Search the RAMMB website
00:00:00:18 – 00:00:07:14
Speaker 1
wildfire. That is, I mean, I’m going. So without further ado, I’ll go ahead and turn it over to Brad.
00:00:07:16 – 00:00:37:00
Speaker 2
Okay. Thank you. Yeah. So I, I’m going to talk a little bit about some work we’re doing to develop capabilities for doing high resolution, trajectory forecasts of where, general wildfire smoke can go. And we’re using this Fort McMurray fire as a chance to validate some of our trajectory forecasts. And in particular, here, we’ll be using beer’s aerosol optical depth observations to initialize the forecast.
00:00:37:00 – 00:01:01:20
Speaker 2
But what I thought I’d do for this visit, satellite chat, is also add a little bit of information about what, just in general, what satellites are telling us about this, particular wildfire, because it’s quite a significant one and fairly early in the season. So this is just a little background about the fire. Started a little over two weeks ago on May 1st.
00:01:01:23 – 00:01:31:05
Speaker 2
And at this point, it was a small fire to the south west of, Fort McMurray. If I may, third, the fire is intensified and air quality in the vicinity, and we can tell, becoming an issue, really flared up by May 4th, at which point they needed to do mandatory evacuations of the area because of other risks to property and lives.
00:01:31:07 – 00:01:59:29
Speaker 2
If I may, if they actually restricted airspace in that area, I don’t know whether that was primarily because of smoke and, interference or because of the, active site firefighting aircraft in the area. The fires continued as recently as May 16th. That evacuation zone was extended to the north, and now it’s affecting some of the, the oil, and, and tar sands camps to the north of that area.
00:01:59:29 – 00:02:18:22
Speaker 2
So it’s a significant fire that’s had a significant impact. I’ve got some loops in here which I’m not sure, played too well in web meetings. So I’m going to start with a still image. I’ve got a link down below where you can find the link to if you want to, but this is on May 3rd.
00:02:18:25 – 00:02:51:04
Speaker 2
This is goes 15 imagery on the top. We’ve got the visible channel on the bottom. We’ve got a 3.9 micron infrared channel which is used to detect the fire hotspots. And, Fort McMurray, you can see it indicated, by the met ups. And this was, when the first pyro cumulus cumulonimbus was observed. You can see this, top of the still image, and it’s just downwind from the fire detection, which is the red spot, to the north, in this case to the northwest of Fort McMurray.
00:02:51:07 – 00:03:28:07
Speaker 2
So now this, should show a little showing, the, explosive pyro cumulonimbus that’s forming, associated with that fire. I don’t I’m not sure whether that goes very well, but we’ll move on. So now this is turning now to May 16th. So two days ago, and the fires are continuing. You can see in the visible image on the left hand side at 1945, the, extensive smoke, columns, going up both to the northwest and to the southeast of Fort McMurray.
00:03:28:09 – 00:04:09:17
Speaker 2
And again on the right, you can see using the infrared 3.9 micron band, you can see the hot spots that are associated with active fire detection. Again, this is a loop for that day. Hopefully you can see that. But again this is a very significant wildfire. Definitely affecting people in the vicinity of Fort McMurray significantly. This is an image now from, this is going to toggle between the, the visible imagery and again, the shortwave infrared from the Sumi NPP VRS sensor, which is a higher resolution polar orbiting sensor, again, for the 16th of May, same time period we looked at before.
00:04:09:17 – 00:04:36:02
Speaker 2
And now you can see, with severe fire detection, you can see the fire perimeter, much more sharply defined with, the high resolution retrieval, of those bright spots. Okay. So now moving on to, the main focus of this presentation is we’re developing capabilities to do high resolution, trajectory forecasts of where the smoke will go.
00:04:36:04 – 00:05:13:27
Speaker 2
The goal here is to provide, low latency, web based forecasts to support National Weather Service. I met, for unsafe decision support during fire incident team. So we’re targeting now very local smoke forecasts, to to support the high network. Again we’re initializing with high resolution peers, aerosol optical depth retrieval. And looking now at high resolution smoke forecast, this is an extension of, an open source, set of software that we’ve developed here at firms called IPA international.
00:05:13:29 – 00:05:40:27
Speaker 2
And we’re doing this testing within the community satellite processing package program, KSP, which is here at SEM. And we will then transcend transition that stop rotation that as the star where it will be part of the current, operational IPA forecast system, because it’s being developed within KSP, it will be available as an open source software, package.
00:05:40:27 – 00:06:07:06
Speaker 2
And so, individual forecast office, if they chose to download it and run it on their own, local Linux machines tailored to their particular, region of interest. So I’m going to go back now to May 4th. This figure, the main the main figure is from the Nasdaq’s, hazard mapping system. And this is a hand in that.
00:06:07:06 – 00:06:34:09
Speaker 2
It’s fire detections in are the red dots. And then the gray shading is hand analysis of where smoke is. The 4th of May. I circled, over Alberta where the Fort McMurray wildfire is located. So that’s the zoom that we’re seeing earlier. And you can see on May 4th, the and analysts are showing this smoke transporting down well into the Midwest.
00:06:34:11 – 00:07:00:25
Speaker 2
In the upper right. This is now the various, aerosol particle depths retrieval. And you can see that it’s retrieving higher aerosol optical depth, which are the colored, the shading, that are just to the, to the north of, of, North Dakota. And these are indicative of high aerosol loading, based on the air slam to go down.
00:07:00:27 – 00:07:26:15
Speaker 2
Let’s just assume now of that same virus, ADR. And we’re going to use these arrows, optical depth retrievals to initialize trajectories to see where, those air parcels are likely to go. So this is now the high resolution trajectory forecast. So we’re initializing trajectories at each one of the first pixels of year retrieval of at seven at six kilometers.
00:07:26:17 – 00:07:59:21
Speaker 2
And for this particular experiment, we’re only looking for trajectories where the aerosol optical depth was greater than 0.5, which is fairly heavy smoke. The upper to run these trajectories, we’re using, the three kilometer nam conus nest meteorology. And so we’re able to resolve the transport of each individual, aerosol optical depth observation. The top panel is just a planner view, showing the the movement of these high aerosol optical depth trajectories.
00:07:59:21 – 00:08:32:14
Speaker 2
And they’re colored by their initial aerosol optical depth. And I’m contour the Nam 600 millibar heights. To give you a sense of the circulation. The lower panels, show cross-sections and longitude and latitude. And what you can see in particular from the latitude cross-section is that this is the forecast. This is suggesting that there’ll be a transport of this fire plume down over Minnesota and over Wisconsin, over the next 24 hour period.
00:08:32:16 – 00:09:00:08
Speaker 2
But if you look in the cross-section, we’re forecasting that this smoke plume will be staying aloft as it’s moving over, Minnesota and Wisconsin. So this slide now is, a freeze, at 23 V as this plume is, arriving over Madison, Wisconsin. This is Madison. We’ve got a, surface instrument, a, sun photometer.
00:09:00:08 – 00:09:22:23
Speaker 2
That’s part of the aero net, observation network. And it measures very accurately an aerosol optical death from the surface. And this is just a time series for, for May 5th, showing what the observations over Madison for aerosol optical, that’s where. And you can see that at 23 V on the fifth, we see a large increase in aerosol optical depth.
00:09:22:25 – 00:09:57:29
Speaker 2
That’s quist with the forecast, the transport of the plume over Wisconsin. And again we’re forecasting that this plume will be aloft and therefore not likely to impact surface air quality. The, the Wisconsin DNR maintains a ground, measurement system that monitors surface, particles, litter, aerosol concentrations. And this is a figure showing for April 26th through May 11th, what, observations are, which are in black and May 23rd in Madison, Wisconsin.
00:09:57:29 – 00:10:23:06
Speaker 2
At the surface, there was no enhancement of, aerosols associated with the passage of this plume. So we consider this a good verification of that particular smoke forecast. The red lines here are from a global, air quality forecast model that that we run here at Madison called Rackham. So now moving on to May 6th.
00:10:23:09 – 00:10:55:13
Speaker 2
Again, the smoke is still very evident, north of the Canadian border. The hand analysis shows the smoke now and the and the observed there is aerosol scaled up to show the smoke reaching as far as the southern tip of Florida. So very significant, kind of continental scale impacts from this fire. Again, zooming in on the veers, optical depth retrieval, you can see this, significant layer of smoke, north of the North Dakota border.
00:10:55:13 – 00:11:26:14
Speaker 2
And in fact, a lot of the smoke is so dense that the air aloft goes up to retrieval, mistakes it for a cloud. And so you can see in the center of this plume, it’s, gray smoke that’s being, not an aerosol because that’s approvals not being done because it’s too thick. So, again, initializing trajectories at the full resolution and virus retrieval, shows that, the trajectory forecast has this moving over much of the Midwest, over a very broad extent.
00:11:26:16 – 00:12:04:05
Speaker 2
And in particular, now it shows that in the latitude, versus pressure cross-section of this smoke, this time will be near the surface and is quite likely to be impacting, surface air quality. This is a slide, report that I got and the Edmund pointed, pointed out and this is from the chat has some Minnesota, National Weather Service forecast office put out a special weather statement at 1:27 a.m. on Saturday, May 7th, saying that smoky conditions will be persisting through the overnight hours.
00:12:04:07 – 00:12:34:03
Speaker 2
And it’s associated with both fires from this Fort McMurray fire as well as some, some more local fires in Minnesota. This is what the satellite imagery looks like on May 7th. And you can see this form of high aerosol optical depth associated the smoke plume is actually, underneath, pretty significant cloud deck. So, the satellites can actually see the aerosol loading over Iowa.
00:12:34:06 – 00:13:10:09
Speaker 2
And southern part of Minnesota. But, on this day. But but the trajectory forecasts are able to provide some insight as to where that smoke may go. May go. This is from a site, that’s been developed by NASA. That’s an enhanced version of the idea website. And the link is down there below. And this is one that we’re rolling out under the, the JCS Fire and Smoke initiative and one that will be specific used for, some of the, the fire and smoke blended products to support the I met forecasters.
00:13:10:11 – 00:13:47:07
Speaker 2
This is the verification from the U.S.. EPA. So this is a national network of surface monitors. And this is a map showing the analyzed, air quality index for particulate matter showing, very unhealthy and, unhealthy for sensitive groups to unhealthy over essentially the full state of Iowa and southern Minnesota. And these are again time series now, some individual surface monitors, and the upper side, we’ve got Apple Valley and Rochester, Minnesota on the bottom panels.
00:13:47:07 – 00:14:12:03
Speaker 2
We’ve got Des Moines and Iowa City and Iowa again, the black line are the observations. The red line is our global model forecast. And what you can see on 12 is on May 7th. You’ve got this very, very significant peak in the observed, surface concentrations. In fact, it reached as high as 240, micrograms per meter cubed, which is well above the threshold.
00:14:12:03 – 00:14:42:02
Speaker 2
That’s, unhealthy, for people to breathe. And thus led to this, extensive smoke, alert that was put out by the National Weather Service. So that’s kind of, the work that I wanted to present today. Again, this is a work in progress. We’re using the current fire season. That’s just beginning to get underway early as a way of testing these high resolution forecasts.
00:14:42:04 – 00:15:11:08
Speaker 2
If you’re interested in looking at, some additional real time, analyzes associated with these fires, you can first of all, there’s the enhanced idea website, which focuses on various airsoft particles, depth and fire detection, and is being developed again under the Fire and Smoke initiative at NASA. Some of the images for example, the image that shows here is from, the 16th.
00:15:11:11 – 00:15:39:26
Speaker 2
This is from FCC is real Earth satellite visualization. I’ve provided link there, allows you to see these, and zoom in on specific regions. A lot of the data is also there. During the last month or so, there’s been this goes 14, sr SLR, one minute, imagery that’s been collected, including a number of, periods when they were collecting imagery for this Fort McMurray fire.
00:15:39:28 – 00:16:03:13
Speaker 2
And then also there’s, satellite, blogs that’s associated specifically with pyro cumulonimbus generated from these fires. And I’ll point you to that one that’s run by Scott Brockmire here at FCC. And so if you’re interested in looking at more details about this Fort McMurray fire, you can look to those websites and, and I’ll take any questions that people have.
00:16:03:13 – 00:16:21:22
Speaker 2
Now.
00:16:21:25 – 00:16:31:18
Speaker 1
Okay. Any questions or comments for Brad?
00:16:31:20 – 00:16:49:18
Speaker 1
Okay. If not, we will have another, short presentation here from, Curtis Seaman, who is here at, Sarah in Fort Collins. And without further ado, let’s turn it over to Curtis.
00:16:49:21 – 00:17:05:26
Speaker 3
All right. I’m going to share some images from viewers of the, Fort McMurray fire and talk about some of the the aspects of, fire detection using years.
00:17:05:29 – 00:17:40:22
Speaker 3
And so, I’ll, I’ll show some images of the fire and, discuss some of the, the issues in some of the, the various bands with, with saturation and fold over. And, I’ll also talk a bit about the fire temperature RGB composite, which is, composite that we developed here at Cirrus specifically for, for visualizing fires and, satellite imagery.
00:17:40:24 – 00:18:15:16
Speaker 3
So here’s a, chart showing all the, the different, viirs bands and, of course, the, the, the medium wave or mid wave air bands and the 3.7, to four micron range are the best for fire detection. They are capable of detecting nearly all fires. But, the 2.25 micron band, or M11 is, is capable of detecting moderately intense or active fires.
00:18:15:18 – 00:18:45:07
Speaker 3
And if you have really intense fires, they can show up in, the 1.6 micron bands, either M10 or I3, and it and some of the, the largest and hottest fires will also show up in the the long wave I r he’s a b m15 and i5. And of course the the day night band can detect the visible light emissions from the fires at night.
00:18:45:09 – 00:19:20:05
Speaker 3
So on this slide we’re comparing the various images from the M13 or four micron band on the left with the high resolution medium wave urban I4 on the right. And of course, there’s a resolution differences between these two bands. But there there are also differences in the the way that they detect fires. So on the on the left you can see the the center of the fire.
00:19:20:05 – 00:19:47:20
Speaker 3
This these are images taken from the the afternoon of May 3rd when the when the fire really became intense. And you can see that the, the fire is actually saturating the color table. So I’m coloring any pixels that are warmer than 340 Kelvin. But if you look at the, the center of that, that hot spot in a high for you don’t see any yellow pixel pixels.
00:19:47:22 – 00:20:27:12
Speaker 3
These are actually white or light gray. And these pixels have a recorded brightness temperature around 200 Kelvin. And so it doesn’t make much sense that that at the inside of this intense fire you’d have really cold brightness temperatures. But what’s going on is this, this fire is so intense that the, the warmest pixels observed by M13 are around 487 Kelvin, and this is a good 120 Kelvin warmer than the saturation point on the I for band.
00:20:27:14 – 00:21:02:24
Speaker 3
And when you have that much radiation hitting the detectors, it can actually trick the hardware into, reporting a colder brightness temperature than is actually observed. And so M13 doesn’t have that because it’s a dual gain band designed specifically to not saturate in fires. And so it’s it’s saturation point is is higher than 600 Kelvin.
00:21:02:27 – 00:21:09:28
Speaker 3
So one of the one of the, imagery products we’ve developed here at Sierra is called the the fire temperature.
00:21:10:03 – 00:21:20:06
Speaker 1
Can I can I interrupt? Yeah. So can you, use the fold over to actually added in and find out what the temperature is?
00:21:20:09 – 00:21:27:17
Speaker 3
I’m not sure if that’s if that’s possible.
00:21:27:19 – 00:22:02:29
Speaker 3
So that the fire temperature RGB composite combines information from the, the shortwave and the the midway of either bands. So the the red component of the image is the M12 or the 3.7 micron band. The green component is M11 or the 2.25 micron band, and the blue component is M10 or the 1.6 micron band. And so how this works is that M12 will detect almost all fires.
00:22:02:29 – 00:22:26:24
Speaker 3
And so these these pixels will show up red. But if it’s a moderately intense fire, it will be detected in both M12 and M11, and it’ll appear orange to yellow depending on how intense it is. The the more intense the fire, the more yellow it will appear. And if you have a really intense fire, it will show up in all three bands.
00:22:26:27 – 00:22:56:22
Speaker 3
And, it’ll be detected as a as a white pixel. It may even appear bluish if this fold over is present in M12. And so the the coloration of fires in the image is very similar to how flames actually appear to the human eye, that cooler flames are red and hotter flames become more yellow or white. And of course, the hottest flames are blue.
00:22:56:24 – 00:23:29:10
Speaker 3
So if we advance to the next slide, here’s a an animation of, that afternoon viewers images for the for the first five days where the the fire was really intense. So the period from May 3rd to May 7th and on the, the lower right you have that that high resolution AI for band. And on the lower left you have the M13 band, which does not saturate.
00:23:29:12 – 00:24:01:10
Speaker 3
In the upper right you have the the fire temperature RGB composite. And then the upper left is the true color RGB composite, which can can tell you where the the smoke plumes are and where the pyro cumulus clouds are. And, so one of the things to to note from this animation is just how quickly this, this fire spread off to the, to the east and southeast over this time period.
00:24:01:13 – 00:24:24:00
Speaker 3
And you can also see these little white pixels that that show up occasionally. And I for where this the saturation in and fold over is occurring. And these are actually the, the pixels that are most likely to, to show up as white in the the fire temperature RGB composite.
00:24:24:02 – 00:24:49:06
Speaker 3
And so after these, this first five day period, it, it ended up getting pretty cloudy over Fort McMurray. So you can’t actually see all that much with the. But if we just compare the last image of the the previous animation from May 7th and we advance ahead to, about a week later on May 15th, I’ll toggle back and forth.
00:24:49:06 – 00:25:12:23
Speaker 3
You can see that the that the fire did continue to, to spread during that time. And, you have a lot of little smaller hotspots around the, the edge of the, the burn scar. And if we advance one more day, this is two days ago, you can see that the that the fire really ramped up in activity.
00:25:12:26 – 00:25:39:00
Speaker 3
And again, you have such duration and fold over in the I band and, it’s saturating the the color table on M13, but not actually saturating the instrument. And of course, you know, the, the upper right shows you the, the fire temperature RGB composite for the same time. And then, finally, here is yesterday’s afternoon images of the fire.
00:25:39:00 – 00:26:11:27
Speaker 3
And it has kind of calmed down a little bit from the the 16th to the 17th. And so if you want to learn more about the kinds of issues that I have discussed briefly here, I have written a number of blog posts discussing, fire detection with viewers and, on two different blogs. One is focused mostly on high latitude applications of yours, but then another blog is focused on on verse imagery around the globe.
00:26:12:01 – 00:26:37:00
Speaker 3
And so I have a blog post where I, I discuss the fire temperature argb and, discuss these saturation and fold over issues which are not unique to viewers. They’re present on other satellites. And also some other examples of, wildfires in Canada. And so with that, I’ll, I’ll take any questions.
00:26:37:02 – 00:26:58:20
Speaker 2
Hey, this is Bill at SPC. I have a question. Can you, let us know the main advantage of making the fire RGB? Why not just look at M13, which isn’t a prime part of the spectrum and has the higher, saturation threshold? Why combine the others? To create the RGB?
00:26:58:23 – 00:27:22:14
Speaker 3
It’s it’s just another way to to visualize the fires. So you can you can certainly use just M13 alone is a very, very good way to detect fires. That the fire temperature RGB can also tell you some things about the, the clouds. You know, that that liquid clouds will appear blue and and ice clouds will appear green.
00:27:22:14 – 00:27:34:09
Speaker 3
So there there are other other uses for it as well. Thanks.
00:27:34:11 – 00:27:39:19
Speaker 1
Any other questions for Curtis and Curtis?
00:27:39:19 – 00:28:00:23
Speaker 2
This is Brad as one of the things that we’re doing in the Fire and Smoke initiative is adding the VA’s, fire rated power detection. So you might actually take a look at that. That might be something that would be valuable to overlay as well, since that does quantify the intensity of the fire.
00:28:00:26 – 00:28:08:04
Speaker 3
Correct? Yeah, yeah. We’ve been thinking about, doing a similar retrieval and, and adding that. Yeah.
00:28:08:04 – 00:28:17:17
Speaker 2
So of that retrieval is available, but you could just add that as an overlay to any one of those imagery products.
00:28:17:20 – 00:28:24:04
Speaker 3
Yeah. Yeah, that’d definitely be a good thing to do.
00:28:24:07 – 00:28:33:23
Speaker 1
Okay. Any questions for Curtis or Brad.
00:28:33:26 – 00:28:42:24
Speaker 1
Okay. If not, we want to thank both, Brad and Curtis for, presenting today. And thank you for attending today’s visit, satellite chat, and have a great day.