Okay, I'm going to toss out a topic for a great research project that I wish I had time to do. If someone wants to pick up the ball and run with it, I think it would be interesting.
On page 2 of the good old (and I mean old -- the last revision was 1984), CAPF 104 you will find several tables for you to use to determine the probability of detection of a crashed airplane based on your altitude, search visibility, track spacing, and type of terrain.
I'm sure that this is probably based on research done in the 1950s or 1960s, but I don't know specifically where it came from, though you can find similar tables in various places.
What I would like to see is a project comparing actual PODs on actual CAP missions to the table.
Ideally, here is what you would need for each datapoint:
1. Actual location of the crash
2. From the CAPF-104s for that mission, you would need to know how many times that grid was searched or how many times it was overflown on a route search.
3. Use GIS to come up with a reasonably accurate description of the terrain in some standardized fashion. For example, describe the terrain and tree cover in a 5 mile circle around the crash site (% forest, % open, % open water, etc.).
Realistically, it may not be possible to get the actual 104s from the missions as this would involve a lot more cooperation from individual wings than you would probably get.
However, you could go to the AFRCC and get copies of the CAPF-122s for each day of the mission, which should list all grids searched. This wouldn't be quite as good since there is no way to tell how often a particular grid was searched in a single day, but you would know if the same grid was searched on multiple days.
So, with this you can come up with some actual PODs and compare them to the table and see if they are still reasonable or not.
Another thing you could do with this data is analyze how long it took before CAP first searched the grid where the crash site was. This would give us some idea of how well we're doing on defining our search area and determining the highest priority grids to search. You could also use the 122s to get an idea of how much time is spent searching far from the actual crash site. Again, this would help us see how we're doing at prioritization.
You're going to need probably 5-10 years of nationwide data to have enough to do a good analysis on, which could take a while to gather.
Go for it. Thank RiverAux in the acknowledgement section of your thesis.
I never took the POD as "gospel", rather a simple measure of the coverage that a particular piece of real estate has received.
(Theoretical): Would a POD of 100% mean that a crash site isn't in the grid if the target hasn't been found? I know that a 100% POD is a fairytale (without the forms in front of me, I'm not even sure that POD is possible - don't thnk so).
I see it as a planning tool. It shows me as a Planning Section Chief where I need to apply assets. It does IMO normalize the difference between open terrain and more difficult situations (rolling, high vegetation, etc).
USCG uses POD along with probability of containment in a formula to establish probability of success:
POD X POC = POS
Now their POC has to do with oceanic drift of a search target in the water, but we work with similar variables: last known point, endurance, winds, etc. Is their POD still viable? Don't know. I suspect they use theirs the same way, using it as a yard stick to make sure they're covering areas and using their resources optimally.
I would think the only time your POD is 100% is when you found what you're looking for ;D
Interesting idea, but the proposed method of validating the POD tables won't work.
The definition of POD from the International Aeronautical and Maritime SAR (IAMSAR) Manual, which the US has adopted is:
The probability of the search object being detected, assuming it was in the areas that were searched. POD is a function of coverage factor, sensor, search conditions and the accuracy with which the search facility navigates its assigned search pattern. Measures sensor effectiveness under the prevailing search conditions.The key is that POD is a function of many factors, far more than the those in the POD tables which are terrain, vegetation, search altitude, visibility and track spacing. The US National SAR Supplement to the IAMSAR Manual lists these additional factors (and there are others) to consider when determining POD:
ability to maintain optimum altitude and airspeed, weather conditions, accuracy of navigation, search crew fatigue and size and characteristics of search object. The last two define detectability.
Factors of detectability are size, shape, color, brightness, contrast with the surroundings, duration of exposure (how long was the object within view) and distance from the viewer.
We don't know how these factors were used when the POD tables were developed (i.e. the "experimental controls"). The only way the tables could be validated would be if the objects that were found matched the controls.
Looking at it realistically, instead of theoretically, the detectability of a wing from a 172, the cargo door from the 172, and the damage to the trees at the crash site all have a different detectability, therefore POD, even for the same factors in the POD table.
How many debriefing officers, after the aircrew circles the POD on the back of the 104, ask "Is that POD for a wing, a cargo door or the broken branches in the tree?" How many aircrews think of POD during debriefing that way? Without knowing what object the POD applies to, the POD number is meaningless.
For the purposes of the proposed study, even if the researcher could get the original 104s, the POD factor or type of object would more than likely not have been recorded. Nor would it have been recorded consistently between 104s. Even if there was a block on the 104 asking the crew to circle wing, door, or broken branches for the factor that they used to determine POD, it would be nothing more than a WAG. It couldn't be compared to the same number on a different 104.
QuoteUSCG uses POD along with probability of containment in a formula to establish probability of success:
POD X POC = POS
Now their POC has to do with oceanic drift of a search target in the water, but we work with similar variables: last known point, endurance, winds, etc. Is their POD still viable? Don't know. I suspect they use theirs the same way, using it as a yard stick to make sure they're covering areas and using their resources optimally.
The POD x POC = POS formula comes from probability theory and is used exactly the same way regardless of what agency or search environment we're concerned with. It's usually associated with the Coast Guard only because the USCG has developed and uses mathematical search theory much more rigorously than the air and land SAR communities.
POC is probability of containment which is identical to probability of area, POA. Whether POA or POC is used depends on where a person received their search theory training. We should be using POS, not POD, to make sure we're searching wisely.
Mike
Very interesting - good info. Thx!
I'm not sure the POD tables really need to be on the 104s anyway as they are really a tool for planning. The track spacing and altitude are assigned already and the info that we need back from the aircrew to fill them out is the ground cover/terrain (obviously a major judgetment call) and search visibility.
Sardak makes some fine points about what is supposedly being detected, but as Gunner says, we're really only using these as guides so that degree of rigor isn't going to really be possible. But we do need to know if our guides are pointing us in the right direction or not. Besides those same issues were probably a factor when they came up with the tables in the first place.
So, what we want to know is whether the PODs we're using are really in the ballpark of reality or not. If the table says your POD in a particular situation is 70%, but it is really more like 10%, then we need to adjust that. Now, if the real POD is only 60%, then its probably ok as that will probably fall within the errors introduced by the factors cited by Sardak.
Now, if CAP wanted to spend some major bucks, we could do it the right way and develop a series of POD tables from scratch using multiple known realistic targets ranging from whole aircraft to holes in the ground and a series of sorties over them by dozens of crews in a half dozen different terrain/cover environments. But, I'm not sure that would really be worth the cost compared to the paperwork analysis I'm suggesting.
Some of the search theorists say that the POD tables shouldn't be used (or be on the 104s) because of the lack of understanding of the tables' development and proper use. These same people grudgingly admit that having them is slightly better than nothing because the tables provide consistency (even if wrong) and are better than searchers making random guesses at what their POD was. This leads to an entirely different discussion on using coverage to determine POD. Let's skip that.
Though the tables are a planning tool, having them on the 104s is good because it keeps all the brief/debrief info for a sortie in one place.
RiverAux's idea of doing tests to develop POD tables has been done in the ground SAR world. The goal was not to develop tables but a scientifically based system for estimating POD. Three studies were funded by the National SAR Committee and the Coast Guard (yes, for ground SAR). The tests were performed in different parts of the country at different times of the year. The final report of the series is the one titled "Sweep Width Estimation for Ground Search and Rescue." All three reports are available at:
http://www.uscg.mil/hq/cg5/cg534/nsarc/Land_SAR_reports.asp
Doing something similar for air SAR would definitely cost major bucks and I don't think anyone is interested in funding any studies.
Mike
And it is still PROBABILITY. That means even with a 95% POD, you can fly 20 search missions directly over the target and still fail to detect it. As Gunner said, it is just a number.
The most critical element of a search is correctly defining the POC - with probababilities for all the sub-areas. EX: POC = 100% POCa = 12%, POCb = 8%, POCc = 9%, etc. The POC and the POCa for a specific area will determne your POS for that area. Then you design your search plan to search the highest POS sub-areas with the most effort. You have to adjust every operational period based on the new POS for each sub-area after that day's search.
Much of this POD/POA came as a result from a book "Theory of Optimal Search" -- from way back when the navy was looking for a sub (or nuke) and wanting to figure out how to best search. In the early days of POD, it was a math formula and a fellow in PA wing put it into a better to understand graph. Many years ago I had a program in BASIC that we used on a Radio Shack Model I to figure POD/POA and other stuff. The program was CASP -- Computer Assisted Search Planning.
BTW the original POD stuff needed to have an ocean movement factor (drift) so when it was applied to aircraft, this variable was always set to zero. It took a few years to realize that we didn't even need this variable as few aircraft crashes "drifted" within CONUS.
Jerry Wellman
Utah Wing
But, as stated above, we do have a Last Known Point, duration, and winds. This is analogous to drift on the ocean.
QuoteYou have to adjust every operational period based on the new POS for each sub-area after that day's search.
I've seen some criticism of that approach based on the idea that you shouldn't necessarily adjust your search areas that much based on previous efforts. If there is an area where you judge that the target is likely to be found that is still going to be the same no matter how many times you search it.
Now, there is an tipping point once you've searched that high probability area numerous times where it probably isn't worth another effort, but apparently searchers often judge that they're at the point much earlier than they should. I'm aware of a recent incident near me that sort of backs that up.
That is where your adjusted POS comes in. REMEMBER - it is only a NUMBER. You use it to allocate your assets for the most possible gain. You are trying to maximize your efforts by searching the areas with the highest POS. The search areas are not adjusted - only the number of assets you are assigning to them.