Ericka Farrell: Network and ORD's Office of Research and Development Research on improving indoor air quality during smoke episodes and where to find helpful resources. And now, our first presentation will be from Mr. Brian McCaughey. 
Brian McCaughey: Thank you Ericka, as Ericka stated, my name is Brian McCaughey and I work for the Hoopa Valley Tribe which is in very far northern California by the Oregon border. I'm the Air Quality Program Manager there and we’ve partnered with the US EPA in the ASPIRE Research Study regarding low-cost air sensors and looking at the effects of wildland fire smoke and residential wood burning stove smoke on indoor air environments. And I think the slide that you're looking at now hopefully is showing a map of the wildland fire potential put out by predictive services last summer showing very, very above average potential through July which continued through the summer and fall. And actually the predictions are out for this summer and they look-- they look very similar to last summer's outlooks, so above normal to above average for the west for wildland fire potential. And the Hoopa Tribe over the last couple decades has been seeing the smoke events on the reservation affecting our residents, 3,500 residents, become an annual event. [UNINTELLIGIBLE] The perfect place to put a research study as we saw last year, we don't root for smoke, but we need it for research so, we got a lot of good information out of it last year. All right, next slide.
Male Voice: There is [UNINTELLIGIBLE]
Brian McCaughey: So in 2020, I'm seeing the introduction slide, but 2020 over 9,500 fires burned 4 million acres and impacted over 50 million people across the western United States with unhealthy air and in that slide you can see a demographic GIS map of the populated areas would be the next slide, maybe it's just loading slowly. And you can see that the monitoring network on the right, I'm on a different slide, so okay. So, yeah, if you can find that slide that's many areas impacted by wildfires
Andrea Clements: Slide four, please.
Brian McCaughey: Yeah, that one I just spoke to, it's the next one
Andrea Clements: Slide nine.
Brian McCaughey: Yep, slide nine, so in 2020 over 9,500 fires burned over 4 million acres and impacted, as I said, over 50 million people across the west with unhealthy air, and the scattered nature of those regulatory monitors are seen there across the west which hinder our forecasters, our air quality forecasters but we can rely now greatly on the PurpleAir to fill in the gaps, these are these low-cost sensors that we'll be talking about today. And I mentioned that we're really excited to be participating in the research project with the US EPA for this second fire season. Next slide, and hopefully it's the right one [LAUGHS] yeah, okay. So yeah, as we're talking about air sensors have been able to fill in these gaps from these larger more expensive and more time-consuming regulatory monitors that we use here on Hoopa but now we're using them more for a collocation and reference monitor, then we're able to utilize the PurpleAir Data and since it has been corrected which we'll go into later in the webinar it's really been helpful to reduce the confusion amongst the public, about the differences in values that were seen prior to the low-cost sensor network between regulatory monitors and the PurpleAir Map, there were big discrepancies in the air quality index prior to last season's web map. So everyone [INAUDIBLE]
Ericka Farrell: Brian, you're muted.
Brian McCaughey: I was muted then?
Ericka Farrell: Yes.
Brian McCaughey: Okay, oh, yeah, so I had spoken with—about air quality sensors filling in the gaps, and yeah, it's been really important, I mean Hoopa is really excited to be a part of the ASPIRE study the low-cost sensor study for this second fire year and later in the webinar, we'll be talking more about the AirNow Fire and Smoke Map which has really been helpful in reducing the confusion amongst the public about the differences in values seen between the regulatory monitoring network map and the PurpleAir Map. The AQIs are corrected and the data is much more easy to translate for the public, so with that I'll turn it over to our next presenter. Thank you.
Andrea Clements: Thanks, Brian, can I have the next slide? So my name is Andrea Clements and I'm in the Office of Research and Development at EPA, and just to make sure that we're all on the same page, I thought we could take a few minutes to talk about the PurpleAir sensor and make sure that everyone knows what we're talking about.So the next slide.So if you haven't seen one of these PurpleAir sensors before it's about the size of the palm of your hand, it's meant to-- they actually  sell a couple versions of this sensor, generally people install them outside, but they can also be used inside as well. The sensor itself contains two different sensing elements, those are the blue boxes that are shown in the bottom of this sensor picture right here, and we call these channels A and B so they independently measure particulate matter and they give us data at two-minute averages. The sensor itself can give us information about PM 2.5, but also a couple other size fractions and it can give us the information about temperature and relative humidity as measured up behind those sensors underneath the cap. So it's a little bit hotter and drier than you would normally see in the ambient environment. Data from the sensor can be streamed to the cloud if you connect it to a Wi-Fi, and you can choose as the user to display that data publicly which means it enters into the PurpleAir Map, or you can keep it private which means only you can see it,  or you would need to log in to see that data, and you would choose who could download and view that data. Some versions of this sensor also have an incorporated micro-SD card, so you can store the data locally and pull out the card and put it in your computer and look at the data later. If you choose to have data sent to the PurpleAir cloud which means you can view the data on your computer, then a user would need to register that device and when they do so, they tell-- they give the sensor a name, they tell people where it's located, they mark it as an indoor or outdoor sensor, and they choose whether the data is public or private. And as you can imagine because a user supplies this information, sometimes there are errors or sometimes the user may choose not to give very descriptive names or location, because they want some privacy, so those are all user supplied information. Next slide. So we do have to get a little bit technical here for a minute and mention that the data that comes off the PurpleAir sensor is actually giving us information about PM 2.5 using two different correction factors, one we call cf=atm and the other called cf=1.This information comes straight from the plan tower sensor that's that little blue box that's inside. And the reason why it matters is because they are not the same. At really low concentrations, they actually are equal to one another, but as the concentrations get higher there is a deviation, there's a different relationship as you can see in the figure here, where the cf=1 channel is actually much higher than the cf=atm channel. Why does this matter? So it matters because the data that's displayed on the PurpleAir Map is actually choosing which correction factor to display for you. So if your sensor is marked as an outdoor sensor, it's actually using the cf=atm correction factor, and if your sensor is indoors, it's using the cf=1 factor, so what are the important takeaways. First of all, no matter what correction factor is displayed and systematically we see that the concentrations reported by PurpleAir are higher than co-located regulatory monitor. And so these are the monitors that we have the highest confidence of data in, so we know that the sensor over reports the concentration, and it can also be very complicated if you're trying to do indoor outdoor studies as Amara will discuss later in this presentation, because the display is different, you're not seeing the same correction factor. So that's a really important consideration if you're doing indoor outdoor studies. Next slide. So I don't know if you've been to the PurpleAir Map before, but this is just a picture of what it was just a couple of days ago, and the PurpleAir Map lets you see all of the publicly available sensors on the map in this display, and in the bottom there's some controls that help you choose how you want to view the sensor data. Next slide. So we'll dig into this display just a little bit more. So the first thing, first drop down menu that we'll talk about is this one in the upper left, here you can see that this allows you to choose what data you're going to view. The default is to view all of the air quality data using the US EPA AQI for PM 2.5, and that's great except that it takes the instantaneous data and it translates that directly into the AQI, and generally the AQI is based on a 24-hour average of data, so the translation is not quite what we'd expect. Also you can use this drop down menu to look at other information so for instance if you wanted to look at the actual concentrations you could do that, or if you wanted to look at temperature or relative humidity, you would just use this drop-down menu to do that. There's a second menu just below that, that helps you look at how you'd like to average the data, so when you come to the map you're actually by default looking at 10-minute averages, but you can use this drop-down menu if you want to view data in real time so that's the two-minute data or if you want to look at longer time averages. So for instance, if you wanted to look at what an hour average would look like or a daily average, you could use this menu. Down at the bottom there are a couple of radio buttons that help us select which sensors we want to view on the map, and as I mentioned, you have the opportunity to install the sensor outdoors or perhaps indoors, and so you can use these controls to decide which sensors you would like to look at. And toward the top there's actually another button that looks at the conversion, so as I mentioned these sensors tend to over report the concentration and so there are a number of conversions that you can apply to bring those back in line and make them more comparable with the regulatory data, and you can use that menu to choose the conversion. Next slide. And I do want to point out quickly that you can find out information about those conversions using that question, question symbol. So here, on this slide, we're talking about those online conversions and when you hit that question symbol, you'll actually pull up this long list that tells you about the conversions, it tells you who developed them and what the equation is that actually converts that data. So you'll see that there are a number of options and it's very confusing to the general public to understand well which one should I use. I'm going to make my personal recommendation that you should probably choose the US EPA correction, because we've done a lot of work to make sure that that correction is appropriate for most places in the United States. So we've taken a lot of collocation data from around the country in order to develop this correction. One thing I will note that currently this correction is good up into concentrations up to about 250 micrograms per cubic meter. So that's actually a fairly smoky environment, but we do know that concentrations can get higher when your area is impacted by a wildfire or something like that. So we will actually be doing some additional work to make sure that the correction works at these higher concentrations. Next slide. So let's talk a little bit about the air--[INAUDIBLE]
Ericka Farrell: Yep, you muted Andrea.
Andrea Clements: Yeah, I just thought it was muted sorry about that. Okay, so we're going to talk about the integration of sensor data onto the AirNow Fire and Smoke Map. Next slide. All right, so many of you may be familiar with AirNow, so AirNow is the traditional place that EPA has shared air quality information. The data that's displayed on this map is from trusted sources of high-quality air quality information collected by certified instruments, by trained staff across the United States. So this is really high quality data. In addition, that data is quality checked so there are calibrations, flow checks, things like that to make sure that data is of high quality. We also make sure that all the instruments that are used in this network are sited properly, so we make sure that they're not influenced by really hyper local sources, and instead give us information about air quality within an area. Can we go back to this slide that shows AirNow and PurpleAir? There we go. Yes, so I also want to mention that the data that's displayed on this website is hourly average data to which the now cast has also been applied so we get a good indication of what the AQI is based on a longer time average, and it also provides health-based measure-- messaging about what people should do. So, we're also going to compare that to the PurpleAir Map which is also a great source of information. This is an extensive collection of crowdsourced data which means people buy these and put them up in their homes or communities, and so everyone gets to participate in the data collection which is really great. It's high time resolution data, so we're getting this data every two minutes, so it can give you information about air quality that might be changing quickly, and it can give you information about local sources, because sometimes they're impacted by things that are very nearby. What we don't know about this data is we don't know exactly how people site them, how they maintain them, and the data quality can also be uncertain. So although these are two great sources of information, it's not an apples-to-apples comparison, and that's what motivated our next piece of work. Next slide. So the Fire and Smoke Map tries to combine these two pieces of data to help people better understand their exposure to smoke from wildfires and prescribe burns. So from the traditional AirNow Map, what we're doing is we're selecting only the instruments that give us PM 2.5 measurements at a higher time resolution, so hourly data. And this will include permanent monitors and also temporary monitors that are placed by air resource advisors who are responding to wildfire incidents or also state and local agencies and they're placed in places where we expect to have smoke impacts. Now we're also going to add sensor data. So this sensor data, we have to do a lot of cleaning in order to get the data that we know is of high quality. So first, we only want to select outdoor sensors because a lot of things can impact the indoor environment, so we're just looking at outdoor sensors. Next, we're going to take this really high time resolution data and we're going to average it up into one hour data, and that's going to make the data between these two maps more comparable. Next, we're going to use those two sensors inside the plant tower, the A and B channels to remove some questionable data. Basically, if those two channels measure very similarly, we have good confidence in the data, if they don't measure similarly, we think that something might be up, and so we're going to exclude some of that data. Next, we're going to apply the correction equation that US EPA has developed based on this collocated data from across the country, and so we can make this data more comparable to our regulatory data. And next we're going to apply that NowCast AQI, so that we're looking at these two data sources very similarly. So the bottom line here is that a lot of work goes into making these two data sources comparable, and to your benefit it's all done behind the scenes on the map. So if you as an EJ community or tribal community, you might not have the capacity to build your own map, you might not have the capacity to do the studies to create a correction, you might not have the capacity to try to make this apples-to-apples comparison, and the good news is, if you buy a PurpleAir sensor and you connect it and make that data public, your data can flow to this map, and a lot of that work is done for you. This map also includes some information about fires, so things like fire detects and smoke plume extents that are largely captured by ground or satellite observations and there's also some incident reports that are created by fire specialists, who are deployed to large wildfire incidents. Next slide. And so this is just a view of the map as it sits today, again this was a pilot project to include sensor data onto this map, we do foresee that the pilot will extend and there's an expected update in 2021, so you can look forward to that before the beginning of the next fire season. But what I will mention here, is that you can see that we have permanent monitors from the AirNow site shown in circles, we have temporary monitors that go out to try to measure for smoke, they're in triangles, and we have all of this great low-cost sensor data in squares on the map, and you can see them all harmonized here. And you can learn more about this using the FAQ button here on the top right, which will give you access to the user guide and a lot of answers to frequently asked questions. Next slide. So we know that there are still some questions about how these two maps compare so we're going to do just a little bit of side by side comparison the next couple slides. Next slide. So first of all, I mentioned that each of the slide, sorry each of the sites whether it be AirNow or the PurpleAir Map had specific characteristics, and that we can bind these two to try to bring the best of both worlds and to the Fire and Smoke Map to help you understand when smoke might be impacting your community. So this table is just a quick summary of some of those differences, including averaging times, what kind of data is supplied, what kind of pollutants are measured and some of the QA procedures. I won't go through it in detail here, but just mention that this slide is a great resource if you have questions in the future. Next slide. So when we look at these two maps, we have AirNow on the left and we have PurpleAir on the right. So the AirNow map-- Fire and Smoke Map, excuse me, is displaying PurpleAir sensors that are outdoors and the data has been cleaned, it's been averaged, it's been corrected, and we've applied the NowCast to that data. On the right you have the PurpleAir Map, and here we are showing indoor and outdoor sensors, they are using different correction factors for each and we're looking at 10-minute averages. And there's also no cleaning of the data, so every sensor that's out there is being shown on this map. Now, if there's some disagreement between the A and B channels, generally the sensor is shown behind other sensors, but it still appears on this map. Next slide. Now, in the PurpleAir Map I mentioned there's a number of ways you can customize it so we could remove those indoor sensors, we could apply the US EPA correction, but there are still going to be differences between the maps. One because currently the US EPA correction is only applied to two-minute data, so we can't actually get the hours and we can't actually apply the US correction to hourly data. Additionally, there's no option to use the NowCast, so those-- there will inherently be differences between these two maps and that's important to understand. Next slide. And I think the biggest differences between these two maps, is really just in the way they look. So if you look at the AirNow Fire and Smoke Map on the left, you can see that we have categories, so green is good, and moderate is yellow, unhealthy for sensitive groups is orange, but when we move to the PurpleAir Map we're not using a solid color for an entire category instead we have a gradient of colors. And so if we just think about the moderate category, the yellow category here, we can see that the colors of the dots on the PurpleAir Map may look anywhere from yellow to an orange, and so in general this makes the map look more orange than it actually is. So comparing some of the numbers might actually be the better way to look at these maps, and I will also mention that this information is accurate as of about December, I think it's still accurate today, but all of these sources of data will evolve in the future right PurpleAir might make their own decisions and change their maps, as I mentioned the AirNow Fire and Smoke Map will be getting an update this year as well. Next slide. So there are a couple of considerations that you should keep in mind when reviewing air sensor data on the Fire and Smoke Map and this is in general because it's coming from crowd source data. Next slide. So there are four key points that we'll discuss in the next couple of slides, one is that sensors can fail, they're not quite as robust, they're not checked as often as higher quality air quality measurement instruments, and so it's important to keep that in mind. It's also important to keep in mind that sometimes sensors can be mislabeled mislocated or poorly sited, and this is just based on how someone an individual might purchase it and put it up at their home, and how they choose to register it.
Third, is that sometimes sensors can saturate at really high concentrations and when we're in a dense smoke situation, this is an important consideration, we need to learn how to use the sensor data appropriately. And four, is that sensors may not always respond to the same sources in the same way and we'll get into an example in the next couple slides. So next slide. So I mentioned that sensors can fail, there's actually been a great look at the ways in which PurpleAir sensors can fail conducted by Mazama Science, and if you visit this website they'll tell you a lot about what they've observed, and how the data looks as a result. Generally, we found that the work that we've done to compare the A and B channels has been very good at helping to clean up some of this problematic data that may happen when sensors fail. Some things that aren't really captured and dealt with yet, is that sometimes sensors can drift or age. And you can think of this as sometimes sensors might get dirty over time and so concentrations might nest kind of creep up, because they're dirty or sometimes sensors may age just because of the concentrations they're exposed to, so for instance the ambient concentrations, the concentrations are generally low and maybe it takes a long time for a sensor to get dirty. But if it's exposed to a lot of smoke maybe it gets dirty faster, and these are still things that we're investigating today, there's not a lot of strong evidence to tell us exactly what's been happening here, so we're still investigating this and you'll see more information in the future. Next slide. I mentioned before that sometimes sensors can be mislabeled or mislocated, and this is often because a user has to register this device, and maybe they change their mind about how they want it to be used, or maybe they've moved and just forgot. So here, in this example, there's a sensor on the map that is clearly mislabeled. This sensor is labeled as an outdoor sensor, but it's actually indoors, and we can use some of our information, the temperature information for instance and the PM information to help us identify what might be going on. So on the left if we compare the PM measurements with other sensors in the area, we see that the concentration is much higher and that there seems to be a pattern in it, it's high during the day and low at night. And in this case, this sensor is likely indoors, so let's check that, we're going to go and look at our temperature measurement. And here we see that the temperature is staying really steady, it's not going up in the day and down at the night like we'd expect. And so this suggests that it's in a building with an HVAC system, and so we can use this data to help us understand these outliers on the map. Currently, we don't have any automated processes to find sensors like this on the map, but if we just review the data periodically we can find these, and we can understand what might be happening. Next slide. So poorly sited sensors is a very common concern, and my colleague Amara Holder who will speak next did this wonderful experiment, where she tried to take some sensors and poorly sight them, and to see how they compared to a sensor that was properly sighted. So in the top right, you see a sensor that's properly sighted, it's out in the open, it's not the flow is not obstructed very good sighting and she had three scenarios, one where a sensor was near a strong airflow from air conditioning unit, one where the sensor is really close to the ground, generally, we like these to be in the breathing zone or higher, and one where the sensor flow is obstructed it was actually behind some patio furniture. And what we did was we compared this data with the properly sited data and that's what's shown in the figure here. And we can see that for the sensor that was close to the ground or sensors that were impacted by that high airflow, they agree very well with this sensor that was properly sited. And the only one that seemed to have some issue and had a lot of cloud and this data that I'm showing you here is the one that had unobstructed flow. And so the good news here is that most citing scenarios will be fine as long as you put them out in the open don't obstruct the flow, even if it is near your home, but we do want to worry about that obstructed flow case. So if you're buying a new sensor keep that in mind. Next slide. The third consideration was that sensors can sometimes saturate when the concentrations get very high. So we were fortunate-- unfortunate that in 2020 there were a number of wildfires in California and Oregon, during which we captured some really high concentration events where we had sensors co-located with reference monitors, and what we saw that instead of a very linear relationship, we saw that the data started to curve and this means that the sensors were overloaded with particles during this time. So we are currently taking the data that we collected in 2020 to develop a new correction, accounting for this curve in the data, and we'll see that upgraded and included in the update to the Fire and Smoke Map this year. Next slide. And lastly-- sorry go back to the map real quick. Lastly, sensors don't always respond to the sources in the same way, so for instance, we really understand how the sensors respond to ambient conditions and to smoke impacts, but what about things like dust. And last year we also captured some data, where we had a Saharan dust cloud come over the southern, east southeast US, and here in this photo our figure, you can see that the permanent monitors are often showing yellow or orange, meaning the air quality has deteriorated, but nearby sensors are all showing green. And this is just because the sensors aren't picking up that signature from the dust as they should. The particles are larger, they scatter less light, and so we don't see them in the same way that we'd see things like smoke. So I think the takeaway here, is that the US wide correction may not be applicable to some sources, and most specifically to dust. So if there's near, if sensors are placed near a roadway that's unpaved or if you're being hit by a Sahara dust storm, the sensors may not be giving you accurate information and that's something to keep in mind. Next slide. And with that, I want to turn it back over to Brian who will talk just a little bit about his experience as an ARA and how on the Fire and Smoke Map helped him in the field. Over to Brian, thanks. 
Brian McCaughey: Thanks again, Ericka and welcome back, yeah, This US EPA partnership with the fire team at US forest service has been a really good thing for the Wildfire Air Quality Response Program which is a interagency team from Bureau of Land Management and National Park Service and US Forest Service and EPA, and tribes, among others, who utilize a lot of different tools to help forecast for wildlife impacts of wildland fire smoke on communities and tribal communities. And very helpful tool the air Fire and Smoke Map for air resource advisors on these incidents, but also for Air Quality Program Managers, as I am for Hoopa Valley Tribe and for the air districts, who also put out their own products for the protection of the public. And, yeah, I was going to talk a little bit later, I think too about how I've seen the confusion, I mentioned it before. Not be eliminated, but definitely reduced in the sense of members of the community looking at the PurpleAir Map and without correction factors. And seeing one reading, and looking at a website that's referencing regulatory monitor of a state or local entity that's reading different, because it's a more robust reference monitor or regulatory-- even regulatory monitor. And the public not understanding that data sometimes needs to be corrected and that it's not-- it's not wrong, because of that reason. So thank you for letting me share about that and my experience, and I will pass it on to the next presenter. 
Tara Weston: Thanks, Brian, can I get the next slide, please? This is just a take home that point that Brian just mentioned, and this is a news report from NPR in San Francisco Bay Area, just noting how the new Fire and Smoke Map was so beneficial and just making a cohesive story between the sensors in between the regulatory monitors providing a data source that people could readily understand and interpret for their lives. Next slide, please. Okay, so we're going to change gears a little bit and we're going to talk specifically about Hoopa and the community monitoring plan that we have there. Next slide. Now, we at EPA wanted to do something, we wanted to do some research that would be able to help communities better respond to wildfire smoke episodes. And so we initiated the wildfire ASPIRE study so that's the advancing science partnerships for indoor reductions of smoke exposures. So we wanted to partner with communities that were facing, this yearly challenge from wildfire smoke-- we found two great partners in Missoula, Montana, and the Hoopa Valley Tribe in California, which Brian is representing here today. We sat down with them and basically asked them what do you need, what can we do to help you in your response to protect your citizens from wildfire smoke. And through these discussions we were able to identify a couple of research questions that we hope to answer. Now, I have them all listed here today, but specifically, I want to talk about the question of what science is available to support the recommendations for communities to develop clean air spaces in larger buildings like schools or community centers. We wanted to find some best practices for these communities to have to give people a place where they could go to that would have clean air, even if they didn't have it at their homes. Next slide, please. So the way that we did this was by developing a sensor monitoring plan for both communities. Now, we follow a general framework that I've shown here, and this is really basic for any type of sensor monitoring plan, You always start out with your question what are you hoping to achieve, what knowledge are you trying to gain from this monitoring network. So for this project we're looking at what's the PM 2.5 concentrations across the community as well as indoors at these public spaces. Now, given that question you have to formulate your plan and figure out what's the most appropriate way what sort of data quality do you need and how can you achieve that, and so we desired to have a smoke monitor in multiple locations across the community we wanted to be in many different buildings all at the same time so what we needed was a precise sensor, and we definitely needed it to be low cost so we could pull it off at multiple locations and due to those constraints we selected the PurpleAir. Then we went on to looking at how we were going to set up these sites and what we needed to set up these sites. So we wanted to make sure that they had power, we wanted to make sure that they would be secure, and then we also wanted Wi-Fi later on-- this is something that kind of goes in hand in hand with how you collect and maintain the data. So for the PurpleAir you can have the data stored on board, or you could have it transmitted through Wi-Fi for--
Male Voice: Providing a [UNINTELLIGIBLE]
Tara Weston: Please, mute.
Ericka Farrell: Please mute your lines.
Tara Weston: So that's something a consideration that needs to be evaluated when you're developing your monitoring plan. And then finally, once you have your monitoring plan, your monitoring network up and running, you need to evaluate your data and so for us we were looking at the PurpleAir data that was displayed privately on the map and that way we could look at the sensors in real time and see what sort of concentrations that we were experiencing in these different areas, communities, and diagnose any issues that might have come up with the monitoring. Next slide, please. Okay, so a little some information here about how to set up these sensors. Andrea gave a nice description of some of maybe the worst practices. So here are some of the best practices for setting up your sensors both indoors and outdoors. The most important thing is to identify a site where it's going to have some air quality significance for your study, for us we were very interested in looking at places where sensitive or vulnerable populations might be congregating. So this might be a school, or a community center, or senior center, where we know people might be especially harmed by wildfire smoke. But other places that might be of concern would be a community center or would be places where there are outdoor workers, places where you might have a pollution hot spot that or the community is concerned about it. Definitely places where there are no monitors and you'd like to have more air quality information, and roadways are a major source of air pollution so that might be another location that you might like to set up at. At each of these sites you want to make sure that you have enough you have the infrastructure requirements that are going to support your deployment, so you want to make sure that you have on-site power, if you need that, or you have sun exposure if you're going to be doing a solar setup, you want to check and see if you've got a Wi-Fi signal, if you're set up at a building that already has internet or a cellular modem and make sure that you have your cell signal. And if you have another-- neither of those, there's also a satellite option, you might have satellite signal although it's much more expensive. It's good to keep in mind trying to get a secure site so that you're prevented from having any tampering, also theft might be an issue, so it's good to set up behind a gate, if possible, to set up out of reach so it makes it more difficult to try and take that sensor down or unplug it. And it's nice to have a supportive host, we found that schools, fire stations, libraries, community centers, have often been willing hosts for us and been very supportive of setting up at their site. And of course, you want to look at having safe access to your deployment site, so you want to prevent any sort of fall or shock hazards when you're deploying these sensors. And then finally, you want to install the sensor for optimal data quality, so this is something that Andrea touched on. What we recommend is to have at least 180 degrees of free air flow around the sensor head. So you can see our outdoor picture there that's a site in Hoopa, it's set up on a tripod so you have almost 360 degrees of free airflow to that sensor which is great. Indoors we have a sensor located at a senior center in Missoula, and you can see it's mounted on this post here, it doesn't quite have 360 degrees of free airflow, but it definitely has more than 180. So we think that provides optimal data quality. Both of those locations we would like to get it near breathing height, because that's where we're interested in understanding exposure. Outdoors you definitely want to be above about a meter, above ground, so that you're not impacted by splashing water or floods, you don't want to get the sensor in contact with water. It's good to stay away from nearby structures, but if you have to be set up next to the wall of a building, it's better to place it on the upwind side versus the downwind side, it's also good to place it away from vegetation, so don't tuck it back behind the bushes, because those bushes can filter out some PM. And depending upon your goals for your monitoring program, you may want to either put it near PM sources, because that's what you're interested in, or you may want to keep it away from PM sources if you're interested in more of the ambient concentrations. So I would say consider dusty roads, cooking appliances, like stoves and grills fireplace, fire pits or fireplaces, you want to just be aware if those are in your environment, and then you also want to stay away from strong airflows. We know that it didn't make a difference in our particular application, but it may be important in other applications, so it's probably good practice to just stay away from exhaust vents and AC units if you can. Next slide, please. So given those general guidelines, let's talk a little bit more specific about the Hoopa monitoring network. This is a map here of Hoopa California that's in northern California, it's a very beautiful location, nestled in a valley with a beautiful forest all around, it is particularly prone to wildfire smoke. So they certainly pay a high price for air quality in the wintertime-- in the summer times when they have wildfires in the area, so that's why we partnered with them to hopefully help them face their challenges, and we wanted to set up in multiple locations across their community to look at a diverse set of buildings. So what we did was an initial collocation with our entire fleet of sensors that were going to be deployed at these different buildings. And you can see a picture here of our collocation at the high school, track and field, this is where the central monitoring station is in Hoopa. We did this initial collocation with their reference instrument, this is-- so we can have a correction factor that's applicable to their unique location. And then we deployed these sensors at buildings across Hoopa, we started our sampling back in November of 2019, and it's currently ongoing and we hope to continue until we get to hopefully more of a post-pandemic society where these buildings are open like normal. We are planning to do a mid-study collocation check here in another month or two this way we can check the health of these sensors, it's really difficult like Andrea said earlier what to determine if these sensors are experiencing drift, and Hoopa experienced a very terrible wildfire smoke season last year. And so they certainly saw some very high concentrations, so we'd like to get a handle on whether they show any wear and tear from that, and we can want to continue sampling there year-round. So we're not only interested in the wildfire smoke, we're also interested in the wood smoke impacts in the winter time, in the springtime there. Because that's a challenge that the community faces as well. Next slide, please. So, this is a couple of pictures here of what the sensor installations look like in a picture of Brian at the daycare room with the sensor we measured at 10 outdoor locations across the Valley, and nearby the Valley and we also measured at 14 public buildings we targeted these buildings that. We thought people might stay for an extended period of time to get some respite from the smoke, so we looked at places like workplaces where people would be working, places where sensitive or vulnerable populations might be gathered, these would be the schools, daycares, senior center, also places that might serve as a potential clean air center so that would be the neighborhood facility or a church, and then a COVID adaptation because most of these buildings were closed all of last year we also set up at Brian's house to see what the impacts were in a residential environment. Next slide, please. And so these are some preliminary results from last year's wildfire season, as I mentioned before we had some very high concentrations in the Valley you can see that this blue line and this time series is the outdoor measurement, and these are corrected purple layer sensor data so we believe that it's accurate, we use that high concentration correction, so that we have some good faith that the data is pretty accurate here, and we're seeing values that are over one milligram per meter cube that are being experienced in the Valley. So that's very-- really quite extreme. And we monitor it at one location this is at the daycare, and you can see that the indoor line that's the pink line there in the time series. Initially, when they were first seeing smoke impacts in the Valley this particular building was not doing a great job of keeping up with cleaning their indoor air. Now, this particular building has a central air conditioning unit, but it just wasn't up to the challenge. There weren't a whole lot of people coming and going through that building daily so we don't expect that there was a lot of door opening and closing allowing air flow to come in, but we just think that the HVAC system was not up to the challenge. The workers at this building noticed the poor indoor air quality and were able to rent an industrial air cleaner, there's a picture of that air cleaner there, that was provided by Brian. And once they turn that air cleaner on the air quality, greatly improved in their facility, you can see it dropped it down to near zero PM concentrations indoors. So this is a success story of what you need to do in order to get clean air in your building when you have really terrible smoke outside. Next slide, please. So this is another example of a building in Hoopa, again experiencing the same sort of outdoor conditions. This building in particular does not have a central air conditioning system. Cooling is achieved in this building using a window AC unit and they have quite a large number of people that come and go through this building daily. Now, I will note that the room that we are monitoring in is in an indoor area it has no doors to the outside, it has no windows to the outside and that may be one reason why the air quality is much better there. and what we determined after the fact is that the people who were working in this building were able to borrow air filters and they did not specify the type of air filter that they borrowed, but that air fuel-- filters were doing a pretty good job of keeping the concentrations much lower than the outdoor concentrations. So again this is a pretty good example of a good response using air cleaning technology to improve the indoor air quality. Next slide, please. Now, here's our final example of a building where the air quality indoors was really quite terrible. This was the high school classroom, some background information about these classrooms, their individual classrooms that have a single door to the outside. So there is no corridor or any other structure around them, it's just a single room and then the outside. Each of these rooms has its own central air conditioning system it does have a MERV filter on it with an unknown rating at this point. Not so many people were occupying these rooms during this time, because they were closed due to COVID, and we're not quite sure if they were operating their HVAC system, there's some question whether it was actually going or not, so that could be one reason why the indoor air quality was almost identical to what was going on outdoors. So this is just giving you a range of experiences that we're seeing in these buildings across Hoopa. So next slide, please. So I just want to take a step back here and talk about our sensor network and some of the experiences we learned while we were monitoring within Hoopa. What we found is that this initial collocation was really critical, it helped us to identify any defective sensors out of the box. We didn't have a whole lot with PurpleAir, that's probably one of the unique aspects that's really great with PurpleAir is that their out of the box failure rate is pretty low. We haven't seen that always be the case with other vendors or manufacturers, and we've also were able to take this data and improve the sensor precision. And so even though most of the purple layer agree very well with each other. We could fine-tune that agreement and improve the precision even more by using these individual sensor corrections. We've always kept the long-term sensor collocated at that reference that way we would know that the correction that we developed during that initial collocation was valid throughout the entire monitoring time, and if it wasn't we were able to correct for that. One thing that we found was that power strips are really important, sensors tended to be unplugged quite a lot at both locations. Outlets are hot commodity, and it just so happened that people would come along and unplug that air sensor whether they wanted to plug in an air freshener in its place, like this picture shown on the bottom right corner, or they wanted to plug in their cell phone to charge it. These sensors would get unplugged pretty often and often times they wouldn't get plugged back in so we'd have to ask Brian or other partners in Missoula to come and put that sensor back online. So putting in a power strip prevents that from happening, it doesn't keep it from happening entirely, but it sure helps. It was really good to have a sensor installation that was somewhat inaccessible a secure site is strongly recommended, we haven't had any tampering, we don't think at these locations, but we have had theft and some of our other studies. It's just kind of an attractive little device there, people don't know what it is, and it can certainly walk away. And finally, we found that the online data reporting that we moved to was really useful in identifying very quickly when a sensor was had gone offline or was failing, and we could send someone out to go and investigate. So those are some of our initial takeaways from this first year and a half effort in Hoopa and Missoula. Next slide, please. Okay, so now I'm going to change gears a little bit and talk about more of the building characteristics, and this is something that we're actively looking into finding what are the best practices for your building and how you operate your building, and how that may translate to indoor air quality. So we have been doing on-site inspections of the HVAC system as well as the building itself, we've done a full set of evaluations with our Missoula partners, and we're hoping to get down to Hoopa and do that sometime the next few months and just go into every building that we've monitored at and look at all of the different features of it. So we want to look at air handling settings and schedules the use of portable air cleaners HVAC systems and filter conditions look at how the doors and windows are kept up whether they have gaps or seals around them. We want to look at the building age and the construction type, because we know that impacts smoke infiltration, look at room and building pressure is it positive respect with respect to the outdoors, is it negative that can also impact infiltration. Note any of the potential indoor air sources or air pollution sources like cooking, or tobacco smoke, or vacuuming or sweeping. And also keep track of the building open and close hours and when people are coming and going, because we think those door openings may be impacting smoke and filtration, we've also put-up door sensors to estimate those door openings and closings. So on the right hand side here, I've shown some of the preliminary results from Missoula and this is just a an evaluation of the air filters and what they've looked like in the different buildings, we've seen that most buildings have lightly used filters which is great. Some of them had filters that were a little bit more used and some of them were definitely past their service life and needed to be changed. And so that's the kind of information we like to identify, how often is this happening and maybe push that up in our best practices and encourage people to put them on a better schedule or pay more attention to when the pressure drop in their system gets too high. Next slide, please. So these are some of the pictures that we were able to gather, this is work done by Tom Javins, he's our building HVAC expert who's been doing these evaluations for us. These are some of the things that we want to help people avoid, so in this first picture we really want people to use right size filters, this was a filter that was too large for the filter rack and so you can see this been crunched into place and it's made some big gaps where dirty air can just bypass that filter entirely, so the filter's not doing its job. We've seen other examples of where the air intake into the HVAC system is completely blocked by other stuff that's going to have some impacts on the HVAC system, and may get more dirty air from other intakes. So that's a-- we want to keep those air intakes freely open and allow air to go in there especially if that's where your filter is located at. And then we've seen a couple of instances of some very dirty filters and so we want those to get-- We want those to get replaced on a more routine schedule with clean filters. Next slide, please. Okay, another great result from the ASPIRE project is that we some of our team have been working with ASHRAE, that's the American Society of Heating Refrigeration and Air Conditioning Engineers to develop a new guideline to provide some of this information for protecting commercial building occupants during wildfire events. Previous ASHRAE guidelines have been focused on have been tied to whether an area was in non-attainment for PM, and so this breaks free from that and also gets more specific into some of the challenges that's unique to wildfire episodes. And so some of the guidelines here are providing, just telling some really basic information like upgrading to the best filters that you can so a higher rating in order to better remove the smoke maintain your HVAC system, identify where you might have leaks or broken dampers or controls, test your system, especially before wildfire season to make sure that you can-- that your new filters or your higher quality filters can work and you maintain a positive building pressure, and you have sufficient ventilation. And then weatherize your buildings, seal any gaps in your doors and windows, and try and limit door openings as much as possible. And then monitor your PM 2.5 levels, we found that these sensors are really critical for providing us real-time feedback for how well that building is doing with its indoor air quality, and so we think that's a valuable addition to any building is to try and get an indoor PM sensor. And then use portable air cleaners if necessary, and so if your indoor air quality is not good rent an air cleaner, if you can or purchase one. Next slide, please.
Brian McCaughey: Amara you guys wrapping up for-- we've got some time we would like to have set aside for the Q&A possible.
Amara Holder: Okay, yeah, I can just go quickly through the next few slides. These are really focused on the do-it-yourself air cleaners, so just to say that we've been working on evaluating some of these do-it-yourself air cleaners, so it's a low-cost option that many people are using, especially, when portable air cleaners, commercial air cleaners are not available. So we have a nice set of instructions provided by the Confederate Tribes of the Colville Reservation. We're currently working on safety evaluations to make sure that these are safe and effective and preliminary data is showing that they are very effective in certain circumstances when you have a smaller room and we do recommend that you use only new box fans, because they have added safety features like fused plugs and thermal knockouts. Next slide, please. This I'll just skip over, this is just an EPA challenge on trying to find more and innovative low-cost solutions to air cleaning and so hopefully we'll have some solvers, submit some new technology that we can use-- that anyone can use at any price point. Next slide, please. And this is our take home summary for this entire presentation so I just want to say a significant amount of research has been done by EPA and others to look at the PurpleAir in its performance, and it's given us much more confidence in the reported measurements from PurpleAir. The EPA PurpleAir correction equation definitely improves its accuracy, and it has shown to work very well for smoke conditions, and it adds a high degree of additional spatial variation on the map. And we found that this map is very useful for communicating to the public and it's also-- these sensors are very useful tools in looking at indoor air quality as well and trying to improve indoor air quality and reducing smoke exposures. And our work will continue to try and identify the best practices and the lowest cost practices for cleaning your indoor air during these wildfire smoke episodes. Next slide, please. These are just some references that you can refer to later that these are some links to the study websites and some of the scientific papers that support the research that we've shown here today. Next slide, please. I'd like to acknowledge the many local air quality, local, state, federal, tribal, air quality agencies that have worked with us on this, it's been a huge effort by a very large team and everybody has done their part here so we really appreciate that. Next slide. And these are some of the resources that you can go to for more information there's a number of great guidebooks and fact sheets and other information that's at a variety of different levels of detail on wildfire smoke, air sensors and indoor air quality. And with that I think the rest of them are just supplemental slides, so we can end here and open it up for questions. 
Ericka Farrell: Thank you, Amara, and I thank you to Brian and also to Andrea for your great presentations, and before we have our Q&A session, I would like to introduce a special guest Mr. Ron Evans from our Officer Air will give us updates to the AirNow and Smoke Map.
Ron Evans: Okay, we'll make this very brief so we can get through the questions. As Andrea said early on, we're in the process of putting out a new edition of the Fire and Smoke Map for next fire season, we'll have that out in the latter part of July and what we're doing here is improving the functionality and the usability of the information that's there for example, if you were to click on a monitor or sensor right now, you get a graph showing what the air quality information is as it relates to the AQI and what we're doing is looking for more user-friendly ways of displaying that information, so that people can understand what the air quality is more in a timely fashion and also adding information about trends and the things like that. This is a-- remains as a pilot product. doesn't mean that we're taking it away at any point it means that we're going to be trying things out here. So we encourage everybody who's using this over the course of time to send comments to us, and tell us what they think about, this is good, bad, whatever it may be. The one other thing I want to emphasize just to build on something Andrea said as it relates to tribes and others going ahead and taking advantage of the Fire and Smoke Map. EPA cannot endorse any specific sensor at this point, we don't endorse commercial products, but we've had a lot of working understanding of the PurpleAir sensors and that's why we include them on this map and the one thing to emphasize is that if you go ahead and purchase a PurpleAir sensor register it with PurpleAir, it will automatically appear on this map, there's no other steps you need to take, to make it appear on this map, and with that I'll turn it back so the questions can be in.
Ericka Farrell: Thank you, Ron and now we will have our question-and-answer session, and this session will no longer be recorded.
Danielle Ridley: Thanks, Ericka, I think Andrea and the team have done a great job answering questions in the chat.