Figure 1.
Study area and position of the bird radar.
Map of the north-western part of the Balgzand intertidal area in the Wadden Sea, the Netherlands, showing the position of the radar (blue dot) and the transects used by visual observers (red arrows into the direction of observation). Concentric circles around the radar position are separated by a range of 1 km.
Table 1.
Altitude and distance classes used in the line transect surveys.
Figure 2.
Number of radar tracks that could be matched to a visual observation, as a function of an imposed time lag between visual observations and the set of radar tracks. This lag curve was calculated for the full visual observation set
. The solid line is a fit to Eq. 5, giving
s and
s.
Figure 3.
Time difference between consecutive visual observations.
Probability density histogram (2 s bins) for the time difference between consecutive observed transect crossings (bottom/left axes). The gray curve shows the fraction of valid observations for different values of , i.e. the fraction observations for which both the following and preceding observation are found at a time interval larger than
(top/right axes). The dashed vertical line indicates
s, as used in this study. This value implies that a fraction of 0.43 of the total number of observations is used for the validation (dashed horizontal line).
Table 2.
GAM models for the probability of detection compared by AIC.
Table 3.
Coefficients of the best GAM model for the probability of detection.
Figure 4.
GAM predictions for the probability of detection (POD).
Top: the effect of range for four altitude categories for the median size of birds in our study area (0.4 kg) above a flooded surface. Bottom: the effect of body mass for two altitude categories, both above a dry and wet surface. To illustrate the uncertainty estimates of POD, the area between the mean and the upper 1 bound of the confidence interval is indicated in grey.
Figure 5.
Average probability of detection (POD) as a function of range.
Each scatter point refers to a distinct distance class of one of the transects and its corresponding subset of visual observations from , drawn on the horizontal axis at its mean range. The modelled POD equals the mean GAM prediction for these observations. The observed POD equals the proportion of these observations that could be matched to a radar track directly. Lines indicate the upper and lower 1
confidence intervals.
Figure 6.
Average probability of detection (POD) per species in the radar range 0–1500 m.
POD values are shown for the 10 most frequently observed species, from top to bottom ordered by body mass: Great Cormorant, Common Eider, Common Shelduck, European Herring Gull, Eurasian Curlew, Eurasian Oystercatcher, Common Gull, Black-headed Gull, Sandwich Tern and Common Tern. The modelled POD equals the mean GAM prediction for all visual observations within this 0–1500 m range. The observed POD equals the proportion of these observations that could be matched to a radar track directly. Black dots indicate the average body mass per species.