Search arc statistical test (altitude)

[rader]

Hey all,

Ran some altitude tests. For those of you who saw the other thread, I set them up the same way: Coral Sea scenario (advanced weather off, but still think there are some weather effects?), 4 TFs (2CA/CL + 1DD x 4) at range 10.

What I did was make a save game of the set up and relaod the scenario for each altitude case.

Nells set to range 10, no search arcs set, 30% search/70% rest.

I tested 1000ft, 6000ft, 29995ft (max alt), 8 samples per case.

I tracked TFs spotted, claimed attacks, and ops loses. Note that no hit claims were real, even when the pilots once correctly reported the ship name (Chicago).

Results:
*=hit claim $=ops loss #=Flak loss

6000ft case (TFs spotted):
(4/3**/2/1/2**/3*$/2/3). Mean spot = 2.5.

1000 ft case:
(3/3/3/1$/3/3/2*#/2). Mean spot = 2.5.

29995 ft case:
(3/4/3$/3*/4**/4/2$/3$). Mean spot = 3.25.

Since there is not obviusly a significant difference, I decided to run an ANOVA between the cases on TF spotting. The result was F = 2.33; p = 01.22 (not significant).

Discussion:
Although there was no significant difference betwen the cases, interestingly (and contrary to my expectations), the highest altitude setting got the best results. Of course, it is much easier to show a result than "prove" a negative result, so I can't say for sure if more samples would have found something more significant. But I do think that it is highly unlikely that there is a strong effect of altitude on searching, at least in cases similar to these (range, target, searching a/c, etc.).

[Apollo11]

Hi all,

ORIGINAL: rader

Hey all,

Ran some altitude tests. For those of you who saw the other thread, I set them up the same way: Coral Sea scenario (advanced weather off, but still think there are some weather effects?), 4 TFs (2CA/CL + 1DD x 4) at range 10.

What I did was make a save game of the set up and relaod the scenario for each altitude case.

Nells set to range 10, no search arcs set, 30% search/70% rest.

I tested 1000ft, 6000ft, 29995ft (max alt), 8 samples per case.

I tracked TFs spotted, claimed attacks, and ops loses. Note that no hit claims were real, even when the pilots once correctly reported the ship name (Chicago).

Results:
*=hit claim $=ops loss #=Flak loss

6000ft case (TFs spotted):
(4/3**/2/1/2**/3*$/2/3). Mean spot = 2.5.

1000 ft case:
(3/3/3/1$/3/3/2*#/2). Mean spot = 2.5.

29995 ft case:
(3/4/3$/3*/4**/4/2$/3$). Mean spot = 3.25.

Since there is not obviusly a significant difference, I decided to run an ANOVA between the cases on TF spotting. The result was F = 2.33; p = 01.22 (not significant).

Discussion:
Although there was no significant difference betwen the cases, interestingly (and contrary to my expectations), the highest altitude setting got the best results. Of course, it is much easier to show a result than "prove" a negative result, so I can't say for sure if more samples would have found something more significant. But I do think that it is highly unlikely that there is a strong effect of altitude on searching, at least in cases similar to these (range, target, searching a/c, etc.).

Thanks! [:)]

BTW, would it be possible for you to modify the scenario and then load both WitP-AE and that new scenario anew for each test?

I know this is big pain but, unfortunately, this is the only way to properly test since RND numbers are not reset otherwise (i..e you would not get absolutely random seeds)...

When I tested UV and WitP this was the norm!


Leo "Apollo11"

[jetjockey]

Hit claims from 29K??? LGBs? Any confirmation of the hits?

[Djordje]

Size of TF influences naval search. Low level search will more easily find single ship TFs while high level search will be better for finding large TFs.
In your test TFs are about average so it is ok as the test goes, I just say that in the game one should select altitude based on what he is looking for.

[rader]

ORIGINAL: Djordje

Size of TF influences naval search. Low level search will more easily find single ship TFs while high level search will be better for finding large TFs.
In your test TFs are about average so it is ok as the test goes, I just say that in the game one should select altitude based on what he is looking for.

That's good to know [:)]

Are you sure? Is this based on knowledge of the code, experience, or is it a guess based on what makes sense?

[rader]

All the reported hits were FOW.

ORIGINAL: jetjockey

Hit claims from 29K??? LGBs? Any confirmation of the hits?

[CapAndGown]

ORIGINAL: Apollo11

Hi all,

ORIGINAL: rader

Hey all,

Ran some altitude tests. For those of you who saw the other thread, I set them up the same way: Coral Sea scenario (advanced weather off, but still think there are some weather effects?), 4 TFs (2CA/CL + 1DD x 4) at range 10.

What I did was make a save game of the set up and relaod the scenario for each altitude case.

Nells set to range 10, no search arcs set, 30% search/70% rest.

I tested 1000ft, 6000ft, 29995ft (max alt), 8 samples per case.

I tracked TFs spotted, claimed attacks, and ops loses. Note that no hit claims were real, even when the pilots once correctly reported the ship name (Chicago).

Results:
*=hit claim $=ops loss #=Flak loss

6000ft case (TFs spotted):
(4/3**/2/1/2**/3*$/2/3). Mean spot = 2.5.

1000 ft case:
(3/3/3/1$/3/3/2*#/2). Mean spot = 2.5.

29995 ft case:
(3/4/3$/3*/4**/4/2$/3$). Mean spot = 3.25.

Since there is not obviusly a significant difference, I decided to run an ANOVA between the cases on TF spotting. The result was F = 2.33; p = 01.22 (not significant).

Discussion:
Although there was no significant difference betwen the cases, interestingly (and contrary to my expectations), the highest altitude setting got the best results. Of course, it is much easier to show a result than "prove" a negative result, so I can't say for sure if more samples would have found something more significant. But I do think that it is highly unlikely that there is a strong effect of altitude on searching, at least in cases similar to these (range, target, searching a/c, etc.).

Thanks! [:)]

BTW, would it be possible for you to modify the scenario and then load both WitP-AE and that new scenario anew for each test?

I know this is big pain but, unfortunately, this is the only way to properly test since RND numbers are not reset otherwise (i..e you would not get absolutely random seeds)...

When I tested UV and WitP this was the norm!


Leo "Apollo11"

If this statement were true, then each test would look exactly the same since each random number would be the same for each place for which a random number is needed since random numbers are chained.

[Apollo11]

Hi all,

ORIGINAL: cap_and_gown

ORIGINAL: Apollo11

I know this is big pain but, unfortunately, this is the only way to properly test since RND numbers are not reset otherwise (i..e you would not get absolutely random seeds)...

When I tested UV and WitP this was the norm!

If this statement were true, then each test would look exactly the same since each random number would be the same for each place for which a random number is needed since random numbers are chained.

It is true... the savegame contains the "RND seed" (at least it was always that way in UV and WitP - we can only assume that WitP-AE is the same)... sorry...


As for not looking 100% the same - that is not the case because there are many RND factors involved [;)] - but the "true" randomness is only achieved through restating game and loading scenario anew...


Leo "Apollo11"

[d0mbo]

It looks like OP losses are higher at max altidude? A fluke or significant?
 
 

[Djordje]

ORIGINAL: rader

ORIGINAL: Djordje

Size of TF influences naval search. Low level search will more easily find single ship TFs while high level search will be better for finding large TFs.
In your test TFs are about average so it is ok as the test goes, I just say that in the game one should select altitude based on what he is looking for.

That's good to know [:)]

Are you sure? Is this based on knowledge of the code, experience, or is it a guess based on what makes sense?

I've read it on the forum, I think one of the devs said it but not 100% sure as it was some time ago. Going to search for it.

[witpqs]

Yes they did say IIRC. Size of TF, size of ship(s), damage to ships (or maybe just system and float) can make them easier to spot, not sure what else.

[Djordje]

ORIGINAL: witpqs

Yes they did say IIRC. Size of TF, size of ship(s), damage to ships (or maybe just system and float) can make them easier to spot, not sure what else.

One of the things I remember was that ships on fire have really high detection level (smoke during the day, glowing at night).

[Pascal_slith]

Now I thought that many moons ago when there was discussion on this topic the devs indicated that optimal altitude was programmed somewhere in the 4 to 7k ft altitude range....

Yes, I do remember the random number generator problem. Yes, you have to reload the game to get acceptable random number sequences.

[Sardaukar]

Historically, Naval Searches (at least before a/c radar) were flown at altitudes of 1000-1500 ft. This was because it allowed ships to show as silouettes against horizon and worked lot better than flying higher.

[tigercub]

Sardaukar as i never realy taken much notice of how high they were flying in my reading of WW2 naval Searches ....but i am an aerial photographer light planes door off stuff 10years and that seems a little low...but not out of the question.

[Sardaukar]

ORIGINAL: tigercub

Sardaukar as i never realy taken much notice of how high they were flying in my reading of WW2 naval Searches ....but i am an aerial photographer light planes door off stuff 10years and that seems a little low...but not out of the question.

You an find the quotes about altituddes from for example "Shattered Sword". It was to get ship siluetted against horizon, which was not only useful for detection but also for recognition. And since one would have to fly below clouds anyway...

[PaxMondo]

ORIGINAL: Sardaukar


And since one would have to fly below clouds anyway...

In the WWII Pacific, I have always understood that this was the key driving force for visual search altitude. And the numerous rain squalls frequently hampered effective searching.