NASA ARSET Data Analysis and Visualization

0:03 Speaker 1: Welcome back to our r SET training series Advanced NASA Earth Observations and Tools for Active fire, smoke and post fire monitoring. Today is part two of our training series. My name is Brock Blevins, a training coordinator for the r SET program, and I'll be your host for today's training. As I mentioned in Part two, or the final part of this training series, we will have an overview of static thermal anomaly data inside firms, how to use static thermal anomaly information to identify thermal anomalies that are not likely vegetation fires, how to visualize active fire information on static graphs and interactive maps, and how to analyze active fire information using a fire count histogram.

0:56 Speaker 1: Once again, the homework is available now on the training web page. It will be due December third, and a certificate of completion will be awarded to those who attend both five sessions and complete the homework by December third. And with that let's begin Part two of the training series, Data Analysis and Visualization. I'm very happy to introduce our guest instructors for Part two. Brad Quail joins us from the US Forest Service where he is the lead for the Disturbance Assessment and Services program, and we also have Dylan Mendez, a senior application developer for Firms.

1:40 Speaker 2: Part objectives.

1:42 Speaker 1: By the end of part two, we hope that participants will be able to recognize how static thermal anomalies associated with industrial and natural sources are identified, use data available in firms to identify locations of routinely detected static thermae WILL anomalies, and use Jupiter notebooks to access and visualized firms data. Once again, if you have any questions during today's training, you can put them in the Q and A box within WebEx at any time. At the end of this training, we'll address as many questions as we can, and then we'll post a transcript of these questions and written and answers to the training web page within a week.

2:27 Speaker 2: Let's get right to it.

2:28 Speaker 1: First demonstration by Brad will be on static thermal anomalies.

2:36 Speaker 2: In particular. His outline will.

2:38 Speaker 1: Show an overview of static thermal anomaly data available in firms, accessing STA data, associated data and tools in firms, and how to use firms to highlight examples of thermal anomaly activity they are not likely vegetation.

2:54 Speaker 2: Fire over to you, Brad. Thank you.

3:03 Speaker 3: All right, thank you. Well.

3:04 Speaker 4: Today, what I'm going to demonstrate is our new static thermal nominally data that we've incorporated firms, along with some associated data sets and tools we've included to use those data. And what you see here as a firm's map of Oregon and western Idaho. And I'm going to zoom in here to northeast Oregon to a little town called the Dirky Organ And as you can see, if I zoom into this location here, just outside that town, there's a large excavated area with an associated area of industrial infrastructure here and this particular area, if I turn on the fire detections, I can see I'm getting some detection activity associated with that infrastructure next to the excavation pit.

3:50 Speaker 4: Well, this area here is a cement plant. And this is a really good example of a static thermal anomaly feature because sment plants have large kilns that they use to heat up material to make cement. And these kilns burn at very high temperatures or have very high temperatures approaching fifteen hundred degrees centigrade. And these features can be detected by satellites and included an active fire detection data, and that's why we can see them here. If I back off a little bit here on the scale, you can see those features and here's some other detected activity here, likely associated with a fire.

4:29 Speaker 4: But if I jump ahead one month to the end of July of twenty twenty four, again looking back thirty one days in the active fire detection record, I have all these other fire detections that are associated with wildland fire activity that was occurring at that time. And as you can see as those detection data fill in, I have a lot of detections for the month of July for that cement plant we just looked at before. Now, if you're a wildland fire manager, you may that may give you pauseing you'd wonder if you're not familiar with the area that might be associated that detection activity might be associated with the wild land fire.

5:10 Speaker 4: So having these static thermal anomaly features inventory and identified allows users to identify these areas of fire detections that might be associated with industrial features or natural features that are giving off heat or thermal activity and can be detected in active fire detection products. So static thermal anomalies what are they In the context of firms. They're basically semi persistent or persistent sources of thermal energy that we can detect routinely in active fire detection products.

5:48 Speaker 4: But they're not vegetation fires. They're associated with industrial features or industrial activity, and in some cases they can be associated with natural features that they're given up heat, such as volcanos and other sources of geneo thermal activity. And what we did in developing this static thermal anomaly feature or layer that is in firms right now is we started with year twenty twenty three cumulative active fire detection data that was acquired by SMPP bears. It is about twenty one million active fire detections worldwide, and then we also included terra and aqua active fire detections for the same year, which was an additional eight million.

6:31 Speaker 4: So we had about twenty nine million active fire detection data that we processed and analyzed and thresholded accordingly, and we summarized those data and we identified features that we determined were static emitters of thermal energy, and we identified well over sixty three hundred of them and they're represented here on this map. So the next thing I'm going to show is we're going to jump to a map in firms for Western Turkey, and what if you want to access the static Thermal Anomalies layer mask, it's under this Static Thermal Anomalies group in firms and you can see I have it drawn up and you can they're very small features at this scale, and they're they're showing up in magenta.

7:23 Speaker 4: But if I change my background layer to a dark gray, they stand out a lot better. You can see them here throughout Turkey and a decrease and whatnot.

7:33 Speaker 3: We also have.

7:36 Speaker 4: Sources of industrial heat or industrial activity identified. Now we didn't develop this these layers in firms, but we went to authoritative sources who inventory and compile this data globally, and we've provided these data and firms for reference. So we have all types of different industrial plants and also power plants which may be emitters of heat. And so if I turn on a couple of these layers here you can see these iron and steel plants, cement plants and so forth. And if I go to the layer info for one of those layers, you get a brief description of what this layer enteils, and you also have a citation, a link to get to the source data if you'd like, and also a citation for the data layer.

8:31 Speaker 4: If you have any questions about these layers, we recommend that you contact the authoritative sources for these data because these data, again are not developed by NASA. They come from these various sources and we're just provide them here as a courtesy in firms for folks to have contexts and reference to these data. So I'm going to switch back to the image base and I'm going to zoom in here on this town in Turkey called Heroes the Turkey, and I've turned on the static thermal anomalies mask and magenta there and you can see that this area corresponds to an industrial area there on the coast.

9:13 Speaker 4: And if I turn on the fire detections for this day, I'm looking at June twenty sixth, at twenty twenty five, and I'm just I have a one day reach back, so we're just looking for the detections that occurred on this day. You can see that there is a fire that's burning in this forested area to the south of town, and we're also picking up thermal activities associated with the static thermal anomaly area. Well, what we've done in firms is provided the capability to filter and identify active fire detections that are associated with the static thermal anomalies layers or features, I should say.

9:49 Speaker 4: And under the static thermalomalies group, there's a static thermal anomalies detection layer, and if I turn that on in Cyan, these detections that are I that correspond or that are associated with the static thermal anomalies feature will be highlighted. And basically what it's doing is features that directly intersect the static thermal anomaly area as well as or within a pre defined buffer tolerance are selected, so we can see that these detections are associated with the industrial activity here.

10:22 Speaker 4: And we're picking up a little bit here that some detections that are associated with that fire, but that's due to the buffer tolerance we're using there. We try not to get too aggressive so we don't pick up detections too far outside of the static thermal anomaly feature. If I jump ahead one day, this fire started the day before, you can see again for that day on the twenty fifth, here's all they have fire activity associated with that forest fire south of the city, and these are the active fire detections associated with that static thermal anomaly feature.

10:56 Speaker 4: Just want to note here you noticed on the previous day, and today we have some detections in this area. This is obviously of industrial origin, these active fire detections, but we our static thermal anomaly feature does not extend out there and capture that. And that's because again for this initial prototype layer we have, we did not We only used that you know, year twenty twenty three active fire detections and so we just didn't have any detected activity in that area. That meant our thresholds and criteria to identify that as a static thermal anomaly.

11:32 Speaker 4: So just you know, for little things that are nuances like that for users to be aware of when they're using these static thermal anomaly features. For the next part of my demo, I'm going to jump to the Hues Fire, which was a fire that burned in southern California in January twenty twenty five. And what I'm looking at here is on the date of January twenty eighth, reaching back for one week, all the cumulative active fire detections for modus and vehers that occurred with this fire. And as you can see here, we have a little cluster of detections here a couple kilometers away from the fire that's associated with this excavated barren area.

12:18 Speaker 4: And what this area is here is it's a landfill area. And I zoom in here on it a little bit and turn off the detections. You can see that area and what it looks like up close there. But I'm gonna zoom back out and turn the detections back on. And if I turn on my static thermal anomalies layer, but that render there is a little bit slow displaying and you can see those magenta That magenta polygon is that our static thermal anomalies mass that was derived from the twenty twenty three active fire detection data.

13:00 Speaker 4: So what is going on here is upon doing a little bit of research in this area, this is a landfill that in twenty twenty two basically an underground fire started there. So the garbage and the waste that's been buried there in that landfill was ignited and the fire has been burning underground and smoldering and whatnot for the last three years. And this thermal activities routinely detected by the active fire detection sources in firms. So having this information would you know, possibly help a fire manager to know that these fire detections that are close to the you know, the Hues fire when it was going on.

13:44 Speaker 4: They are not associated with that incident. But as you can see as I jump through time here and maybe if I jump into February, well after the Hues fire was over, we're still picking up these detections three times. So just to show you that that you know, relatively small, subtle smoldering thermal activity that's happening there with that landfill fire is being picked up and detected. And of course I can turn on my static thermal anomalies detections and isolate those detections with that activity if I'm interested in doing so.

14:25 Speaker 4: On this last example of working with static thermal anomaly data, we're going to focus on a natural heat source. And where I'm at right now with this map is this map. It shows the East African Rift Valley and you can see here we have a couple of volcanoes that are present and where we have some active fire detections that are occurring with those volcanoes. And this area here in general is the Democratic Republic of Congo. And to identify where these volcanoes are at, if you look under the overla, we have a volcano layer, and you can filter these volcanoes for what's displayed based on their most recent activity.

15:10 Speaker 4: We have here by default on those volcanoes have been active in the last sixty years are displayed, but if I toggle it to display all of them, you can see we have all these other dormant volcanoes that have not also erupted in the past sixty years. So I'm going to toggle that back to just the ones that have been active in the last sixty years, and I'm going to zoom in on this volcano here. It's called Nayamalazra, and this is the most active volcano in Africa. And as you can see on this particular date for July thirtieth, with a one day reach back for the active fire detection data, we have a lot of active fire detections that are associated with the volcanic activity that's occurring on this volcano that day.

15:54 Speaker 4: And if I draw the static thermal anomalies layer and turn off my active fire detections from moment, you can see the static thermal anomaly feature that we discerned using the twenty twenty three active fire detection data is pretty pretty large. It extends about three to four kilometers from the main crater of the volcano, and if I turn my detections back on from Modus and Beers for this day, both the day and night detections, and if I turn on my static thermal anomalies detections, I can easily filter and identify those active ore detections that are associated with the static thermal anomaly feature.

16:32 Speaker 4: But I'm going to turn that out for a moment, and also the static thermal anomaly's mask, and I'm going to turn off all of the active fire detection layers except for the Noah twenty one beers data for that day, and I'm also going to turn off the night time detection, so we're just looking at the daytime detections associated with the volcanic activity. And if I turn off the volcano layer here so you can see it a little bit easier, you can see these are the detection is right here on the volcano, on the volcano itself that we got some activity there, but we also have some activity here off the west slope as well as trending down the east slope of the volcano.

17:13 Speaker 4: So if I look at the Noah twenty one beer's false color composite imagery on this day and turn off my active fire detections.

17:23 Speaker 3: For a moment.

17:25 Speaker 4: Again, this is relatively coarse resolution imagery at this scale, but you can see with the shortwave in for red response, we have some thermal activity going on here in faint that you can see some thermal activity out here far on the west, on the west slope as well. I'm going to turn my beer's active fire detection back on and also turn off my Beer's imagery, and I'm going to take a look at the HLS imagery and it happens on this particular day, July thirtieth, both lands At and Sentinel to imagery were acquired for the area.

17:58 Speaker 4: And if I drop up the true composite imagery, this is what it looks like on that day, the Sentinel imagery. And now I'm going to draw the false color composite imagery for the Sentinel to acquisition. Here it comes in for patients here. As it draws in again, it repeating me some of what we learned with the working with the HLS data before. The true color composite imagery draws in very fast, but the false color composite imagery is rendered a little bit slower than the true color composite imagery.

18:38 Speaker 4: And if we take a look close look at this imagery.

18:44 Speaker 4: If we take a close look at this imagery in the context of the active fire detection data from BEERS, you can see with the if I turn off the active fire detection data, I have all these areas that are easily visible in the short wave in forread data that are resulting from the thermal activity, the lava and the eruption activity that's going on around the part of the volcano here, and it's coming down the east slope as well as all this activity that's easily seen in coming down the west slope. So these are lava flows coming down from the main crater of the volcano, and so that's what we're seeing when we see those detections farther out to the west and also creeping down the east slope.

19:28 Speaker 4: And this is a good example of how you can use the HLS imagery, which is much higher spatial resolution than the Veers and Motus imagery to discern and determine maybe what is going on there on the ground and is the source of the thermal.

19:45 Speaker 3: Detected activity that you can see there.

19:48 Speaker 4: This wraps up my demonstration here today with the stat of thermalomaly data, and I just wanted to say again that this current stat of thermal anomaly layer that we have in firms is a prototype layer. It needs more work and improvements and the firm's team is working on doing that and we plan to release an updated static thermal anomalies layer in the future.

20:12 Speaker 3: Thank you.

20:19 Speaker 1: Thank you very much Brad for that thorough demonstration of static thermal anomaly data and firms and its functionalities.

20:27 Speaker 2: It was great to get a tour of how to use it.

20:32 Speaker 1: In this next demonstration, Dylan will be showing us how to generate a firm's map key, how to visualize active fire information on a static graph and interactive map, and how to analyze that fire data. If we would like to replicate the demonstration provided by Dylan, please see the link to the GitHub which can be found under part two of our training web page, as well as a link and how to generate a map key. We'll make a record of this demonstration available by tomorrow so you can go through the process at your own pace.

21:11 Speaker 5: Amazing well, thank you for the more technical part of this training. Here is where your start for the coding portion, and this is a gub repository. There should have been more materials provided if you'd like to check out how it works and all, but it's pretty simple. First you all need to go to the firm's website here get an API key. When you come over here, you'll just click get map key, put in your information and click get map key. You'll then get an email where the body looks something like this.

21:51 Speaker 5: This is just a demonstration key for the webinars to what we do. But you'll take this key right up here at the top and that will be your MAP key that you will need to use the notebook that is linked down here. To get to the notebook, all you have to do is hit open an colab here. This will take you over to Google co lab and then you'll be able to make a copy and then do some work from there. So I'll take my.

22:21 Speaker 3: API key.

22:22 Speaker 2: There's some.

22:24 Speaker 5: API overview information and different information on our sources and all the different satellites that are available. But the code starts down here in the setup and authentication part. And then you will take your API key and where this big green bar is you put it in between these two quotes. Once you've done that this will enable. You'll click run and then this will ask you if you're connect if you'd like to connect to Google. They can't from GitHub, so you can click run anyway. After that is done, it will connect to some server provided by Google and it will import all of these packages that are used in this training and you'll see setup complete.

23:17 Speaker 5: API key is loaded. Then we can come down here to the world example. This function I wrote to query the firm's API for the world with all of the information on the different sources, day range, daystring, all of the things like that. Pick play just to load this function. Once you load this function, you can pass the details into it, and we're looking at the world here for the date of August twentieth, in twenty twenty five for the VIAS NOAH twenty nearal time.

23:55 Speaker 3: Data sise.

24:02 Speaker 5: It'll print out exactly what a FI link you used here. You can click on that and then it'll bring you into a new tab with the raw data itself. But this will load the data into a data frame from PANDACE and you can see an example of the first couple of lines here. The information included below there is all of these columns and this little backslash is showing me that it's continued on the next line. So these are the rest of the columns, and now I'll move on to showing you a bounding box example. Only to run this to load the function so that it can be called in the next block.

24:51 Speaker 5: This bounding box is in the Amazon basin and I prepared it ahead of time just so that everyone could get a quick example.

25:01 Speaker 5: Once you run this, it will show you the data Bougust twenty six, twenty twenty five for the data inside the bounding box, and then it will also print out an example of the data frame here, which is similar to the one above for the world. In this case we're using the Veers SNPP satellite in the new real time.

25:30 Speaker 5: These are our available sources below with information on their real time status standard status where they cover in the case of Landsat. And that just shows you how to get the data in a bounding box and a function to use that to pass the bounding box and to ingest the data into a geo data frame from the firmas with lat long coordinates. Will run this function that turns it into a geo data frame and we can run the data on the data frame area from the world example, which is up here labeled data frame area.

26:22 Speaker 5: This will take in that data frame area and then turn it into a geo data frame as an example shown here with the points. You can do an exercise here to see how do some continue to do some coding and see the amount of the number of points or active fire detections in this example with the confidence above eighty, you can easily filter on the confidence which does not look like there's one on the Beer's s PP satellite. But manipulating the function up here or a different source or calling it in. When you call the function itself, we'll be able to reveal some different confidence values that are not available for all of our satellites for various reasons.

27:21 Speaker 5: So that we can view some of these satellite detections, We're going to use the geodata frame and then a function in that so that we can show these active fire detections in the world for the same date of August twentieth plus one day, so there's two days total. Once we run this, it will show the exact thing loot that we were showing earlier, showing the world and you can kind of make out Africa over here and North America into South America over here, and.

28:00 Speaker 2: Europe and Asia over here.

28:04 Speaker 3: This will.

28:06 Speaker 5: This is a static plot, so you can't zoom in or click into it. However, we can use the folio map to show the same data on a interactive map so that we are able to zoom in and such and see different areas. In this case, it's colored by confidence.

28:32 Speaker 5: It's taking a little bit to run.

28:34 Speaker 3: Right now.

28:44 Speaker 5: Now that it's been executing, we can scroll in the bottom box below the code and see the different parts of the map. We can zoom in or out send that backwards and take a look at the different fire detections based on confidence, and we can go to part of Brazil over here and see that it is colored in a way.

29:08 Speaker 3: That we wanted it to be.

29:15 Speaker 2: All right, We.

29:20 Speaker 5: Will now count fires by the day, so, using the same data frame earlier on the acquisition date variable, we will convert into daytime and then create some grouped by date time or group by the acquisition date, and then the size of those acquisition dates, so it will be the number of fire detections by the acquisition date. And then we can use PLT to plot the figure. Right now, it's only showing one day go up and show let's go change this to two days. We will now do some manipulation based on the date, and you can do this for any of the different columns that are available in our output, or additional data that you have brought in to do some subsetting or some charts on.

30:35 Speaker 5: Right now, we're going to take the data frame the acquisition date. We're going to turn it into a daytime objects that I can be grouped by the date, and the size in this case is the number of fire counts. I just showed this earlier, but now you can see that I went from one day to two days worth the data of August twentieth and August twenty first, showing that there were more fires on or more active fire detections on August twenty first in the world and there was on August twenty As a conclusion, we set up authentication for the firm's API, retrieve data using the firms API and bounding box queries, and for the world.

31:22 Speaker 5: I ingested the data into PANDAS and GEOPANDAE, visualized the fire detections with the static and interactive maps, and then analyze some temporal patterns. That is just some basic manipulation of our firms active fire detections using our API. You can expand this to historical trends over time, overlaying fire detections with different information that you may have about various things that have to do with lands such as land cover or population data sets. And you can also developed automated monitoring pipelines from things like this.

32:02 Speaker 5: Below is the homework section which is where you'll be able to do all the homework here below. And yeah, that's it for me. Thank you all for watching.

32:21 Speaker 1: Great, Thank you very much, Dylan, And as I mentioned, we're going to make this recording available by tomorrow so you can go through this on your own time. But let's summarize what we learned here today. First, Brad showed us how to access and use stas and firms, which included the display of STA layers, a display of the STA feature source types such as industrial heat sources or natural heat sources, and identify active fire detections associated with STS. And we also saw how to access and just and visualize first day using the super or notebook.

33:01 Speaker 1: And just to summarize this training series, this two part training demonstrating two participants several new capabilities. Within firms. In Part one, we looked at burned area mode in firms new veers, burned area product and the use of HLS imagery and indices to perform burned area assessments. This is useful as monitoring post fire impacts of vegetation is essential for evaluating burned severity, which informs post fire de brief flow risk, understanding patterns of ecosystem recovery, and identifying areas required for restoration intervention.

33:38 Speaker 1: We also looked at smoke and aerosol mode in firms, the use of aerosol index and PYROCB layers to identify and track wildfire smoke and as we learned while fare, smoke monitoring is critical for air quality and public health applications as exposure to find particular matter within smoke can cause respiratory issues and other health complications. And then in part two today we looked at static thermal anomaly information to identify satellite active fire detections that may be associated with industrial courses and then also how to access ingest, visualize, and manipulate firms data using a Jupiter notebook.

34:20 Speaker 2: Homework is available now on the training webpage.

34:22 Speaker 1: It'll be due on the third of December, and for those who did book live webinars and completed the homework by the deadline, you will receive a certificate of completion the email approximately two months after the completion.

34:37 Speaker 2: Of this course.

34:40 Speaker 1: Here we've included the contact information for myself and today's speakers, as well as links to our set website and YouTube channel. I would like to take another opportunity to acknowledge our fantastic guest presenters Jenny Houston, Brad Quail, Dyln Mendez, Diane Davies, as well as as and Radoff and Upmar Belsina from the firm's team. Thank you very much for joining us on this training series. And now we'll move over to the question and answer session. We've been monitoring the Q and A box for questions and we've been moving them over to this document.

35:28 Speaker 1: If you have future questions, please locate the three little dots on the lower right hand side of the application or web browser.

35:38 Speaker 2: For this WebEx and you'll see a Q and A section.

35:41 Speaker 1: Please answer your questions there, but the ones that we have we've tried to start answering, so we can start off with question number one.

35:52 Speaker 2: What are the data we rely on for fire detections. Is it only thermal.

35:56 Speaker 1: Data or smoke or is there any other data that helps fire false positives like wind congestion or accuracy metrics. I think I got the gist of what this question was asking.

36:13 Speaker 3: Hey brod this Brad. I can take that question.

36:15 Speaker 2: Oh, hey Brad, great dear, Yeah.

36:18 Speaker 3: Yeah, great question. Yeah.

36:20 Speaker 4: The way the active fire detection works it's it's reliable or it's reliant on thermal data or thermal detection of activity, and it can be affected by a heavy cloud cover, heavy smoke cover, as is noted in response here. And the way it works is, uh, you know, when it identifies obvious uh, you know, thermal activity, you know, above average temperatures there and a big given pixel, what it does is it slides into what we call a contextual analysis part of the algorithm, and it looks at neighboring or adjacent pixels to assess their thermal response in order to figure out what that pixel would look like in the absence of fire.

37:01 Speaker 4: And if we have a large enough departure, there's statistically from that thermal response based on the response of its neighbors which are not also obvious fires, the algorithm can then determine with a certain level of confidence whether or not that's an active fire detection or not. And the second part of your question, you know, as far as using information that can help you identify if there's you know, potential false positives on a detection, you can use the confidence information that I mentioned, Like on the Modus active fire detections, there's a confidence a level that ranges between zero and one hundred percent, and with the veer's data, it's a categorical rating of low, moderate, or high confidence and that those confidence levels are affected by you know, where the detection is, like if it's adjacent to a water body or a cloud edge or whatnot, which may cause some confusion in the thermal response and determining the.

38:03 Speaker 3: Statistics that I mentioned before.

38:05 Speaker 4: So you could look at those confidence levels for an active fire detection and maybe you know, consider excluding some of those very low confidence level detections which may be actual false positives.

38:21 Speaker 1: Great, thank you, and I put the link that you see right here to the firm facts in the chest. Question number two, how about aerosol layer this is it possible to estimate spatial distribution? Have we ever use other satellites like Tempo or paste to integrate into the results.

38:42 Speaker 2: I brock, this is Jenny, and I can take that one.

38:46 Speaker 6: So Firms currently makes available aerosol index layers from AMPS which is about smpp as well as NOWHA twenty and this was covered last week by Diane Webinar one. This particular, this set of aerosols index layers are useful for both both the detection as well as the tracking of aerosols and aerosols from from smoke from fires as well as interestingly dust storms. And I also wanted to highlight the NASA's overarching near real time system, which is called LAMPS and within which the Firm system is situated, continually investigates additional data sources that could be useful for a range of applications, including wildfires for example, and has recently made available multiple near real time products from Essay's Triponi sensor, including an aerosol index product, as well as selected neural products from PACE.

39:58 Speaker 6: And we are currently working to make available selected near real time tempo questions which I know that the question has specifically asked about, so I will provide the links for those as well for the PACE on.

40:19 Speaker 2: Thank you, Thank you very much.

40:22 Speaker 1: Yeah, so it sounds like you are already on top of a lot of this other data being integrated.

40:29 Speaker 2: Question three, what is the definition of data ingestion?

40:32 Speaker 1: And I think this was asding Dylan's demonstration near the end.

40:39 Speaker 2: Yeah, and the answer here just a means of acquiring the data.

40:45 Speaker 3: Yeah.

40:46 Speaker 1: If there's a follow up question from that participant, please please let us know if there's more that you wanted to know. For another question, is it possible that firms captures temperature anomalies caused by other sources rather than fire? I saw areas on the map, I know, and suffer from urban heat islands by high levels of infermeable surfaces such as asphalt or buildings.

41:14 Speaker 4: Yeah, Brock, Brock here, I'm happy to take that. Yeah, that that's another really good question.

41:20 Speaker 3: Yeah.

41:20 Speaker 4: The fire detect algorithms are used for you know, modus and bears and you know the other sensors are available uh in in in firms. Uh. They're they're designed to detect and measure uh thermal radiation that you know, are at very elevated or high uh temperatures that are associated with you know, combustion and fire and whatnot. So uh, these types of you know, thermal anomalies that you're describing here with urban heat islands and whatnot. They're they're there's there's not designed to uh, you know, detect activity at that particular temperature and whatnot.

41:56 Speaker 4: In some cases, you know, in some sensors, you know, particular a should I say it's approaches have ad to be taken to exclude those areas so we don't end up getting false positives all over the place, like from urban heat islands and you know, desert surfaces and whatnot. But yeah, the algorithms are designed to not capture those areas as thermal anomalies.

42:24 Speaker 2: Great, thank you very much, Brad.

42:26 Speaker 1: It sounds like, yeah, the sensors aren't just designed for that particular detection.

42:32 Speaker 2: Question five. I was wondering if the static.

42:34 Speaker 1: Thermal anomaly masks we saw on the firm's website are made available somewhere as shape files. I would be interested to use them from through the mass I use to distinguish vegetation fires from non vegetation fires in my own algorithm.

42:51 Speaker 2: And it looks like there's a start of an answer here.

42:55 Speaker 5: Yeah, so I can take that one. So we do not distribute the static thermal MoMA later. But the information tab, the little I icon on the firm's website gives you all the information about where we are getting our static thermo anomalies information from.

43:16 Speaker 2: Okay, thank you very much. Question number six, what sort of time intervals? For sure? With the WRT?

43:31 Speaker 1: I guess what kind of temporal resolution or flower vacations are are you working with your data? And I'm going to assume this is going to vary depending on what centsor platform we're talking about.

43:44 Speaker 5: Yeah, I can take that. There are various temporal sampling based on the satellite you're looking at. It's different for modus and different for beers, different VI GEO stationary satellite, it's all across the world. Our frequently asked questions would be the best place to find information about that or in the information tab on whatever layer you're looking at.

44:13 Speaker 2: Okay, great, thank you very much.

44:14 Speaker 1: So I'm the layer and then you can find some information on how often that is being sampled, thank you.

44:23 Speaker 3: Yeah.

44:23 Speaker 5: And one more note in our advanced mode, I think is where it is there is orbit overpasses up the polar orbiting satellite, so notice and yours that can help you identify when the satellite has overpassed your area or area of interest.

44:44 Speaker 2: Oh great, thanks for pointing that out.

44:45 Speaker 1: If I remember correctly, the intro to firms earlier this year had a section where that was being displayed. Okay, I have a question. I haven't moved over here yet, but I'm going to throw it in here so everybody can see it. What about solar farms? Is it possible that they will show a map like a false positive?

45:18 Speaker 4: Yeah, Brad, here, I can take that one, Brock, Yes, potentially solar farms. Can you know due to sun glint cause false positives in various active fire detection products? These polar urbaning active fire detection products, like for motus and beers, the NASA algorithms are very robust. You know.

45:40 Speaker 3: They use additional bands.

45:43 Speaker 4: In the in the visible and neurinfrared wavelengths to help identify those areas of sun glint and reject those false positives. So the the beers and Modus data are pretty good in rejecting or keeping.

45:57 Speaker 3: Those false positives out of the the data.

46:00 Speaker 4: However, the geostationary active our detection algorithms are are not as quite as good as motus and fairs in rejecting those identifying and rejecting those false positives, so you would have a tendency to see those false positives from sun glint from solid farms showing up in the geostationary active fire detection data.

46:27 Speaker 2: Thank you.

46:30 Speaker 1: I have another question here, what what what STA detect peat fire in context of a below ground peat fire and would it be obscure by heavy smoke from vegetation fire.

46:53 Speaker 4: Yeah, I'll take that one again, Brock. The short answer would be generally, no, we're trying to exclude these natural fire events or you know, vegetation fires from the STA, and you know, we're doing that by looking at persistence of the fire activity. For example, you know, an industrial source that would have detections you know, over a long period of time throughout the year, whereas you know a wildfire you know, pete fire or whatnot would not be it, you know, persistent throughout the year. So we're trying to you know, exclude those areas wildfires and whatnot by looking at the persistence of those detections.

47:40 Speaker 2: Great, thank you.

47:44 Speaker 1: It's like there was a question in the chat. I'm not sure what the question was, but the answer from Jenny.

47:54 Speaker 1: There's been a lot of coverage of the pollution in India and the news that uses firms. Oh, I am from India and I have a yearly problem of farm fires that can contribute to the pollution. How can we distinguish general fire from double burning through veers or other natura products?

48:15 Speaker 6: Thanks Broke. I saw that question in the chat, so I just started answering it. So interestingly, there has been a lot of coverage of the pollution in India in various news sources and terms are being used. Data from firms are being used to help inform what is happening and as far as can we distinguish general fire from strubble burning. So additional contextual information so spatial extent of the farms, etc. Could be used to further kind of refine the locations and potential sources of the fires.

48:59 Speaker 6: But these fire is that are certainly being captured by both veas about NOWER twenty SMPP and now twenty one, as well as by modis about terror and Acwan.

49:16 Speaker 2: Great, thank you very much.

49:19 Speaker 1: I did see a question regarding to the homework once again. That link can be found under part two on the training web page and it is live right now. Question number ten, how does static thermoonomaly filter out thermonomaly from industrial facilities.

49:39 Speaker 2: So it's more like, how does it work. Are they using coverage binding from the file or with the nearest nerve neighbor algorithm?

49:49 Speaker 1: Tell us a little bit more about the features.

49:54 Speaker 2: Like we have a link here, we're putting a chat.

50:00 Speaker 3: Yeah, yeah, brocc I'll be happy answer to that. Yeah.

50:03 Speaker 4: As as noted in response there, we have an earth Data article posted about the s t A and how the you know, general description of how they're derived. But as mentioned before in the demo, what we what we did is we you know, took the cumul of the factor fire detections for at least sepp veers over the course of the year and also Tara and Aqua modis. We used Modus primarily take advantage of Terra's mourning observation capabilities, and then we'd have Aqua and as mpp vers for the afternoon observations.

50:39 Speaker 4: And uh yeah, we we identified those uh persistent areas where there was persistent fire detection activity within or i should say, over you know, extended period of time again looking at that persistence measure, and we're able to you know, identify those areas or locations as static thermal anomalies which may be due to industrial heat sources or natural sources.

51:09 Speaker 3: Like volcanoes and whatnot. Like as we discussed during a demo, we.

51:15 Speaker 4: Provide the sources of industrial heat activity infirms as a reference because you know, we're we're not trying to capture all industrial sources of heat out there on a landscape. We're just trying to capture the ones that are, you know, or at least identify the ones that are routinely identified in the modus and beers active fire detection products. And that helps users, you know, identify those particular detections that are associated with those features.

51:48 Speaker 3: So yeah, that's all pretty much. Have that sound that one, Brock.

51:56 Speaker 1: Here's a question, is there a way to a de linear surfs versus atmospheric thermal energy?

52:02 Speaker 2: Up Here in Canada we have a large smoke columns.

52:05 Speaker 1: That carry lots of heat and appear to artificially inflate the size of the fire. But once the smoke subsides, their hits.

52:12 Speaker 2: Are no longer valid.

52:13 Speaker 1: It often appears in a repeating grid pattern of hits along the direction of the smoke column is heading. It looks like we're gonna get back to on that one. Look for this answer to be build in, but we're going to consult our science team and get a little bit more of an accurate answer for you here.

52:35 Speaker 6: Yeah, thanks bro, And just to note that again webinar one kind of touched on this and so that would be another source of information and as well as we will consult our science team and provide additional information.

52:55 Speaker 2: Great, thank you very much. Okay, that seems to be last question right now.

53:01 Speaker 1: We have a few minutes left before the top of the hour, so if anybody has any more questions, feel free to type them into the an A box. And as it comes to the homework, it's due by the third of December, and then it's going to take a little bit of time for us to process attendance and homework submission. So typically you can expect to see your certificate.

53:30 Speaker 2: Of completion sent you as a PDF in your email about two months after today.

53:43 Speaker 1: In the meantime, this is an advanced training that builds off of one that we did earlier this year that was offered in English and Spanish. There are plans to deliver this in Spanish as well and then your future

54:02 Speaker 1: so please look for that if there's anything any of our guest speakers want to offer in general, to the participants here today, I welcome you to do so well.

54:16 Speaker 2: Thanks Broke.

54:18 Speaker 6: I would like to say thank you to everybody for participating and we hope you found this information very useful.

54:31 Speaker 2: Well, thank you very much. It's great to be back.

54:33 Speaker 1: And we thank everybody for hanging in there with us, and look for more wildland fire training to be offered by the our SET program in the future. I have a great rest of your week, everybody.

54:46 Speaker 2: Thank you.