Eyes on Earth Episode 135 – Exotic Annual Grasses and Sheep Grazing

[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] EAGs, or exotic annual grasses, have degraded thousands of square miles of sagebrush rangelands in western states. [00:39] Eros, using Landsat and other satellite data, tracks every bit of it. [00:44] In sagebrush habitats, cheatgrass is one of the big ones. [00:47] It outcompetes native grasses. It has different seasonal timing or phenology than native grasses. What's worse is that it is very flammable. [00:57] It's also expanding into higher elevations. [01:00] Dr. Kelly Hopping, an associate professor at Boise State University, is studying whether sheep can be used for targeted grazing in these areas to help reduce the amount of cheatgrass. [01:12] The idea is to steer the sheep to the right place at the right time to do the most good. [01:18] She has been setting up plots to test this and she uses EAG maps and phenology maps from Eros to do this. We're going to talk about those maps and about the benefits of this study. [01:30] to the landowners and to the sheep producers, but one of the biggest economic benefits of this strategy is managing fire risk.

[01:38] Another thing that will come up in this discussion is NDVI, the Normalized Difference Vegetation Index, which is really just a measure of the greenness of plants. The team at Eros develops NDVI measurements from harmonized Landsat and Sentinel satellite data. [01:56] They are now in the middle of releasing weekly EAG maps. They do this for a 10-week time period during the season when EAGs green up. [02:07] Callie, introduce yourself real quick. [02:09] Hi, I'm an associate professor at Boise State University in Human Environment Systems. I have a project where I used you guys' products and mentioned it to someone at the Society for Range Management meeting recently, and I think that's why I'm here. [02:23] And thank you for doing this. Matthew, tell us about your role real quick. Matthew Riggi, I'm an ecologist here at Eros. I lead the RCMAP fractional component effort, also lead the EAG staff. [02:36] Trenton, will you introduce yourself real quick? [02:38] I'm Trenton Benedict. I work on the EAG phenology portion of the project, and I've been working on this for a few years since we started with the phenology. [02:49] Dev, go ahead and introduce yourself, too. All right. Thank you, Tom. My full name is Devinter the House. [02:56] Everybody call me Dave. And I'm working [03:00] One of the person who worked on the easy mapping project we have here for the last almost a decade, more than a decade now actually, I think it starts in 2014.

[03:10] Kelly, can you introduce us to the study and what's the problem that you're trying to solve? [03:16] So we're trying to solve the problem of cheatgrass invasion. Cheatgrass is an exotic annual grass that was introduced to the United States in the 1800s and has spread throughout the West and now spreading quite rapidly in sagebrush ecosystems in particular. And there's been some work showing that it's also moving up in elevation into places where it wasn't previously. So we have a project looking at whether we can try to help manage cheatgrass [03:46] sagebrush ecosystems and trying to use grazing as a way to potentially do that. [03:51] Give us the brief rundown on what cheatgrass is. What's it like? Why is it bad? [03:57] Well, it's an annual grass and it has a little bit different phenology or seasonal timing than a lot of our native plants here. And so when it comes into an area, it competes really well with some of the native vegetation once it's established. It can start to turn an ecosystem from being one with more diverse native plant community to being more dominated by cheatgrass. [04:27] this really good dry fuel that's very flammable that can carry a fire. So if there is some sort of ignition, it's more likely to take off when it's got all this tender essentially in the ecosystem. And then after a fire, cheatgrass comes back quite readily. And that's where a lot of that competition with the native plants can occur. And so it can be sort of a slippery slope where the more invaded an ecosystem is, the harder it is to get it back out of that state toward a healthier

[04:57] try to prevent it from getting all the way to totally cheatgrass dominated, we might have more chance of restoration success. Anyone who recreates in it at the wrong time of year is familiar with getting the pokey seed head stuck in their socks or their dog's ears. And so it's something that we ideally would like to keep from getting to new places, but it'd be great if we could also learn to live with it and manage it in places where it already is as well. It's not possible to eliminate cheatgrass, is it? Or is it just more about managing it? [05:27] I think the general consensus at this point is it's probably more about managing it. We can try to protect core areas of sagebrush habitat that aren't already inundated and try to keep them safe from it, essentially. But there are other areas where we're going to have to figure out how to live with it. [05:43] You had mentioned earlier about how it is expanding into higher elevations. Can you talk more about that? [05:50] Like here in Idaho, we have the Snake River Plain, which is in the Great Basin. It's lower elevation. And if you ever look at a map of cheatgrass, it's just this red swath across Idaho, southern Idaho. And so that's what comes to mind for a lot of people is just these kind of lower elevation like carpets of cheatgrass. But where I'm working in the Sawtooth National Forest, it's a bit higher and it's not maybe where people classically think of it being. And yet even there, there's hillsides that are just covered with it. [06:20] was a big turning point was that there was a really large wildfire there in 2013. And so a lot of places then where there might have been a bit of cheatgrass there before, that was a turning point for the ecosystem to become much more dominated by it.

[06:32] Why not use cattle? Why use sheep for this grinding? [06:36] Yeah, that's a great question. And there's been a lot more work done with cattle grazing cheatgrass because they tend to have grasses as a larger portion of their diet, whereas sheep are also eating forbs or wildflowers and shrubs. And so they're not maybe the obvious choice. But as cheatgrass moves into these higher elevations, the terrain tends to be a bit more rugged. And so when we're on public lands like this, the area where I'm working, it's only permitted for sheep grazing. It's not considered suitable for cattle. [07:06] And so we had the idea that sheep are already there. They're permitted to be there. So kind of administratively, it's an easier lift to be able to work with them. But then another thing that I think is so cool about sheep is that they have herders with them. And so without having to add any extra infrastructure or fencing or virtual fencing or any of these other kind of technologies, there's just a person and dogs who can guide them where we want to target on the landscape. [07:36] that's why we thought it'd be worth it to try and see if we could do it with sheep. [07:40] Sheep don't need fences. No, they just move with a person or a couple people and dogs. All right. [07:46] Do sheep like cheatgrass? I can't help but imagine them being brought to this area and they look at each other like, wait a minute. [07:54] This isn't what we normally eat. Are they okay with it? [07:57] Yeah, it's a great question. We have spent a lot of time staring at them saying, do they take a bite? Do they take a bite? And the herders have said that the sheep don't particularly love cheatgrass, but they'll eat it if other stuff that they prefer more isn't available.

[08:27] And, uh, [08:29] be part of the team, really. So not only are the ewes who are coming back year after year learning where to go, learning that they're supposed to, to some extent, stay in the areas where the herders have been holding them for the plot. So it becomes more a question of how do we time it? Because for sheep and cattle, they don't want to eat it once it's too far along in the season. It starts drying out. The seed heads become really spiky and unpleasant. So nothing wants to eat it at that point. But in addition to timing it just where it's green and palatable, it's also a matter of [08:59] to them and trying to hit windows where the cheatgrass is a relatively better thing for them to eat. And so in these higher elevation systems, there are a lot more other diverse plants. And so they do have choices. We're not putting them into a place where there's absolutely nothing to eat except cheatgrass. And so the challenge is, can we get them to have enough of a treatment effect on what we want to target without going overboard and having unintended consequences for the native plants? And so that's a lot of the balancing act that we're trying to do. [09:29] can keep working on but I think there's a quite a trampling effect occurring as well so it goes beyond just what they want to eat but also so indirect um or non-consumptive I should say impacts on tea grass because sheep are smaller and they're really gregarious or like to stick together um you end up having many more hooves in a small area than you do with cattle and then another difference is the lambs when they're there with them in the spring they have quite different

[09:59] them a lot. I've heard multiple different people refer to them as like gangs of teenagers running around together and you do see it like you'll have the moms, the ewes are being pretty calm and then the lambs are just like racing around in little packs and so I think that [10:13] All of that adds up to differences that are, I really do think, impacting the effect they're having on cheatgrass. After they've grazed an area, can you tell if the native grasses are coming back? Are you seeing benefits in the areas where they're grazing? [10:28] Yeah, the way that we set up this study was that we have a baseline year in 2022 before we did any grazing treatments. And then we did fall and spring targeted grazing. And then we're heading into our third season. But right now we have two years post-treatment in hand. And the weather and the phenology is very different in those two years. The cheatgrass production is also very different in those two years. So it sort of depends on what year we're talking about. [10:58] grazing in both years by around like 12 or so percent, 12-14 percent. But where we saw the biggest reduction so far was with the fall grazing. So something we didn't talk about earlier, but cheatgrass can green up again in the fall when the native plants. [11:11] pretty much are on their way to dormant. So if the rain comes at the right time, that cheatgrass can green up. And one of our years, we hit that window where it rained, we got the sheep out there, and they were able to reduce the cheatgrass by around 20%, which we were pretty happy with, because the first year when it wasn't green, they really didn't have an impact on much of anything. And so I think we're starting to learn a lot about how we can time it most effectively. So we'll keep

[11:41] make management recommendations that could say, okay, when the conditions are like this, it's worth going and doing it. If not, maybe not. So, so I think that we're, [11:50] not only learning a lot, but we're starting to get some good outcomes from it that we're [11:55] Making a dent. [11:57] I think it's time to turn to talking about the data a little bit more now. Dev, what data goes into these EAG maps? The first thing is we use Harmonized Landsat Sentinel HLS. [12:11] Okay, both Landsat and Sentinel data. [12:15] Those are harmonized, developed by NASA weekly at GELSA, actually. So we grab... [12:21] individual scenes and create weekly NDVI from those. [12:26] individual scenes. That's the main input for us. Okay. [12:30] We also have other soil data, weather climate, innovation, DEM, [12:37] solar radiation, [12:39] So multiple data set goes into it. [12:42] Okay, it's not as simple as saying we use satellite data, we use just Landsat. There is more that goes into it than that. So that's the work that goes on. And you end up producing EAG maps weekly. [12:56] But that's only a certain time of year that they're weekly that you produce, right? Yes. It's only from the... [13:02] peak growing season from early April to late June. When we started, when the project started, we had only two maps, one in the May, first week of May, and the first week of July.

[13:16] So from there, we started bi-weekly and weekly. And from last year, we are developing 10... [13:23] weekly maps from April to July. [13:27] Every week for 10 weeks during this time frame. That started last year? [13:31] Yeah, that started last year. Are these maps showing just the presence or absence of cheatgrass? Actually, this is... [13:39] The percent cover, that means... [13:42] 30 by 30 pixel. [13:44] How much is the sear grass? [13:47] How much of cheatgrass is in a pixel? [13:50] Yeah, in a picture. Not only chillgrass, chillgrass covers around 90% of the total EEG. We cover other... [13:59] other 15 different species. [14:02] Oh, other species as well. Okay. But Chikar is the highest one, so we can call it Chikar anyway. [14:08] Okay. In that 30 by 30 meter pixel, how much is the EEG grasses? Okay, I get it. Is there any use of AI, artificial intelligence, in the production of these maps? [14:20] What we get from the satellite will have very often, especially in this early spring, [14:27] gaps because of the clouds and snow and everything. So we use artificial intelligence model to fill those gaps so it can learn from the previous years and all the different weeks and tell us, okay, what's [14:41] what should be there at that time of the year. [14:44] or that week. So we use that kind of modeling to predict the NDVI, and then that feeds to the EEG modeling. Let me ask Trenton, if you have anything to add to that part of it, or is there anything you can add about the maps themselves, the work that you do on that? Yeah, so we developed another machine learning.

[15:05] model that tries to estimate the phonology of [15:09] our 16 group of exotic handle grasses. These maps produce our phenology, which is... [15:18] Basically, the sustained growth of these exotic animal grasses, which show when it starts to green up, [15:24] how long it stays green for, when it greens down, and then everything in between, those greenness. And it's about... [15:33] eight different products, the timing, and then we have NDVI products for when that timing occurs, what the NDVI looks like. [15:42] So the NDVI basically shows when it starts to green up those profiles. [15:49] And then their maximum greenness in the NDVI. And that's also with the HLS 30 meter pixels. [15:57] The phenology part of it, is that in just the weekly product for this 10-week time frame? Sorry, that gets produced annually. That's annual? Yeah, so we try to produce it at the end of the growing season for the year prior, but I think we might start looking into more of a near real time, predicting it or estimating it like the year of maybe. But we're still in the works of looking into that. [16:27] that just a prior year that we're going to produce just so we have the full NDVI yearly profile and produce based on those NDVI profiles basically.

[16:39] Kelly, how were the EAG maps used in your study? So when we started this project, we had selected two grazing allotments, which are pretty big areas. And one in particular is a very, very big area where the sheep are spending anywhere from like a month to several months. We had a lot of criteria on our plate of what we had to consider when we were choosing where to do it. So we had to think about just what's good experimental design. [17:09] And two of the big ones were we couldn't put them on slopes that were considered too steep to be suitable for sheep grazing because we didn't want to have like creating erosion problems and that kind of thing. We also needed to be a certain distance from trails. The sheep have guardian dogs with them. And so there's a lot of recreation in this area and just in general, wanting to not put the sheep and the dogs and all of that right on top of a trail is another priority. [17:39] the extent that we can. But then the sheep themselves have needs of a big one in their case is needing to be able to access water daily. Also from the science perspective, like we have a research team that needs to be able to actually get to these places. And so we're looking at these huge allotments and trying to think of how to prioritize all these different things. And so I loaded up a bunch of geospatial layers of where are the trails, where's the water

[18:09] sort of clip out everything that's above a slope that's considered too steep, and then had a number of these fractional vegetation cover products to try to know where is the cheatgrass that actually falls in meeting all these different criteria that we have. And this EAG product for our area was really, I think, the best, the most accurate. But this EAG product worked really super well for being able to go to a place where it showed a high cover of cheatgrass, and then sure enough, [18:39] did just a tremendous amount of hiking that first spring before setting it up and it made all the difference in the world to be able to know where to go and have a pretty good chance of finding something that would fit our criteria. It was just tremendously helpful for that. It made a big difference for why the whole study is designed the way it is. [18:57] What are the economic benefits? There are a lot of different avoided costs that could come into play here. And so if you do it this way, where we're just using sheep that are already there and herders that are already there, then that's essentially cheaper than spraying herbicide or doing some other sort of treatment, which would take a lot of human labor, might not even be feasible at all. And then because we're changing the timing of when they're out there, in our case, they're coming earlier in the spring or staying later in the fall. So they're getting access to areas that [19:27] of their normal permit dates. So that's a huge benefit for them, like at a time of year when they might, in some cases, depending on the setup, like feeding them hay back at a ranch, they're now able to stay out on the land longer because there's this different phenological window that's now available to them to do the grazing. And so that lowers cost.

[19:45] And then the biggest one is that if this could be successful in reducing wildfire risk, then you've avoided the tremendous cost of potentially managing those fires and restoration and the indirect toll that it's taking on society to have large areas burn in catastrophic fires. And so we are looking at whether these treatments are reducing wildfire risk. And I think that's something that we'll see build more slowly over time because we're not trying to do it as a fuels treatment. [20:15] a restored ecosystem that would be less likely to burn or to have a fire go through it quite as quickly. And so if we can [20:24] keep chipping away at it and avoid those large, more frequent fires and the economic benefit of that would be huge. [20:31] Is there a benefit to the sheep themselves? We were really concerned that we didn't want to put them at a nutritional deficit. And this is a particular concern when the lambs are out there because raising a lamb to a certain weight is the goal of the sheep owners. And so we don't want there to be a negative impact for them at the end of the season if their lambs end up lighter as a result of participating in the study. And herders are really concerned about that. They're really thinking about their animals' health and wanting to prioritize that. [21:01] them on a patch of land in a plot for a really long time. Instead, they're kind of making multiple passes through the area. So they're spending a lot of the day grazing the way they normally would, choosing what they normally want to eat. But then they're making sure to come through the plots in sort of a more concentrated than usual type of way, but still not just holding them there for really long periods. It's more a matter of trying to just make sure that they stay healthy and really keeping a close eye on them, which has been working so far.

[21:27] It's fun to just think of them out there just having fun helping us out with this study. Yeah, I heard someone who's doing this like as a private operation where it's like people can hire her sheep to do targeted grazing in California. And I heard a podcast with her and she talked about the sheep as part of the team. And I really liked that because it's how it feels like, yeah, we have a team of, you know, thousands working on this. And it just happens to include a lot of sheep. [21:56] Are the sheep herders on board with this too? Are they cooperating with this study? [22:02] Yeah, the sheep herders and the sheep owners that call producers, like basically the ranchers, they've all been really, really great to work with. I think for the sheep owners, they have a big incentive to do this to try to help restore these areas and hopefully lower the wildfire risk because that can be devastating. And so to be part of the solution, I think, is highly motivating for them. And just to be able to use their animals for something that's useful to protect. [22:28] the world as well, I think is attractive. And then the herders, they've just been great to work with when we've done the interviews with them. There's even some plots where [22:38] the combination of spring and fall grazing that we do, like some of those plots are on some areas where they're reducing the cheatgrass and there's not as much native plants left. We're getting starting out a lot more exposed soil, which isn't necessarily a bad thing if other plants can come back in. But the herders are telling us, like, we're concerned that, are you going to make us do that again? Like, we're not sure that this is the right move for the environment. Like, we really care about the impact we're having with the sheep. And so they're totally cooperative and

[23:08] They're keeping an eye out for, oh, maybe we should back off on these plots. I think we're having too much impact. Anyway, they've been really... [23:17] amazing to work with as well. It's been a really nice collaboration. And there's also many other scientists and Forest Service partners, and it's a really large team that's working together to make this happen. We did our best to set up this study from a scientific perspective and a Forest Service perspective and incorporated the ranchers' perspectives to an extent. But if I had it to do over again, or if this is going to be something that others could use as a management tool, I think working even more closely with the sheep producers and talking with the herders about what will [23:47] where will the sheep go? What can you realistically accomplish with your herding? I think then it could just get even better. Kelly, your study seems to be getting a lot of attention. Why do you think that is? [23:58] I think that at least around here and probably in other places where cheatgrass is a big issue, the public is aware of it and cares about it and is concerned about it. So I think having a positive perspective, [24:10] story in the sense that it's a pretty diverse group of people who's working together on this, trying to come up with a solution. I think that's maybe interesting to people. I think people are kind of interested in sheep. And so to have all the sheep out there with their lambs, it's, I think, also intriguing to people. And it's just a really beautiful area where a lot of people go to recreate. They care about it. People who live in that valley are right up against the

[24:40] What's next for the EAG data? Are there any refinements that you're working on? You're in the midst of this 10-week, you know, weekly release right now. After that's done, are there improvements, refinements, and things like that that you will make? [24:55] Absolutely. It's always an ongoing process, actually. For this year, we are on week five right now. [25:02] processing started this morning. So yeah, okay. And once we are done with this expedited release, [25:11] And we will spend another few months making it better in every aspect. [25:17] accuracy and everything. Trenton, does that sound good? Is there anything else that you have to add in that area? I'm hoping to try to get out the phenology products a little sooner in the years. I think we might try to [25:30] Expand our study area, too. Maybe utilize the RC map study area, and that will increase more of our range and our elevation as well. Just always trying to refine that phenology work, trying to make sure everything looks how it should. But, Kelly, if you ever see anything lining up correctly, incorrectly, reaching out really helps. And, yeah, just... [25:57] Making sure everything looks green when it's green. [26:00] This is a little silly to say, because like I work with remote sensors, like I know you're normal people, but it's a little bit like celebrities. Like you're the ones who made this product that is so useful. And to know that you actually invite, you know, ecologists out on the ground to interact with you about it is is really helpful to hear because it doesn't even occur to me until I met you and talked with you about it. So thanks for saying that.

[26:24] Well, thanks for calling us celebrities, but I do not feel like a celebrity. We appreciate it. We certainly welcome, if you've got sort of Landsat equivalent vegetation data, [26:36] That sort of thing can be integrated into the model. Like you said, Kelly, it's hard to get every pixel just right. There's a lot of pixels to get right. [26:46] And these models are [26:48] trained over a large region, [26:51] there's trade-offs made that way as well. I mean, if you trained a model for cheatgrass over just the Sawtooth National Forest, it might perform somewhat better than a regional model would. But then you've got regional discrepancies between models, and that creates a mess for that sort of scale analysis. And remote sensing is always a game of trade-offs. I was curious if you would ever consider doing something similar for the fall to catch if it is greening up again. [27:21] And what we're seeing, it could be [27:23] really helpful from a management perspective to be able to have that data to complement what's going on in the spring, because it's such an important window in the lifecycle of this particular grass. [27:33] Yeah, we tested, but we don't know how good we did. [27:37] But we just did last fall. In some ways, the fall map might be easier than the spring map because there's less other grasses green at that time. Does that track with your observations, Kelly? Yeah. [27:51] Yeah, it does. It'd be really neat to have that as a product if it's possible.

[27:56] Yeah, I mean, that's something that we can certainly look into. It sounds like, Dev, you've done some research on that that I wasn't even aware of, so that's good to know. [28:05] One year we had the pass that the sheep needed to walk over to get to the study area didn't melt out as early as we needed it to. So the sheep got there kind of late by calendar days. But that also meant the phenology of the area was later that year. And so being able to look at the phenology product, the EEG phenology product, has been really illuminating to show what we think we're seeing on the ground, [28:35] every year that's going to be the most effective. It's really timing the grazing treatment to when is the right time for the plants. When is it going to be where they can graze or trample the cheatgrass most effectively and then hopefully minimize the less desirable impacts to other native species. With these kind of products, we can get closer to making those recommendations for other types of livestock like sheep or other areas and try to really hone in and understand [29:05] didn't even try because, you know, you missed the window or the window wasn't here this year. And so it's not worth putting all the resources into getting all the animals out there in a year where it's not going to work. Conversely, how do we target it in a year that is potentially going to be having a high impact, like how to get them out there at the right time to accomplish the desired effect, then make more useful recommendations for others wanting to apply it.

[29:29] It's just so helpful to have this product ready to go. I think if we can use studies like ours to identify the right window of when the treatment can be effective, then managers could really use these products to make decisions, like seeing what the cheatgrass is doing from year to year, and then know the right time to send out the livestock to have the most beneficial impact. So I think it's really exciting to have this kind of work available. [29:59] Thank you, David. [30:29] of the U.S. Geological Survey, Department of Interior.