DW sound propagation model measurements - Page 2

Pisces · 07-20-07, 11:34 AM

Quote:

Originally Posted by Dr.Sid

... But upgrading SoPro is number one priority now (shorter name for sound propagation .. how do you like it ? lol).

Cool! But how about "SoProp-er"? Proper in a UK dictionary means better or right. "It's the proper way to simulate sound propagation"

Dr.Sid · 07-22-07, 06:07 PM

Uh .. so exact filling of area between rays showed little more complicated than it seemed son I only solved some basic cases and SoPro 1.1 is not ready for release yet. But it looks promising !

Dr.Sid · 07-24-07, 05:24 PM

Another night behind me and I'm getting close to new release. There are some glitches where beams flips (those bright dots and jaggies) and the bugs when editing SSP are not taken care of yet, but whole brand new beam based simulation (rather then ray based) is done. I've also fixed some bugs in intensity, now I'm few orders closer to simple spherical spreading if I try to simulate it.

All random numbers are gone, all non-covered pixels are gone.

Small delta time now only improves bending simulation, not coverage.

Delta angle can be no much higher (I use 5 degrees now !). There are two mechanisms which allows it. First I have beams with non-zero width now, and beam is divided into more if it gets too wide. Like this I can use small count of beams where it is not important and more beams where it is needed. All this can of course be controled by user.

Also interference simulation got better because of better coverage.

Here is quick peek of actual results .. note that delta angle is 5 degrees and delta time is 0.01s !

Other things I'd like to make yet (tomorrow I hope):
- fix known bugs (thanks to all for reports)
- Session save/load (options,ssp)
- individual rays overlay - you will be able to specify another delta angle (let's say 10 degrees) and individual rays will be displayed at this angles, with full intensity (and different colors), so you can see what the sound is doing.
- Intensity overlay - you will be able to display intensity for given depth as graph where intensity will be on Y axis (logarithmic scale)

Possible changes:
- logarithmic display of brightness (not during the simulation, but after it).
- better spline function for SSP definition to allow linear gradients and better control.

Pisces · 07-25-07, 06:32 AM

Congratulations on the progress DrSid. Really looks promissing indeed. *Applause*

But in all honesty, we sort of hijacked your topic. It's not about your DW test anymore (which is interesting enough itself), but about the propagation tool. We best dig up that first Soundpropagation tool topic and continue there.

Dr.Sid · 08-18-07, 10:56 AM

I updated the first post with new convergence zones measurements. Nothing unexpected was found, but the data now are more complete. I've updated both image and text.

In short I can say that CZ are about the same as surface duct. Except:

1) layer is more shallow, shadow zone starts closer.
2) there are the convergence zones, exactly at 30 and 60 nm. That one on 30 nm makes signal stronger from 18 to 39 SNR. That on 60 nm cannot bring the signal above 0 SNR, but on NB display it can change practically invisible line into 2 quite strong lines.

Convergence zones both are pretty narrow, you can cross them in seconds (depending on speed difference) so they are easy to miss.

Pisces · 08-20-07, 06:04 PM

Excellent work DrSid!!

There's one thing I was wondering about the shadowzone(s) from your previous graph and this one. Is that front-edge within 5.5-11Nm below the layer a parabolic shape? The previous graph had more datapoints IIRC. I can understand if it is impossible to say with these results. But it would be nice to know how far you can get into that shadow zone corner.

Dr.Sid · 08-21-07, 04:25 AM

Quote:

Originally Posted by Pisces

Excellent work DrSid!!

There's one thing I was wondering about the shadowzone(s) from your previous graph and this one. Is that front-edge within 5.5-11Nm below the layer a parabolic shape? The previous graph had more datapoints IIRC. I can understand if it is impossible to say with these results. But it would be nice to know how far you can get into that shadow zone corner.

Will do more tests. Both edges have 3 points, but it is not so obvious for the the new one, and both show slight parabolic shape. It's quite steep anyway. For getting as close as possible to moving target you would need some safe margin (you also don't know the distance exactly), so I guess the curve is not so important.
How the start of the shadow zone depends on the layer depth would be more interesting knowledge, but again it will change in tenth of mile.

Dr.Sid · 11-11-07, 08:11 PM

I did some last test of submerged-submerged situations. I have finally cleared all my doubts and I updated (for the last time I hope) the first post once again.

It showed that in the end in DW the sound transmission is symmetric. It means when A hears B, B will hear A at exactly same quality. So if surface listener hears sub bellow the layer with some transmission loss, the sub will hear the surface boat with same loss. If sub bellow the layer is in the shadow zone, surface boat will not be able to hear it and also the sub will not be able to hear the surface boat. Surface won't be able to ping the sub: sub won't hear the pings, and there will be no returns from the sub. This all is correct and realistic.

Please note that this can seem to be broken sometimes, since for example FFG uses long towed array. So it can seem to the array that sub is lost, but for FFG it is not in shadow zone yet and it still can ping it. That's what fooled me few times.

So what we already now, but inverted: sub above the layer can be heard normally. Sub bellow the layer can be heard somewhat worse (30 SNR to 20 SNR, 3 lines to 1 line). Active sonar is affected in similar way. The 'dot' is smaller and you can hear the return. Sub in the shadow zone can't be heard nor pinged.

Now what we didn't know. When the listener is under the layer too, it can hear the sub-layer target even in the shadow zone and the detection range is limited only by distance and gradual signal weakening. That was to be expected. But what is little surprise, the signal too will be weaker then if both target and listener were above the layer. Simply bellow the layer there are worse conditions.

Check the first post for final conclusions.

MarkShot · 11-11-07, 08:53 PM

Okay, I went back to your first post. So, what are your conclusions for the non-expert players; casual players like me?

(1) Except for the shadow zone, thermals do not play a significant role in combat. Fast deductions to accelerate ID and solutions and/or evasive maneuvers to frustrate a solution are more important than the small factors of thermal loss/enhancement on acoustic behavior.

(2) Thermal affects in DW are symmetric unlike in real life.

(3) Acoustics affects are from array to hull. So, it is possible for a hull to be in a shadow zone while its towed array is not. Thus, potentially yielding an asymetric situation with regards to detection.

(4) Both passive and active sonar follow the same model.

Are the above correct?

---

Also, I think one other piece of information you should add in your other notes:

The wash out created by explosion is purely a graphic affect provided to the player to enhance atomosphere. It is not actually part of the acoustic modeling at all.

I have seen this in a number of ways:

(1) Dynamic objects created as the result of a destruction of another object are able to gain contact immediately after an explosion; not at all impeded.

(2) I have assigned trackers immediately following explosions by assigning at known bearings despite whatever is displayed on the screen.

(3) TMA station shows tracking lines being generated during periods of explosions.

What this means for the player is that immediately following an explosion (for the next few minutes), you are a very vulnerable to the AI. (although I haven't tried NB to say if it is degraded during this time ...) This is especially true when close dynamic units are instantiated.

---

Thanks for your detailed work to improve understanding of the game engine!

---

Reference:

http://www.subsim.com/radioroom/showthread.php?t=122769

Deamon · 11-11-07, 08:58 PM

Quote:

Originally Posted by Dr.Sid

Or here you can check picture from my ray-tracer (download here).

Good stuff Dr.Sid but the link doesn't work!

Dr.Sid · 11-12-07, 04:49 AM

For the new version of SoPro utility go here:
http://www.subsim.com/radioroom/showthread.php?t=120487

I'll to simplify my conclusions.

There are 3 SSPs in DW. Bottom limited (BL), Surface duct (SD) and Covergence Zones (CZ).

In BL there is no layer and there is no difference no matter what depth you or the other ship is. Signal gets weaker with distance and that is all.

In SD a CZ there is layer. In CZ it is more shallow, but otherwise CZ behaves very much like SD. Layer can TOTALLY hide you. To achieve this, you must be under the layer, and you must be distant enough. My measurements shows this distance is about 6 nm with SD and 4nm with CZ. Also the one who is trying to detect you must by above the layer ! I call this situation 'being in the shadow zone'.
If you are closer than this distance, but under the layer, your signal will only get weaker, and for subs, it can complicate identification. However you will be detected (as you are very close now).

This applies for both active and passive sonar. If you are trying to evade active sonar into the shadow zone, get under the layer and get distant enough. You know that you are hidden when you no longer hear the pings.

It is possible to detect target under the layer, but the listener must get under the layer too. And even then the transmission of sound under the layer is much worse then above the layer. I guess the detection range bellow the layer can be even half of that above the layer, and ID of good sub can be certain only at very short distances (under 5nm).

Note that all transitions between these states (normal signal, signal weakened by layer, shadow zone) are sharp. There is no gradual signal weakening. Once they hear you, in the moment you cross the magic border they don't.
Depth also plays no other role in detection. You either are below the layer or above it.

There are some gradual changes in signal strength in DW, but they are not caused by SSP.

With CZ there is also minor signal increase for surface to surface at 30nm and 60nm.

feld · 08-22-08, 01:57 PM

Dr. Sid,

Do you happen to recall water depth during these tests?

-feld

Dr.Sid · 08-24-08, 08:42 AM

Don't have missions anymore but IIRC it was 10.000 to 12.000 feet and in different sets of test I verified that water depth has absolutely no effect on sonar (at least passive one).

feld · 08-28-08, 02:21 PM

Quote:

Originally Posted by Dr.Sid

Don't have missions anymore but IIRC it was 10.000 to 12.000 feet and in different sets of test I verified that water depth has absolutely no effect on sonar (at least passive one).

That's deep enough: I think your own data shows some water depth effects on sonar. My source for much of what follows is here. But I think that Ulrich's Underwater Sound has similar discussions and I seem to recall that you mentioned the book in a previous post.

Spherical to Cylindrical Spreading:
This means that the exponential tail on the left side of the graph is probably spherical spreading. Ulrich mentions it in the "transmission loss" chapter of his book. Around pg 120 I think but I don't have mine handy. Anyway spherical spreading is used to compute transmission loss when the range is short (for example, my source link above uses less than half the water column height) in simple SONAR equation based propagation models. 10,000-12,000 feet is 3 to 4 kiloyards or roughly 1.5 to 2 nm which matches up nicely with the transition from exponential tail to linear slope at the far left side of your graph. I'll bet that, if you went someplace really really deep (like marianas trench) that this change from exponential (actually 1/R^2) to linear would occur at a longer range. The transition should occur at the

holy Bottom Bounce Batman!
More exciting, I just realized that the large "hump" on the Bottom Limited curve and the smaller hump on the "Surface Duct" curves could be bottom bounce transmission loss!

Take a look at this picture from the FAS website below:
It's a curve of transmission loss (dB) versus range (km). Note that the zero for TL is at the top of the graph. This means that TL numbers near the bottom of the graph mean higher losses and lower probability of detection for a source/receiver combination at the range on the bottom.

Source: http://www.fas.org/man/dod-101/navy/...P/snr_prop.htm

Note the "hump" structure at ~10km. That's bottom bounce "gain" i.e. a range at which less sound energy is lost in transmission. If you imagine the TL vs range line above subtracted from a horizontal line measuring signal strength and plotted it might be easier to see...

@Dr. Sid: Do you remember how the "Surface Duct" curve looked when the source and receiver were below the layer? There should be a BB hump there too...if I'm right.

-feld

edit: to add the stuff on bottom bounce

Dr.Sid · 08-28-08, 04:02 PM

Cool .. somebody really looked at my data

Unfortunately I can't remember about the situation where both listener and source are under the layer. This calls for a another test.

As for the bulge .. it could be even bottom bounce, in case that reflection angle would be taken into account (or phase interference). Otherwise the bottom effect would be uniform for all distances. Tests with different water depth would solve this. I recommend you try it, it's great fun, but little time consuming.
I'm quite busy and I invest all spare time in the 'other project' (see my signature).
It's a pity we most probably will never know how exactly DW model works, and it still can surprise.

11-11-07, 08:53 PM	#24
MarkShot Ace of the Deep Join Date: Jul 2002 Posts: 1,134 Downloads: 93 Uploads: 0	Okay, I went back to your first post. So, what are your conclusions for the non-expert players; casual players like me? (1) Except for the shadow zone, thermals do not play a significant role in combat. Fast deductions to accelerate ID and solutions and/or evasive maneuvers to frustrate a solution are more important than the small factors of thermal loss/enhancement on acoustic behavior. (2) Thermal affects in DW are symmetric unlike in real life. (3) Acoustics affects are from array to hull. So, it is possible for a hull to be in a shadow zone while its towed array is not. Thus, potentially yielding an asymetric situation with regards to detection. (4) Both passive and active sonar follow the same model. Are the above correct? --- Also, I think one other piece of information you should add in your other notes: The wash out created by explosion is purely a graphic affect provided to the player to enhance atomosphere. It is not actually part of the acoustic modeling at all. I have seen this in a number of ways: (1) Dynamic objects created as the result of a destruction of another object are able to gain contact immediately after an explosion; not at all impeded. (2) I have assigned trackers immediately following explosions by assigning at known bearings despite whatever is displayed on the screen. (3) TMA station shows tracking lines being generated during periods of explosions. What this means for the player is that immediately following an explosion (for the next few minutes), you are a very vulnerable to the AI. (although I haven't tried NB to say if it is degraded during this time ...) This is especially true when close dynamic units are instantiated. --- Thanks for your detailed work to improve understanding of the game engine! --- Reference: http://www.subsim.com/radioroom/showthread.php?t=122769 __________________ War games, not wars! --- Only a small few profit from war (that should not stand)!

08-22-08, 01:57 PM	#27
feld Loader Join Date: Sep 2003 Posts: 84 Downloads: 22 Uploads: 0	How deep was the water? Dr. Sid, Do you happen to recall water depth during these tests? -feld

08-24-08, 08:42 AM	#28
Dr.Sid The Old Man Join Date: May 2005 Location: Czech Republic Posts: 1,458 Downloads: 6 Uploads: 0	Don't have missions anymore but IIRC it was 10.000 to 12.000 feet and in different sets of test I verified that water depth has absolutely no effect on sonar (at least passive one). __________________ ComSubSim forum at CADC

08-28-08, 04:02 PM	#30
Dr.Sid The Old Man Join Date: May 2005 Location: Czech Republic Posts: 1,458 Downloads: 6 Uploads: 0	Cool .. somebody really looked at my data Unfortunately I can't remember about the situation where both listener and source are under the layer. This calls for a another test. As for the bulge .. it could be even bottom bounce, in case that reflection angle would be taken into account (or phase interference). Otherwise the bottom effect would be uniform for all distances. Tests with different water depth would solve this. I recommend you try it, it's great fun, but little time consuming. I'm quite busy and I invest all spare time in the 'other project' (see my signature). It's a pity we most probably will never know how exactly DW model works, and it still can surprise. __________________ ComSubSim forum at CADC

07-22-07, 06:07 PM	#17
Dr.Sid The Old Man Join Date: May 2005 Location: Czech Republic Posts: 1,458 Downloads: 6 Uploads: 0	Uh .. so exact filling of area between rays showed little more complicated than it seemed son I only solved some basic cases and SoPro 1.1 is not ready for release yet. But it looks promising !

07-24-07, 05:24 PM	#18
Dr.Sid The Old Man Join Date: May 2005 Location: Czech Republic Posts: 1,458 Downloads: 6 Uploads: 0	Another night behind me and I'm getting close to new release. There are some glitches where beams flips (those bright dots and jaggies) and the bugs when editing SSP are not taken care of yet, but whole brand new beam based simulation (rather then ray based) is done. I've also fixed some bugs in intensity, now I'm few orders closer to simple spherical spreading if I try to simulate it. All random numbers are gone, all non-covered pixels are gone. Small delta time now only improves bending simulation, not coverage. Delta angle can be no much higher (I use 5 degrees now !). There are two mechanisms which allows it. First I have beams with non-zero width now, and beam is divided into more if it gets too wide. Like this I can use small count of beams where it is not important and more beams where it is needed. All this can of course be controled by user. Also interference simulation got better because of better coverage. Here is quick peek of actual results .. note that delta angle is 5 degrees and delta time is 0.01s ! Other things I'd like to make yet (tomorrow I hope): - fix known bugs (thanks to all for reports) - Session save/load (options,ssp) - individual rays overlay - you will be able to specify another delta angle (let's say 10 degrees) and individual rays will be displayed at this angles, with full intensity (and different colors), so you can see what the sound is doing. - Intensity overlay - you will be able to display intensity for given depth as graph where intensity will be on Y axis (logarithmic scale) Possible changes: - logarithmic display of brightness (not during the simulation, but after it). - better spline function for SSP definition to allow linear gradients and better control.

07-25-07, 06:32 AM	#19
Pisces Silent Hunter Join Date: Dec 2004 Location: AN9771 Posts: 4,904 Downloads: 304 Uploads: 0	Congratulations on the progress DrSid. Really looks promissing indeed. Applause But in all honesty, we sort of hijacked your topic. It's not about your DW test anymore (which is interesting enough itself), but about the propagation tool. We best dig up that first Soundpropagation tool topic and continue there.

08-18-07, 10:56 AM	#20
Dr.Sid The Old Man Join Date: May 2005 Location: Czech Republic Posts: 1,458 Downloads: 6 Uploads: 0	I updated the first post with new convergence zones measurements. Nothing unexpected was found, but the data now are more complete. I've updated both image and text. In short I can say that CZ are about the same as surface duct. Except: 1) layer is more shallow, shadow zone starts closer. 2) there are the convergence zones, exactly at 30 and 60 nm. That one on 30 nm makes signal stronger from 18 to 39 SNR. That on 60 nm cannot bring the signal above 0 SNR, but on NB display it can change practically invisible line into 2 quite strong lines. Convergence zones both are pretty narrow, you can cross them in seconds (depending on speed difference) so they are easy to miss.

08-20-07, 06:04 PM	#21
Pisces Silent Hunter Join Date: Dec 2004 Location: AN9771 Posts: 4,904 Downloads: 304 Uploads: 0	Excellent work DrSid!! There's one thing I was wondering about the shadowzone(s) from your previous graph and this one. Is that front-edge within 5.5-11Nm below the layer a parabolic shape? The previous graph had more datapoints IIRC. I can understand if it is impossible to say with these results. But it would be nice to know how far you can get into that shadow zone corner.

11-11-07, 08:11 PM	#23
Dr.Sid The Old Man Join Date: May 2005 Location: Czech Republic Posts: 1,458 Downloads: 6 Uploads: 0	I did some last test of submerged-submerged situations. I have finally cleared all my doubts and I updated (for the last time I hope) the first post once again. It showed that in the end in DW the sound transmission is symmetric. It means when A hears B, B will hear A at exactly same quality. So if surface listener hears sub bellow the layer with some transmission loss, the sub will hear the surface boat with same loss. If sub bellow the layer is in the shadow zone, surface boat will not be able to hear it and also the sub will not be able to hear the surface boat. Surface won't be able to ping the sub: sub won't hear the pings, and there will be no returns from the sub. This all is correct and realistic. Please note that this can seem to be broken sometimes, since for example FFG uses long towed array. So it can seem to the array that sub is lost, but for FFG it is not in shadow zone yet and it still can ping it. That's what fooled me few times. So what we already now, but inverted: sub above the layer can be heard normally. Sub bellow the layer can be heard somewhat worse (30 SNR to 20 SNR, 3 lines to 1 line). Active sonar is affected in similar way. The 'dot' is smaller and you can hear the return. Sub in the shadow zone can't be heard nor pinged. Now what we didn't know. When the listener is under the layer too, it can hear the sub-layer target even in the shadow zone and the detection range is limited only by distance and gradual signal weakening. That was to be expected. But what is little surprise, the signal too will be weaker then if both target and listener were above the layer. Simply bellow the layer there are worse conditions. Check the first post for final conclusions.

11-12-07, 04:49 AM	#26
Dr.Sid The Old Man Join Date: May 2005 Location: Czech Republic Posts: 1,458 Downloads: 6 Uploads: 0	For the new version of SoPro utility go here: http://www.subsim.com/radioroom/showthread.php?t=120487 I'll to simplify my conclusions. There are 3 SSPs in DW. Bottom limited (BL), Surface duct (SD) and Covergence Zones (CZ). In BL there is no layer and there is no difference no matter what depth you or the other ship is. Signal gets weaker with distance and that is all. In SD a CZ there is layer. In CZ it is more shallow, but otherwise CZ behaves very much like SD. Layer can TOTALLY hide you. To achieve this, you must be under the layer, and you must be distant enough. My measurements shows this distance is about 6 nm with SD and 4nm with CZ. Also the one who is trying to detect you must by above the layer ! I call this situation 'being in the shadow zone'. If you are closer than this distance, but under the layer, your signal will only get weaker, and for subs, it can complicate identification. However you will be detected (as you are very close now). This applies for both active and passive sonar. If you are trying to evade active sonar into the shadow zone, get under the layer and get distant enough. You know that you are hidden when you no longer hear the pings. It is possible to detect target under the layer, but the listener must get under the layer too. And even then the transmission of sound under the layer is much worse then above the layer. I guess the detection range bellow the layer can be even half of that above the layer, and ID of good sub can be certain only at very short distances (under 5nm). Note that all transitions between these states (normal signal, signal weakened by layer, shadow zone) are sharp. There is no gradual signal weakening. Once they hear you, in the moment you cross the magic border they don't. Depth also plays no other role in detection. You either are below the layer or above it. There are some gradual changes in signal strength in DW, but they are not caused by SSP. With CZ there is also minor signal increase for surface to surface at 30nm and 60nm.