Eyes on Earth Episode 109 – Fall 2023 EROS Poster Session

TOM ADAMSON:
Hello everyone, and welcome to another episode of Eyes on Earth, a podcast produced at the USGS EROS Center which celebrates its 50th anniversary this year. 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, and on today's podcast we'll hear from a handful of the EROS staff members and university graduate students who took part in our fall poster session. A poster session is essentially a way for scientists to share their work with their colleagues in a public forum. The posters in the EROS atrium for our session covered a variety of topics on remote sensing, invasive species, mapping cover crops, drought monitoring, and more. I pulled aside a few of the authors during their site visit to hear more about their work. First up, we'll hear from a student who studied the benefits of using remote sensing to map cover crops.

ADAMSON:
Go ahead and introduce yourself.

BELINDA APILI:
Uh, so my name is Belinda Apili. I'm from the geography department, geography and geospatial department, South Dakota State University. My research is on winter cover crops.

ADAMSON:
You might have to explain real quick what a winter cover crop is.

APILI:
So winter cover crops are crops that are planted during the off season, when farmers are no longer planting their crops. So the land is plain. It's just left there bare. So farmers tend to plant these crops to conserve their, you know, their land. So the winter cover crops gets its name from cover, right, because it covers the land. It's to protect the land during the winters, right? So my research is basically using satellite imagery to do to map and then to detect and map winter cover crops over a period of seven years to see the changes that have happened in as far as adoption of winter cover crops is concerned.

So, the USDA and the Natural Conservation Services have been trying to encourage farmers to adopt this practice because of its sustainable and known benefits. These cover crops like perform a lot of things. For example, they help to prevent the lands from erosion. They reduce effects of erosion. So when ice is melting off, so it protects the land and keeps land away from erosion effects. It also adds more organic matter to the soil. It's also known for something called carbon sequestration, so carbon sequestration is like some of these carbon crops are known for extracting carbon from the atmosphere and then taking it back to the soil which is climate resilient. So it's also known for fixing nitrogen to the soils, so they really do a lot to the land. So that's why the USDA have been really trying to encourage farmers to adopt to this sustainable practice.

ADAMSON:
It sounds like it makes it better for the crops when it is the growing season. OK, lots of good things with cover crops. In your study, you mapped cover crops. Is that what you did?

APILI:
Yeah. So I'm basically using satellite imagery. So before like other some studies have showed that surveys have been done before, like going during physical survey. But that has shown that it is kinda expensive, time consuming. But with the advancement in remote sensing gives us the opportunity to just map these cover crops using satellite imagery.

ADAMSON:
Are you aware of who might find this valuable?

APILI:
Of course. The farmers themselves. Of course, USDA themselves. It will be easier for them to monitor the usage, since they're the ones trying to encourage this practice. So it will be easier for them to monitor if farmers are really adopting to this practice as they want because it's really valuable.

ADAMSON:
What's the best part of your results?

APILI:
I'm using what we call random forest classification method, so it's like a machine kind of learning a machine learning classification method. I'm using both ground truth, ground truth data and with the satellite imagery. So as you can see here, I collected my samples last year in November, cover crop samples, and I'm training my model based on three classes that is basically cover crop, noncover crop lands, and then the noncrop lands. So the noncrop lands are basically lands like urban roads, things that are not croplands, right. Basing on so far the samples that I have, I trained my model and then my results show that the model—this is my accuracy assessment—so it really shows that the model is performing well.

ADAMSON:
The model is doing well. That will mean this mapping that you're doing will be valuable.

APILI:
It will be applicable and it will work out.

ADAMSON:
This is really great. Thank you for talking to me about this. 

APILI:
Thank you so much.

ADAMSON:
Our next student also studied cover crops and other agricultural practices used to prevent soil erosion using the European Space Agency's Copernicus Sentinel data. So tell us who you are.

JAIN:
I'm Khushboo. I'm from University of South Dakota. I'm currently a PhD student there and I am doing work in agricultural sustainability. I'm trying to understand different management practices which farmers are doing across eastern South Dakota. Agriculture is the number one industry in this area. More than 89% of the land is under agriculture here, and people are mainly doing corn and soybean. So since the Dust Bowl like which happened in around 1930s when a lot of soil was eroded, NRCS Soil Conservation Service was formed, and they advocated the use of no tillage. Earlier, farmers were doing conventional tillage, in which the soil is disturbed a lot and which decreases the soil organic matter. It decreases soil fertility, decreases the soil microorganisms, and it has caused a lot of soil erosion. So they have been advocating the use of minimum tillage, no tillage around this area so that it can help in increasing the soil organic matter over time. And also it can help in maintaining the soil cover even in regions which have very high soil erosion.

And one of the major practices which is advocated with this no tillage is cover cropping. So generally in winter months people don't grow any crops around here, but if they are growing any crop, it will help in preventing soil erosion because there is a cover on the ground for the for the whole these winter months when the soil is generally left barren. So that's why my study is focused on understanding the distribution of these practices across eastern South Dakota. So currently I have been mapping these practices for 2022 and 23 and I hope to scale down my factors to 2011 because the impact of these practices can be seen over the years. It can take five to six years for soil organic matter to build. So I want to look at how these practices have helped in building the soil organic matter, how the distribution of these practices have changed over time, and what factors might be impacting these changes. For example, I have noticed that regions which have very high precipitation, generally, the people around there are doing more conventional tillage. They're disturbing the soil more around those regions, because if the soil is too wet, they need to dry up the soil so that they can do plantation. So in conventional tillage, the soil dries out quicker. So I'm trying to look at the factors which are affecting the distribution of these practices.

ADAMSON:
OK. How are you able to measure that?

JAIN:
I'm using satellite imagery, so Sentinel-2 satellite is available at 10-meter scale, so my prediction is at 10 meter and then I'm using it, converting it at field levels, all my predictions are happening around field level around this region, so I can say that this farm, for example this region is doing no tillage. And then this one is this farm is doing like minimum tillage and this one like this whole plot is doing. So I'm I can measure how many farms are doing conventional, minimum, and no tillage using satellite imagery which is available online.

ADAMSON:
It's freely available, so you can make use of that. It's a lot quicker that way than going there on the ground.

JAIN:
Yes. Yeah, that's true. That's why that, that that's why satellite imagery is so helpful. And I'm doing also some field data collection to validate whatever I’m mapping is correct. So my field data is helping me in validating my observations. 

ADAMSON:
There is still some field data involved. 

JAIN:
Yes, yes. So I need to validate whatever I'm predicting is correct. So field data helps me in establishing that, OK, if I'm predicting conventional tillage, it is actually happening in that place. And I'm not giving out a wrong output out in the world. Yeah, it's helpful.

ADAMSON:
OK, that sounds really great. Thank you. Our next graduate student uses MODIS and VIIRS phenology products. MODIS and VIIRS are sensors on NASA satellites and data from those sensors are distributed from the EROS Archive. These products can be used as a proxy, or secondary data, for field crop progress data. He also mentions NASS, the National Agricultural Statistics Service, which is part of the U.S. Department of Agriculture, or USDA, who could find his work valuable for validating field data.

MALIK:
Hi, I'm Naeem. I I'm a PhD student at Department of Geography in South Dakota State University. I joined the PhD program in fall 2021 and now I have been working on using remote sensing satellite data from MODIS and VIIRS phenology products and to use them as a proxy for field crop progress data. So usually what happens is the NASS of USDA has to collect a lot of field data, field crop progress data, on number of crop phenological stages how much percentage of that crop has achieved a certain stage. So that is reported by around a lot of reporters around all the States and that is accumulated and that percentage is reported and published on a weekly basis, but that is a very hectic task. To overcome that, we used MODIS and VIIRS phenology products to and develop relationship between those data sets and the crop progress data that is collected from the field, and we concluded that those data sets that we produce from the MODIS and VIIRS data, they are more comparable with the field progress data and these can be used as a proxy to overcome those challenges.

ADAMSON:
OK. Do you have an idea of who might find this data valuable?

MALIK:
Yeah, the NASS itself, which is collecting the crop progress data, so it can be benefited from our results because that equations can be used to predict for the number of years to come. For example, we use this linear equation that we developed across from MODIS from 2001 to 2021, and for VIIRS from 2013 to 2021, we used those equations to predict for the year 2022 and the results were quite good. So these equations can be adjusted and can be modified as well and can be used for next coming year for 23, 24 and we can refine those equations as well.

ADAMSON:
OK. That sounds good. Thank you. The VIIRS sensor has also proven useful for modeling evapotranspiration, or ET, for monitoring drought conditions around the world. OK. I'll let you introduce yourself.

KAGONE:
Hey, my name is Stefanie Kagone, and I'm a research scientist here at the EROS Center as a contractor for ASRC.

ADAMSON:
OK. And tell us about your poster. What is your study about?

KAGONE:
So my poster’s titled actual Evapotranspiration modeling and Mapping for global drought monitoring using VIIRS thermal data. And I work for a project called Famine Early Warning Systems Network where we model evapotranspiration, or short ET, using the SSEBop model and other variables for monitoring and assessing the extent and severity of droughts in the United States and around the world.

ADAMSON:
And tell us how this is valuable.

KAGONE:
So we really we show the here we show the newest updates to the model and including the VIIRS thermal data instead of data from the MODIS sensor due to the decommissioning of the satellites. We also updated the model to improve our ET estimates using the forcing and normalizing operation or pheno approach in areas where water and land are not easily distinguishable by the remote sensing images that we use. ET estimates over time show reliable results on the, especially on the poster for Ethiopia should be showing here to monitor possible drought conditions from year to year and all this data really and much more is freely available from the USGS FEWS NET data portal. It's available on climate engine and ESRI's living Atlas for the world.

ADAMSON:
OK, this sounds really valuable. Thank you. We are talking a lot about the data that these researchers are using. Much of it comes from the EROS data holdings. So let's hear from our customer services team at EROS about how to get the vast amount of data that EROS offers.

MILLER:
Hi. I'm Abby Miller. I'm from user services and I'm going to be talking about USGS data access in some of our cool tools and websites that we have. 

ADAMSON:
At EROS here, we have a lot of data. It looks like this is how you get the data. 

MILLER:
Yes, exactly. So this is just breaking down some of our really cool websites and tools. We talk about EarthExplorer, our bulk downloading web application, machine to machine, and even the commercial cloud data access, and overall it breaks down some of those links. And if you're not familiar with the tools, it gives you a great overview of kind of some of the ways you can get data that you might not know exist. And it's a great way— user services is really working on putting out some awesome content webinars updating some of those YouTube videos on USGS trainings and if that interests you, you can always write to join our listserv at custserv@usgs.gov.

ADAMSON:
That sounds great. Thank you. Invasive species turned out to be a common topic at the poster session. A USGS scientist describes mapping exotic annual grasses in sagebrush areas.

BOYTE:
My name is Steve Boyte. I'm a research geographer at the USGS EROS Center.

ADAMSON:
OK, now I see in the title of your poster it says something about Exotic annual grass. Can you explain what that is?

BOYTE:
An exotic annual grass in this case is a grass that is non-native in a sagebrush biome and a little beyond in the in the Western rangelands of the United States.

ADAMSON:
OK. It's something that maybe isn't supposed to be there. It's not native.

BOYTE:
Right, it's non-native it. It doesn't really belong.

ADAMSON:
What is your study about then in relation to these non-native grasses?

BOYTE:
So basically what we're attempting to do is to generate, in an expedited manner, reasonable spatial estimates of exotic annual grass invasion. That would be the distribution and abundance, and we do this so the land managers and fire personnel have access to early growing season, fine fuel load data that is contributed by these exotic annual grasses. And fine fuels are extremely ignitable and they spread if they're in a in high enough abundance. They spread fire through these sagebrush biomes or these sagebrush biome very effectively.

ADAMSON:
Why is that such a bad thing?

BOYTE:
Well, fires are bad thing because it destroys the native plants. These native plants provide habitat for native wildlife species and some of the species. There's about 7 species that are obligate species and they require the sagebrush to survive. The Bureau of Land Management uses our data. They use it along with other data sets to pre-position firefighting resources and they also use it to conduct timely evaluations of vegetation treatments. 

ADAMSON:
Thank you.

BOYTE:
You're welcome.

ADAMSON:
One of the students also studies an invasive plant. She is using Landsat and Sentinel to map and monitor yellow sweet Clover, which actually has some benefits.

SARAF:
Hi, I'm Sakshi. I'm a third year PhD student from University of South Dakota.

ADAMSON:
And I see that your poster says something about invasive yellow sweet clover. Can you tell us more about what that is first?

SARAF:
Yellow sweet clover is an invasive, biennial legume. 

ADAMSON:
A legume?

SARAF:
Yes, it's a nitrogen fixing plant and it was actually brought here for, you know, soil stabilization purposes and especially during 1930s when there was like a Dust Bowl event, the soil became very poor that time. So when people actually forced it over here so that you know the soil quality improves, so it's actually used for like roadside stabilization. You can find it everywhere like across the roadsides during summers, you'll see yellow colored flowers blooming. It's very good for bee production like habitat bee habitat as well, and used for forage as well, so it has pretty good economic value, lot of positives.

ADAMSON:
It does have some good things about it, but you're describing it as an invasive. That sounds kind of bad.

SARAF:
Exactly. Of course, we are making positive out of it like we are this plant, we are not getting rid of it. So my work is kind of focusing on reducing its negative impact or keep a check on it and the methodology that I'm using to map or monitor this plant species can be implemented on other invasive species also very easily. So it's just like a framework that we are creating to map and monitor one particular invasive plant species which is kind of a difficult to be honest because it's a biennial species, it grows over the period of two years. So in the first year it's not very detectable through the satellite imagery, but in the second year it blooms, and then you can see. So lot of important stuff like regular monitoring and which year it has been blooming and which year it hasn't been. If we don't have information on the phenological stages and kind of like, yeah, I'm somewhere in the middle of mapping them. And also like kind of finding its impact on evaluating its effect on other plants.

I had very interesting results. I found overall because it's been a dry year like most of the years has been like kind of a drought over here. So in the long term period it kind of decreased shrub and forbs richness, but at the same time, like in the specific wet years like 2018, 19, and even in 2015 and even I think 2023, it was like average precipitation, like I would say low precipitation but still like we saw massive blooms of yellow sweet clover. And then it kind of increased for benchtop richness in these specific years. With the coming years, I predict or like I presume that we're going to be seeing more of these blooms.

ADAMSON:
Did you say what data you're using to map the locations of this? 

SARAF:
I mapped them for both Landsat and Sentinel, which is like 30 meter and 10 meter freely available data. And kind of I got similar results for both.

ADAMSON:
This was specifically about yellow sweet Clover, but this method can be used for other invasives?

SARAF:
Yes, definitely.

ADAMSON:
OK. Well that sounds really good. Thank you for talking to me about this. Our next student describes something called object-based image analysis using aerial photos that are available from the EROS Archive. OK, we're coming up to another poster here. Can you tell us who you are?

SCHILD:
My name is Zach Schild. I'm a masters student at the University of South Dakota. 

ADAMSON:
OK, great. And now give us a quick summary of your study.

SCHILD:
So for my study, I wanted to look at mapping Russian olive distributions along the Missouri River using object-based image analysis. To do this, I've chosen 2 segments of the Missouri River downstream from Gavin's Point and Fort Randall dams, as they represent remnant floodplain ecosystems that still see some amount of flooding and the dynamics that correspond with that. I'm working with National Agriculture Imagery Program imagery. It's taken about every other year on a state-by-state basis.

ADAMSON:
This is aerial photography, right?

SCHILD:
Yes, and it is aerial photography that comes in 1 meter and 60 centimeter resolutions containing red, green, blue, and near-infrared bands. I'm using an object-based image analysis approach on mapping because typical pixel based will pick up on a lot of noise on high resolution imagery, so I'm hoping that by using objects we can kind of keep each classification to an individual tree or a patch of trees.

ADAMSON:
Are those the kind of objects you're talking about when you say objects?

SCHILD:
Yeah. So ideally, I'm using a segmentation process and I've tried to make it so that the smallest level of object represents one Russian olive tree of a smaller size, so some newer Russian olive. And then maybe a big patch of Russian olive just broken up into a few different objects.

ADAMSON:
So what value will this study have? Or do you have any idea who will find this valuable?

SCHILD:
Russian olive has historically had some control methods implemented, although those have since kind of decreased along the river. But there's ongoing studies on how well birds are able to use Russian olive, and there's a high potential that insect studies in the future on Russian olives. There's very few insects that use Russian olives, so seeing how that kind of can play a role on insect populations along the river, so future researchers this data could be useful as they go to study local species on the river.

ADAMSON:
That sounds good. One student even got to use drone data, referred to here as UAV, or unmanned aerial vehicle data, to monitor a disease that affects wheat using machine learning techniques. OK, so tell us your name.

JANJUA:
My name is Ubaid from South Dakota State University. I'm graduate student there.

ADAMSON:
I see in the title of your poster it says wheat Fusarium head blight. Can you just start by explaining what that is real quick?

JANJUA:
Yeah, thanks, basically fusarium head blight is one of disease into the wheat that especially into South Dakota and that disease affect mostly the yield of wheat, so.

ADAMSON:
The yield of of what? 

JANJUA:
Wheat. 

ADAMSON:
OK, this this affects wheat—got you.

JANJUA:
So, especially into the South Dakota, North Dakota, these states. So our research comprehensively mostly focus on this disease and we have some testing fields in our department. So we test this in different genotype of wheat and the farmer using the traditional way to assess the disease. Basically, they go to the field and they will assess the field and it's time consuming and needs lot of resources to assess these disease and for the temporal time. They also need such kind of— For us, using the machine learning, remote sensing, and deep learning in combination of these ways, we can assess this disease in one day and without lot of resources without consuming lot of money. Yeah, this is a special thing.

ADAMSON:
That sounds like a good idea. What types of remote sensing data goes into this?

JANJUA:
In this we have used UAVs. 

ADAMSON:
UAVs—drones. 

JANJUA:
Yeah. UAV, drone data. And we use—we flew the drones and that area specific area process that data and prepared for the machine learning and deep learning using some spectral and textual feature. After that we have applied some machine learning and deep learning models and we have classified it as well as regressional analysis. And we have applied sport vector machine, random forest DNN, model to assess disease and especially we have used their spectral feature analysis from that drone data. Also we have used some textual feature. We have assessed disease… spectral features as well as textual features. Then we combine it. The combination of spectral and textual worked amazing.

ADAMSON:
And do you have an idea of who might use this data? Who might find it useful?

JANJUA:
Yeah, basically if we have a high resolution data, for example 1 meter resolution data for the study, we can use it globally. We can assess this disease on a globally, we just need some training samples and we can apply this model onto the globally.

ADAMSON:
You're able to use the training samples right here in South Dakota, but then this will be effective everywhere. 

JANJUA:
Yeah.

ADAMSON:
I just have to ask, are you the one driving the drone or who does that?

JANJUA:
Yeah, I was driving in this year. Yeah. My other colleagues here. So yeah.

ADAMSON:
I mean, it looks kind of fun, but this is for scientific value. That's really neat.

JANJUA:
Yeah, that's really amazing.

ADAMSON:
OK. Well, thank you for talking to us about your study.

JANJUA:
Thank you so much for your time.

ADAMSON:
Thank you for joining us on this episode of Eyes on Earth, where we touched on a lot of research that uses remote sensing data. This poster session provides local students the opportunity to present their work alongside EROS scientists, and they all get to learn from one another. Check out the USGS EROS social media accounts to watch for our newest episodes, and you can also subscribe to us on Apple and Google Podcasts. 

VARIOUS VOICES:
This podcast, this podcast, this podcast, this podcast, this podcast is a product of the U.S. Geological Survey, Department of Interior.