More data, more fly-bys

I’ve been spending the past few weeks working with more data and trying to make the direction finding software work more efficiently and be a little more automatic and reliable. 

Bat calls that cover a wide frequency range are generally easier to locate than the narrow-band calls. (This makes sense as the slope in phase across the band gives an unambiguous delay estimate if it covers enough frequencies, but doesn’t do so well for narrow coverage). So some data are hard to interpret and there have been some fly-bys I haven’t been able to track. This could be improved by a better arrangement of microphones I think.  

Here’s a screenshot of some software I built to help with localising the signals and resolving ambiguities. For every chirp I calculate the coherent average of the signal strength and find directions on the sky where this is a maximum, In many cases there are multiple maxima of similar amplitude so I’ve been using The SciPy optimisation routines to find local maxima across the sky. The top panel is the spectrogram, below that is amplitude vs time for the 5 strongest maxima (the darker the red, the greater the amplitude). The next panel is a plot of possible azimuth solutions vs time, then elevation below that. Finally, the two plots at the bottom show sky distribution in two different projections. Usually I can use the brightest signals to ‘anchor’ the remaining chirps and find the trajectory. But you can see the ambiguities, and I need to work on that. The movie shows the final processed result.