A curious thought on DW passive sonar detection range

[goldorak]

SCS could base DW 2 on a totally new codebase, keeping secret the core engine while at the same time providing modding tools for inserting new models, new database entries and why not a plugin architecture so that for instance the graphics engine is decoupled from the core naval engine.
Keep the secret stuff secret for their institutional customers, and let the game community have access to "public interfaces".
But this will never happen since all their games are based on the same old old old last century old code.

[DrMilton]

I don't see how something "classified" could be allowed to reach a commercial market, especially from a military contractor (who plays by the rules). It all looks more like marketing strategy or business arrangements between the companies that brought DW to the market (its not just SCS).

Instead of a conspiracy theory about secret hidden codes maybe we should just accept the fact that business decisions are made according to profit and government contracts are more promising than computer games.

So, just another product that gets abandoned. It has happened before.

[SeaQueen]

OH PUH-LEASE!!!! There's nothing classified in DW. If there was, everyone here would need a clearence, and your hard drive would be stored in a safe. You couldn't play DW over the Internet either, you'd have to use the SIPRNET. Oh, and the window behind your computer? You need to put some blinds on that. And nobody talk about the game, for heaven's sake without turning on the white noise generator! That'd be a classified conversation. People have been saying things implying that a given subsim contains classified information since Microsoft's Red Storm Rising. They didn't have anything then, and they don't have anything now.

Somehow, they draw on a small family of TL curves and use it to compute a very simple signal-to-noise ratio. There's nothing classified in that. Even the Navy's official sonar models are unclassified. At most they're maybe FOUO. The database feeding values into them is the classified part. Here's some screenshots of their sonar models:

http://www.d-a-s.com/imat.html

Originally Posted by sonar732

The only thing about that SeaQueen is, and I'm sure you're aware, is that if they even got a small fraction of anything classified in their sonar model the US Navy and any other nation they provided software for will have them for lunch.

This has been my stance eversince Sub Command and DW's early start. We don't know how much classified material they used for any of their modelling and can't make judgements as such.

[Dr.Sid]

Originally Posted by SeaQueen

OH PUH-LEASE!!!! There's nothing classified in DW. If there was, everyone here would need a clearence, and your hard drive would be stored in a safe. You couldn't play DW over the Internet either, you'd have to use the SIPRNET. Oh, and the window behind your computer? You need to put some blinds on that. And nobody talk about the game, for heaven's sake without turning on the white noise generator! That'd be a classified conversation. People have been saying things implying that a given subsim contains classified information since Microsoft's Red Storm Rising. They didn't have anything then, and they don't have anything now.

Somehow, they draw on a small family of TL curves and use it to compute a very simple signal-to-noise ratio. There's nothing classified in that. Even the Navy's official sonar models are unclassified. At most they're maybe FOUO. The database feeding values into them is the classified part. Here's some screenshots of it:

http://www.d-a-s.com/imat.html

Hmm .. this terrain influence sound propagation .. looks more interesting then I expected :hmm: I wonder if this could be done realtime. Will try, you know me ..

[SeaQueen]

Originally Posted by Dr.Sid

Hmm .. this terrain influence sound propagation .. looks more interesting then I expected :hmm: I wonder if this could be done realtime. Will try, you know me ..

Sound in the ocean is REALLY complicated. People in video-game land don't always seem to realize it, though, because they have ray traces stuck in their heads of convergence zones, surface ducts, and bottom limited environments. The thing is, ray traces are not the best way to model sound in the ocean. They're only really good for relatively high frequencies and they don't really take into account things like interference.

What you really need to do is compute the normal modes of the parabolic equation. There's different algorithms for doing that. There's a book called Computational Ocean Acoustics which is all about just that.

Most of what you see on the web page is a graphical representation of transmission loss as a result of computing the normal modes of the parabolic equation.

As for real time.. it really depends on what you mean by real time. Once every few minutes? I think you could do that. I doubt you could use the parabolic equation for a video game, though. It would take a lot of horsepower. It'd be neat if you could, though. People would definitely find it a lot harder to estimate how far they could see because it could vary so much for no obvious reason.

[Dr.Sid]

Hmm .. this book is on google books . Awfully incomplete, as usual. You cost me a lot SeaQueen, you know that ? :rotfl:

Well I know raytracer is not enough. But I'm having troubles finding the differences. Let's say comparison between raytrace and other methods, on pictures, side by side.

For the game all this level is not needed. We just would use some of the features, like fluent changes (compared to steep changes), changes in time, sound channel, more effects of the bottom in active sonar, and some other stuff.

First we must select what effects are important and needed. Then we can talk about how to do it. I can do the later. I hope you SeaQueen will help me a bit with the former.

[Castout]

..okay.

Now can we enhance the passive sonar detection capability:rotfl:

[Molon Labe]

According to most publications, RL detection range is far shorter than in DW now. So, NO.

[Castout]

Really? I just tried a test with DW. This time a Type 42 DD at 2900 yards only shows up as thin line in 688i's normal BB. I went to NB to try to classify it and it was classified as either 2 torpedoes...weird.

In the end the DD detected me much faster than I did and launched 2 torpedoes at me missing me because I was too close.......

So I say most publications LIE. DD more quiet than a sub?

I have some screenies but too lazy to put them:rotfl:
I was travelling at 5 knots bottom at 9,000 somethin probably feet. No thermal layer.

[SeaQueen]

Originally Posted by Dr.Sid

Hmm .. this book is on google books . Awfully incomplete, as usual. You cost me a lot SeaQueen, you know that ? :rotfl:

Well I know raytracer is not enough. But I'm having troubles finding the differences. Let's say comparison between raytrace and other methods, on pictures, side by side.

The biggest difference I've noticed is that raytracers tend to be pretty good approximations for TL in the thousands of Hertz range, and awful in the tens and hundreds of Hertz range.

There's also some lecture notes that go well with that book on MIT's OpenCourse website.

First we must select what effects are important and needed. Then we can talk about how to do it. I can do the later. I hope you SeaQueen will help me a bit with the former.

It's like I said, the phenomenta that are important are frequency dependent. The sonic layer depth might matter a lot for a sonar operating in the thousands of Hertz band, but have an almost immeasurable effect on a sonar that operates in the hundreds of Hertz. How to capture that in the game, is tough, though, because you'd have to solve the parabolic equations somehow (See the book I recommended).

[SeaQueen]

Originally Posted by Molon Labe

According to most publications, RL detection range is far shorter than in DW now. So, NO.

You know... back in the day, much was said about how sonar could detect things at ranges of hundreds or sometimes even thousands of miles. That's still true. The thing is, it all depends on the target and the acoustic conditions at the time. People don't like to believe that the really is that much variability but there is. In the '80s they were worried about hordes of NOVEMBER class submarines forcing their way through chokepoints in the northern latitudes, now they're worried about smaller numbers of KILO class submarines in tropical and subtropical latitudes. It's not that the sonars have changed, is that the environment and the targets have.

[Molon Labe]

Originally Posted by Castout

Really? I just tried a test with DW. This time a Type 42 DD at 2900 yards only shows up as thin line in 688i's normal BB. I went to NB to try to classify it and it was classified as either 2 torpedoes...weird.

In the end the DD detected me much faster than I did and launched 2 torpedoes at me missing me because I was too close.......

So I say most publications LIE. DD more quiet than a sub?

I have some screenies but too lazy to put them:rotfl:
I was travelling at 5 knots bottom at 9,000 somethin probably feet. No thermal layer.

none of this information is any good unless you tell us the acoustic conditions, which sonar you're using, and what depth. And telling us what the filter classifies a contact is, is worthless with regard to sonar performance no matter what.

[Dr.Sid]

Originally Posted by SeaQueen

I said, the phenomenta that are important are frequency dependent. The sonic layer depth might matter a lot for a sonar operating in the thousands of Hertz band, but have an almost immeasurable effect on a sonar that operates in the hundreds of Hertz. How to capture that in the game, is tough, though, because you'd have to solve the parabolic equations somehow (See the book I recommended).

I found some more papers on the web, especially here:http://oalib.hlsresearch.com/PE
But man, this really goes beyond my math skills. In very very rough terms I understand what they talk about, but I'm pretty far from grasping the problem at the moment.

Will buy the book, anyway. It costs quite a lot, but I never regret buying Urick (which YOU made me to buy ), which was not so cheap too.

[Dr.Sid]

After thinking about it I suddenly came to a theory what it's (I mean parabolic equation) all about and by looking on the formulas again it seems I'm right. I still will need lots of study, but at least I have SOME idea now. I'm also trying some very naive and rough wave based simulations. Which leads me to this: we really need just approximate what's going on. It means faking. For example if you say SSP has not much effect on low frequency, which means passive sonar, we can say OK, let's ignore SSP for passive. Level of the approximation is one thing. We can have simple methods, which will be frequency dependent, as well as complicated methods which wont. I'm trying to understand as much as possible, so I can decide 'this effect can be approximated by such and such simple curve, no need to simulation' or 'this could be simulated quite well, even quite fast, and it will help the game' or 'this is impossible to simulate or fake, and/or unimportant for the game, let's ignore it (but state in manual, of course)'.
It's not that frequency dependent effects are impossible or would require complex simulation. We don't need to be much true to the real world, we don't design real-world sonar set or something. We just need to catch the basic nature, the most important factors.

Which again reminds me one cool idea. Sonar is about listening right ? Even these days. Then why the hell does DW use few samples for all that boats and subs ? Why those samples are speed independent ? Why there are no transients ? It would be great fun to be able to HEAR: 'it's typhoon, doing lets say 10 kts .. ah ! changing depth now .. opening silo hatches !' Imagine tutorial missions teaching you to tell 2 screw from one and so on.

[SeaQueen]

Originally Posted by Dr.Sid

I found some more papers on the web, especially here:http://oalib.hlsresearch.com/PE
But man, this really goes beyond my math skills. In very very rough terms I understand what they talk about, but I'm pretty far from grasping the problem at the moment.

The phenomena I descibed, where low frequencies aren't ducted but high frequencies are is described on p. 151 of Urick. Absolutely most basics of the normal mode theory (basically what the papers are getting into) for sound propagation is on p. 122 of Urick, although honestly, he really doesn't get into it there. He also does a little bit on p. 174. Urick, though, came from an era when ray-tracing was the general rule and computing power was limited. Now-a-days, computing power is significantly greater and so people parabolic equation models of sound in the ocean are favored. The results I showed you earlier come from parabolic equation models.

Basically, all of those models have different tricks for approximating the solutions to differential equations governing the behavior of a sound wave in the ocean given all sorts of complicated boundary conditions.

The bad news is, even though the models are very good approximations, the oceanographic data going into them typically isn't, so the output is still subject to great uncertainty. It captures a lot of phenomena that ray traces don't get, though.

Will buy the book, anyway. It costs quite a lot, but I never regret buying Urick (which YOU made me to buy ), which was not so cheap too.

Urick is wonderful. There's just so much in there. Sometimes it's weird because every time I think there's something that's not in Urick, I end up re-reading it and realize that I was the dumb-dumb you had forgotten something, not him.

Don't buy a book you can't understand, though. I used to do that in highschool and didn't really catch up with myself until after I'd finished my undergraduate degree.