Yield monitors and yield mapping technology have been available for several decades, but the practice of leveraging the data gathered from these tools to make better farm management decisions is still immature. One critical aspect of using yield data to make beneficial decisions is ensuring data quality. Producing high quality yield data relies on following proper calibration methods, engaging in best practices during harvest operations, and knowing how to post-process yield data to adjust for errors such as lag time, header overlap, and speed changes. Dr. Joe Luck, associate professor at the University of Nebraska - Lincoln, joins the "FarmBits" podcast for this episode to discuss these topics, as well as provide his insights into opportunities created by the modern prevalence of digital yield data. We hope that this episode will bring value to you no matter when you listen to it, but especially if you're in your combine for harvest this fall. "If that data is being used in the future – for analysis, prescriptive maps, nitrogen, whatever – I think it’s worth it, personally, to clean the data." - Joe Luck
Opinions expressed on FarmBits are solely those of the guest(s) or host(s) and not the University of Nebraska-Lincoln.
On this episode
Joe's E-Mail: email@example.com
Joe's Twitter: @joeluck_unl
Samantha's Twitter: @SamanthaTeten
Jackson's Twitter: @jstansell87
Jackson: Welcome to the FarmBits Podcast: a product of Nebraska Extension Digital Agriculture.
I'm Jackson Stansell,
Samantha: and I'm Samantha Teten,
Jackson: and we come to you each week to discuss the trends, the realities, and the value of digital agriculture.
Samantha: Through interviews and panels with experts, producers, and innovators from all sectors of digital technology, we hope that you step away from each episode with new practical knowledge of digital agriculture technology.
Jackson: We're excited to welcome back Dr. Joe Luck, associate professor at the University of Nebraska-Lincoln, for Episode 3 of the FarmBits podcast. Dr. Luck is a precision ag extension specialist and has had experience working with yield monitors and data since his time as a graduate student at the University of Kentucky.
Samantha: For those of you who listened to our first episode, he and Laura Thompson dove into the relevance of digital agriculture and technology adoption. On today's episode, we are going to be taking a look at yield data quality and calibration, and we'll even get to a discussion of how yield data is being used,
and what to be on the lookout for in the future. Some of you may be listening to this podcast from
your combine or while driving a grain cart, so hopefully this is timely information for you.
Jackson: And if it is, happy harvest to you and now here's our interview with Dr. Joe Luck.
Samantha: So, Dr. Luck, what do you think is the current most common practice for Nebraskans across our
state when it comes to how they currently calibrate their yield monitor?
Joe: Yeah, it's a tough question just not not being able to get out and visit with a lot of growers, but
I think most of the growers out there are trying to do a pretty good job of calibrating
you know against manufacturer specifications, but I also know that I've talked to a lot of
folks that they kind of think more is better when it comes to calibration. So, one of the key points
of good yield monitor calibration is when you do a calibration load and remember that,myou know if
I've got some high moisture corn,I want to do a calibration for that crop and that moisture. I
want to try and hopefully my yield monitor allows me to do multiple calibration loads so something
from a really high flow to a really low flow. I only need about 3 thousand to five thousand pounds
of grain per calibration load and whether it's just access to a scale a lot of times, I think that's part of the problem some people don't have you know a grain cart with a scale or a wagon something like that. A lot of folks will harvest maybe a thousand bushels. They'll harvest an entire truckload and take that to the when they take it to the elevator they'll weigh it, and then they'll put that in for their calibration load. And that is, that's a mistake when it comes to the calibration process. You know, if you think about that high flow calibration load that I want to run, I want a quick again, you know that you're talking 50 to 90 bushels, that's all you need, I want that run at a very high flow rate, very consistently high flow rate over that short
period of time. And I want to get that weight and get it in the system. And if you can do that for
each load then technically the calibration process shouldn't take too long, but again I think a big
part of it's just access. Do we have access to a weigh wagon or a grain cart with scales?
Jackson: You know, follow up on that, you know you say you want these smaller batches, and I guess when
we're thinking about yield monitor calibration, we're looking at ideally a three-point calibration
for yield monitors at least. And so is the best way of getting that different amount of grain flow
in those periods of time to get those different size loads is the best way to do that changing
your swath width, or is it changing your speed. How do you recommend doing that or both of
those equally valid methods?
Joe: I think they're both equally valid. So, in an ideal situation there's still you know still a lot of machines out there that only let you do a two-point calibration. So again, you know, you're trying to get the high flow and then get a lower flow rate. The ideal situation is if you can do four or more points per load. So again, you know, if I'm doing high moisture corn- I want to look at let's just take the constant cut width. I want to say I'm going to harvest it six miles an¬†hour, that's the fastest I anticipate traveling, so I'm gonna run a quick calibration load, you know, three thousand pounds/six miles an hour full header cut width. The next load down I might do you know three thousand pounds at five or four mile an hour, and then I drop down to three maybe, drop down to one to two miles an hour to generate that four point or more. So that's kind of the procedure cut width versus speed- it really doesn't matter if you think about it. Each one of those scenarios, I want to get that three to five thousand pounds. So, each one I should be about the same flow through the machine. It's typically for me it's whatever's you know easiest for the grower. The one thing I will say
is though you know when we when we say varying speed, you want to be a constant speed while
you're harvesting that one calibration load. That's the key, is stability. So, that's the only advice I'd give there, but either way you end up with the same result, same amount of time to grab that amount of grain and get it entered into the system.
Samantha: Sure, and how many times should they be doing this calibration throughout the season, or do you have any rules of thumb for growers?
Joe: Well yeah, it's I would say a minimum: there's a few things to think about. First, is I have to have it for different crops. So, if I'm doing a lot, which is fairly typical here at least in the eastern part of the state, high moisture corn, I need to I need to create a calibration for that. A lot of people switch over to soybeans. They've got to do a calibration for that. And then they'll go back to corn and it'll be lower moisture. And we need to do that, you know, actually interestingly a little data out of our friends over at Iowa State noted that if you have a 2.5% swing in moisture content, that could be a 5% error in your yield data. So, that there's a lot of challenges there because you could have that much of a swing in one field. How do you deal with that? But again in general that's the minimum. I would look at doing and if you can do an extra calibration or two in there, it's worth it. The other thing I would mention is test weight can also have an impact. So, some of the data we collected back when I was at University of Kentucky working in the yield monitor test facility,
we had two different hybrids of corn. One weighed about 56 pounds a bushel the other one was 62 pounds per bushel. It's a pretty big swing at the same moisture content. Six pounds. That ended up being a two and a half percent error, and so if we calibrated with one and switched over to the other and didn't change anything, it was two and a half percent error. So really you know, keeping an eye on test weights is important as¬†well. So, if you have hybrids of really different test weight, I would consider that too. And so, you know you were talking about you may have different moistures within one field and obviously you can go back and forth between a lot of crops, I mean is there is there any way to kind of find this best compromise between time and accurate calibration? Especially since harvest is, I mean there's logistics and like it's a time crunch
I think for a lot of growers. It's a kind of a catch-22. You don't want to spend all your time calibrating, you know if you're doing that. You know, we do we see some new technologies out there like ActiveYield from John Deere that's an on it's basically on-the-go calibrating you know on the machine you don't need a scale necessarily. So, we'll see new technologies like that coming out that I think help out with that a lot people. I could say okay, I'm switching over a new field, I just want to run a new calibration, I don't have to stop and do that with the grain cart. So, you know until that time where that's widely adopted it's you know it's just have to recognize that you know. I've got to make that decision of you know okay I've dropped four or five percent moisture, and I'm still harvesting corn maybe it's time to take a new calibration just to update that.
Samantha: (Sure), so switching gears a little bit so if let's say your combine, you're all calibrated once, you're now harvesting the rest of your field, what are some operational best practices to ensure that your data is coming through to the yield monitor accurately?
Joe: Yeah, that's a great question. Especially when we're doing on-farm research plots, you want to get the best data you can. We can do a lot with post-processing but the number one thing I tell people is once you've gotten calibrated and start running, speed changes, try to minimize any speed change you can because even though we know we can post-process some of that out, just taking that out of the equation is really
helpful. You know, try and run at a constant speed, let the calibration take care of
everything else. If you've done a decent you know, load of high flow rate calibration.
The second thing I would tell people to think of really about is looking at your header sensor.
Making sure that that is when you lower the header and start cutting crop that's
activating appropriately and when you pick it up, it's stopping that logging. Because even if
that system is working and you're you know you're dropping 10-20 feet out from the crop,
you're telling the system, I'm harvesting crop there. And so, if we really want to line up well,
you know we need to drop that right at that last instant start engaging the crop. Pick it
up right when I'm out of the crop. Make sure that system knows that it's starting to stop
and that's especially important, you know as we as we start the passes and in the passes.
The third thing I would get people to think about is there's some great animations out there
on YouTube, Claas, Deere, have these internal simulations of what's going through the machine.
Most of the farmers know what what goes on inside the combine, but just recognizing that there's so
many paths. That if I'm driving along and I cut one foot, you know, travel path of grain there's
at least three or more different passes that grain could take through the machine before it gets to
the mass flow sensor at the top of the clean grain elevator. One of the farmers I worked with
in the central part of the state, they would actually look at their tailings
elevator sensor and try and minimize and again they're slowing down a little bit. Because the
more crop you try and force through the machine, typically that's when you'll see more go through
the tailings elevator. But they would actually try and slow down a little bit in their plot areas
to try and make sure that they minimize that last path. That's the longest path material is going
to take is if it goes all the way through the machine, goes through the tailings elevator, and
goes right back through the machine again. Little things like that could make a difference,
but those are the big things. Of course, moisture sensors are important. We need to check make
sure those are operating normally because again that's going to give you the idea of marketable
yield as you move across the field. Which is we really want to see that in the yield maps.
Jackson: Sure, I guess you know, for the typical grower out there and you know I've talked to a few of
the growers that you know cooperate with us on research about how much they usually spend time
cleaning their yield data. You know we've said that we can do a lot post-processing wise,
so I guess for the typical grower out there in what scenarios do you think it's really worth it to spend that time cleaning their data? You know if they have a particular use case and you know if you do think it's worth it for growers to spend time cleaning yield data? What tools would you recommend to them to use for that process?
Joe: Yeah it's a a great question it's a huge, huge answer I think for that one so I'll try and go be as brief as I can. The definitely, if that data is being used in the future for analysis for prescriptive maps, nitrogen whatever, I think it's worth it personally to clean the data. You know if you have if you've done a really
good job those first few points we talked about, once you get calibrated and you get in the field
speed is minimized speed changes are minimized, you know that start at the beginning of the harvest and you know a row and at the end really being quick with the header, you're going to minimize a lot of the errors we want to take out. Now if in a lot of cases if you have point rows and you don't have some type of auto-swath monitoring on the combine. Those point rows are going to generate errors because no not many people are going to go in and manually change cut width. You know in that 20 or 30 feet of field length and that's going to show up in your errors. But step one is if you're using the data and you're not looking at it, you know if you'll take the data out and actually look at each yield map, and say yeah I don't see a lot of errors, you know you can you can bring that into farm management software and quickly do a query and see where yield values are really high or really low, If you're not going to visually inspect that and you're going to use it in the future, I think that's when we need to do the post-processing. You know you can set a lot of the farm management software for instance, uh you know we use Ag Leader SMS, some you can go in and set filters
and filter out some of the data internally and you can save those and reuse them. We also use yield editor software from USDA. Once you get the data extracted out of a program like SMS, and you can get it into the right format you can run that through yield editor, it automatically processes a lot of those things. I think that overlap they've got an overlap filter for point rows in there, I think that's one of the greatest things I've seen. You know I didn't grow up in Nebraska and I know everybody thinks that Nebraska has perfectly square fields except the circular ones and that's not true. I just I'll tell people out there you know I grew up in a
in a state in an area the state where the average field size was maybe you know 40 acres 30 acres.
Really weird shape fields um that you really see it a lot of swing in those points. So, you know if
you're out there and that's and you have a lot of those issues a lot of point rows things like that
that's where again, if you're using that data for a further application you're going to benefit
from removing those errors.
Jackson: So, and yield editor is a freely available software right? Growers can get that you know just from the USDA website, and I know from using it myself it's a pretty intuitive program to use.
Joe: Yep and it's great you can you can view what errors exist, you know what the filters are pulling out you can look, and see you know it's always a great step when you're using that software is run the automatic filters. Let it pull out what it thinks need to be pulled out and then go in and look you can say, okay show me all of the acceleration/ deceleration errors. If they're pretty sporadic, you know that it's probably doing its job. If you see a particular filter that it's taking out a huge chunk of data right in one location that probably needs a little bit more more, uh you know visualization you look at a little bit closer. (Sure) That's I think that's one of the key things is a lot of the tools today we almost don't even have to look at the data anymore, we can just dump the data into a program run it through a system never really see it. That's when I think we need¬†to take this extra step of the post-processing.
Samantha: What different challenges exist for the different crops that we use when it comes to yield mapping and measuring that yield?
Joe: Well that's a great question. A couple of different challenges that come to mind one, is in corn and unfortunately for a lot of folks in the central Midwest the wind damage they're gonna experience for the people that didn't have to go ahead and remove that crop down corn is a huge challenge. In corn you know, in the corn crop. You have to slow down you typically have a lot of losses there if you're in that. That's one of those examples where you if you haven't calibrated it that low of a flow you expect you got to go back and do another one at a really low flow so you get a good average around that. And of course, that's nobody wants that situation in beans one of the biggest challenges, we've seen is they dry down during the day a lot of times so people you're really frustrated. I've seen a lot of yield maps where you start out harvesting and then they come back later in the day the same area and the yield values are way off and a lot of times that's moisture related. Again moisture content affecting you know the in the way the grain impacts the plate different things like that in a mass flow sensor. That's always been a big challenge, not sure we've ever come up with a great
solution for that. But those are two things that come to mind pretty quickly.
Jackson: Sure, and you kind of brought up and alluded to these different types of yield monitors that are out there with the mass flow sensor that you were talking about you know, so you have your scales, your mass flow sensors, your optical sensors, and then I guess over in Europe, they kind of have the nuclear, which we
don't have here in in the US, but what are the pros and cons of each one of these different yield monitors and are there any special considerations for those different styles of your yield monitors out there?
Joe: Yeah, not having any experience unfortunately with the nuclear- type yield monitors but with mass flow and optical, you know you can get good data from both of them. We've kind of covered the mass flow sensor impact plate style. We kind of talked about calibration for that. The only other consideration and we've talked about test weight and how that can affect that sensor performance. The only thing I would caution people is if you are using an optical sensor, you probably already know the test weight is critical in the calibration process for that. Because we're measuring essentially volume flow on the machine now and then we're calibrating that with a weight a scale weight. So, we need to know what the density of that grain was to go
back you know into the into the yield monitor. So, test weight is in incredibly critical when it comes to the optical the optical flow sensors. And you know most of the times I've or at least I've heard when you buy those, you know your free gift with that is a little scale where you can take your test weight out in the field. You know the little bucket you know recalling that to my knowledge you're not supposed to just take the little bucket and scoop up a lot of grain, you know there's a you want to kind of fill that very slowly, very evenly, and things like that little things can make a difference. But that'd be the biggest difference, you know we've actually seen data from both checked it versus scales, you can on average you can have very good data from both systems.
Samantha: So, you talked about a little bit earlier how important it is to clean your data if you're going to be using it for a later purpose, so when you think about using that yield data, what are the best opportunities for getting value out of that?
Joe: Yeah, I think just understanding where industry is going and service providers you know. More and more that data is going into future prescriptive efforts, so if it's nutrient removal, people are using it for that if it's future nitrogen application, you know we're using it for that. Just the understanding of you know, and if you have three to five years understanding of historically how that crop yield has varied across the field, can be a really important part of that. That's incredibly powerful. You know people as we move forward just having that baseline with which to compare to if you can quantify yield variability across your field,
you can put a price on that and now you know what range. Of you know in other words economic benefit or cost you have to work in work within if you're gonna you know have a solution to what crop. So, figuring out what problem it is you're trying to solve if what do I put into that problem, am I going to have a potential return for it, and so to that point I think the on-farm research to me is one of the most powerful tools
we have moving forward. And as we've discussed you know the precision ag technologies, the GIS
analysis that allows us to do those comparisons every year, we could be out the field doing
small plots here and there to test out different management strategies, and we have to have as good
of yield data as we possibly can to make sure we have a lot of confidence in the results we get
from those, but that really tells us you know, hey I made a change this year but you know it didn't
didn't pay off, I'm not going to try that again, or maybe it did, and I can move forward with that
in the future.
Samantha: So, when you talk about also like farmers using their own data can you speak on
at all like what companies are doing with yield data if they get access to your personal data?
Joe: Well unfortunately a lot of that would be speculation. But you know there there's
yeah, we I'd have to speculate too much with what some of the folks are doing out there, but
we know that there's power in for instance aggregated data. You know, some of the companies
that are they're remotely logging data are able to use that to learn about their hybrids, their crop protection techniques, all these different things, so it's all part of this big this future of big data and agriculture. And can companies, you know if they're collecting data from a wide geographic region you know they know the rate of seed, the type of seed as much more information they can collect it what happened during the year they
can use that to improve their products or maybe the recommendations. That's what you hope to see.
So yeah, it's hard to say exactly what they're doing, not being privy to some of those discussions but again there's a lot of there is a lot of value in that aggregated data. And I'll probably get in trouble for saying this, but I but I think my opinion is precision ag data on farms is much more valuable than what we're leading it on to be. You just think about what it takes the investment to do a research study in some of these fields, if we could if we can turn the farmers' fields into those which is what we're seeing out there. In other words, we can log all the information what went in when, where every input, we can monitor the output, we can monitor weather data, pretty well remotely now. Imagine the value of that replacing all these research studies with precision ag technology. That's a lot of value, so personally I think you know that's something we need to
need to think about and talk more openly about in the future.
Samantha: Especially when growers really value that full field data versus plot data. You know people are really skeptical so there's a plot data so that's good to hear too.
Jackson: Yeah, it's absolutely critical to be a production scale. You know to convince somebody that something works and you've kind of already been alluding to this the big data future of agriculture and how yield data can pertain to that, but what are some other things that we may see coming you know within
yield data in the future as far as kind of yield data sharing technologies? You know machine
tracking technologies within the field, machine to machine communication. And you know I guess even in
other crops like cotton you know what maybe some of the yield technologies that are coming down
the pipeline? I know there seems like there's been some innovation there in recent years. So.
Joe: Yeah, you know, they have essentially flow sensors now for cotton harvesters so they're able to do
some site-specific management with in cotton as well now so I think you know we're going
to continue to see a push for improving the quality of the yield data and you know we're
going to start to see efforts at sub header with estimates of yield. And we've already
you know we've seen some research you know you can you can go out and look at patents that have
been you know applied and awarded to companies that there's information there that hasn't been
really commercially made available yet. You know that gets typically filed pretty early
you know how do we how we break down a you know a 12-16 row header and maybe use imagery or
some other machine vision type technology to break that down, and say well we could actually
look a little bit you know more a higher resolution at. What's going on in that machine we know the total flow, let's try and scale that a little bit across the header. I know you know I have a lot of friends over in you know the agronomics and plant
sciences and I know they would love to see more resolution and some of that data you know not just
you know longitudinally as we move through the field but laterally as well. And so, I think
we're going to see some push but it's been great you know the data quality piece has been
really challenging as we look at the future, which is going to have artificial intelligence in it,
machine learning techniques you know you cannot say enough about data quality and a lot of the
literature that I've read would tell you that most of the projects that fail it's because of data related you know in quality issues. So, as we look more for those techniques to take off you know data quality has got to be number one, garbage in garbage out that's the saying.
Samantha: So that was a lot of awesome information. So, to tie it all together what would be the biggest
piece of advice or message you want to leave the listeners with?
Joe: Well, the one thing I would do I would say is you know don't be afraid to reach out if you need assistance with anything at all. You know especially the group you know the digital ag group and at Nebraska Extension you know we're working for the University we're always willing to help out if it's an issue with
you know calibration or I've got a data issue. You know, we learn when with the folks
that are out there that we work with and so you know we've always prided ourselves on being as unbiased a source as we possibly can, and I think we'd all do a great job of that. And so that's the main thing is just you know don't be afraid to reach out if you have questions if you're interested in something you know if you're listening to this that you're interested in something. So, don't be afraid to reach out to any member of the team and see you know if there's something we can help with yeah that'd be my one piece of advice is
you know don't hesitate to reach out to folks for assistance.
Samantha: And we'll put Dr. Luck's email and contact in the show notes for any of you guys who want to reach out.
Jackson: Thank you again to Dr. Joe Luck for joining us on the FarmBits podcast.
Samantha: As his students, we talk to him often about yield data quality for On-Farm Research but we still
managed to learn a lot of new information from that interview. Personally, I really enjoyed
when he talked about how those yield maps are being used for things like prescription maps,
or multi-year analysis of field patterns.
Jackson: For sure, and following on your favorite part Sam, I thought his discussion of the interaction between harvesting practices in the field and how those relate to the importance of post-processing your yield data for future use is really important information for people to consider, especially during this harvest time that we're in right now. So, we hope that you join us next week as we do our first Farmer Focus episode.
Samantha: We are heading out to the field to hear from a few farmers and learn about their technology use during harvest. Thank you for taking the time to join us today on the FarmBits podcast.
Jackson: We would like to thank Nebraska Extension for their support of this podcast and their commitment to providing high quality informational material to members of the agricultural community in Nebraska and beyond.
Samantha: If you enjoyed this episode and it sounds like something you'd listen to each week, subscribe
to the podcast and set your notifications to let you know each time we release a podcast.
Jackson: We would love to hear from you with your feedback so if you have comments or questions for us, please reach out to us over email at NEdigitalag@unl.edu, on twitter @NEdigitalag, or in the reviews
section of your favorite podcast platform.
Samantha: See you next week on another episode of FarmBits.
Transcripts are generated using a combination of speech recognition software and human transcribers, and may contain errors. Please check the audio before quoting in print and write firstname.lastname@example.org to report any errors.