Where are all these crashes coming from?

Post by **dima** » Wed Jan 02, 2019 9:43 pm

So let's say you're walking around in the mountains, and BAM! You run into a crashed airplane (pretend this happens all the time). You wonder "what is this?" "When did it go down?" "How?"

It turns out that most (all?) such incidents are investigated by the NTSB. And they then write a report. And then somebody puts all the reports into a database and makes them available on the internet. How thoughtful! For our purposes, we mostly care about the location each impact site. Some reports actually have a lat/lon pair, while others have a range/bearing from some "observing" airport. Many reports don't have either, but there isn't a lot I can do about that.

In any case, I just did some typing, and produced this:

https://caltopo.com/m/N13E

This is a map of all the incidents after 1982 that have a location in the report. The lat/lon pairs are represented by a red circle, while range/bearings are shown as a red line (the ranges are accurate to a nautical mile, so the line shows a +- 0.5 nm of error). If you click on the line/circle you get a URL that gives you the NTSB report. The circles/lines aren't very accurate, it looks like, but the text of the reports often has more specific information. I have all the data in the US, but that link is just the general area of the San Gabriel mountains; the web interface is way too slow to show more. I can easily generate any other data, so ask me if you want something else.

Happy hunting!

Post by **dima** » Wed Jan 02, 2019 9:58 pm

I forgot to mention that the data appears to not be particularly accurate, and probably is useful only in narrowing down the possible incidents if one comes upon something; the coordinates themselves aren't sufficiently precise to actually find stuff. And I'll do a writeup at some point in the near future.

Post by **dima** » Wed Jan 02, 2019 10:20 pm

Oof. Two more things:

I just found a bug in my stuff, so the latlon coords were plotted in the wrong place. Fixed in this new map:

https://caltopo.com/m/N13E

And I found a caltopo bug: the circles aren't showing the ntsb report URLs. Not my bug, so not a ton I can do about it. The file being displayed in that link (that has all the URLs) is this

stonehillnews · Post by **stonehillnews** » Wed Jan 02, 2019 10:38 pm

Neat, is it possible for you to update your map objects without having to create a brand new map? Curious because it seems there's a missed opportunity to use the NTSB Accident Number (e.g. LAX02LA180) as a label for your objects. I followed the url you added to the description of the one that happens to be nearest Mt Bliss, and it opened an accident that seems to have been on the freeway in Alhambra, so it's kind of confusing. https://app.ntsb.gov/pdfgenerator/Repor ... L&IType=CA

What's even more confusing is that report contains coordinates 34.083333, -118.333333 of the wreckage, but that seems to be sort of Hollywood, not Alhambra where the crash is given as a named location, nor Monrovia, which is where the bearing line is on the map. I'm not sure if the data is mixed up, or if I am, but an accident number could help me figure that out.

stonehillnews · Post by **stonehillnews** » Wed Jan 02, 2019 10:41 pm

Ah, just saw your comment about the bug after I posted, but from what I think I see, my observations are still the case in the newer version. Also, small usability improvement would be to make the urls a hyperlink in the description (if caltopo supports html in there).

stonehillnews · Post by **stonehillnews** » Wed Jan 02, 2019 10:42 pm

If you think you have a real caltopo bug, I've found the developer to be extremely responsive. I made a feature request on his forum, and he implemented it almost right away.

Post by **dima** » Wed Jan 02, 2019 10:48 pm

stonehillnews wrote: ↑ If you think you have a real caltopo bug, I've found the developer to be extremely responsive. I made a feature request on his forum, and he implemented it almost right away.

I haven't tried to report it yet. Want to do it for me?

Bug:

1. Create a marker with some comment
2. Export to gpx
3. Start a new caltopo session
4. Import the gpx you just made
5. Observe that your comment isn't there anymore

stonehillnews · Post by **stonehillnews** » Wed Jan 02, 2019 10:53 pm

dima wrote:I haven't tried to report it yet. Want to do it for me?

Lolz, there's a bug in my todo list, it's not accepting proxy bugs at the moment. ?

Post by **dima** » Wed Jan 02, 2019 11:01 pm

stonehillnews wrote: ↑ Neat, is it possible for you to update your map objects without having to create a brand new map?

Not in any way I could figure out. It'd be nice.

stonehillnews wrote: ↑ Curious because it seems there's a missed opportunity to use the NTSB Accident Number (e.g. LAX02LA180) as a label for your objects.

I could do that, but I'm unconvinced it'd be helpful. It WOULD make the already glacial interface ever slower, though. Making the urls into actual clickable links would be really nice., as you sy That maybe is possible, actually. I haven't tried, though.

stonehillnews wrote: ↑ I followed the url you added to the description of the one that happens to be nearest Mt Bliss, and it opened an accident that seems to have been on the freeway in Alhambra, so it's kind of confusing. https://app.ntsb.gov/pdfgenerator/Repor ... L&IType=CA

What's even more confusing is that report contains coordinates 34.083333, -118.333333 of the wreckage, but that seems to be sort of Hollywood, not Alhambra where the crash is given as a named location, nor Monrovia, which is where the bearing line is on the map. I'm not sure if the data is mixed up, or if I am, but an accident number could help me figure that out.

That's just what the data says, unfortunately (unless I'm not interpreting it correctly). This particular report has actual latlon coords in Hollywood. You see a nice red circle at those coords. This report ALSO lists an observation facility (EMT: El monte airport), which is reported to be 7 nautical miles away at a bearing of 210 degrees. And unless I messed something up, that is the line near Mt Bliss, which is clearly bogus. I looked at some other reports whose location I know of, and the way I'm interpreting this information is correct for those. If you can figure out how I'm messing this up to make this line up better, that'd be great

Post by **dima** » Wed Jan 02, 2019 11:11 pm

This is pretty interesting though: I'm finding out about events I've never heard of. For instance: https://app.ntsb.gov/pdfgenerator/Repor ... L&IType=CA

stonehillnews · Post by **stonehillnews** » Thu Jan 03, 2019 8:15 am

Accident number labels would be helpful for two reasons. First, you could differentiate the objects in the list on the left (right now they just show NA), and second, you could correlate the bearing lines and the lat/long markers that go together, as in the random case I was looking at.

Post by **dima** » Thu Jan 03, 2019 10:10 am

stonehillnews wrote: ↑ Accident number labels would be helpful for two reasons. First, you could differentiate the objects in the list on the left (right now they just show NA), and second, you could correlate the bearing lines and the lat/long markers that go together, as in the random case I was looking at.

Ah. I agree that some sort of label would be good. Just meant that the accident number specifically is cryptic and unhelpful. Let me update it to use the year and vehicle type. And maybe color-code on year, or something. Give me a few days.

headsizeburrito · Post by **headsizeburrito** » Thu Jan 03, 2019 4:32 pm

Wow, airports are dangerous. Planes should avoid those in the future to reduce crashes... ?

Nice work! I wonder if there is anything interesting in my part of the country...

Post by **dima** » Thu Jan 03, 2019 4:44 pm

Send me a lat/lon pair of where you would like to find stuff, and I'll look that up

Post by **dima** » Sun Jan 06, 2019 11:39 pm

Alright, I cleaned this up, and updated the map. And I uploaded all the code and data. The previous caveats still all apply, unfortunately: this is useful, but not accurate enough to find stuff all by itself.

And I just learned that in 2016 a Boeing 777 cleared Mt. Wilson by 1000 ft, not counting the towers. Some people live dangerously.

Girl Hiker · Post by **Girl Hiker** » Thu Jan 10, 2019 7:55 am

Wow Dima, looks like you have a lot of time on your hands

stonehillnews · Post by **stonehillnews** » Sat Jan 12, 2019 9:50 am

Girl Hiker wrote:Wow Dima, looks like you have a lot of time on your hands

Nah, he’s just a whizkid ?

Post by **dima** » Sat Jan 12, 2019 5:23 pm

To be fair, the computer and the NTSB investigators did all the actual work. I just facilitated

Post by **dima** » Tue Feb 12, 2019 9:29 pm

OK, I worked on this some more, and it's now done. "Done" meaning "I'm not working on it anymore". Final product: http://ntsb.secretsauce.net/map/

It's similar to before, but doesn't use caltopo, so is MUCH faster. And you can move the map around, and get reports from anywhere in the US, not just in our local mountains. Enjoy. And don't break my server