Detection and Measurement of Rain at Sea   Page 2Page 4

In order to measure rain at sea, scientists at the University of Washington’s Applied Physics Laboratory designed and built Acoustic Rain Gauges (ARGs). The ARG consists of a hydrophone (underwater microphone), some electronic circuitry, a low-power sampling computer, and a battery package designed to operate the ARG without servicing for up to a year. The ARG is attached to a mooring line, and can be placed at any depth in the ocean, although practically the depth is limited by the crushing strength of the instrument case. Every few minutes the ARG "wakes up" and evaluates and records the underwater sound field. Currently, the ARG design is autonomous from the surface float, and the recovery of data awaits recovery of the mooring. In the future, real-time transmission of the data will be needed to provide useful data for weather forecasting.

When listening for rain in the ocean, the first step is to identify the sound as rain. There are lots of other sounds underwater, including the sounds of waves breaking, man-made sounds and biological sounds. Biological and man-made sounds are sometimes very loud and, if they contain frequency components that overlap the rain-generated sound, then they can prevent acoustical measurement of rain. These noises are usually intermittent or geographically localized. Some locations where persistent "noise" is present includes harbors (shipping and industrial activity) and snapping shrimp colonies. Snapping shrimp are from a family of shrimp species that make very loud "snaps" and that inhabit shallow tropical waters. Fortunately, the frequency content of most sounds is unique to their sources, and can be used to identify the sources, including rain, drizzle, and whitecaps. Some examples of oceanic sound spectra are shown in the graph below.

  Acoustic Rain Gauge
Dr. Jeffrey A. Nystuen holding an Acoustic Rain Gauge (ARG). This instrument is designed to be clamped onto an oceanic mooring and will record the underwater sound for one year. (Photograph courtesy Jeffrey A. Nystuen)

graph label

sound level vs. frequency

Most of the time it is not raining and no man-made or biological noises are present. When this is true, the sound is from the whitecaps generated by wind and can be used to quantitatively measure wind speed (Vagle et al. 1990) as the number of whitecaps is proportional to wind speed. The shape of the sound spectrum generated by breaking waves is controlled by the distribution of bubble sizes generated by the breaking wave (Medwin and Beaky 1989). An interesting feature of the wind-generated signal is an apparent limit to the loudness of the sound at higher frequencies. This is due to quiet adult bubbles absorbing the higher frequency sound levels (Farmer and Lemon 1984). Because of their smaller size, bubbles that absorb high-frequency sound stay in the water longer and can form effective layers of sound-absorbing bubbles.


The graph at left shows examples of underwater sound spectra recorded from an oceanic mooring in the South China Sea. The sound spectra from wind-only conditions (green) show a uniform shape and a sound level which is proportional to wind speed. The sound of drizzle (light blue) shows the characteristic peak associated with the sound generation mechanism of the small raindrops. The sound of heavy rain (dark blue) is louder and includes lower frequencies. The sound of extreme rain includes sound generated by very large raindrops and is very loud. It also shows the effect of "quiet adult bubbles." Two spectra from extreme rain (200 mm/hr) are shown. The first (dark blue) shows extremely high sound levels at all frequencies. The second (burgundy) shows relatively lower sound levels above 10 kHz. This spectrum was recorded five minutes after the first, and yet the rainfall rate was still the same. A layer of bubbles had been injected into the sea surface. New "rain sound" has to pass through the bubble layer to reach the ARG sensor, and is partially absorbed by the bubbles. Since smaller bubbles (higher resonance frequency) are less buoyant than larger bubbles, they stay in the water longer and thus this bubble effect is most noticeable at higher frequencies. (Graph by Jeffrey A. Nystuen)

  Using the graph above, the differences between wind-only and rain-generated spectra often appear to be subtle. However, by presenting the data in a different manner (below), acoustic identification of different weather conditions becomes apparent. The sound of rain and drizzle contains relatively more high frequency sound than the sound from wind-only conditions. Furthermore, rain is much louder. Even drizzle, under low wind speed conditions, has sound levels which can be orders of magnitude louder than wind-only conditions. The characteristic sound of drizzle, the 13-25 kHz peak, is sensitive to wind and has not been detected when the local wind speed is more than 8-10 m/s. On the other hand, the sound from heavy rain is very robust and can be detected even in very high wind speed conditions (over 20 m/s) (Nystuen and Farmer 1989). Extreme rain (over 100 mm/hr) is even louder, and can generate an ambient bubble layer that will distort the recorded sound spectrum.
Graph Label

Graph comparing sound level at two different frequencies

An example of the acoustic interpretation of the underwater sound field is shown below. During this three-day period a strong atmospheric front passed over the location of an ARG. When it was not raining, the acoustic estimate of wind speed matched a nearby mechanical anemometer to within +/- 1 m/s (very good agreement). Because rain is so loud underwater, acoustical wind speed measurements are only possible when is not raining. During the peak of the storm, heavy rain was detected. This acoustic observation was ’confirmed” by near-surface (1 meter depth) salinity measurements. Similar records of acoustic measurement of rainfall have been obtained from ARGs on drifting buoys (Nystuen and Selsor) and from an oceanic mooring in the South China Sea.

  Acoustic weather classification uses features of the underwater sound spectrum to identify the sound source: wind (green), drizzle (light blue), rain (medium blue), extreme rain (dark blue) and to detect ambient bubbles. (Graph by Jeffrey A. Nystuen)

Graph Label

Ambioent Sound Record Graph

next Conclusions and References
back How do raindrops make sound underwater?


A history of the sound field over three days at four different frequencies from the North Atlantic Ocean (ASREX Experiment, sponsored by Office of Naval Research). The lowest frequency (500 Hz) is not affected by precipitation, while the highest frequency (20 kHz) is affected by rain, drizzle and ambient bubbles. The acoustic interpretation of the sound record measures wind speed (light green), rainfall rate (dark blue) and detects drizzle (light blue). Comparison data for wind speed (dark green) and near-surface salinity (green, bottom graph) from a nearby surface mooring are also shown. Note that the surface salinity dropped during the rainstorm. (Graph by Jeffrey A. Nystuen)

Print this entire article