What sources of magnetic fields in urban society are strong enough to lure a compass off north? I'm writing about the usage of magnetometers in navigation applications. As part of my reasoning about why you should be careful with it's usage (I've shown in experiments that some environments bring it completely off course), and I need to give a few examples of why this is the case.
For one example, I measured the magnetic field strength using a 3-axis magnetometer in a car. It certainly didn't show north, and the values changed significantly when I turned on the engine.
I also noticed strange values in some places around the office.
A few things that come to mind are electronics, electrical infrastructure and metal objects (like construction beams, table legs, car chassis).
From my own (non-specialized) physical knowledge, I would think that an electrical cable wouldn't produce much of a resulting magnetic field (since current runs in both directions they should cancel eachother out?), and electronics that have some field would oscillate quickly (50 Hz as this is the frequency in outlets in my country). By elimination, only metal objects remain, but I have a feeling this isn't the full story.
So what could throw a magnetometer off in a normal urban environment?
What specifically is it in a car (engine, chassis, electronics) that does it?
(To specify, I'm really looking for static offsets, not noise. You could assume a low pass filter that removes anything with a period less than a few seconds).
 A: Two big factors: ferromagnetic objects (steel beams in buildings, structural elements in cars, etc) can bend the local magnetic field; and the stators of DC motors/alternators (where a significant current oases through many windings, deliberately increasing the magnetic field far above what you get from a single cable). When you have a DC magnetic field generated "nearby" there will be a significant fringe field - enough to throw your compass by a meaningful number of degrees.
If the magnetic field can be mapped (move through the environment with a magnetometer in known orientation) you can not only correct for it - it can become part of the navigation: "The field vector just tilted away from the horizontal by an extra 15 degrees and increased by 0.12 Gauss - we must be just outside the boiler room!"
A: You can consider the earth magnetic field as a a pretty uniform field when considering distances of the order of meters, even kilometers (uniform means constant magnitude and direction of magnetic field vector).
This uniform field is locally distorted (direction and magnitude of the field vector changes) by magnetic materials, i.e. materials with a permeability different from 1 like iron, steel etc. The picture below shows how a sphere with a permeability >1 distorts the otherwise uniform field.

The same happens in you car: although it is not a spherical object it will locally distort the field in such a way that both magnitude and direction of the field are changed.
This is surely the most important disturbing factor.
If you look at other factors that might disturb the field, the most obvious one is wires carrying important currents. The magnetic fields generated by these currents might become important compared to the earth magnetic field, which is of the order of 40$\mu T$ (microtesla). 
To see under what conditions the magnetic field of a current in a wire starts competing with the earth magnetic field, we can use the following formula for the field:
$
B (\mu T)=0,2 \times\frac{I(A)}{R(m)}
$
, where $I$ is the current in ampere and $R$ the distance between yourself and the wire in meter. From this you see that you need at least a current of say 200$A$ to generate a field at 1 meter of the wire that is comparable to the earth magnetic field. In a lot of cases the wire through which the current 'returns' lies next to the first wire. This 'return wire' will generate an opposite field and the two will cancel each other (more or less). Note as well that this field is alternating 50 times per second in case of AC (your compass won't notice such an alternating field). As a conclusion, we can state that only in exceptional situations (huge electric DC motors for example) electrically generated magnetic fields will be a significant disturbing factor.
