How many of those collisions on final involved a plane doing a straight-in?
91.126(b)(1) - Yes, Virginia, it is regulatory.
- Thread starter pburger
- Start date
wanttaja
En-Route
There were 25 cases (2008-2021) where the NTSB description said either "landing" or "final." This is all accidents, not just homebuilts. Four of the cases involved aircraft taking off in front of landing traffic. Again, I use "cases" to refer to the overall accident; not the count of aircraft involved. "25 cases" involved 50 aircraft (Didn't see any three-ways....).
Three cases involved aircraft which flew straight-in approaches. Two additional cases where one of the planes entered on base...one of which was a spiraling base entry ("He reported that his intention was to perform a simulated engine-out maneuver over the airport then “spiral down to set up for [the] left base” leg of the traffic pattern for runway 32.")
There were two cases where aircraft were on mirror patterns...one entering on left downwind, the other on a right downwind for the same runway... and collided on final. However, the cases involved dissimilar aircraft which may have justified this. One case had a helicopter on a right-hand pattern, the other case was a glider on right pattern that eventually collided with its own tow plane.
One case (ERA15LA084) where the planes involved flew standard patterns of different dimensions. Cherokee 140 in a broad pattern, Pitts in a tight one.
So pattern issues to varying degrees were associated with eight of the midairs. That's about 38% of the midairs that didn't involve aircraft taking off.
Other midairs:
Two cases where the aircraft were operating from different airports, close enough to allow intersection on final.
Four cases involved NORDO aircraft... in one case, BOTH aircraft were NORDO (two Ag aircraft).
Four other cases where pilots had radios, but either didn't use them, were on the wrong frequency, or the radios seemed to have issues.
The last bit of trivia is a surprisingly high survival rate...in over 70% of these accidents, no one was killed. In one case, the two pilots didn't even realize they'd been in a midair until after they landed. Most of these collisions occurred at relatively low altitude, which probably contributed to the survival rate. Overall, about 42% of midairs don't result in fatalities.
Ron Wanttaja
That’s amazing.
wanttaja
En-Route
Ha! Spoke too soon. In 2010, CEN10FA115 involved a Cirrus hitting a tow plane that was pulling a glider.
Looking past the era my analysis ran, I see that the year 2023 also had an odd number of aircraft involved in midairs. That one was a suspected drone strike; no NTSB entry on the drone itself.
Ron Wanttaja
flyingcheesehead
Touchdown! Greaser!
Ron,
With all due respect, your graph here is somewhat misleading in this case. You didn't break down "Maneuvering" and "Enroute", but approach and landing are broken down quite a bit. If "only" 36 of the midair cases occurred in the pattern, that's 25% which is nearly as many as you show for the Enroute bar.
Also, I'm guessing that number doesn't include "Approach" which is at least partially in the pattern (on final), "Initial climb" which is at least partially in the pattern (depending on your definition of "initial" it might all be), and either of the landing numbers.
What are the actual numbers for each category in the graph? Also, in the case of one aircraft being VFR on final and the other being IFR and on approach, how did you categorize that case?
That's not where all the merging is happening. It might be if you forced the 45, but right now most of the merging is likely at the base-to-final turn where all of the straight ins and IFR approaches meet all the VFR pattern traffic.
dbahn
Pattern Altitude
Another way to look at the numbers is "midair collisions in cruise phase of flight". I recall that number to be low compared to phases that involve climbing or descending, but I don't have the figures.
wanttaja
En-Route
I do combine several categories under the header "Enroute," etc. Here's the raw numbers:
| Occurrence_Description | Cases | _ | _ | _ | _ | _ | _ |
| Approach Midair collision | 10 | _ | _ | _ | _ | _ | _ |
| Approach-VFR pattern base Midair collision | 2 | _ | _ | _ | _ | _ | _ |
| Approach-VFR pattern downwind Midair collision | 4 | _ | _ | _ | _ | _ | _ |
| Approach-VFR pattern final Midair collision | 18 | _ | _ | _ | _ | _ | _ |
| Enroute Midair collision | 15 | _ | _ | _ | _ | _ | _ |
| Enroute-climb to cruise Midair collision | 4 | _ | _ | _ | _ | _ | _ |
| Enroute-cruise Midair collision | 21 | _ | _ | _ | _ | _ | _ |
| Enroute-descent Midair collision | 1 | _ | _ | _ | _ | _ | _ |
| Initial climb Midair collision | 5 | _ | _ | _ | _ | _ | _ |
| Landing Midair collision | 5 | _ | _ | _ | _ | _ | _ |
| Landing-flare/touchdown Midair collision | 2 | _ | _ | _ | _ | _ | _ |
| Maneuvering Midair collision | 37 | _ | _ | _ | _ | _ | _ |
| Maneuvering-aerobatics Midair collision | 4 | _ | _ | _ | _ | _ | _ |
| Maneuvering-hover Midair collision | 1 | _ | _ | _ | _ | _ | _ |
| Maneuvering-low-alt flying Midair collision | 7 | _ | _ | _ | _ | _ | _ |
| Other Midair collision | 1 | _ | _ | _ | _ | _ | _ |
| Uncontrolled descent Midair collision | 2 | _ | _ | _ | _ | _ | _ |
| Unknown Midair collision | 5 | _ | _ | _ | _ | _ | _ |
The single "Maneuvering" row in my graphic combines the four "Maneuvering" items above, ditto the "Enroute," etc.
Ron Wanttaja
Attachments
flyingcheesehead
Touchdown! Greaser!
Ah, that makes sense.
It seems like the NTSB likes to change up how they do things with their reports every 20 years or so, and in the process make it REALLY difficult to actually get good information out of the accident database!
Last couple of times I've done a study of the database I've thought that the NTSB, or even AOPA ASF, should just pay a couple of interns to spend a summer cleaning up the database. It's one of the dirtiest datasets I've ever worked with, and that's saying something. (I used to joke that I should put "Data Janitor" on my business cards.)
flyingcheesehead
Touchdown! Greaser!
Oh, and one more thing:
I would say that there are only two valid ways of looking at this data.
The only practical one is to combine all of the "Approach" categories, "Landing" categories, "Enroute" categories, etc... Because at different times, the NTSB has used different fields and different descriptions. So, there are certainly some "Approach Midair collision" events that happened on downwind, base, and final but were categorized with less detailed categories at the time. The raw data as is leads to conclusions like @Ed Haywood had: That there aren't many midairs on downwind. Unfortunately, that is likely not the case, but the data as the NTSB entered it is inconsistent.
The other one, that would require significant up-front effort, would be to read all of the reports and re-categorize things consistently.
wanttaja
En-Route
Yeah, it's kind of messed up. Just last year, they did some subtle change that messed up a bunch of my standard queries. Had to basically rebuild them.Ah, that makes sense.
It seems like the NTSB likes to change up how they do things with their reports every 20 years or so, and in the process make it REALLY difficult to actually get good information out of the accident database!
Last couple of times I've done a study of the database I've thought that the NTSB, or even AOPA ASF, should just pay a couple of interns to spend a summer cleaning up the database. It's one of the dirtiest datasets I've ever worked with, and that's saying something. (I used to joke that I should put "Data Janitor" on my business cards.)
Also, they did SOMETHING in 2008. Through 2019, the downloadable database included accidents back to the dawn 'o time. But starting in 2020, the downloadable database only included accidents since 2008. Oddly enough, the online search tool still goes back to the early days.
I'm skeptical of the flags in the database that your last approach is probably what I'll have to do. For the data I posted recently, I did actually go through all 145 reports and summarize what had actually happened. You can see it in the "Comments" column on the far right.
This not excessively time-consuming; I can set up a custom query with radio buttons ("Final", "Base", "Downwind", "straight in", "NORDO", etc.)
I did something very similar a couple of years back when I looked at the survivability of engine-failure accidents based on the type of ground where the pilot set down the plane. The material on the left is the standard database output and the radio buttons and check boxes are used based on reading the narrative.
I'm thinking of looking at midairs for 2008 through 2022. 247 midairs in that 15-year time period. That's not too taxing; it's only a bit more than the ~180 homebuilt accidents I look at every December.
Ron "But gad, do I need another hobby...." Wanttaja
flyingcheesehead
Touchdown! Greaser!
I thought when CAROL came out they were eliminating both the download and the query of older accidents, which is BS. There's still plenty to be learned from old accidents.
Do you still have a copy of the old database? I was able to grab the old XML/text versions of the flattened one, but a few years ago when I last did a big safety study I got a copy of a relational database from somewhere. Interestingly enough, the XML/Text ones can still be downloaded from archive.org.
Well, it's excessive in that you shouldn't have to do it in the first place, and that you do have to do it every single time to be able to get reliable results, which means that insights into safety are much harder to come by, which kinda seems contrary to the NTSB's mission...
wanttaja
En-Route
Yes, I always keep the ZIP files that I download. I'm in Oshkosh right now, away from the computer I do the work on, but will be back this weekend and will put the file on my web page for download.
Ron Wanttaja
Clip4
Touchdown! Greaser!
That is fundamentally incorrect and who ever taught you that is wrong. The pilot doesn’t control G airspace.
Class E to the surface is untowered, but is controlled airspace with the same VFR weather minimums as Class B,C and D.(not 1 sm clear of clouds)
The definition is contained in FAR 1.1.
“Controlled airspace means an airspace of defined dimensions within which air traffic control service is provided to IFR flights and to VFR flights in accordance with the airspace classification.
Note:
Controlled airspace is a generic term that covers Class A, Class B, Class C, Class D, and Class E airspace.”
You might want to read FAR 91.155(c) and (d).
Last edited:
- Joined
- Aug 18, 2017
- Messages
- 2,474
- Display Name
Display name:
Hang 4
Think the terminology (Pilot Controlled) came from the reaction of non-pilots to an "uncontrolled" airport. For all practical purposes, it's also pilot controlled, in the sense that pilots are responsible for separation, following the FAR's etc. It doesn't mean that on pilot has control over another.
Daleandee
Final Approach
- Joined
- Mar 4, 2020
- Messages
- 6,717
- Display Name
Display name:
Dale Andee
Spot on! The unflying public hears a pilot use the term "uncontrolled" and no doubt negative thoughts arise about airplanes and those lawless, reckless, cowboy type pilots.
Pilot controlled simply means that we use all available resources i.e. see & avoid, lighting, radio communications, ADSB, recognized patterns, pattern entry procedures, proper altitudes, windsocks, AWOS, ASOS, yada, yada, yada, to control our aircraft within the airspace. We do not control someone else's airplane or the airport.
dbahn
Pattern Altitude
Thanks an excellent reason for the FAA to use "non-towered", which is a descriptive term that's hard to argue what it means.
How would you have TIME for another hobby?
- Joined
- Jul 3, 2012
- Messages
- 15,453
- Display Name
Display name:
Velocity173
So much easier to proceed straight to the ramp.
flyingcheesehead
Touchdown! Greaser!
Ooooh, that would be awesome! Thank you!
John Collins
En-Route
it would be bad policy if they did so. I don't think the FAA has any inclinations to do so and FAAST teams and SAFE don't speak for the FAA.
John Collins
En-Route
And the vast majority of those occur on final approach or the runway because both aircraft are lining up for the same point and one overtakes another. The aircraft in front can't see the one at their 6, even if both have a landing light on and strobes and the higher speed over taking airplane can be at a different altitude profile and not be able to see the aircraft in front, particularly if it is a low wing. The most dangerous part of the pattern is while on final. On 45 entry, downwind, and base there is lesser convergence of overtaking aircraft because of differences in performance. I would argue that the total safety would be improved if circling was done on the opposite side as the pattern because only one IFR aircraft can be circling at a non towered airport at a time, so upwind is safer, turning base is safer too because if you were to encounter another VFR aircraft also on base leg, you are likely both facing each other and both have a greater chance of seeing one another. Once you turn final, there is no advantage to either method. Flying the pattern on opposite sides of the runway is standard to account for Ultra light and helicopters due to their differences in performance and would be a safety improvement IMHO for one in/one out type of IFR operations where circling is required. One other note is that circling MDA is often half the normal pattern altitude and this makes circling in the VFR pattern direction more problematic as I quote from Appendix A of AC 90-66C "The use of a common altitude at a given airport is the key factor in minimizing the risk of collisions at airports without operating control towers."
The FAA has also updated their guidance in the Instrument ACS to allow the DPE and pilot to use an agreed to MDA at or above the charted MDA to meet the circling task demonstration requirements. Some DPE may insist on using the MDA on the chart rather than in VFR conditions using the pattern altitude, but this reduces safety if there are VFR aircraft in the pattern. Circling in the direction opposite the pattern may be required by some instrument approach procedures, but I question how this would be demonstrated on a VFR day at a non towered airport, particularly if the procedure is a practice approach and not operated under an IFR approach clearance, as what gives the DPE the authority to break the regulations if operating under VFR and not under IFR?
I was just trying to draw attention to the discrepancy between the regs and the FAA's non-regulatory guidance, which may result in pilots in the pattern and pilots on a straight-in both thinking that they have the right-of-way. Personally, when I'm in the pattern, I treat traffic on final as having the right-of-way regardless of how they got there, but when I'm on a straight-in, I don't assume that people will yield the right-of-way to me.
Shepherd
Final Approach
KPOU (Poughkeepsie) tower alternates left and right pattern simultaneously, especially when the local college flight school is filling up the pattern.
They just call everyone's base leg to keep separation.
They just call everyone's base leg to keep separation.