Riding a bike on the edge of a seismic cliff: between signal and noise


Wadi Degla is a canyon in Cairo, Egypt: less than 40km long, its drainage system collects the scarce water falling over 260 square kilometres of dry Egyptian land. Only 13 millimetres of rainfall per year! I enjoyed more than 1000 mm of rain in the first six months in France in 2013.

The Wadi Degla canyon is very popular amongst foreigners in Cairo, and not only for its beautiful and dramatic landscape, and the wonderful biking tracks. You can also go to Wadi Degla just to have a moment of silence: in Cairo, silence is as rare as rain.

Cairo is a wonderful, noisy city: a ‘city where you can’t hear yourself scream’, as says an article in the New York Times of 2008 (http://www.nytimes.com/2008/04/14/world/middleeast/14cairo.html?ref=world). The cited study indicates an average noise of 85dB all-day long; in some neighbourhoods, an incredible 95 , like a chainsaw. 

At nights, yes, it goes down a bit: but the city never sleeps.


In this megacity more than 20 million people live and work, and more than 2 million cars drive and honk. All day long. Two million drivers driving with one hand on the horn, with a flexible interpretation of traffic law, and with an extraordinary ability of squeezing four lines of cars in two lanes.

At the beginning you think that the concert of horns is just noise. Then you realise that the horn is the radar system used by everyone to communicate its position to others, when you are out of their sight: and you start using the horn, you start making noise to signal your presence.

Actually there is a true language of honking, a sequence of honks which, as in a Morse communication, can be decoded by others: it turns out that most of the sequences translate as insults, often insinuating doubts about the morality of other drivers’ mothers.


Driving in Cairo makes you reconsider your ideas of noise and signal: and it goes beyond the honking, it goes all the way to geophysics.

In geophysical signal processing, a signal is any time and space varying function that we can analyse and process: but we have to change this definition of signal to be able to define the noise. The signal is the part of the data carrying meaningful information: the noise is the part of data that we are not able to use, or that we decide not to use.

So what is noise evolves together with our analysis and imaging methods, with the complexity of the physical model that we define and use to explain our complex data. With sea-bottom nodes, for example, the multiples (the downgoing wavefield) are noise only if you decide not to use them in the mirror imaging.

And on land, we have completely reconsidered the ground-roll. Let’s go back to the cliff: the final, noise-free seismic image of the seismic cliff is obtained processing synthetic noise-free data. If the cyclist is our source, ideally he would sends into the ground pure signal, to receive it at the surface.


In reality, we know that he creates much more surface waves than body waves: 2/3 of the energy he sends into the ground is probably Rayleigh waves, the event that was mysteriously called ground-roll. And it would be the same for more normal, non-cycling seismic sources. The gathers below are synthetic for the Wadi Degla outcrop of the photo: with only acoustic phenomena on the left, and with the elastic surface waves on the right.


I reckon that this is becoming quite boring, and we cannot get into geophysical details here. To cut a long story short: in the past, we used to call it ‘ground-roll’, and tried everything we could to remove it. Today, surface waves cannot be considered just noise anymore. They propagate in the near-surface, and carry lots of detailed information about it: since the near-surface characterization is critical in land seismic, they are signal that we must extract, analyse and use. Only after this step, we remove it from the processed wavefield that we use to image the deeper subsurface.

When you drive in Cairo, you learn that noise is signal: and when you are in a taxi, you actually worry more if the driver doesn’t honk enough.


If you want to read more about surface-waves and near-surface:




Leave a Reply

Your email address will not be published. Required fields are marked *