Calibration of visually estimated distances to migrating seabirds with radar measurements

Abstract

Censusing seabirds from coastal areas requires reliable estimates ofbird numbers and the distances of the birds from the coastline. Logistical constraints make visual estimation of distances the only feasible method in many studies. We tested the accuracy of visually estimated offshore distances of six migratory seabird species in the Strait ofGibraltar using simultaneous measurements obtained by radar. Most birds (91%) were detected within 3 km of the coast and we truncated our calibration at this distance. We found a strong correlation between radar and visual estimates (R2 from 0.08 to 0.20 for different distances and observers. Among the factors potentially affecting the accuracy of = 0.83, P< 0.0001). The magnitude oferrors in visual estimates was moderate and ranged visual estimates of distance to seabird in our study were observer identity, bird species, bird behavior, and weather; the most parsimonious model in our study included observer identity as the only predictor, and no model with more than one predictor had a smaller Akaike’s information criterion value. Radar can be used to help train observers and to reduce biases in visual estimates of distances by means of calibration. When no other methods are available to accurately measure distances to seabirds, visual estimates of distances, recorded by experienced observers and once calibrated with radar (or other ground-truthing methods), may be acceptable for different species under a wide range of environmental conditions.