AndrewJ -> RE: MAD detection of subs (3/28/2019 12:52:08 AM)
Here's a repost of some information I'd added to another thread last year, with a bit more added.
I have to admit that as a player I love MAD on my side, since it makes sub detections so easy. Flying over a sub within a little under a mile slant range (half that for non-magnetic hulls) will give you a perfectly accurate positional detection. I hate it when it's on the enemy side, since it's pure luck of the draw whether an aircraft on its random patrol path happens to pass nearby, in which case you almost always lose the sub if the enemy's armed.
Up to this point I'd not seen any hard figures about the range to which actual MADs worked in the real world, but the other day I accidentally came across a report titled "AR70-14 Soviet Antisubmarine Warfare: Current Capabilities and Priorities". This is is a formerly Top Secret report from the CIA's Directorate of Science and Technology, which was published in 1972, and declassified for public release in 2017. You can read it on the CIA's website here:
As of the early 1970s, it gave the following summary:
"Presently operational saturable-core magnetometers are sensitivity limited and can provide detection ranges of 1,000 to 1,400 feet under favorable conditions. The newest optically pumped magnetometers have a sensitivity of 0.1 gamma, and can obtain detection ranges of 2,000 feet under good conditions."
"Present Soviet aircraft MAD gear is estimated to have a detection range of 1,500 feeet. The Soviet's extensive effort in MAD sensor technology should allow them to extend this somewhat. Evidence from recent naval aircraft operations indicates that improvements have been made. Their weakness in signal and data processing will probably limit them to at most a 50% increase in range in the near future."
What strikes me about this is that the ranges are much shorter than those currently modeled in CMANO, where reliable detections occur out to ~1 nm, which is 6076 ft. Here they're talking about operational and newly developed sensors with ranges of only 1,000 to 1,400 feet (0.23 nm), 1,500 feet (0.25 nm), or 2,000 ft (0.33 nm) and that's only when conditions are good. Presumably the situation would be worse in shallow waters, where the seafloor could interfere, or near other naval vessels, which would generate false signals. The report does mention potential technological improvements to improve magnetometer sensitivity (superconductivity!), but also seems to imply that it may be difficult to make use of such improvements given the natural environment.
It turns out that sensor platform speed (helicopter vs MPA - it's easier to spot the signal when you pass over it quickly), wave (magnetic) noise, altitude (better to be further away from the waves), geomagnetic noise (space weather), geological noise (local anomalies) all have effects on MAD detection ability. This 1976 report (Speed and Depth Effects in Magnetic Anomaly Detection) for the US Navy suggests these factors combine to give useful detection ranges of approximately 1350 feet.
A 1994 thesis for the Naval Postgraduate School, "A Limited Analysis of some Nonaccoustic Antisubmarine Warfare Systems", suggests that these factors would give an effective range for the P-3's AN/ASQ-81 of a little under 500m (1640 feet) against a typical diesel sub.
The most recent MADs seem to have improved their performance somewhat; no doubt advances in signal processing have helped greatly. CAE, makers of the later AN/ASQ-508 MAD, and the recently developed MAD-XR, mentions detection ranges of "approximately 1,200 metres" (3937 feet) in their most recent promotional literature, although this is somewhat at odds with lack of recent enthusiasm for MADs in general. (Treat all promotional literature with caution, of course.)
One complication is that it turns out that the strength of the magnetic anomaly varies greatly depending on the relative orientation of the searcher's flight-path, the orientation of the sub, and the specific local magnetic field. A system which may give a good detection range in one pass, may have a much smaller detection range in another. See these papers for detailed mathematical examples:
This recent (2015) Chinese paper on MAD detection highlights the variability. A sub detection which happens at nearly 800 m (~2,600 feet) flying at very low altitude in ideal orientation can drop to less than 100m (~ 300 feet) if you fly higher and the sub changes direction. (Yes, vertical separation does matter.)
That's a lot of variation!
So based on my highly informed opinion (translation: after few hours of bewildered Googling), it looks like the current MAD model in Command is somewhat optimistic. Even the most modern figures I found don't reach the current 1 mile range that's in the database.
I think that think that reducing the range for the generic MAD in the game would be a good idea. Although identifying the performance of individual MAD systems would be very difficult, given the scarcity of information, it might be a good idea to replace the one 'Generic MAD' sensor with several generations of generic sensor, each covering a decade or two, and gradually stepping up detection ranges from generation to generation. Given the amount of variation MAD sensors seem to encounter, I'd suggest stepping back the detection range claims for the most modern systems a bit. I'm not sure if it would be worth modelling different magnetic signatures depending on sub size too.
I want to add that there is a game-mechanics caveat about this.
Currently MAD detections seem to be be checked every 15 seconds. (Hover a helicopter near a sub you have detected by MAD only. I think you'll see the white contact age numbers counting up to 14 seconds and then resetting in a continuous cycle.) For a fast-moving aircraft, such as a P-3 loitering along at 205 knots, this means one detection is resolved every 0.85 nm. If the MAD had a more realistic range of 0.25 nm, for example, then there would be gaps in the search pattern. Since the MAD sensor is presumably receiving continuously, a quicker detection check interval would be needed to simulate the continuous coverage.