Inadvertently flew VFR into bad IMC today, never again

I

IFRTrainingNow

Guest
Did a flight today that I've done dozens of times before, except that I took a long detour to avoid nasty rainy weather/mist accompanied by lower visibilities. Most of the trip was uneventful, with rain and visibility fluctuating throughout the trip but always at safe levels. As I got under the bravo shelf near my destination things got a little more dicey. Ominous cells of light/moderate precipitation were like a fortress wall guarding me from a direct path back home.

When I was on the home stretch, only ~20 miles from home, I decided to thread the needle between 2 areas of light/moderate precip rather than take the longer route to the south. The gap was maybe 2 miles wide, each cell on my ADS-B radar looked like it was 5 to 8 mile diameter; the gap was also clearly visible to me visually (w/o radar) with rain clearly pouring down on either side. As I was just banked to shoot the gap and began to weave in between the cells I noticed they had gone from green to red on the radar, visually I could see the intensity looked less like a haze and more like an impenetrable wall. However, I felt committed to stay the course confident that I could get through it if I just stayed down the center.

You can probably guess what happened next.
It closed on me. Big time.

In a matter of maybe 1 minute it went from a comfortable 5 miles visibility with light rain to heavy rain with zero visibility. Zero visibility as in I couldn't see anything except for the raindrops on the windshield. At one point I'm fairly confident I couldn't even see my wingtip. I'm a VFR pilot only, and I've never been in conditions like this before, so this was discomforting in the extreme. I didn't panic, but I'm positive my heart rate was triple what it had been the prior minute. I made the call that I'd execute a turn to the south, which is what I'd decided prior to entering that system -- since better weather prevailed to the south rather than 180 degrees behind me.

During just a brief foray into the IMC, I can understand how all the textbook bad decisions are made (including my decision that put me there). Within 1 minute of entering the zero viz I had begun an unintentional slight left bank and begun descending. I didn't catch it until I noticed the DG was sweeping at about 2degrees/sec and then overcorrected (too much bank opposite). I had a little bit of a downdraft that caught me off guard, too. After stabilizing the roll and noticing my altitude was lower I pitched up, but it didn't stop the slow descent. That got me worried and then I focused a little too much on the descent and I started a slight bank again. Luckily after that I was able to gather my wits and balance my attention across instruments. I added power, then did a slow gentle turn to the south, and then once the turn was complete I climbed back to a better altitude. I then focused on nothing but straight and level all way out. About 5 minutes later I was out the other side where the sun was miraculously shining. Never seen such a welcome sight in my F life.

Doing a post-mortem on this, I know what went wrong. Multiple things:
1) Somehow I felt a deceiving and utterly foolish sense of comfort at my proximity to home. There was a sense of "I've done this trip 30x before. I'm almost there, not much bad stuff could happen at this point. I'm literally 10 minutes from landing. In 3 mins I'll even be able to see my airport."
2) The fact that I'd flown through light rain for nearly all of the trip, with visibilities constantly changing between 5-10miles, I'd made the sort of unconscious evaluation that any reduction in visibility would be short lived. Stupid, I know. But I can tell you now that is kind of what I was unconsciously thinking.
3) It's one of the few flights I've done without flight following. Given my location I easily could have contacted approach and asked for support. In retrospect I should have done this as soon as practical. Having a second pair of eyes on me and the traffic around me would have been the right move for both me and others (I only had traffic+wx up on the iPad). I didn't, I was too focused on getting out of the mess. Once I got it into straight and level I wasn't task saturated and it would have been a good time to get on the radio with them.
4) I made excuses to postpone my instrument training too long and it nearly cost me big time.

Things I did right (these do not excuse the things I did wrong, but they did help my situation):
1) For the first 80% of the trip I had taken the safer route and avoided the less predictable areas that were still technically VFR/MVFR, but with more rapidly changing conditions. Despite the substantially longer trip, it pays to take the path that leads to you actually arriving safely. If I had taken the shorter path, it could have been even uglier with fewer "outs".
2) I didn't panic when I hit the IMC and zero visibility. My heart rate spiked for sure, and I quickly knew I was in trouble, but I didn't panic. I focused on instruments only with only an occasional glance out the rain-covered windshield to avoid getting distracted/disoriented. I did make some unforced errors in scanning the panel and flying instruments, but I still did keep my fear suppressed and focused on what was needed to get out of it.
3) Before "threading the needle" between the systems, I had an "out" planned. I didn't have to think a whole hell of a lot about where I should go to get out of this, I was able to realize I was in over my head and put myself on the fastest path out of it. In an alternate universe things may have turned out really badly if I had to fumble around on the iPad and figure out where to go while also trying to fly solely on instruments for the first time.

I will be finishing my instrument rating with urgency and never making that mistake again.

I'll be grateful if my dumbas* mistake and lesson learned here can help anyone else avoid walking into the same trap.
 
Plan continuation bias is real. I've fallen prey to it myself. Bear in mind that even if you were instrument rated, trying to squeeze between two cells like that is still a bad idea.
 
Thanks for sharing.

Learn from yourself obviously, but also others.

Study the latency mentioned above and make sure you understand it.

Also, you might be looking to either side and try to thread between rain-shafts, but have no clue what’s looming directly overhead about to drop on you.

Keep a wide berth from storms.

Get your IR.
 
Really good to hear that the Good Lord was looking out for you! There is a real danger with the time lapse of in cockpit weather. Here's an excellent video from the Air Safety Institute:

 
Good name @IFRTrainingNow. You’ve already kind of begun. It’s obvious you got some very good previous training on inadvertent IMC encounters
 
TCABM said:
ADSB or other data link weather is not a tactical tool
since there can be up to 15 mins latency. From a trend perspective, the question is whether the conditions are improving or deteriorating.
Click to expand...
15 minutes seems optimistic. Maybe there’s a trick I don’t know? Tia
 
Thanks for sharing. The vast majority of VFR pilots don't realize how much EVERYTHING changes when you pop into a cloud. After the ~3hours of instrument reference we do in private training, in clear VMC and an instructor tugging on your shirt, it's easy to take the instruments for granted when your entire windscreen is a gigantic attitude indicator 99.999% of the time. Add the bumps and tumbles of even quite mild turbulence and your brain starts throwing false body signals all over the place.

Even once you finish instrument, 50-odd hours of foggle/hood time is not nearly the same as actual IMC. Getting into real goo isn't as simple as "just look at your instruments instead." I think what struck me first was seeing the outer wisps of the cloud whipping by as my brain tried to figure out depth perception vs the true size of the cloud, suddenly aware I was moving way faster than it felt before. Then everything outside goes gray and you have zero visual references with no foggles to take off. Even more exciting when you have different layers forming crooked lines and false horizons. "Looking at your instruments" and "Trusting your instruments with your life" are two very different things. Especially when the latter means believing your AI saying "straight and level" when your inner ear is screaming that you're in a 45* bank towards the ground.

Then there's the turbulence. There's a reason that airliners will deviate around cloud buildups that aren't just storms. Those friendly, fluffy looking cotton wads can get pretty aggressive once you're inside. Their entire basis of formation is unstable air after all, and it only intensifies once the rain starts falling.

And yeah, as others have said, never use the wx radar(especially ADS-B) as a means to tactically pick your way through cells. It's horribly lagged and only to be used to strategically for "big picture" decisions. I.e. deviate west to go around that line of eastbound storms, or choose to land and wait them out. It's not about knowing what the weather's doing now, but being able to interpret what it's gonna be doing next. And as you've seen firsthand, that changes far too quickly to try and thread needles.

Great debrief. Fly safe, get that IR, and stay well practiced!
 
Problem with in-flight radar is it can led you down dangerous roads. Guilty as charged, by the way. Airborne raadar is old before you get it, you can use it strategically but never tactically. Glad the OP didn't become a statistic.
 
“Did a flight today that I've done dozens of times before, except that I took a long detour to avoid nasty rainy weather/mist accompanied by lower visibilities. Most of the trip was uneventful, with rain and visibility fluctuating throughout the trip but always at safe levels.”

Sounds like your personal VFR weather minimums are too low and it finally caught up to you. You can also get into trouble setting IFR minimums too low.
 
Randomskylane said:
15 minutes seems optimistic. Maybe there’s a trick I don’t know? Tia
Click to expand...

The product itself is updated less frequently than it is transmitted. IIRC, the regional product is updated every five mins but transmitted ever 2.5 mins. That’s latency in production and transmission. Once the product is transmitted from a ground station, your device still has to receive, process, and display it. Most systems will retain an old product until a new product is received.

Particular to ADS-B, you will have to have line of sight to a ADS-B ground tower to receive the product. SiriusXM overcomes that problem at a financial cost, but the products are still latent.
 
steingar said:
Problem with in-flight radar is it can led you down dangerous roads. Guilty as charged, by the way. Airborne raadar is old before you get it, you can use it strategically but never tactically. Glad the OP didn't become a statistic.
Click to expand...
On board RADAR is immediate. But what we get on ADS-B or XM or something that is transmitted to the aircraft from the ground or satellites, that can and will be out of date.
 
drummer4468 said:
..it's easy to take the instruments for granted when your entire windscreen is a gigantic attitude indicator 99.999% of the time.
Click to expand...

This is my favorite part of your post.

Why would anyone want to be an IR pilot after your description? Kidding, of course.
 
^^ I really, really wish everyone understood that.

Unless you’ve got something like a GWX 75 onboard you’re getting a weather product, not real time RADAR.
 
TCABM said:
The product itself is updated less frequently than it is transmitted. IIRC, the regional product is updated every five mins but transmitted ever 2.5 mins. That’s latency in production and transmission. Once the product is transmitted from a ground station, your device still has to receive, process, and display it. Most systems will retain an old product until a new product is received.

Particular to ADS-B, you will have to have line of sight to a ADS-B ground tower to receive the product. SiriusXM overcomes that problem at a financial cost, but the products are still latent.
Click to expand...



Understood, but why is the latency so bad? What are they using to process the data, a 1979 TRS-80? It seems ridiculously slow.
 
Thanks for taking of your time to post a story of your experience, IFRTrainingNow. I can still remember when I realized you have to trust your life to some little dials and needles in front of you. A lesson I have never forgotten. I doubt you will forget now either!

Ignore any negative, critical responses here. We know you have already “learned your lesson!”
 
Half Fast said:
Understood, but why is the latency so bad? What are they using to process the data, a 1979 TRS-80? It seems ridiculously slow.
Click to expand...
For one thing, nexrad composite radar telies on multiple stations and takes several sweeps of any given radar dish to build the full image. It can take 6-10 minutes to fully produce the image. It's updated constantly, but depending on what's happening, the storm you're looking at is building up high while the radar is on a low sweep.

I'm not an engineer or meteorologist but this is my understanding of why there is delay. On onboard radar is only scanning 120(?) degrees ahead of the aircraft on whatever plane the crew selects, whereas the nexrad radar systems are scanning based on an algorithm selected by the system operator.
 
In recent years I've thought about this sort of thing a bit. I think most instruction does not give nearly enough real world experience with this real weather flight decision stuff.... Actually threading that weather needle for real in practice.
It can be done safely in many cases. Folks do it all the time... VFR only weaving through buildups. But where the "limit"?

in my experience, most instructors...even flight review level stuff, won't fly when the weather's like that...not even close...because I think most of them don't have the experience either....
but it seems like it would be helpful to practice a lot....careful to not build false confidence but to help learn what a person can get away with safely, and what it might look like if you don't and end up like ifrtrainingnow did....with the safety net of a CFII helping.
 
Half Fast said:
Interesting. I'd love to see a throughput diagram that depicted all the latencies. It still seems awfully slow.
Click to expand...
There's a lot of data to be processed. It's not the latency, it's the processing.
 
EdFred said:
There's a lot of data to be processed. It's not the latency, it's the processing.
Click to expand...


A processing pipeline is part of latency. I understand there's a lot of data to be processed, but computers have a lot of processing power.
 
Half Fast said:
A processing pipeline is part of latency. I understand there's a lot of data to be processed, but computers have a lot of processing power.
Click to expand...

And then it has to be broadcast and received and there's more latency introduced.
 
Pugs said:
And then it has to be broadcast and received and there's more latency introduced.
Click to expand...


Agreed. So what does the entire pipe, all the way to cockpit display, look like? What’s the delay introduced at each element? I’ve never seen it in print, so I wonder if anyone even knows.
 
Half Fast said:
A processing pipeline is part of latency. I understand there's a lot of data to be processed, but computers have a lot of processing power.
Click to expand...

Maybe we use different definitions of latency. Processing isn't latency in my book.
 
I don't know anything about anything, but if they're using MRMS, then I'd think it would decouple the front half (acquisition/processing) of the pipe from the back half (dissemination) half; all you need to do is hose out the latest model output. Maybe @scottd could say more.

But, to get the thread back on track, to the OP: good job staying cool, now go get the rating.
 
EdFred said:
Maybe we use different definitions of latency. Processing isn't latency in my book.
Click to expand...


In a data processing system it isn’t. In a real-time (or near real-time) system it is. For example, consider an autopilot, a real-time system. Processing is in the pipeline between inertial inputs to the system and outputs to the control surfaces. Processing time has to be considered as part of the latency in control commands. You don’t want the ailerons responding to something that happened two minutes ago.

I suspect that no one who created the ADSB wx system ever thought of it as a real-time, or near real-time, system. It’s been treated as a mere data processing system and latency hasn’t been a design consideration. But, pilots are trying to use it as though it were real-time and that’s dangerous.

If the designers were to consider how wx displays get used they might make minimizing latency a higher priority. Right now I think every just accepts it for what it is.

How much delay is acceptable in an ADSB traffic alert?
 
My guess is that the largest piece of latency is the last hop to your plane. Unless you're in contact with a tower, that could take a bit.

Also you have to consider bandwidth. There's a bunch of data about traffic, and metars, etc that it could get stuck "behind". A snake choking on a cow.
 
Salty said:
My guess is that the largest piece of latency is the last hop to your plane. Unless you're in contact with a tower, that could take a bit.

Also you have to consider bandwidth. There's a bunch of data about traffic, and metars, etc that it could get stuck "behind". A snake choking on a cow.
Click to expand...


Good point. But five or ten minutes still seems awfully slow.
 
Half Fast said:
Agreed. So what does the entire pipe, all the way to cockpit display, look like? What’s the delay introduced at each element? I’ve never seen it in print, so I wonder if anyone even knows.
Click to expand...

Doesn’t really matter if the .gov performance standard says rebroadcast the product once every 2.5mins and refresh the product once every 5 mins. The performance standard only has to be met, not exceeded.

ETA: AC00-63A covers FIS-B broadcast details.

https://www.faa.gov/documentLibrary/media/Advisory_Circular/AC_00-63A_CHG_1.pdf
 
Last edited:
Half Fast said:
Good point. But five or ten minutes still seems awfully slow.
Click to expand...
If this delay is unavoidable, why not make the processing interpolate and project a 'now' image based on all of the recent images, perhaps in a different color scheme to signify its 'guesstimation-ness'. Maybe it has fizzled out or intensified but based on the movement and recent images here is where it would be NOW! IDK?
 
“Processing time has to be considered as part of the latency in control commands. You don’t want the ailerons responding to something that happened two minutes ago.”

In modern flight control systems of the DoDs relaxed static stability airframes, 100 msec, is to much latency and could cause control issues. We usually try to keep the input to output latency to under 10-15 msecs.
 
EdFred said:
Maybe we use different definitions of latency. Processing isn't latency in my book.
Click to expand...

It's not network latency but it's data latency.

I doubt there's an easy way to calculate total end to end latency. Way too many dependencies with any accuracy. Sites with different radar and processing rates and comms infrastructures, receivers of different types in different weather with different settings. Too hard.
 
Last edited:
This story really scares me because I think that once the OP has a instrument rating he would try to shoot the same gap again. ADSB is for big picture planning only. Convective cells can be moving at 30 plus knots. A 2 mile gap between cells is not a correct choice. Many airlines require 20 miles from cells at altitude, 10 miles below 15,000 and 3 miles in the approach phase. I would take those numbers to heart. Cells can spit hail out the sides, lightning anywhere and turbulence that can tear a light airplane apart. STAY AWAY, FAR AWAY!
 
Jeff767 said:
This story really scares me because I think that once the OP has a instrument rating he would try to shoot the same gap again. ADSB is for big picture planning only. Convective cells can be moving at 30 plus knots. A 2 mile gap between cells is not a correct choice. Many airlines require 20 miles from cells at altitude, 10 miles below 15,000 and 3 miles in the approach phase. I would take those numbers to heart. Cells can spit hail out the sides, lightning anywhere and turbulence that can tear a light airplane apart. STAY AWAY, FAR AWAY!
Click to expand...
6 hours of Archie Trammel should be mandatory for an instrument rating.
 
Half Fast said:
Agreed. Why was that the standard, rather than something shorter?
Click to expand...

Real answer? It’s what the prime contractor and the government felt was both achievable and economical for a a certain level of performance reliability, probably something along the lines of 5 9s.

But that’s just to transmit, which is where .gov’s responsibility ends. They/the system can’t guarantee you will receive what’s transmitted.
 
Old Thread: Hello . There have been no replies in this thread for 365 days.
Content in this thread may no longer be relevant.
Perhaps it would be better to start a new thread instead.
Back
Top