Data processing a bat fly-by

It’s been a year since I last posted! Where did the time go? 

The MarkII has had a few changes over the year. I’ve moved to three D batteries for the main power supply, which gives a couple of nights continuous recording, and I’ve improved the wiring which has stabilised things a lot. The microphone array is now on the end of a ~1m cable and the electronic interference I was picking up in the data has pretty much gone. I’ve also started on coming up with a suitable enclosure for the microphones that lets ultrasound in but doesn’t cause unwanted reflections. Still some way to go there, so for now I’m waiting for good weather and recording without any covering.

The main issue this year though has been the data. The problem I found with the phase breaking up past the leading edge of a chirp is present in the MarkII and I’ve also seen it in stereo data from Wildlife Acoustics SM2 recorders. In order to get a good phase measurement and therefore a direction, I’ve written some algorithms to just pick out the leading edge. These work reasonably well, but I’m finding at the moment that inspection of individual chirps is needed in order to fine tune things and get the right delay.

Data fitting and inspection software for localising the audio signals. The left-hand panels show amplitude vs frequency on the top with phase vs frequency for each baseline below. The map in the top right is an all-sky map with the colours indicating how well the data phase matches a sky position.

Anyway, the good news is that I am able to measure directions with the MarkII! Here’s a graph showing a call recorded in my garden last month. The spectrogram is at the top, and below is a sky map showing the path that the bat took (Imagine you’re lying on your back with your head pointing north. The sky map shows the path the sound took, each point is an azimuth and elevation measured from the chirp). The bat mooched around near the neighbour’s fence to the east, flew west and took a right almost over the recorder, then headed off to the north.

Spectrogram (top) and sky map showing a bat flight through the back garden.

So some really good progress at last! The next steps are to improve the processing algorithm and hopefully automate the position analysis, re-arrange the microphones into a configuration that provides better positions (i.e. play with the separations and orientations of the mics), find a good weather protection solution, and do some more field tests! At some stage I’d like to change to microphones that work better at ultrasonic frequencies, for more sensitivity and so I can push the sample rate up to 384 kHz.