Soil monitoring

Abstract

A method for identifying an acoustic signature representing a stridulation signal comprises generating a spectrogram from audio data recorded from a subterranean region of interest. The spectrogram comprises a representation of a variation in radio frequency spectrum with respect to a predefined time period. A portion of the spectrogram is analysed using a learned convolutional neural network CNN to identify the presence of a stridulation signal in the portion of the spectrogram. The stridulation signal comprises a set of rhythmic auditory patterns relating to a species of interest e.g. Coleoptera Elateridae, click beetles in the subterranean region of interest. A soil testing bioacoustics apparatus 201 for insertion into soil 200 is disclosed, comprising a microphone or accelerometer 206 on a probe 202. [Figure 2 to accompany the abstract]