The ice that answers a stone with a falling whistle
Assumes: The plate that carries two kinds of sound · The packet that moves at another speed than its own crests
People who skate on clear lake ice in early winter describe a sound nobody expects. A stone skimmed across the ice, a skate striking it, or the ice cracking somewhere far off under its own stresses, produces not a knock but a long, high, falling whistle, a “pew” that sounds electronic and seems to come from everywhere. Recordings of it circulate as curiosities, often with the suggestion that something is wrong with the ice or that the sound is bouncing off the bottom of the lake.
The whistle is the ice’s own physics, and it is the clearest everyday demonstration there is of dispersion: of a medium that carries different frequencies at different speeds and so takes a single sharp event apart. The packet that moves at another speed introduced the distinction between the speed of a wave’s crests and the speed of its energy. A frozen lake makes that distinction audible.
A sheet that is stiffer to short ripples
Ice on a lake is a thin plate, a few centimetres to a metre thick, floating on water. A disturbance in it travels mainly as a bending wave: the sheet curves up and down as the wave passes, like a carpet shaken at one end. The plate that carries two kinds of sound found two families of wave in a metal plate, one stretching it and one bending it, and found the bending one to be the slow one whose speed depends on the frequency. Ice behaves the same way, with two differences that change the sound: the ice floats on water, which it has to drag up and down with it, and the frequencies that matter are the ones a person can hear.
The restoring force in a bending wave is the plate’s stiffness against curvature, its bending stiffness , which grows as the cube of the thickness . A ripple of wavelength curves the sheet by an amount that grows as for a given height, and the restoring force by the stiffness times the fourth power of the wavenumber . Short ripples are therefore pushed back much harder than long ones and oscillate much faster. For a free plate the result is : the frequency goes as the square of the wavenumber, the crests travel at a speed proportional to , and the energy travels at exactly twice the crests’ speed.
A floating sheet also has to move water. The full relation, for ice on water of depth , is
with the bending stiffness and the water’s weight in the numerator resisting the motion, and the ice’s own mass and the water it drags along in the denominator providing the inertia.
At audible frequencies the waves in ten-centimetre ice run at hundreds of metres a second, rising with pitch: the energy of a 300-hertz wave travels at about 760 metres per second, and that of a 4-kilohertz wave at over 3,000. Somewhere in the hundreds of hertz the crests themselves overtake sound in the air above. Above that frequency, called the coincidence frequency, each crest drags a pressure wave in the air behind it like the wake of a supersonic aircraft, and the ice radiates sound efficiently into the air. Below it, the air is pushed to and fro locally without carrying much away. So the ice broadcasts mainly its higher frequencies, and broadcasts them as it goes, from every point the wave passes.
A knock sorted into a glide
A stone striking ice delivers an impulse, and an impulse contains every frequency at once. At the point of impact they are all together. As they spread outward through the sheet, each travels at its own group speed, the speed at which its energy moves, and by the time they have gone a hundred metres they have separated.
The high frequencies arrive first and the low ones later, so a listener hears the pitch fall. On ten-centimetre ice at a hundred metres, the four-kilohertz part arrives after thirty-three milliseconds and the three-hundred-hertz part after a hundred and thirty-two; the whole glide lasts about a tenth of a second, which is the length of the “pew”. On thinner ice the waves are slower, the separation larger and the glide longer and lower; on thick ice it is short and high. Farther away it stretches in proportion to the distance, because every frequency’s arrival time is proportional to the distance it has covered.
The shape of the glide is close to a simple law. For a free plate, the group speed goes as , so the arrival time of frequency goes as , and the frequency heard at time falls as : steeply at first, then slowly. That inverse square is the acoustic signature of a bending wave, and recordings of singing ice show it. It is, remarkably, the same law that the whistle that arrives sorted found in radio whistlers from lightning, where waves guided through the plasma along the Earth’s magnetic field also have a group speed proportional to the square root of their frequency, and so also arrive with their pitch falling as the inverse square of the time. A plasma in a magnetic field and a sheet of ice have nothing physical in common. They share an exponent, and an exponent is all the ear can hear.
How long the whistle lasts says how far it came
Every frequency’s arrival time is proportional to the distance it has travelled, so the length of the glide is too. On ten-centimetre ice the gap between the arrival of four kilohertz and of three hundred hertz is about one millisecond for every metre of distance: a tenth of a second for a stone thrown a hundred metres off, a whole second for ice cracking a kilometre away. A listener on a frozen lake at night, hearing the ice crack and sing, can tell near cracks from far ones without seeing them, by how drawn-out the song is: short sharp chirps are close, long slow swoops are distant.
That is the same inference the swell that dates its own storm made from ocean waves, where swell born in a single distant storm arrives as a sweep from long periods to short over several days, and the rate of the sweep gives the storm’s distance and the time it happened. Ice runs the sweep the other way, because in ice short waves are fast and on the ocean long ones are; and it runs it a hundred thousand times faster, because the waves are a hundred times quicker and the distances a thousand times shorter. The arithmetic is identical: a dispersive medium turns distance into a spread of arrival times, and anyone who knows the dispersion can turn the spread back into distance.
The packet that will not keep its shape described the general spreading of a pulse in a dispersive medium as a loss, the blurring of a signal over distance. Here the spreading is the signal. A knock that stayed a knock would carry no information about where it came from; a knock pulled into a whistle carries its distance in its length and the medium’s thickness in its pitch. The long pulse that arrives as a spike showed the converse, a long chirp compressed into a sharp pulse by a filter with the opposite dispersion. Played backwards, the song of the ice would arrive as a click.
Timing the glide measures the ice
Because the speeds depend on the thickness, the whistle carries a measurement of the ice in it.
The group speed at a given frequency grows as the fourth root of , which is the square root of the thickness, so the delay at a known distance falls as one over the square root. Doubling the thickness of the ice shortens the delay by a factor of . A microphone or a geophone on the ice, recording a knock made at a measured distance, can therefore read off the thickness, and the method is used on sea ice and on frozen lakes, where drilling holes to measure thickness across a wide area is slow. The catch is the stiffness. Ice’s Young’s modulus depends on its temperature, its crystal structure and the brine or air trapped in it, and for floating ice in bending it is commonly quoted between about four and ten gigapascals. The thickness inferred from a delay goes as one over the square root of the assumed stiffness, so a factor of two in stiffness is a factor of 1.4 in thickness. The figures here use nine gigapascals; a calibration against one drilled hole fixes it for a given lake.
The water the ice carries
The water under the ice slows every wave, and slows the low ones most.
When the sheet bends, the water beneath it must move too, and the water involved reaches down about a wavelength divided by . A long, low wave drags a deep layer of water with it, many times the ice’s own mass; a short, high one drags a thin layer, less than the ice’s mass. This added mass, per unit area in deep water, sits in the denominator of the dispersion relation beside the ice’s own , and it is why the floating curve falls below the free one at low frequencies. At a hundred hertz, ten-centimetre ice runs at seventy per cent of the speed it would have in air. At several kilohertz the added mass is negligible and the ice behaves as a free plate.
The added mass makes the glide longer than a free plate would, because it slows the late, low-pitched part of the knock more than the early, high-pitched part. It also moves the coincidence frequency up, since slower crests need a higher frequency to overtake sound in air. A sheet of ice on a lake is in this sense a different instrument from the same sheet lying on the ground: the water is part of what vibrates. The same principle is why the drum that has no harmonics needed the air around a drumhead counted into its modes, and why boats feel heavier to accelerate than their weight suggests.
The speed a truck must not drive at
Read at long wavelengths, the same relation describes something more serious than a whistle. Gravity enters through the water’s weight, in the numerator, and for waves tens of metres long it is the main restoring force: the ice is lifted and the water’s weight pulls it down. Long waves are then fast because gravity drives them, short ones because stiffness drives them, and in between there is a slowest wave.
That minimum is the same construction that the speed below which nothing on water leaves a wake found for open water, where gravity and surface tension meet at a slowest wave of 23 centimetres a second. Here the stiffness of the ice plays the part surface tension played there, and because ice is enormously stiffer, the minimum is at a wavelength of tens of metres and a speed of tens of kilometres an hour. A load moving over the ice makes waves, and a load moving at the minimum speed makes a wave that travels exactly with it, so the deflection it causes builds up instead of spreading away. The ice under a vehicle at that speed bends several times more than under the same vehicle parked, and can crack where a parked one would not.
This is why ice roads, the winter highways over frozen lakes and rivers in northern Canada and Scandinavia, carry speed limits, and why the limits are often lower close to shore. Over shallow water the long waves are slowed by the bottom, to , four and a half metres per second short of the deep-water minimum over four metres of water, and the dangerous speed falls with them; the wave a vehicle makes when approaching or leaving the shore can also be reflected from it. Drivers on ice roads are told to keep below a speed set from the ice’s thickness and the water’s depth, and to slow further near shore, which is the figure’s long-wave end turned into a rule.
What the plate model leaves out
The figures treat the ice as a thin, uniform, elastic plate. That is right when the wavelength is much longer than the thickness, which holds for the audible frequencies in ice up to about twenty centimetres thick; at higher frequencies or in thicker ice, the plate’s shear and its rotational inertia slow the waves below the thin-plate law, the same correction the plate that carries two kinds of sound found in the full theory of elastic plates. Real lake ice is not uniform: it is layered, clear black ice under cloudy snow ice, and it has cracks, which scatter and reflect bending waves. Ice is also not perfectly elastic; it absorbs high frequencies faster than low ones, which shortens the top of the glide with distance.
The sound a listener hears is not the bending wave itself but the air’s response to it, and that depends on where the listener is. Standing on the ice, the listener’s feet feel the wave directly and the ears hear the air near the surface; standing on the shore, the listener hears mainly what the ice above coincidence radiates. And the stone’s impact is not a perfect impulse: a light pebble excites high frequencies and a heavy stone more low ones, which shifts the balance of the glide. None of this changes its direction, which is set by the sign of the dispersion: in any plate the high notes come first.
At the long-wave end the model treats the ice as floating freely, with no cracks or shores nearby, and the load as a point moving steadily. Real ice-road loads are long trucks whose length is comparable to the critical wavelength, which smooths the resonance, and the ice has internal damping that keeps the deflection finite at the critical speed rather than letting it grow without limit. Field experiments with vehicles driven over floating ice at a range of speeds have found the deflection rising sharply near the speed this kind of model predicts, which is why engineering guidance for ice roads uses formulas of the same shape.
Still open: what sets the shape of a real glide
Field recordings of singing ice show glides whose timing matches the dispersion of a bending plate, but they also show structure the simple model does not: several arrivals, repeated glides, and changes of character over a day as the ice warms and cools. Some of this is reflection from shores and cracks, some is waves guided along the boundary between ice and water rather than within the ice, and some may be the ice’s own thermal cracking far away, which produces knocks that travel kilometres. How much of the sound of a frozen lake on a cold night comes from each source, and how its spectrum relates to the state of the ice, has been measured in a few places and not systematically. Monitoring sea ice from the sound it makes, its thickness and its cracking, is being tried with networks of instruments in the Arctic.
The habit worth carrying away is to ask what a medium does to a pulse rather than to a tone. A floating ice sheet carries bending waves whose speed rises roughly as the square root of their frequency, so a stone’s knock 100 m away arrives as a falling glide, 4 kHz after 33 ms and 300 Hz after 132 ms on 10 cm ice, whose timing measures the thickness; and the same law at long wavelengths, with gravity and the water’s weight, has a slowest wave — 10.8 m/s in 30 cm ice on a deep lake, less over shallow water — at which a truck keeps pace with the wave it makes and bends the ice far more than its weight alone would. A knock contains every note. Ice decides the order they arrive in.
Part 11 of 11
This essay is one argument about Guided waves. The others:
The objects named here
The third axis, after the field and the reading path: the things themselves, and every essay that touches each one.
Added massBending stiffnessCritical speedDispersionFlexural waveGroup velocity
- How long the crossing takes dispersion, group velocity
- The constant that depends on how fast it is asked dispersion, group velocity
- The dispersion made of angles dispersion, group velocity
- The momentum light carries into glass dispersion, group velocity
- The packet that bends with nothing pushing it dispersion, group velocity
- The photons that raced for seven billion years dispersion, group velocity