Low cost accelerometers that could detect the ripples on a pond?

Imagine a sensor sitting on a small float bobbing about on a pond like this.

Would any of today's low cost accelerometers be able to detect such motion or would the signal be buried in the noise floor?

Interesting. However I see some problems.
1: Your float would need to be very small - look how the leaf is not affected by the ripples. A tiny float would not have enough bouyancy to support a (relatively) massive sensor.
2: An accelerometer is no good on its own - you would need to receive and process the signal. So you either need a wireless sender - adding weight; or a wired sender, adding stiffness.

You need to find a solution where the system used to make the measurement does not significantly affect the value being measured. Looking at the video, perhaps an optical system would be a better solution?

Lets say they are about 1mm in amplitude, coming in at say 10 Hz. That's an acceleration of (102pi)^2*0.001 = 3.9m/s/s, or about 0.4 g (g being the acceleration due to gravity, which is commonly used in accelerometer datasheets).

Seems well within capabilities, if you can mount one in a small/light enough float, and run flexible enough wires to the main circuit. Conceivably a flex pcb might be possible to both mount the accelerometer chip and provide suspension, but might be too stiff.