Follow-up #2: Ok, I stuffed my son in the trunk with the iPhone and the MagnetMeter app. It is rainy and wet here, so we couldn't peak out acceleration, but here is what we found. With regular cruising around the neighborhood, the uT are in the 100-200 range, but when I accelerate from a stop the values go way up and seem to be strongest in the center (3 inches off center to drivers side) and about 6 inches from the backseats - measured on the trunk deck. Accelerating 0-40 with about 75% pedal we hit 1600 uT. About half the strength of my very strong small magnet. 0-20mph with half the power was about 1000 uT, so it felt proportional. We didn't venture out to bigger roads, since he was already bouncing off the walls of the trunk testing acceleration and it was raining. It would be interesting to see the values with full throttle up a hill. I suspect they might almost double. Again, these are short bursts -- not sustained, but I suspect they are enough to occassionally mess with a sensitive compass like the DJI on longer trips.
Going down hill with full regen didn't generate much. Also, other places in the trunk were all substantially less. Strib, I did retest and I found high values with the car off in the same places you did. The lower passenger front bumper and lower driver front bumper each gave me about 300-400 uT, although the reading was from outside the car and not strong enough to do much. The trunk motorized latch (in the hatchback) gave a reading of about 300. Stronger than other "off readings", but not enough strength to matter much.
So, using a very crude back-of-the-napkin approach, it appears there is a meaningful amount of magnetic force generated while driving (above the coil in the trunk) that could interfere with a sensitive compass, like on the DJI, but in relative terms the EMF is very small. Too small to erase a credit card or hard disk, but I wouldn't put a pack of 5.25" floppy disks in the trunk.