific counting errors. We first assembled a data set composed of paired observer counts from two observers looking at spawning pink salmon. We developed an observer calibration (adjusting the counts of the assistant to those of the primary observer) to make the two observers counts comparable. We examined the calibration using more reliable on-the-ground mark-recapture estimates. The results indicated the observer calibration was not functioned as intended, and degraded the quality of the estimates. We could not justify implementing a between-observer calibration. We therefore suggest that aerial count data not be corrected in an observer-specific manner to calculate the escapement.