AoR 46: Matt Reeves, Fuelcasting: A West-Wide Rangeland Fuel Assessment

[ Music ]

>> Welcome to the Art of Range, a podcast focused on range lands and the people who manage them. I'm your host, Tip Hudson range and livestock specialist with Washington State University Extension. The goal of this podcast is education and conservation through conversation. Find us online at artofrange.com.

[ Music ]

My guest today on the art of range is Matt Reeves. Matt is a research ecologist with the Forest Service Human Dimensions Program at the Rocky Mountain Research Station out of Missoula, Montana. Matt has been a scientist with the Forest Service for a while and has done a lot of work on practical applications for remote sensing data on range production and fuel, especially herbaceous plant amounts. Matt, welcome to the show.

>> Hi, Tip. Thank you for having me.

>> Matt is a repeat offender. We had him once on Episode 10, where we talked about some of the technology behind the effort that we're going to visit about today. And so if you want to hear a little bit more about that, you can take a look at Episode 10 at the artofrange.com. Matt, I think people are sometimes critical of scientists for, you know, having their head in the clouds, so to speak. Maybe that's especially true of climate scientists and remote sensing gurus, but I you've been trying to get that information to people in a way that they can do something with it in the real world, on the ground, and we're gaining significantly in, you know, the technical ability and the accuracy of tracking real time changes and things like forage or fine fuel depending on which way you want to look at it. So you've been doing a monthly webinar on changing fuel conditions called Reading the Tea Leaves. What is that about?

>> Well, Reading the Tea Leaves is a monthly webinar that is seven minutes or less. And it covers, as you said, the basic fuel conditions. And by conditions, I'm talking about fuel amounts that we are sensing. You know, it's a program that that uses remote sensing and daily weather information and also a drought index to estimate the yield of fuel. And that can be found on fuelcast.net. And the whole purpose of Reading the Tea Leaves is to evaluate the weekly changes in fuels and the projections about fuel on the Reading the Tea Leaves. So we are providing the analysis for the listener. And I liken it to, you know, it's kind of like a weather forecast from Al Roker or something except it's for forage and fuel. So it covers, you know, the US in seven minutes or less. It's not meant to be in depth. It is meant to give you a snapshot of where things are interesting around the country both in a positive direction. So the fuels are accumulating or in a negative direction, where fuels are much less than normal.

>> Yeah, I think that's part of what's interesting because, you know, fire risk has more to do incorporates all kinds of things, you know, conditions that maybe a rancher would see as positive also mean high fuel. So there's more going on than just, you know, how much standing stuff there is out there. You know, what are the some of the different ways that you can crunch that?

>> Well, we offer a variety of products on fuelcast.net, including the surface fire behavior fill models. And if you read or if you watch or listen to the last webcast, we discussed the linkage between fire return interval and fuel amount. And it's almost a -- well, it is a very good predictor in some of our rangelands of whether or not we're going to have a good fire season with above average fire activity. But that only works in some rangelands. And I'm calling those the, you know, in general, the fuel limited systems. So think about the Mojave, the Sonoran Desert country and those places where by the mid of summer, it is hot enough and dry enough to produce significant fire behavior. The key is whether or not we have enough fuel. So that's the one area in the country where, just by knowing that there's an abundance of fuel. If you look back in time, that's been a very good indicator fire activity, especially if you look back to 2005, where we had significant fuel and very significant fire activity, that doesn't hold in other parts of the country where they're not as fuel limited. And the fire activity depends more on the weather or the conditioning of the fuel. So that's one way to look at it. We also on fuelcast.net in the process, we break out the standing dead component. And that is a separate product in and out of itself, so people can kind of look at what the previous year's accumulation would be, and tie that into their fire behavior projections as well.

>> Is the standing dead based on a current image or a data analysis or is that based on a combination of what's being picked up right now, compared to let's say what was showing up in October in November?

>> No, it is it is an estimate of what was left over from the previous year's growing season, which of course varies around the country. But we track that and so you can imagine, you know, up in your country in the Palouse region, if we think about sage steppe or Palouse prairie, it's all the stuff that is, you know, brown in February and March that we are evaluating in that standing dead product. And that's has a direct linkage to how much was produced in the previous year. And of course, we get that through the arrangement production monitoring service. So that enables us to keep tabs on or estimate how much standing dead we think there is in a given system.

>> And what are some of the applications of that that land managers or livestock producers have been using or do you know?

>> Oh, yeah, I think there are three rapidly growing use cases that are worth noting. The first time would be for the long term analysts on fire events and the fire behavior analysts on fire events. We have a growing user community there. They seem to like the product because it gives them a pretty good indication of the fuels and what kind of changes have happened to the fuels in the last couple of months. So we had good success this fire season, good reports in terms of the accuracy and utility of those data, so that's one use case. Another use case is with the nation's fire decision support systems such as the [inaudible], the wildland fire decision support system and the FTD tools. Those are big, big programs, but they rely on accurate and timely fuel information which up until, you know, recently has been elusive. So that's another use case. The other use case that is growing is -- and I think more rapidly than any would be the use of looking at forage reduction. So flipping the tool on its head a little bit and evaluating this from a forage perspective. We're working with the NRCS and also significantly the FSA Farm Service Agency to develop some agency specific tabs on fuelcast.net, so that they can quickly evaluate the forage loss estimates for their area, which are official forage loss calculations will be available in mid November. And that is then used by the various analysts particularly in the NAP program, N-A-P, to provide a second unbiased estimate which is required by the legislation and the policies.

>> In real time say, you know, the information that you have available right now. Does that just provide standing fuel volumes or does it also provide some indication of how dry the fuel is in various regions?

>> No, it does not provide the -- how dry the fuel is, you know, other teams have been reasonably successful in that, although I should mention fuel moisture is one of the key variables that would indicate the potential for large wildfire activity. And I know that -- I don't remember the scientists but some NASA scientists have developed a fuel moisture product. We're looking at weaving that into the fuelcast.net for a more complete picture, you know, teaming up with them. It seems like they've cracked that nut a little bit. So yeah, it's worth looking at. I think you could Google it and find it. I just don't remember the exact name.

>> Yeah.

>> We're just primarily interested in the fuel amount. We do keep track of the phenological aspects of fuel, which can kind of be related to fuel moisture, but it's not one-to-one.

>> Tell me a bit about the data behind that. Is this coming from your rangeland production monitoring service or is there more than one data source feeding into the fuel cast?

>> Now, fuel cast is part of the rangeland production monitoring service, which as I mentioned a while back in a different podcast has two components. The first is the retrospective viewpoint from 1984 to 20 19, soon to be 2020. Fuelcast.net does not rely on that information. It's its own entity. It uses weekly E-motos. We get weekly remote sensing information. We get the daily precept, the daily eddy information which is the evaporative demand product from NOAA. And we allow up to two years worth of that information to influence the present weeks projection of conditions and the projection start four months in advance of the peak. So for example, in your neck of the woods, it would probably start somewhere, I don't know, in early February. And it uses that information that it acquires daily and then weekly, in a machine learning process to crunch the numbers and to decide, you know, for example, what did an inch and a half of March precept do over the last 36 years to our annual production? And so the computer unravels these relationships using machine learning.

>> Mm-hmm. And to what extent are you able to do any prediction from that at this point?

>> Well, the predictions, we will be hosting a publication coming up in -- Oh, I don't know, three to five months that documents the error rates that are accompanying those projections. And recognizing that there's going to be error because no one knows what the future holds, which we can do is estimate the uncertainty using a 95th percent prediction interval. For example, at a given pixel, we might estimate a thousand pounds per acre, plus or minus 250 pounds. And so that 250 pounds would represent the 95th percent prediction interval. So it gives us a sense of confidence, sometimes a false sense, but you can use that to get a handle on the uncertainty of the projection. I should mention and it's critical for listeners to understand. The closer we get to the peak, the more accurate those projections become. Likewise, you know, after the peak, it's no longer a projection. It's an observation, right. So that's an important distinction to make.

>> And the peak is different depending on where you're at in the country.

>> The peak is drift. That's exactly right. And we let the data speak for themselves. So if there are, for example, by modal systems to be accounted for, the way we deal with that is slightly different. But in our experience, usually, in those by modal systems, one leg of the mode, you know, one of the modes, let's say maybe the early one or the late one usually dominates the other. And you usually don't get two giant pulses. One early and one late, I guess it can happen, but it's pretty rare.

>> Mm-hmm. You mentioned the pixel and uncertainty or variation within a pixel. What's the spatial resolution of that tool or that data?

>> T he ultimate output is 250 meters on a side so that's 6.25 hectares coverage or roughly 15 acres for each pixel. We do have a 30-meter product that would be Landsat based that we are in the process of unveiling? It requires more computational power, but that's OK. Google can handle it, so 250 presently.

>> If a land manager or a rancher is interested in trying to quantify those fuel amounts, should they take a look at fuel cast or at the RPMS data? And what is the, you know, what's the user interface right now for the rangelands production monitoring service?

>> Well, there's a couple of different interfaces, but I encourage people just to look at fuelcast.net. And you'll see there's two tabs, one for the application and one that just says forget the application. I just want the data. So we serve both of those masters. If a producer has an interest in looking to see what's happening in their neck of the woods presently, they can, you know, zoom in. It's one of these interactive types of maps. We're in the process of working through our range limb technology partnership, which includes a consortium of producers and agency personnel to build out a few tabs. That would be, you know, producer specific, one for the Forest Service, one for FSA, and RCS, so on and so forth, because each of those entities has different requirements. And potential use cases, for example, a producer might have an interest in pastures. It might have an interest in different soil types to see, you know, what's been happening. And so therefore, we would enable that kind of viewing, say they want to differentiate their sandy soil production from their loamy soil production and how it's doing in a given pasture.

>> Mm-hmm. The other thing I wanted to ask you about is to what extent can you measure within your disappearance for lack of a better term? Is there anything out there that would allow someone to estimate grazing utilization? So, you know, say the satellite data on July 15th shows X and most of the area that the person is interested in was grazed for the whole month of June. What's the -- I guess, how -- would it only show up -- would that only be picked up by satellite data if it had been nuked? Or would it show up if it had been grazed say it, you know, 30% utilizations.

>> I know that the literature is not well developed in this area. And I know that ARS, I've recently spoken with some people in ARS that are working on this sort of issue. And it is my feeling that a reduction of yield depending on the ecosystem, a reduction of yield, I'm just guessing somewhere around 30% would be required to be assessed. Now, I can't -- I couldn't guarantee that. But that's in data that I have looked at, that's my general sense. The other thing to remember is there are other types of tools out there that will view two different pastures one side by side. And they can statistically compare those two pastures and determine if the yield reduction is significant. We have some code that is used in the Forest Service to help us determine that. And it's meant to identify things like, you know, significant prairie dog activity. It compares like kind of sights, and determines the statistical relationship between them to identify areas where the vegetation is underperforming. So in that kind of statistical approach, you might be able to, you know, quantify how big of a deal the reduction really is. But I haven't looked into the utilization aspect of this enough.

>> Mm-hmm. That's still pretty impressive. I was not aware of that.

>> Well, it's -- it -- the idea is born of the need to evaluate the performance of different, you know, pastures or paddocks. And so we want to know, you know, how is my area or how is my pasture compared to my neighbors? And this is an important question for especially public land managers, is, you know, being able to enter compare within their allotments and see, you know, are there significant trends? Or are there significant conditions where one pasture should be behaving in another way but it's not. And the only way we can do this is by quantifying that performance based on or compared with like kind sight.

>> Right. And then the kind of a large landscape, it's really not technically feasible to do it. I don't think using ground-based monitoring, you know, whether or not the ways in which utilization monitoring is used as a subject for a different podcast, which is a great idea. But the ability to read that, to detect that or quantify that over a large landscape using remote sensing, I think has a lot of potential just by virtue of the difficulty of trying to do that with any degree of accuracy on the ground, given the variability that's inherent in most of Western rangelands.

>> Well, that's exactly right. And you would have to control for the type of vegetation that may be occurring in the different grazed areas and the different phonologies. I mean, it gets pretty tricky, as people know that are in the business. For example, let's say you're in a tall grass prairie scenario. And you've -- it's a very good year. So two pastures side by side, you have reduced, let's just say 1,500 pounds on one pasture. Well, that's really not enough that doesn't get you below the detectable threshold of a lot of our indices to even sense any difference due to the, you know, the nature of some of those vegetation indices. So it is pretty tricky and I do believe that it's ripe for research.

>> I think you mentioned on your website or one of the emails regarding the Reading the Tea Leaves webinars that you would possibly be increasing the frequency of those webinars as we get a little bit further into the summer, are we at that point yet or is that not in the cards?

>> Well, it's not in the cards this season. This idea was born out of the request of some of our users, particularly in the fire community, where they wanted to have more timely summaries of the information. And that will start next, next year.

>> Got it. So for now, if people are interested, they can take a look at fuelcast.net is that right?

>> Yeah, they can Google fuelcast.net and look and see what's happening in their neck of the woods. They can also view any of the webcasts that have been performed. And if they wish, they can download data and bring it into their GIS, whether that's the surface for a behavior field models, the standing dead, the estimate of production, so the pounds per acre, but also the percent change product, do how am I compared with average. Any of those would be available now.

>> OK. This episode is probably going to come out toward the end of August in 2020, so not very far from our recording date. But can you give just a brief overview of what those conditions look like this year, the kind of thing that you'd be reporting in these webinars in case folks don't end up tuning in this year. Where are we standing right now in August, today's what the -- 14th, the 12, 2020?

>> Right. Well, can keep in mind that this does cover, you know, the western rangelands, so we're looking at somewhere around 600 and 50 million acres plus or minus that we're reporting on. So it's hard to get too specific but there are there are three notable trends. The first would be that on average, the Northern Great Plains had a good year. There was a good and timely moisture. Not everyone got it, but it was a significant year for the Northern Great Plains. In the southwestern US, it was a very good year. It produced a lot of fuel. And some of the Mojave Desert especially on into Southern Nevada, throughout Western Arizona had very, very timely moisture. And produce much larger than average yields some cases in excess of a thousand pounds per acre. Unfortunately, most of that, as, you know, would be invasive annual, such as red brome, which doesn't help a lot of things. Some emerging topics and scenarios I think are worth noting would be the conditions across most of Northern Colorado aren't good, especially in the Northwest and the northeastern part of the state, so those two corners. We see significant reductions in both fuel and forage occurring compared with the long term norms. And then there are areas in Central Southern Oregon, I think, where we've noted some pretty significant declines in fuel amounts this year compared with average. So I guess there's also a trend in Northern Nevada, we saw where, you know, we're looking at 15 to 30% reductions, sort of on average across that Northern Nevada Country. So I think those kind of capture the highlights of what we saw this season.

>> Well, we're going to be a little bit shorter than our average podcast length today, but it was kind of a tight topic and we felt like it was timely and wanted to get this information out to people about this fuel cast program even if folks aren't able to use it much this year. I think it's going to be increase in applicability. And we'd like to have it out there so people can use it next year. Matt, is there anything else that you wanted to cover or talk about with regard to this fuel casting system?

>> No, I would just mention that the user base has been growing. In fact, last week in one 24-hour period, we had 260 views. And the average time on site was four minutes. People are, I think, a little reticent to download the data overall, but, you know, you can't hurt it. If you just take a look at what's happening your neck of the woods, pull in those data if you do have the GIS. And I always appreciate feedback. So anything people would like to mention. I do get a lot of email and I would love to hear some feedback.

>> Great. And is your email available from that website?

>> Yeah, the contact information is there.

>> Got it. Matt, thanks again. I encourage people to go take a look at fuelcast.net and watch out for those webinars.

>> We'll copy that. Thank you, Tip for doing this and I hope to talk with you again soon.

>> Thank you for listening to the art of range podcast. You can subscribe to and review the show through iTunes or your favorite podcasting app, so you never miss an episode. Just search for Art of Range. If you have questions or comments for us to address in a future episode, send an email to show at artofrange.com. For articles and links to resources mentioned in the podcast, please see the show notes at artofrange.com. Listener feedback is important to the success of our mission empowering rangeland managers. Please take a moment to fill out a brief survey at artofrange.com. This podcast is produced by Connors Communications in the College of Agricultural Human and Natural Resource Sciences at Washington State University. The project is supported by the University of Arizona and funded by the Western Center for risk management education through the USDA National Institute of Food and Agriculture.