These maps were made by detecting small differences in sea surface altitude, and inferring the gravitation field. But changes in gravity can also be measured directly, using a gravimeter.
However, they tend to be finicky to work with--difficult to calibrate and to eliminate sources of error. After all, we know from Einstein that there is no measurable difference between gravitational acceleration and inertial acceleration (like, say, bumping a sensor).
Still, gravity seems like an area of immense promise for sensors of the future. We've spent immense effort improving and miniaturizing sensors for things like magnetic fields, electrons, and photons. But we've recently discovered--using gravity--that these things might represent only 5% of the total mass of the universe. We haven't even detected a gravity wave yet. We're just barely getting started at using gravity to observe and understand the universe.
However, they tend to be finicky to work with--difficult to calibrate and to eliminate sources of error. After all, we know from Einstein that there is no measurable difference between gravitational acceleration and inertial acceleration (like, say, bumping a sensor).
Gravity gradiometry is used in exploration geophysics. Typically this is done as a ground survey, taking measurements on a grid. I don't know how they do the error correction, but aerial gravity surveys can also be done. Obviously, the more stable the platform the higher quality the data, which is by De Beers used a zeppelin for surveys in Botswana looking for kimberlite pipes. This company [1] uses a BT-67 (an updated DC-3) as their fixed wing platform. The company I used to work for had a small gravity survey flown in the mid 2000s. When chatting with the technician from Bell, I facetiously asked whether they had considered an Antonov An-2 [2], and was surprised to hear that it had been (briefly) considered it as a platform.
When I studied geology we did a gravity survey along a road. The gravimeter was a big metal can; inside there was a weighted arm suspended by a spring. Measuring the differences in the static deflection of the arm gave us a reading of local gravity at each point along the survey.
The can had to be placed perfectly level, and sit for a couple minutes (to allow any vibration in the sensor suspension to dissipate) before taking the reading. We also couldn't take any readings when a vehicle was going past. Luckily this was along a fire road in the mountains, so there was very little traffic.
I've heard of both aerial and underwater (submarine or towed sensor) gravity surveys, and it's so impressive that they get usable data from a moving platform. I think one trick is to do multiple passes so that chaotic sources of error (like turbulence) average out at any given point--while real differences in gravitation would persist.
Edit to add: Going back to my comment above, I can't imagine how a truly portable (like, arbitrarily hand-held) gravity sensor could be developed, the way we have portable sensors for light, radio, sound, ambient pressure, magnetic field, etc.
Most of the moving measurements aren't gravimeters. They're gradiometers.
You don't actually get the same data out, and you can't use it in the same way.
The key part isn't just the acceleration due to motion. It's that you have to know your absolute elevation very precisely if you're using a gravimeter. Otherwise, the data you get can't be corrected relative to your other measurements and is more or less useless. (The method mentioned in the article actually measures the geoid directly by measuring the sea surface, which is an equi-potential surface.)
However, if we don't worry about the absolute acceleration due to gravity, and instead measure the local rate of change in gravity, we don't need to know elevation precisely.
That's referred to as gradiometry. You can't use the data in the same way, but it's still very useful.
In a nutshell, most of the movable gradiometry sensors work by using multiple accelerometers. Acceleration due to motion affects both equally (with some caveats when rotation comes into play). The differences in acceleration between the two accelerometers is therefore purely a result of the "tilt" of the geoid locally.
It's difficult to integrate this back into an accurate picture of what the free air or Bouger anomaly would look like, but it's still useful information. We can't necessarily calculate the same things from it, but it's a great edge detector.
>> After all, we know from Einstein that there is no measurable difference between gravitational acceleration and inertial acceleration (like, say, bumping a sensor).
I think it's better to say that we know it from Newton's laws of motion and gravity - they are older, much easier to understand, and can still explain the principles of the technique.
However, they tend to be finicky to work with--difficult to calibrate and to eliminate sources of error. After all, we know from Einstein that there is no measurable difference between gravitational acceleration and inertial acceleration (like, say, bumping a sensor).
Still, gravity seems like an area of immense promise for sensors of the future. We've spent immense effort improving and miniaturizing sensors for things like magnetic fields, electrons, and photons. But we've recently discovered--using gravity--that these things might represent only 5% of the total mass of the universe. We haven't even detected a gravity wave yet. We're just barely getting started at using gravity to observe and understand the universe.