64 
Teil C - Annex 
Figure 10: Bird detection probability as a function of distance at sea (n = 694) (unpublished, data: 
Institute for Avian Research “Vogelwarte Helgoland”). 
See Table 3.2.2: Visual observations/recording of flight calls. 
Process instruction for night-time automated survey of bird calls for identification of 
species in the offshore region 
Call recording should be carried out with an omnidirectional microphone of high sensitivity 
and a very good signal-to-noise ratio. To minimise disturbance by wind and wave sounds and 
to increase the microphone’s life, it should be water-proofed by wrapping in thin plastic film 
and set up in a basket-type windscreen with fur cover; if possible, it should be suspended by 
rubber bands, thus mechanically decoupled from the mounting. If possible, the microphone 
system should be set up with free range in all directions (tip of microphone vertically upwards). 
At fixed locations, an alternative may be use of a directional microphone with known cha 
racteristics, since, as the case may be, mathematical correction of the recording range is 
possible only under stationary circumstances. For data recording, the microphone should be 
connected to a computer via a microphone pre-amplifier with phantom power for the micro 
phone. As a rule, connection is via a corresponding sound card (internal, external or on the 
computer’s motherboard). After set up, the system should be regularly calibrated at least 
twice a year by playing previously recorded bird calls using speakers at various distances and 
low volume. The recording level must be set so that the calls are recorded by the microphone 
system while at the same time they can be heard by an experienced ornithologist. 
The recordings should be saved as uncompressed WAV files (16 bit, mono); a sampling rate of 
22 kHz is sufficient. The used recording software must be AROMA (Automatic Recording of 
Migrating Aves), a software for the automatic recording of bird calls based on the script lan 
guage Tcl/Tk, which was developed by Dr Ommo Huppop from the Institute for Avian Research 
“Vogelwarte Helgoland”. Based on the audio processing tool kit “Snack”, this software con 
tinuously examines the incoming audio signal to detect peaks, that is, sounds that steeply 
pitch above a previously set minimum frequency (maximum of 1,500 Hz; cf. Hill & Huppop