Regional and Mesoscale Meteorology Branch

00:00:00:00 – 00:00:10:13
Speaker 1
And, again, thanks for Matt Elliott for agreeing to to, lead today’s visit satellite chat. And without further ado, I’ll turn it over to Matt.

00:00:10:15 – 00:00:40:08
Speaker 2
Thanks, dad. Yeah. I’m glad to be here. Not so in circumstance in terms of what we’re talking about as outages with lightning data. But it looks like things are starting to get fixed, and hopefully things will be, you know, improving over the next few weeks or so. But in the meantime, you know, we thought it was important to at least, you know, documents show the documents, so that as we get the convection season, the early portion of the Leafs know what’s going on and maybe kind of how to handle it, things like that.

00:00:42:01 – 00:01:05:17
Speaker 2
So just to kind of kind of, release the situation. You know, both, the network’s the heartland and LCN data, are both experiencing some, some data outages and they’ve had some it’s been going on for, for, you know, since we’ve been getting the data and, we thought, it was very, you know, the radio did some stuff last time that fall.

00:01:05:17 – 00:01:22:28
Speaker 2
We thought maybe you fixed it. But, you know, it’s tough to tell in the winter season when it’s out. A lot of lightning. And as we’ve gone into the early portion of the spring season here, we’ve it’s been real obvious that these, these issues are still going on. And we do have a lot of, data collection, a lot of different regions.

00:01:23:18 – 00:02:05:04
Speaker 2
The, the members of that, the Lightning Initiative working group, and, all the websites are impacted by this and the all the data that we’re seeing in the and they left it’s all the same that each, you know, the the time the place that you know, it’s basically occurring at each site in the same way at the same problems and the same errors are occurring, and, the heat excites, and just just to note that, network’s data is, experiencing a little bit more of outages and LDN data, on average or so I would say a 24 hour period, there’s generally maybe 24 hour or so,

00:02:05:13 – 00:02:28:15
Speaker 2
maybe 16 to 24, one minute outages and network data and maybe you have 6 to 10 on average and low end data. And that’s with one minute, basically, missing data. And I’ll be talking about showing some things, about showing that, and, you know, this is what we’re here really, is that this could impact, warning decisions.

00:02:29:05 – 00:02:53:27
Speaker 2
So you know, obviously using total lightning now, to lose aid and decisions and, you know, sort of people looking at it and, yes, sudden decrease. You know, they could certainly catch a forecaster off guard and they can start questioning the data. So it’s important to keep in mind. So I mentioned it does affect coming on as they have, 16 .1.2.

00:02:54:04 – 00:03:27:26
Speaker 2
That’s scheduled for a mid-late April for deployment. So it might not fix all the outages and it probably won’t, especially with the AMCON data. It’s having a few more issues. Some other problems. Thankfully, at least the the majority of the problems are the ones that are going to get fixed in this new bill. So there might be a few lingering ones that get fixed over the next few months, and we’re certainly working on that as the lighting working group and with the developers, and got to try to get that fixed.

00:03:29:06 – 00:03:47:20
Speaker 2
And I should add a little bit more, you know, the end the majority of their outages, are actually of the type that they will get fixed in this, in this build. So it’s hopeful that at least at the end of the end, data that the majority of their outages should, should go away for the next few weeks.

00:03:47:22 – 00:04:09:13
Speaker 2
So just going to jump right in here and show you, what a typical kind of outage presents and outages and a nine hour periods of a random, day on March 8th where there was convection over to the, over Texas. And, you could see several outages there. I don’t know, one, two, three, four, maybe 7 or 8 during that nine hour period.

00:04:09:22 – 00:04:38:28
Speaker 2
You can see the outages by that. There’s a vertical white line, where there should have been data. You know, lightning is a very, steady saying. It shouldn’t be a while just where it goes from 400 to 0. It just it just it just doesn’t are very unrealistic. And then what is it happening? Is, you look at the blue arrow, and in the networks data, what ends up happening is the next data frame after that loss of data, is actually a perfect simulation.

00:04:39:29 – 00:05:05:04
Speaker 2
I could talk about that more in a second, but essentially you’re only getting a portion of that next minute. And anyway, that, that that is where you get a zero decrease. And what you see when you visualize an eight. If I show that kind of what you see, you know, and what you see on TV, this is just zooming in on a storm, over Texas, it was a marginally severe storm.

00:05:05:10 – 00:05:25:24
Speaker 2
It’s a 12 panel, essentially each minute of data, you know, just gridded lightning data, one minute by eight kilometer data. And the max value is just. I can’t read the value of the max in the top right. Just shows that where the cursor is sampling the highest value, of, the flashes, per minute for eight kilometers.

00:05:26:18 – 00:05:54:09
Speaker 2
And so what ended up happening here, there were two, two, minutes of missing data in this, 12 minute frame. So there’s a frame here, at 13, ten and then at 13, 17 and, the way Aves handles, missing data is essentially it just skips the frame. So if you’re looking at one minute data, you’re going to see it just skipped from 1309 to 1311.

00:05:54:10 – 00:06:15:13
Speaker 2
You’re going to miss that 1310 frame. So it actually isn’t even display. But when that happening is that 1311 frame, at least we talked about the Earth Network’s data. That becomes a partial accumulation. So essentially you’re going from you know, 18, 17 flashes all of a sudden at 1311, you had this massive decrease down to three.

00:06:15:23 – 00:06:40:07
Speaker 2
And then again at 13, 18, you go from 28 to 1 and that’s a substantial decrease. And to certainly, you know, the forecasts are doing more progress on this, which is quite confusing to the forecaster. And, especially as he’s back up again. If you see a 13, 19 seconds, 14 flashes. So in the forecast, you start thinking that a decrease, you know what’s going on.

00:06:40:07 – 00:07:04:18
Speaker 2
Meteorological. And it’s actually a problem with the data. So move on to the next frame just to kind of show it again, this kind of, this graph form. So it’s a 3720s and there’s no data to either house here in the next frame. If you have the perfect simulation where it’s if, if have a three for of that next minute, I change the threshold to six.

00:07:04:20 – 00:07:26:22
Speaker 2
So it’s to kind of show what’s really going on with that partial accumulation. I looked at some of the Hdf5 files coming out of, you know, what was coming and, and compared them with or networks, raw data. And I grabbed them basically and put them in just in a second bed. So this is at one UTC.

00:07:26:24 – 00:07:50:07
Speaker 2
It’s a two minute period from 34 to 36 UTC. And if we were getting the exact raw data into the Hdf5 files, all those lines should match up perfectly. And, they do. But there’s a portion there where you see they job and that’s where there’s one minute of missing data that basically is it’s just not existent in the CSI files.

00:07:50:18 – 00:08:18:03
Speaker 2
It should be there, but it’s not getting it. For whatever reason. And then what is it happening? Which I still don’t totally understand, but it doesn’t plot that first. You know, it’s always 48 seconds, 48.5 seconds ago, a plot that that that part in red. But what it does plot is that after you get one minute later, there’s about a 12, a half second period where you see in blue there.

00:08:18:05 – 00:08:49:04
Speaker 2
And that’s what the partial simulation, there’s an ellipse. It says 12.5 seconds in the following minute. Where it kind of kicks back in and and able to be able to recover. So that’s what you actually see. And they looked at the blue and you don’t see the red. So it’s important to know that the partial simulations are not occurring with the, you know, the end data threshold where they are essentially, missing one block of data that goes from the top of the minute to the next to the bottom of the minute.

00:08:49:06 – 00:09:18:03
Speaker 2
They don’t have this 48.5 seconds, to 40.5 seconds the next minute, issues that sort of networks is, experiencing. So. Well, what really makes it a problem is that it’s missing it, and it’s actually getting into the the larger accumulation stage that, but this doesn’t really show that here. I just just to show you that, you know, the five minute, the 15 minute and the 16 minute, intervals are all being impacted by the issue.

00:09:18:21 – 00:09:38:13
Speaker 2
The the missing minute will be not included in those, and, it’s much harder to tell, though, that that you’re missing a minute by just looking at those, you know, because I think it’s comparatively, you’re not missing that much. It’s within the one minute data.

00:09:38:15 – 00:09:58:09
Speaker 2
So just a kind of a quick thing. So this is just, like, too fast for the busy meeting, but it’s it’s, you know, it’s essentially this is just a guess loop here. Showing a way to identify which and the best way that we found, you know, working with, with Steve. And so my office here is that, where the temperatures have the lightning peaks, whatever.

00:09:58:12 – 00:10:23:15
Speaker 2
You survey whiskers and essentially just loop through the times. And what you end up seeing is I’m going to stop and then go to the next slide. This is a minute for the outage. Time to go to the minute of the outage. What happened? Okay. It’s the minutes ago. You see here I went to 21, 41 to 21, 43 and you end up getting a not loaded message in the one minute old product.

00:10:23:18 – 00:10:44:27
Speaker 2
And if you look up in the top left where the counts are, it’s just not there. There’s no place to wait because they’re minimum, but they search, for that. So now what is that happening? Is all these accounts become part of the simulator out here? You know, this is all prime, for example, essentially, basically, they’re 12 months of data that’s regular variable.

00:10:44:27 – 00:11:00:18
Speaker 2
You know. Okay. And by 2144, this is the system kind of recovers. But that missing data is still there as well as the partial accumulation. That is also still there as well.

00:11:00:20 – 00:11:11:28
Speaker 2
And as you can see, it moved to the next minute that the, the the data from 2008 values put a quote over that pretty similar I’m talking there.

00:11:12:00 – 00:11:12:27
Speaker 1
I think yeah that’s.

00:11:12:27 – 00:11:14:09
Speaker 2
Come on please mute their phone.

00:11:14:12 – 00:11:19:11
Speaker 1
Yeah. That sounds like State College. Could they move their phone.

00:11:19:13 – 00:11:47:20
Speaker 2
All right. Sorry about that. Essentially, as you see that that’s missing up there. And you can see it’s really you have to tell that you have missing data when you have this sequence like that. And that’s, again, the partial accumulation of his remains in that particular frame. So what makes it really difficult, though, is when you might know, if you look at the if you just pull up the default, lightning sequence starting with made up with this jumbled mess in the top left.

00:11:47:22 – 00:12:11:09
Speaker 2
But you can’t read anything. I have no idea how that’s supposed to be helpful for us. I don’t know if you guys have seen that or been frustrated by that, but, I can have Brian send out a bundles that I’ve created that separate those so you can see the full blast count in the sequence plots. If you’re interested, or you can contact me and I can send them to you as well, either way.

00:12:11:28 – 00:12:30:16
Speaker 2
And I got I also had an end of the end where essentially when you load them off the default menu, you get this jumbled mess in the top left in the flash counts are, in separate those and then you can see your flash counts for each of the the past five minutes and do some comparison things like that if you’re interested in that.

00:12:30:29 – 00:12:50:10
Speaker 2
So again, if you want those you can email me for free hour talk with Brian. But there’s really just a few last things. The best practices with these outages. Yeah, these are all going to be decreased is essential. Essential. But you might have the forecaster looks at it as a decrease. You might see an increase in the next frame.

00:12:50:12 – 00:13:12:21
Speaker 2
So you know, it’s really it’s really good to be that to be wary of unrealistic increases in lightning. Yeah. There’s lightning jumps. But if you have a just an enormous decrease over a short period, especially if you’re looking at the Conus data, where you’re going from 200 flashes or 400 flashes per minute down to some, you know, really unreasonable value, like 20 or a 12.

00:13:13:18 – 00:13:32:17
Speaker 2
That’s something that happens there. And then, you know, if you’re using it for the warning decisions, make sure you’re looking for these outages so that you know, you’re aware of them when they happen. And, you know, you certainly want to keep that in mind. So this is summary to summarize everything as I kind of go through it.

00:13:33:15 – 00:14:00:13
Speaker 2
All the sites are impacted by this issue right now. Works through scraping a little bit more of them. So that at the end is, those networks does have that partial accumulation in the in the minute following that, the outage. And that can certainly impact the, the, the forecasters decision making. And it’s important to keep in mind, all of the accumulation products are impacted generally looking at 15 minute, 15 minutes, they’re all impacted.

00:14:00:16 – 00:14:25:06
Speaker 2
And again, it’s going to impact planning decisions. And we hope that the build helps, significantly, though there could be a few remaining problems. And that’s pretty much it. A lot of stuff. Sorry I threw a lot at you guys, but if you have any questions, let me know now. Or offline. Either way.

00:14:25:08 – 00:14:47:18
Speaker 2
Thank you for the presentation. I did have one question about, one of your slides. And really, I’m just curious if there was a slide you had showing, raw data at the top and then at the bottom, you know, you were matching that with, the data that was in a website with the Hdf5 files.

00:14:48:00 – 00:15:14:24
Speaker 2
My question is, what was the what was your source of the raw data, or did you put or did you find that that’s from the networks themselves. So it’s the same basis. But then I would like to know not what they sent us, but we we were doing a special, data collection to document this. And so we got this as, what the raw data is.

00:15:14:27 – 00:15:35:15
Speaker 2
Okay. Thank you. Well, I, I guess I should mention a few things I didn’t mention. A good thing is, is I did compare the raw data set to the minutes that we were getting data, and and they all lined up very, you know, everything lined up very well. So, you know, when we get the data and the data is good.

00:15:35:17 – 00:15:58:11
Speaker 2
And that’s good to know, obviously. The other thing to keep in mind is, is, is right now the way it was to set up, there’s no keepalive message. So if there is no lightning anywhere within the domain, right. It doesn’t matter if it’s in your Home Depot. Just even within the domain that, that networks are sending their data.

00:15:58:14 – 00:16:22:18
Speaker 2
And same with the end of the end. If there’s no lightning during that particular minute, then no product is written and they wait for, you know, it’s tough to tell if you’re missing data, when it’s not very active across the whole country. But on days like today or, you know, any active day, if you try it out to be very obvious and, you know, you need to go build it now in your system, I guarantee you’ll see them.

00:16:22:18 – 00:16:32:21
Speaker 2
They’re they’re they’re they’re there every day, throughout the day.

00:16:32:23 – 00:16:49:07
Speaker 2
And basically each one would impact the whole domain. So like when you have that missing data, it’s not affecting just your region. It’s affecting the whole, domain that, that, that, that you gave the data.

00:16:49:09 – 00:17:03:20
Speaker 2
And there was a drop out at 1850 today. Yeah. It’s amazing how easy they are to be once you kind of look through that at the site.

00:17:03:22 – 00:17:15:12
Speaker 1
Okay. Any other questions from Matt?

00:17:15:15 – 00:17:25:07
Speaker 1
Okay. Brian, did you have anything to add or, want to say anything about this material being, available later? We can put it on FTP somewhere.

00:17:25:16 – 00:17:57:03
Speaker 2
Yeah. We’re happy to provide, Matt slides. We also have, we have Jason Jordan on the line from Lubbock. I’m here. Okay. Jason also has a presentation that we’ll talk more about the NLP and data. So, Dan, if you can make, Jason the presenter. Okay, that’ll be part two of this, presentation.

00:17:57:05 – 00:18:26:28
Speaker 2
Hey, Brian, while we’re waiting for the switch, can you just update on exactly which build? And, this, correction or fix is going to come in. It’s coming pretty quick, right? Yeah. The the fix has already been developed. It’s in testing. The testing is supposed to be complete by April 14th, and then the patch will go to all the, operational websites.

00:18:26:28 – 00:18:42:20
Speaker 2
The week after that, the fix will go into the baseline, starting with 60 to 1, which becomes available to the field on May 2nd.

00:18:42:22 – 00:18:47:13
Speaker 1
Okay. And I can see your screen, Jordan. So you can, start whenever you’re ready.

00:18:47:15 – 00:19:19:04
Speaker 2
Okay. Similar to what Matthew described, there are issues with the Italian data from Isola. It’s a little bit different, though, in that, like he said, we still are experiencing an occasional issue with the data feed coming into alerts. However, there is something that is, has become apparent since we started looking closely at the lightning data, that was found back on February 23rd, and that was a day that we had, supercells across the Louisiana and the southeast.

00:19:19:06 – 00:19:41:02
Speaker 2
And one storm in particular started to head towards the New Orleans office. And, the, the, the AG there. Yeah, this is the image. It’s, 2248 UTC from the New Orleans radar. And, they did have several tornado warnings out on it. And the storm started to move towards the radar. And then the last scan came in at 22, 51, 39.

00:19:41:08 – 00:20:11:23
Speaker 2
And about that time, one of the employees there at the New Orleans or Slidell office sending a message to the Southern Region Rock NWS chat room saying, hey, I want to let you know that the New Orleans 88 just got struck by lightning. And we’re down. So, since we were watching the storms, trying to see and get more detail on the outages, I said, well, this is going to be a perfect example for me to go in and pull out this case and see, the strike on the lightning data.

00:20:11:25 – 00:20:39:26
Speaker 2
So here’s the start of the hey, let’s plot again, thanks to Matt Elliott. He actually captured it. You’ll see real hard because it’s pretty small, but, up here in the top left, we’ve got, one positive strike. In the end, it’ll be in data. There’s a few, dots, which are the cloud flashes from the Earth Network’s data, and this is at 2248, and we’ll step forward minute by minute.

00:20:39:28 – 00:21:11:12
Speaker 2
So next frame, 2249. We do have one negative CG about, 2 to 4 nautical miles west northwest of the New Orleans radar, 2050. There really wasn’t much here, but we do see another Earth Network’s lightning flash up across the, again, well off to the west. And so since the radar went down at 2251, I was expecting to see a lightning flash right on top or very close to the map location of the Dorian’s radar.

00:21:11:12 – 00:21:45:15
Speaker 2
But there were only two flashes, one about to hit two nautical miles northwest, and another one about two and a half to three nautical miles to the southwest. And knowing since we’ve had some issues with the minute by minute accumulations, I figured it’d be best to go ahead and look at the 20 to 52 to see if it came in, and there was still no it’ll be in flash near the New Orleans radar, but there was one Earth Network’s flash, less than a nautical mile closer to about half nautical mile south of the New Orleans 88.

00:21:45:18 – 00:22:15:13
Speaker 2
And that falls within the error radius that the Earth Network’s detection, has. So that started the hunt to try to figure out what’s, what happened with the flash. And, Texas Tech University has access to the I saw it, you know, the Navy thrilled again for collaborative research purposes. And we have access to that. Not in a WIPs, but we have to go on to their system and pull out the raw data.

00:22:15:16 – 00:22:36:12
Speaker 2
So I did and then loaded up the data that we received in a report from the Hdf5 file. And there were two flashes, one here west of the radar that matched up. You can see the two times the, push pin is from the Texas Tech feed, and the square is from the Awacs feed. Another matching one that was up off to the northwest.

00:22:36:12 – 00:22:58:02
Speaker 2
So those are the two that, showed up in a WIPs, but up here in the top right, just behind and to the left of the legend, there is a push pin that has no, matching across here to it. And then right on top of the 80, there at the bottom center of the screen were several flashes.

00:22:58:24 – 00:23:19:07
Speaker 2
The first one occurred right at 2251, right after they sent the message. And so the feed from Texas Tech showed that the lightning flash was in that data. And, yet it was not showing up in way. So that started another hunt, trying to figure out where this data is going and why it’s not showing up in hours.

00:23:20:12 – 00:23:40:09
Speaker 2
Matt Elliott went back in. And BI solar does have a web portal so that you can look at their lightning data in real time as another method. And he pulled this image off and compared it to a website. So here in red are at the top. Image from the thunderstorm manager. Are all the flashes from, that hit the 8080.

00:23:40:12 – 00:24:04:13
Speaker 2
And of course, it’s not showing up anywhere near the New Orleans radar. And then we also have the three other flashes in the last hour that did show up, in circled in green. And those matched up perfectly. So when we start the process of getting Vaso involved to figure out what’s happened and why, this flash did not show up in a report, and after a day or two, it finally became apparent.

00:24:04:13 – 00:24:29:05
Speaker 2
And they told us that the feed that goes into a. And Brian may have to correct me if I say this wrong, is basically a legacy feed that was created when we first brought in alerts back in the late 1990s. And of course, back then, bandwidth was, much more of an issue. The processing power of the old HP UX workstations was also an issue.

00:24:29:07 – 00:25:01:04
Speaker 2
So my soul was sending basically a deprecated feed where not all the flashes are being transmitted across the median to WIPs. So that now introduces a new problem that you really don’t have a way other than looking post event or for those of us that do have access to alternative feeds from by solar or the Texas to the thunderstorm manager, that there could be lightning flashes that are being detected that are not being transmitted over to a website.

00:25:01:06 – 00:25:35:17
Speaker 2
So, this is something that the working group is aware of. It’s something that we will probably be putting in as a request in the next contract. When it goes out prepared that, all the data that’s available coming in. So if here are in your office and you get a report of a lightning flash striking, the facility, unfortunately, if it strikes a person or equipment and it doesn’t show up in alerts, that could be the reason why is that I saw is not transmitting that data to the National Weather Service.

00:25:36:00 – 00:26:02:17
Speaker 2
Before you ask any questions, we don’t, at least I have not heard as to what the criteria are for them to send flashes over or not. It’s something we’ll need to get some more clarification on. But, I will say that if you do have, report of lightning striking your equipment or, fatalities or anything else that happens, and it’s not showing up in a report, please make note of it.

00:26:02:27 – 00:26:26:08
Speaker 2
The time date, rough location. And that way we can go back in and get in touch with Isola or look into the archives and and see if it’s in there or not. And that’s all I’ve got. I’ve got. Okay. It’s this is Tim Oram at Southern Region headquarters. I guess I have a question because I want to make sure I understand what what you’re saying.

00:26:26:21 – 00:26:57:09
Speaker 2
And having worked with this data, as part of my research from the white collar, and it really goes back to the old NLB network, there’s very clear terminology on things. There’s a return stroke, which is a place where it strikes the ground, and a flash, which can consist of multiple return strokes that are coincident basically in time and roughly in geography, that they grouped together as a single lat long location.

00:26:57:11 – 00:27:24:07
Speaker 2
So are you are and what and what we’ve always in the way I’ve always seen it because we had independent fans at the Space Flight Meteorology Group. Is that what we were get today, which was purely the flash data. So I want to understand I clearly what you’re saying is it is are you saying that they are sending all flashes but not all return strokes or are you saying that they’re even filtering the flashes so that we’re not even seeing everything that they call a flash?

00:27:24:09 – 00:27:53:15
Speaker 2
That. No. Yes, it does make sense. And you are correct. That’s a very important clarification to make the, you know, the in feed that the weather service receives is only the flash data and they do no filtering on that. But there are certain strokes that get recorded. For instance, there were 7 or 8 strokes that struck the 80 and none of those were grouped into a flash to be sent over to the weather service.

00:27:53:17 – 00:28:23:11
Speaker 2
So they have those two different feeds. The research feed is the NLB and stroke feed. And then the weather service gets the end of the in flash feed. And if their grouping algorithms do not classify it as a flash, it will not be sent over to Amos. Okay. Thank you I appreciate that. And so then we’re we’re supposed to be getting every flash and then we have this drop out problem on top of it, which is compounding some of these problems.

00:28:23:13 – 00:29:03:26
Speaker 2
All right. Thank you. Correct. And like Matt said, the dropout problem with it all the data is nowhere near as bad as it is with the Earth networks. And we’ve already diagnosed what the issue with that is. And it’s a problem with the WMO header decoding, which should be fixed in 16 one two. And for instance, the date that we looked at on the eight, we basically did a test where several offices across the weather service dumped all the data from the eighth off, including the log files off of it in the 24 hours worth of the Hdf5 files across the weather service as a whole, everything matched up perfectly.

00:29:03:28 – 00:29:25:21
Speaker 2
And looking through the log files, you know, the Earth Network’s is dropping out sometimes multiple times per hour. You know, 20 or 30 times per day. The it’ll be it is only dropping out maybe once or twice every 2 or 3 hours and maybe 4 or 5 times a day, if that. So overall, the, you know, the infield is a little more reliable.

00:29:25:21 – 00:29:31:18
Speaker 2
We’re seeing less the drop outs in the Earth Network’s data. Data.

00:29:31:20 – 00:30:01:26
Speaker 2
Hey, Jason, this is Steve Zuber. I put you on the spot. Maybe, Matt could help too, but could you just clarify for the NLB and what the differences between the flash data that we get and the stroke data that we don’t get? Yeah. Like Tim said, you know, a lightning flash can consist of multiple strokes, and those happen, usually in pretty short order on the order of milliseconds or even shorter time frames than that.

00:30:01:28 – 00:30:38:02
Speaker 2
And it’ll be in one an algorithm to group all those strokes together into one flash. And based on some exchanges I’ve had with them, the way they send that data over to alerts is that in the flash feed, it’s usually the strongest return stroke based on current. So if you go in and look at the Hdf5 files, whatever the peak current was on any of the return strokes during a flash is what’s going to be reported in the alerts.

00:30:38:14 – 00:31:02:15
Speaker 2
Again, the question is, in this case, you know, the the peak current in the first return stroke was almost 20 -20kg, which is close to average. And yet it still wasn’t classified as a flash to come across in our system. And by solar is working on that to try to figure out why it was not sent over since obviously it took out a radar.

00:31:02:15 – 00:31:27:12
Speaker 2
And that’s pretty important. Flash not to lose, but I haven’t heard anything back since they started taking a look at it as to why it didn’t make it. So. A flash is not necessarily a cloud. The ground strike in the Nhlbi in data, I believe it is according to their classification. We are not receiving the cloud data. Like Earth networks yet.

00:31:27:14 – 00:31:53:05
Speaker 2
So I think when they say it’s a flash, it is a CGA flash. So it was -20kg and and yet and it took out the radar. And yet their algorithm did not classify it as a flash although there were strokes there. Correct. I guess I’m just having trouble seeing, you know what. Why are we seeing this from them?

00:31:53:05 – 00:32:14:11
Speaker 2
And, you know, I guess it involves their algorithm and what they define it as. Correct? Correct. And like I said, I haven’t heard anything back from them since February 24th, the day after this happened. As to why that could be. So, I’ll just need to hit them up and say, hey, do you have any more information as to why I didn’t make it?

00:32:14:13 – 00:32:46:08
Speaker 2
Yeah. And clearly, you know, if if, say, you were out west and you were monitoring and looking for you know, fire starts from lightning, you would have never seen this, this flash, this, you know, and it clearly hit the ground, or had, well, hit power, but. Well, I think this term I think what we really need to ask, and hopefully they have it in their data is a listing of all return strokes that went into the flash.

00:32:46:08 – 00:33:10:22
Speaker 2
Or did they confirm to you that, yeah, they saw those return strokes, but they were never grouped into any flash. Which I would be surprised by. Yeah. Just went back to the frame that has at the very bottom has all the return strokes that were recorded at the location of the 8080. And so there were seven return strokes that took place.

00:33:10:22 – 00:33:38:06
Speaker 2
So they did record, all seven of those return strokes. It just for some reason was not classified as a flash. And again, that’s something on their sorting algorithm and, how they pair all that, but for whatever reason did not get shift into the eight speed to come to the weather service. But it did show up in the, Texas Tech feed, which is the LDN stroke feed.

00:33:38:08 – 00:34:03:17
Speaker 2
So the data did go out. And just for our purposes in the weather service, it did not come across right. And at least on eight weeks, one and I don’t know if it’s the same case in the net in the Hdf5. Yeah. But in the netCDF files for the, you know, the flash data, there used to be a field called multiplicity, which is actually the number of return strokes associated with that flash.

00:34:03:17 – 00:34:25:21
Speaker 2
I don’t know if that’s available on the HDV, but it is the dump that if it is, I mean, I mean, if you can dump that is like text, you can at least have some indication that there’s more than one, you know, ground strike locations aka return strokes associated with that flash to at least give you a heads up that there’s, you know, that there’s some more points that may have been impacted.

00:34:25:24 – 00:34:46:17
Speaker 2
Yes. And that data is available in the Hdf5 files. We’ve got a, a script here that passes that out for the Gibson Ridge plate increase. It creates an orange place file. And and that’s one of the ways again that I found this flash is that, we also create the place files at Texas Tech. And I look at it and I could see that there was a difference.

00:34:46:22 – 00:35:04:16
Speaker 2
And, off the top of my head, I do know that there have been some flashes recorded across two that have come out with 24 and 25 and five, return strokes. And, so we are getting that information. There’s just other than parsing it out, there’s no way to actually display it in cave or the day to day perspective.

00:35:04:16 – 00:35:21:01
Speaker 2
You actually have to create that file or look at it after the event. So just a follow up question for Steve here is so on this multiplicity, that’s the number of return strokes. So in this case what what if there’s 7 or 8 at this location.

00:35:21:03 – 00:35:42:20
Speaker 2
Yes. There were seven recorded at that is on top basically for the eight. Yes. So you know a no no. Not looking at that parameter obviously is seven something that’s like a maybe a kind of a an average number or that, you know, we never see that. But I just don’t know. I mean, you know, maybe we see multiplicity of 2 or 3 and it calls it a flash.

00:35:42:20 – 00:36:06:00
Speaker 2
So it’s probably a question that you can’t answer is probably there algorithm. But you know, why do you suspect that their algorithm says, well, if I only see three co-located return strokes or less, I’m not going to call it a flash. Yeah, no clue on that. I’ll have to ask myself one, right? Yeah. And you know, I’m a member of the of the National.

00:36:06:00 – 00:36:27:12
Speaker 2
So like, in working, lightning implementation working group, you know, and going and and pure raw, you know, they’re they’re on it. That’s on it. You know, it sounds like something that we need to kind of push up and work through this group that you see, because it sounds like a legacy system. As Matt pointed out, that we’re just seeing it.

00:36:27:12 – 00:36:53:08
Speaker 2
We’re getting stuff that was set up back in the 90s and because of bandwidth, and we could probably, like you were saying, probably better at it. Yeah. And I just pulled up the email that one of the gentlemen at Vassar sent me about the investigation into this particular case. And it’s interesting because says because he says that the National Weather Service only takes the Nhlbi and slash historical slash bandwidth legacy.

00:36:53:11 – 00:37:23:06
Speaker 2
So I’m wondering if the only reason that they haven’t changed the feed for us is because this is the way it’s been since they were started back in the late 90s, and we’ve been looking at this data for almost 20 years now. And all of a sudden, we may change it to the latest and greatest feed and see, and it could be a pretty significant uptick in, cloud to ground activity just because we’re now getting all these other, bits of information.

00:37:23:06 – 00:37:44:24
Speaker 2
And so that’s definitely something that we need to bring up in the working group. And, and see what my soul is going to, how they’re going to handle that and that. Yeah. I mean, it obviously sounds like it involves contractual type issues. And so it’s, you know, it’s something that is way beyond us.

00:37:44:27 – 00:37:57:00
Speaker 2
But something we have to push for, you know, to, to be investigated first and find out what’s available. And, you know, make the right decision.

00:37:57:02 – 00:38:27:25
Speaker 2
Yeah. Jason, let me, this is Brian. Let me just confirm what, Steve and and you have said is true. And there’s the main, total lightning processor that I saw has has a rich data set which includes total lightning, similar to what Earth networks provide. But what they do is they reformat that data into a legacy format on a different server.

00:38:27:25 – 00:39:12:03
Speaker 2
And then transmit that to a web. So we’re getting sort of a reformatted, which may be less rich, if you will, than, what their native solution data set has. So that’s the difference between what, you know, the two things that Jason Jason showed are able to produce. So, it’s sounds like it’s a fixable problem, but, we’ve had several events where the data just didn’t get sent to where the service from on.

00:39:12:05 – 00:39:36:25
Speaker 2
Right. And that’s, you know, that’s kind of a separate issue, I think. But, that’s clearly when you look at the lat line that he’s got, he’s got displayed here. I mean the, you know, there’s, there’s probably a couple of tens of meters of, of distance, you know, between some of these return strokes and and I don’t know how the algorithm works.

00:39:36:25 – 00:39:54:19
Speaker 2
I would just kind of think that that might be within the realm of these are all basically near or at the same ground point, within measurement error.

00:39:54:21 – 00:40:20:12
Speaker 2
I think it’s reasonable for us to say that there should have been like a plus or minus minus sign, you know, at this location, I guess a minus sign, actually, in this case, right? Yes, I agree, I totally agree because, you know a head on not been monitoring NWS chat and not heard that the A.D. got struck. This would have gone by unnoticed unless right.

00:40:20:13 – 00:40:50:02
Speaker 2
New Orleans called up the the NLL, the NFI to see. Hey, let’s see if we can see it in here. So, this is only one case. I don’t know of any others that, have been investigated or looked at yet. And then also there’s other issues with is it hit a tower or an 88 is an actual tree cloud to ground, because then we start getting into all the details about waveforms and how they classify that.

00:40:51:00 – 00:41:13:04
Speaker 2
And the only reason I bring that up is that, that becomes a little more problematic during the winter season when you start to get thundersnow. Because going back and looking at a case we had back here in Lubbock in 2014, there were some flashes that were detected in the, the lighting mapping array that never showed up in, you know, the end data.

00:41:13:04 – 00:41:48:03
Speaker 2
So was that because it was never classified as a cloud to ground strike, or was it just never reported? And fortunately, we didn’t have the Texas Tech feed. And I can’t go back and look. But you know, it’s something that we need to be concerned about because if it happens and I hope it doesn’t, but if we get a lightning fatality and it does not show up associated with anything else, that instantly brings into question, at least for a forecaster, perhaps that wasn’t really a lightning strike, because I’m not seeing anything in my display.

00:41:48:06 – 00:42:14:03
Speaker 2
Yeah, and we get calls, couple times a year from, fire investigators at our CWA who know that we have lightning data, and they ask us to look at, at a particular time and location to see if there was some type of lightning activity that could have caused the fire. And they’re doing some type of, you know, fire investigation, you know, possibility of arson.

00:42:14:03 – 00:42:52:08
Speaker 2
And they want they want to try to rule out, you know, was with lightning and and most of the times I’ll just pull up the NLB and data and go to the time when they back and tell them, you know, if there were any flashes, you know, plus or minus signs near the location, they give me. So that kind of now brings that into question, you know, not only from the drop out standpoint, which is a kind of a, you know, a possibility, but just from the fact that there could have been a flash there and it just wasn’t, you know, plotted today because we didn’t get it in it.

00:42:52:10 – 00:43:01:14
Speaker 2
But there actually might have been lightning activity there, obviously, just like there was with the radar here.

00:43:01:16 – 00:43:13:28
Speaker 2
I, I would just like to see all of the data basically.

00:43:14:01 – 00:43:27:19
Speaker 2
Okay. Thanks, Jordan. Jason, do we have any other questions?

00:43:27:21 – 00:43:35:29
Speaker 2
Going once, going twice. Hey, Brian is not a question, but can you and Tony hang on for a second? Sure.