The underwater sound of rain.

From deep underwater, amid the rumbles, squeaks, whistles and other noises often heard, researchers are begining to pick out the distinctive sounds of rain, hail and even snow striking the water's surface. This newly conirmed technique may make it possible to detect and measure rainfall over the oceans, filling a wide gap in knowledge about global weather patterns.

The experiments show that measuring and detecting rain over the oceans using buoy- or bottom-mounted acoustic sensors is, unexpectedly, feasible. Currently, rain gauges on ships provide unreliable, spotty data, and microwave measurements from satellites can't be properly calibrated.

One experiment, reported in the Dec. 19 NATURE, was done in Cowichan Lake on Vancouver Island, British Columbia. There, says Joseph A. Scrimger of Jasco Research, Ltd., in Sidney, British Columbia, "you get as much rain as you'd ever want." In winter and at night, when the measurements were made, the lake is also relatively free of human-made and fish-made noises, he says.

Scrimger mounted a hydrophone in 35 meters of water about 300 meters from shore. A cable carried the hydrophone's signal to a shore based instrument, where it was recorded as a spectrum showing how the sound's intensity depends on its frequency. Great care was taken to ensure that the detcted signal was tuly "the signature of the rain and not of the equipment," he says.

Scrimger managed to observe several rainstorms and by chance, hail and snow episodes. "We were flabbergasted," he says, "to find that [the rain's signature] was so introduced."

Sound spectra for rain, under calm conditions, have a sharp peak at 13.5 kilohertz. Wind tends to round and spread out the peak. Hail, on the other hand, has a broad peak at 3 kHz, while snow tends to get "louder" with increasing frequency. However, these snow sounds are largely at frequencies beyond thos detectable by human ears.

Scrimger's results are similar to those obtained by Jeffrey A. Nystuen, now at the Institute of Ocean Sciences in Sidney, British Columbia. While a graduate student at the Scripps Institution of Oceanography in La Jolla, Calif., Nystuen measured rain-generated underwater noise in an Illinois lake, then developed a computer model of a splashing drop to try to explain why the spectral peaks fall at a particular frequency. Nystuen reported his results at a recent American Geophysical Union meeting in San Francisco.

The effect is like that of a "water hammer" banging into the surface, says Nystuen. The impact of large, floppy drops of rain produces a lot of white noise, similar to the buzz heard on a badly tuned radio. Smaller drops produce less white noise.

But an instant after the initial impact, water must begin to flow. For drops of any size, ths appears to take aout 0.06 millisecond. After this time, no further sound is generated. This means that rainfall spectra should have a peak at close to 15 kHz. Through an earphone, a listener hears a kind of snapping or crackling noise.

"All of the drops contribute to the peak," says Nystuen, "but only the bigger drops cause a rise in the spectral level at low frequencies. That may explain the change in the spectal character from the heavy rain when big drops are present and light rain when small drops are present."

The spectral differences are often quite noticeable. "At Scripps," says Nystuen, "I could look at the spectrum in the lab and tell you what was coming down outside."

Scrimger says he has heard stories about lakes that "sing" whn there's a very fine drizzle. These little drops are like explosive charges, he says. "When they hit the surface, they go off with a little ping." In the case of gently drifting snowlakes, the sound is probably generated by a process associated with the melting of snow.

Scientists are already starting to measure the speed of ocean surface winds by detecting their sounds underwater. Rainfall may be next.