Detection of rain in tropical cyclones by underwater ambient sound

Abstract

Rain in tropical cyclones is studied using eight time series of underwater ambient sound at 40 Hz–50 kHz with wind speeds up to 45ms−1 beneath three tropical cyclones. At tropical cyclone wind speeds, rain- and wind-generated sound levels are comparable, so that rain cannot be detected by sound level alone. A rain detection algorithm based on the variations of 5–30 kHz sound levels with periods longer than 20 seconds and shorter than 30 minutes is proposed. Faster fluctuations (<20 s) are primarily due to wave breaking, and slower ones (>30 min) due to overall wind variations. Higher frequency sound (>30 kHz) is strongly attenuated by bubble clouds. This approach is supported by observations that, for wind speeds <40 m s−1, the variation in sound level is much larger than that expected from observed wind variations, and roughly comparable with that expected from rain variations. The hydrophone results are consistent with rain estimates by the Tropical Rainfall Measuring Mission (TRMM) satellite and with Stepped-Frequency Microwave Radiometer (SFMR) and radar estimates by surveillance flights. The observations indicate that the rain-generated sound fluctuations have broadband acoustic spectra centered around 10 kHz. Acoustically detected rain events usually last for a few minutes. The data used in this study are insufficient to produce useful estimation of rain rate from ambient sound, due to limited quantity and accuracy of the validation data. The frequency dependence of sound variations suggests that quantitative rainfall algorithms from ambient sound may be developed using multiple sound frequencies.