Eyes on Earth Episode 136 – The EROS Test Site

[00:08] Hello everyone, and welcome to another episode of Eyes on Earth, a podcast produced at the USGS Eros Center. Our podcast focuses on our ever-changing planet and on the people here at Eros and across the globe who use remote sensing to monitor and study the health of Earth. My name is Tom Adamson. [00:30] we often talk about the fact that Landsat is the gold standard for Earth observation. [00:35] One of the reasons is the calibration and validation work that goes on at Eros to ensure its accuracy and reliability. [00:43] One of the ways the team validates what they have so carefully calibrated is to take measurements on the ground while Landsat passes by overhead.

[00:51] We have one of those validation sites here at Eros. [00:54] I spoke to Jeff Irwin and Travis Karpinski who take care of the Eros test site and who head out there every time one of the Landsat satellites passes over the site. [01:03] Well, not quite every single pass. We'll let them explain that when we get to it. [01:07] Most importantly, we're going to go out to the test site during a Landsat pass so you can get a sense for what that work is like. [01:14] Jeff Irwin is the Eros Survey and Field Campaign Lead.

He oversees all field work that happens for the Eros CalVal Center of Excellence, in addition to elevation-related projects that we do for the National Geospatial Program and the Coastal National Elevation Database. Travis Kripinski, contractor to the USGS Eros, is a field surveyor and data processor, and he carries the mobile field spectrometer for this Landsat field validation work. [01:44] that you use when you go out to the test site? Yeah, so we have two field spectrometers. They're called ASD field spec fours. Each one of those is actually three instruments. It has a v-near channel.

Visible near infrared. And then there's two shortwave infrared instruments as well, so they have a pretty high spectral resolution for measuring parts of the electromagnetic spectrum [02:14] is mobile and one of them is static. The mobile one measures eight transects that we have out at our site and the static one measures a calibration panel continuously while we're doing that. We have eight transects spread out at our site. They're 150 meters long and they're 20 meters apart. So we sample along those transects so that we can aggregate that information to compare to the [02:44] at the same time that Landsat is passing overhead.

Yeah, ideally we want to be within plus or minus 20 minutes of an overpass. Okay. So we've established our routine, so we start 13 to 15 minutes before the satellite overpasses, and we're generally halfway through our collection when the overpass happens. So you're walking back and forth? Well, Travis is walking. Travis does the walking? Okay. Well, why don't you tell us how that works? [03:14] strapped to my back. It's got a fiber optic cable leading to a four optic, which I hold out ahead of me, not have many shadows. You're careful not to cast a shadow on it?

Okay, glad you thought of that. If I'm picking up shadow, I'm not doing my job. Yeah. And I'm going to walk these north-south transects that are about 150 meters long. The instrument measures light as it's reflecting off the ground, and of course you're not very far off the ground. I mean, that might be a meter and a [03:44] at the same time is 438 miles above the ground. How does that work? How are we kind of comparing what you're collecting on the ground versus what Landsat is collecting way up in space?

We're sampling what Landsat is collecting. I'm just measuring a small portion at the same time. So then here's what we got. What do you got? There's three sensors in there, and it's measuring electromagnetic spectrum from 350 nanometer to 2,500 nanometer, which includes the visible portion. Okay. [04:14] So, yeah, every nanometer we're getting a measurement. Additionally to those instruments, we have these calibration panels. They're made of a material called spectrolon that is a near-land version material, so they disperse light equally in all directions. We've had them characterized. [04:35] What that means is for several different solar geometries, we know how the panels behave.

[04:44] And then, so during different times of the year, the sun is in different spots for us. So that allows us to deal with those different solar positions throughout the year. Okay. What do those panels look like? They're essentially basically a one-foot square. They're both at 99% reflectance, so they're quite white. One foot by one foot. Is that okay to do since the Landsat pixels are 30 meters? [05:14] across? Yeah, so we're not relying on seeing the panels themselves in the imagery. Yeah, okay. What we're doing is we're getting our four optics quite close to the panels themselves, so we're just sampling a very small portion of the panel, and the light that's being reflected back, we are pretty confident that we understand the light that is coming off of the panel, and that allows us to [05:44] transferred to the spectrometers, which we then transfer to our field site.

What's the first thing you do when you get out to the site? First thing we do is drive out there and pull out our panels and instruments, set up the panels where they need to be initially, and then we start setting up our instruments. They need around 15 minutes of time to let them warm up and get up to operating temperature. Okay. [06:14] The computers that store the data that the spectrometers are collecting and get our files all ready to go for storing that information. What if it's cloudy? Then do you still do it or not?

We consider this validation work, not calibration work. Oh, okay. So that's part of the reason why we have a grass field site because it's not easy. [06:44] like deserts where they have nice bright reflectance and they're well understood, and they use those for calibration. But this is a validation site, and so we do have a tolerance for cloud cover. They kind of have a threshold of 40% cloud cover forecast that is kind of our go, no-go decision. Where do you get your information about whether it's 40% cloud cover or not?

So we use the National Weather Service. [07:14] forecast for up to seven days out. So raining is a no-go? Raining is a no-go. Another thing we don't do is when there's snow on the field. The instrumentation does have limitations in terms of operating temperatures. I don't think we've done anything below 40 degrees, I would say. How about weekends and holidays? Landsat doesn't care if it's a holiday or a weekend. That's right. Landsat is imaging the aero site, assuming [07:44] Free. Every eight days. So we'll do weekends. We will do some holidays, depending on how our collection season has been.

There are definitely times of year, like, we have a lot better time getting good collections in September than we do in, say, March and April. Yeah, yeah. We get more rainy weather in the springtime here. Yes. And June can be variable. [08:14] with wildfire smoke. Got some interesting results that I'm not sure we've been able to tease everything out with yet. Yeah, we're on the northwest corner of the property. Essentially, it's an area that is free of trees and just grass, and it's got some slope to it, but it's gradual.

Okay, well, Travis is the one who has to walk in this field. Are you wearing boots or bug spray [08:44] because the site is, you know, there's tripping hazards out there. Yeah, okay. It's dynamic. New holes show up occasionally. Have you run across any surprises, you know, critters and things like that? A lot of times I've went down, tripped in a hole, get up, keep moving. Oh, no. In May and June, the fawns will lay really tight in the grass. You're talking about baby deer? Yes, yes. [09:14] move until you are darn near on them.

It startles you. The pheasants are the same way. So that's an interesting challenge. At least there's nothing dangerous. It's not like we have, you know, poisonous snakes or anything like that. I have a running tick count every collection. [09:35] Ticks are probably the most dangerous thing we have here, right? Yeah. How long has this site [09:44] the level two product for Landsat, so the surface reflectance product. It was decided that we should probably have a way to validate the surface reflectance product, so we started working towards that. In 2021, we first started working with the Aeros facilities folks on where we might try to do this.

[10:08] Part of the reason we selected the northwest corner was, again, directed by facilities, because at some point we thought that we may ask them to mow the site to make it more uniform, if you will. They still don't mow it, though. We're just letting the grass grow. Is that still okay? Does it still work well as a good site? Yeah, I think it's a good site for us. Again, we're trying to do validation, so we want the, if you will, the dirty sites. [10:38] Yeah, okay. Because, you know, calibration is where you want the best information possible so you can make the sensor work optimally.

But validation, we're looking for the dirty areas so we know how good our calibration is, if you will. What benefits does this add to the validation work? I guess the blunt way to put it is, why do this? We're producing a surface reflectance product. People around the world are consuming this product. [11:08] is considered the gold standard in terms of its calibration. We want to find out if there are issues, be it in the atmospheric correction algorithm or in any of the specific bands that Landsat measures in different types of land cover.

Vegetation is difficult because it's darker and there's not always a lot of return of the signal that the system is measuring, [11:38] That's why we're looking at vegetation primarily and trying to ensure that we're delivering a good product that's reliable. Besides Landsat, are there other times that you're out there taking measurements as well? Sentinel goes over every five days. The European Copernicus Sentinel satellites? Correct. Sentinel-2A or 2B and 2C at this point it is. Ah, okay. [12:08] are kind of a priority collect. You know, we're getting a good cross-comparison between two platforms.

And NMAP, being as it has a variable, it has a cross-track degree of about 30 degrees out from ground track, and so they overpass more often. NMAP is a satellite from where? That's the DLR Germany. And you do the exact same thing for all those? [12:38] same routine? Same routine. How much of this work is also in preparation for Landsat Next? You can view the supporting the Sentinel and NMAP missions as in some ways part of that because Landsat Next architecture will be quite a bit different and some of the bands will be very similar to the bands that Sentinel uses so it behooves us to understand those parts of the [13:08] bit of a different animal.

It's the same spatial resolution as Landsat, but it's a hyperspectral mission, which means it's collecting a lot more bands than Landsat will. But as we transition to Landsat next, we'll also have a lot more bands than we currently do. So again, we're supporting the commissioning of these European instruments by collecting data for them, but it's also potentially [13:38] At this point, since a Landsat pass was coming up, we needed to head out to the field test site. [13:50] It really couldn't be a nicer day for this though, right? Is there better conditions than we have today for these kind of collects?

[13:55] This is about as good as it gets in eastern South Dakota. Yeah, it's probably like 78 degrees. The sky is nice and blue. Not really any clouds to speak of. Nice and dry, too. [14:15] Travis, can you tell me about what you're setting up over here? Sure. We're using two reflectance panels today, and I'm setting up the A panel on the tripod. [14:25] by 12 inch reflectance panel with 99% reflectance. And they do tend to degrade over time, and so we keep them covered until... [14:35] Until needed. Okay. [14:41] Set up geometries all the same the same face at the Sun all the time.

So there's consistency to our setup and to our measurements then. [14:50] the same spot that you always set it up. [14:52] Correct. We've got one face of the box that is pointed at the sun, and it's a consistent measurement. [15:03] Thank you. [15:04] Thank you. [15:05] For those who aren't aware of where Eros is located, we're about 10 miles north of Sioux Falls, South Dakota, sort of in the middle of cornfields and soybean fields. Eros is on a half section of land, so there is some space to accommodate one of these test sites.

[15:22] Okay, to the extent that you can, can you tell me about what you're wearing? Sure. This is an ASD Field Spec 4 spectroradiometer. I've got it strapped to my back, but it's essentially the same spectroradiometer that is set up as the fixed base unit. This one is mobile with me. [15:52] Yeah. [15:53] It's kind of a heavy backpack. Does that weigh you down? Does that wear you out out here? It's not terrible, but you notice it, I'll say. Okay, tell me what you're holding in your right hand right there.

Attached to this painting roller stick. [16:11] Okay, we just rig up whatever we can, don't we? We've repurposed it, see. No, we've got an 8 degree 4 optic, which is where the light enters to measure. Okay, great. And I'll hold that out ahead of me, out of my shadow, so we're always getting clean light. [16:29] I like to think of it as, you know, it's the same as the Landsat satellite, only... [16:34] smaller in my hand. Only a meter and a half off the ground and just getting a really, really small area at a time.

[16:40] That's right. But it's measuring the same. [16:42] The same stuff. We're sampling what Landsat is getting, and hopefully what we're measuring is going to... [16:48] Validate. [16:49] what Landsat has. [16:51] Why did you call it an 8 degree? What does the 8 degrees mean? That means that there's an 8 degree... [16:56] A cone. [16:57] Um... [16:58] radiating out from the center of that four optic. That measures the cone that you're measuring straight down. [17:05] The Air National Guard is also busy. [17:10] We got time. We're just waiting for the satellite pass to start anyway.

There's the other one. There's usually two. [17:28] So this is looking straight down and then that's the size of the cone as the circle hits the ground. Correct, correct. Okay, sounds good. How far do you walk, do you figure? Okay. [17:38] 150 times 8 is 8, 1200 meters. Plus I'll forget about the little side steps in here. Definitely more than a kilometer anyway is what it comes down to. You're getting a little bit of a hike in here today, but the terrain looks pretty... [17:55] Easy. It's not bad. There's this, you know, I like to keep an eye on the ground because it can be dynamic.

It's not exactly flat. There's just a little slope. Yeah, not a terrible uphill. We've got a little minor ditch through that adds some flavor. [18:16] Do you walk the same path, the same transect? Do you have that kind of figured out? You have to walk this path every time. It's not like worn down, though. It's not terribly worn down. The grass is somewhat resilient. It seems to handle that. I try to stay consistent. [18:32] Being my survey background, I'm always counting my steps, and I can't really go anywhere without counting steps.

[18:41] Travis started by getting a reading of the white panels. What's the purpose of that? [18:46] So the purpose of that is to provide another known reference point as he's collecting, and then as he goes along, we'll have him collect this panel again occasionally. Oh, okay. And then we can do a, assuming the atmosphere is the same, we can do a linear interpolation between the two white panel collects. Sure thing. Okay. There's a couple other instruments out there. Can we talk about what those are? Are they related to what you're doing today?

Okay. [19:12] So those are what we call ground viewing radiometers. They were designed and built by the University of Arizona, one of our university cooperators. They are an eight band instrument. They're essentially doing the same thing that we're doing, except they're measuring just one spot. And each of those has a field of view on the ground that's roughly a four. [19:33] foot and a half [19:34] that it's staring at [19:37] so the all have solar panels on them and some [19:41] other instruments like a moisture sensor and so [19:46] They know [19:47] essentially when it's light out [19:49] one that's warm enough to operate and one that's raining.

And so, you know, given the proper conditions, they will take a measurement every two to five minutes, depending on how we program them to do that. [20:01] That data is all collected and processed over [20:05] The Internet. [20:06] cellular information and University of Arizona processes it [20:11] bit for us and passes it back and they give us a reflectance value. The challenge with those things is they're literally only looking at three different spots on the ground throughout the field where we're trying to get a more comprehensive view by walking. [20:28] transects through the field.

[20:29] Okay. Well, we're kind of gathering lots of data. Anyway, lots of different types of data. That's kind of the point here, isn't it? Yeah. Yeah. We're also experimenting with UAS through our colleagues at the National Long-Trude Systems Office in Denver, again, to see if we can expand this sort of work to... [20:51] areas that are less accessible for walking, [20:53] We also have university partners at SDSU analyzing similar technologies to the ground being radiometers that are small portable radiometers that a farmer could use in his field and are low cost.

And so they're looking at putting a calibration on those and seeing if they can do the same thing that way. [21:21] Well, he's got a little bit of a path here to walk down. [21:27] I wouldn't say he's walking fast, but he's moving along pretty good. [21:31] *sad music* [21:36] There are a few little pathways kind of cut through here like you've been here before. Yep maintaining our consistency. Okay, very good very good the uh, what do you call it at the end the four optic? That's correct. Okay, he's got that off to the right side a little bit So it's out of shadow [21:56] You're trying to walk at about the same speed all the time too, right?

That is right that uh We've got a roughly a five by five [22:04] pixel area and we try to get the same amount of shots per pixel. Oh, okay. And sometimes the instrument will mess up and, and, uh, [22:18] so I count my paces so I kind of know where it missed and I can go back if needed. [22:23] Oh, okay, okay. I'm messing up your count now, I realize. Sorry about that. Oh, you're fine. [22:29] You're just gathering as much data as you can. You can kind of check while you're out here to make sure it's good.

Yeah, yeah. We've got a little bit of time. We could correct something minor, but we'd hate to lose a full collect, you know. Right, right, especially on a nice clear day like this. Right. Well, you do have to watch your step at the same time, don't you? Yeah, the laptop isn't the best transparent device. [22:53] front of you. Yeah. [22:57] Some of the burrowing critters are... they'll get active and all of a sudden they might... there might be an obstacle in your path. [23:09] Yeah, you probably get used to these pathways out here [23:12] But, the wild animals, they don't really care.

No, no. [23:16] They don't understand. [23:17] - Yeah. [23:19] Okay, I'll leave you to it. Keep walking. [23:25] Thank you. [23:28] What were you doing with your phone there, Jeff? So I take... [23:32] Photos periodically throughout the collection so that if anything looks anomalous in the data processing, they can go back to the photo and see if we had clouds move through or something else. [23:43] Just a little more verification to be ready for. [23:48] How do we know what time Landsat is passing over? [23:54] There is actually a Landsat acquisition tool out there on the web where you can go and find out anywhere in the world what time Landsat will be overpassing your part of the world.

Oh, okay. That's public information, right? That's all you're using, right? Yes. Okay, cool. It's not like you can look up and see it. No, no. It's doing an orbit every 90 minutes. [24:18] I don't know what that translates into miles per hour over the ground, but it's pretty fast. It is fast. I'm always telling people it moves about 17,000 miles per hour. [24:28] Travis, on the other hand, isn't going that fast. [24:31] How long is this pass going to last? It's going to take us roughly 26 to 28 minutes to collect the entire field.

Oh, okay. At the same time Landsat's passing over. It's not like it's taking the image of this area for 26 minutes. [24:46] No, I'm just saying it's going to take us 26 minutes. It takes you that long. That's it. Okay. Right. [24:51] The actual Landsat Pass is only about 15 minutes over North America, something like that. [24:56] I believe that's correct, yeah. [25:00] you're not getting tired yet are you not yet okay good good [25:07] We're in the last mile. Last mile. [25:10] The home stretch, here you go. 469 NZ.

Got it. What did those numbers mean Jeff? So those are the numbers of measurements that he took during the transects and over the panel and we'll use those to process the data. I don't know. [25:30] I told you're not, but he's got a little GPS puck on his shoulder. No. And that's how we correct for where the measurements are located at in the field. Oh, okay. How accurate is that GPS? Plus or minus 10 meters. [25:43] Okay, okay, that's still pretty good. Yeah. Okay, that gives you enough... [25:46] enough accuracy for that.

Yeah. We're trying to compare to a 30 meter Landsat pixel. Definitely, you're within that. Alright. So I don't want to jinx it here or anything, but what if you find something that doesn't match in the data? What if you find some surprises in the data? We don't like surprises. So then, yeah, we have to dig in and track it down and try to figure out what happened, what went wrong, what it was. Yeah. [26:16] We don't often get surprises. It'll go back to the CalVal team anyway to look into to investigate.

Travels to process the data after we get back in, it takes about a half a day to do that. And then we pass it off to... [26:31] more of our analysts who [26:33] grab the imagery and look at the reflectance measurements from the imagery and compare that to what we collected on the ground. Yeah, okay. Okay, sounds good. [26:42] Thank you. [26:43] Thank you. [26:44] Thank you. [27:14] Check out our social media accounts to watch for all future episodes. You can also subscribe to us on Apple and YouTube podcasts.