We’ve had an exciting day today. We have finished shooting with the airguns (top left) and brought in the streamer (middle top) along with seaweed (top right) – this bit is from the tailbuoy. Autosub (bottom left) was launched again this evening, it is controlled by a tablet (bottom right)
Below: Autosub spirals down to the seafloor some 3500m below the surface.
Professor Roger Searle has very kindly written this next piece explaining how Autosub magnetism data is processed.
When first recorded, scientific data often are contaminated by noise, artefacts and other undesired effects, which have to be removed before the data can be interpreted. This process is called data reduction, and can be an extensive and time-consuming part of research.
As an example, look at the magnetic data from one of our Autosub dives. The top three plots in the following figure shows the ‘raw’ data, measured as three separate field components at right angles to each other. The x, y, z directions are fixed to the vehicle, and move with it. These are the data from a single dive, which spent almost 24 hours at the ocean bottom and collected 35,000 samples of data – approximately one every 2 seconds.
Note that each component is different, and they step up and down by large amounts. These steps are caused partly by the sub’s changes in direction (for example, if the sub changes direction by 180º, the ‘new’ x direction will be the opposite of the old one, and the field component measured along x will reverse its sign). Partly the steps are also caused by the sub’s own magnetic field, which is added to the measured Earth’s field. It’s complicated, because Autosub’s field has both a ‘permanent’ component (fixed and constant, like a fridge magnet) and a component ‘induced’ by the Earth’s field, and this varies with the sub’s direction.
In the bottom plot, we have combined the three components to calculate the so-called ‘total’ field – this is the length of the vector: √(x2+y2+z2). Note that the scale is a bit smaller, because the total field is always positive, while the components range from plus to minus.
The other plots show how the sub’s heading, pitch and roll varied around the figure of eight – quite a lot!The next step is to remove the big steps. To do this, we need to measure the effects of the sub’s changing direction. We do this by driving it round in a few tight circles while it pitches and rolls. This manoeuvre is called a ‘figure of eight’, and the track (in latitude and longitude) is shown in the top left picture below.
The next figure shows how the magnetic field varied, both with time (sample) round the figure of eight (top) and as functions of heading, pitch and roll (below):
The heading correction is quite simple, and can be modelled well by a two-term sin function: y = a1 sin(b1 x + c1) + a2 sin(b2 x + c2). When this correction is applied it reduces the variation a lot (from about 10,000 nT to 400 nT!), but there is a residual field component that is roughly linearly proportional to the pitch angle (third figure above), and this is also removed
Finally, we have a (fairly) clean signal with most of the vehicle effect removed. When we plot that, for example as profiles along the track (below), we can clearly see areas of higher and lower field (so-called ‘magnetic anomalies’) which we can begin to interpret. However, if you look closely, you will see there are many small spikes and other ‘glitches’ in the data that we will try to remove.
Thank you Roger 🙂 Very interesting indeed.
For more information about Autosub do look at: codemap2015.wordpress.com
ROV – Remotely Operated Vehicle is also in the above blog. We don’t have one on RRS James Cook this time but I thought you might be interested.
Above: The beautiful vast ocean. I still feel incredibly privileged to be here.