OFR21: Precision Nitrogen Management with Laila Puntel
In this video, Laila Puntel discusses how farmers can use crop models to generate a variable-rate nitrogen recommendation for their fields. These precision nitrogen management technologies can help farmers reduce nitrogen and increase their yields.
icon search Searchable Transcript
Toggle between list and paragraph view.
[00:00:00.399](bright upbeat intro music)
[00:00:10.820]Okay. Well, since we already solved the technical part.
[00:00:15.170]Good morning, everyone.
[00:00:16.710]Thanks for joining.
[00:00:17.940]So today I'm gonna try to give a brief introduction
[00:00:23.650]to one of the technologies that we are testing
[00:00:26.100]through our precision nitrogen project.
[00:00:31.640]And so we are gonna dive into crop model based tools here.
[00:00:36.930]So just to give you an idea
[00:00:38.800]of how these tools might become very useful and helpful is,
[00:00:44.430]and I wanna try to link it to some of the questions
[00:00:46.670]that you might have during your growing season,
[00:00:49.960]like for example,
[00:00:51.120]how much water you have left in the soil, right?
[00:00:54.080]If we wanna try to predict
[00:00:55.710]how many days can I go without rain,
[00:00:58.600]crop model based tools
[00:00:59.800]kind of allow you to make that prediction into the future,
[00:01:03.730]or also things like,
[00:01:05.650]well, how much nitrate am I having in the soil
[00:01:08.740]in that particular season?
[00:01:10.330]So crop model based tools kind of integrate all the factors
[00:01:13.970]to kind of know how much nitrate you have.
[00:01:17.710]And then also, are helpful
[00:01:18.990]because they combine all these things
[00:01:21.500]and try to give you an assessment
[00:01:24.020]of how much nitrogen you might need for the crop
[00:01:27.720]during that growing season.
[00:01:29.570]So on that graphic, on the right,
[00:01:33.330]what I'm showing you
[00:01:34.163]is kind of all the factors that interact
[00:01:36.950]to define that optimal N rate.
[00:01:39.610]And for those that are not familiar with that term,
[00:01:42.160]it means the nitrogen rate that maximize your profit
[00:01:46.030]for that particular year
[00:01:47.280]and so that particular soil in your field.
[00:01:50.420]And so as you can see,
[00:01:51.840]it's a balance between all the nitrogen
[00:01:55.450]that's coming from your soil.
[00:01:56.970]So the soil supply and the crop demand.
[00:01:59.550]So how much nitrogen you need in order to satisfy the yield.
[00:02:04.290]So crop model based tools integrate all those buckets
[00:02:07.780]that you see see in that figure
[00:02:09.480]and try to allow you to have a prediction.
[00:02:13.800]So what is in a crop model?
[00:02:16.496]So imagine like it has different boxes
[00:02:19.800]that have multiple equations that run every single day
[00:02:23.680]to kind of come up with an estimation
[00:02:25.680]of how much your plant is growing
[00:02:28.280]and how much nitrogen it's gonna need.
[00:02:30.900]So for running some of those tools,
[00:02:32.980]you might need some management data.
[00:02:36.180]So like planting day, the fertilizer day,
[00:02:39.550]what kind of, sort of, so fertilizer are you using
[00:02:43.200]and things like that.
[00:02:45.220]And then we'll also incorporate things like
[00:02:49.980]all your weather variables.
[00:02:51.390]So temperature, radiation, and precipitation.
[00:02:54.940]And as you know,
[00:02:55.800]those are key to guide
[00:02:57.420]how much nitrogen I will have in mineralization in the soil
[00:03:00.970]or how much leaching I'm gonna have.
[00:03:03.180]So all important factors.
[00:03:05.150]And then as well,
[00:03:06.280]we will have all the soil characteristics.
[00:03:09.070]So this is another key point of these tools
[00:03:11.790]that will allow us to account for all the organic matter,
[00:03:15.000]values, our texture
[00:03:16.950]and different nitrate conditions in the soil
[00:03:19.870]when we start the season.
[00:03:22.960]Another thing that these tools allowed us to do
[00:03:25.440]is to account for the spatial variability.
[00:03:27.960]So here in this picture,
[00:03:29.230]what I'm showing you
[00:03:30.170]is like two contrasting soils in a field.
[00:03:33.380]So we have an upper hill and like a more low in lane area.
[00:03:39.490]So crop model based tools allowed us to put
[00:03:43.270]an input of different organic matters
[00:03:45.350]at the top of the hill
[00:03:46.600]and maybe a different organic matter value at the bottom.
[00:03:50.340]That way, when we run that model,
[00:03:52.790]then we can predict two target N rates
[00:03:55.850]for the two parts of the field.
[00:03:59.700]And another key point,
[00:04:00.840]and I think that's one of the maintenance strengths
[00:04:02.940]that we have for these tools
[00:04:04.440]is that they account for weather.
[00:04:06.290]So the more closer we get
[00:04:08.960]to when we are gonna do the decision in the growing system,
[00:04:12.280]the more weather data we can incorporate
[00:04:14.880]to know either if we lose nitrogen in the soil,
[00:04:18.380]if my crop is really good that season,
[00:04:20.970]and if I'm gonna need more or less nitrogen.
[00:04:23.410]So those are kind of the benefits.
[00:04:26.380]And in this graph,
[00:04:27.213]what I'm showing you is sort of like the predictions
[00:04:30.070]of one of the models that we have run in the Midwest,
[00:04:34.160]and in the top panel,
[00:04:35.260]I'm showing you the prediction of soil nitrate
[00:04:37.940]in the growing season,
[00:04:39.180]and in the bottom panel,
[00:04:41.690]this is yield as a function of the date.
[00:04:44.520]And the distance between that green line and the red line
[00:04:48.410]is sort of the uncertainty that you have
[00:04:50.550]predicting the yield when you start your growing season.
[00:04:54.090]And you can see that that panel when you move to the right,
[00:04:57.600]so closer to your harvest date, it's getting smaller.
[00:05:01.380]So that means that the more information we know,
[00:05:05.730]the less uncertainty we have to predict our yield.
[00:05:08.130]The same happen with how much nitrate we have in the soil.
[00:05:12.640]So what model is the best?
[00:05:14.960]And here I'm just giving you some idea
[00:05:17.850]about what are the commercial models
[00:05:19.510]that we are currently using or testing
[00:05:21.530]in our On-Farm Research Network.
[00:05:24.090]And so we have Maize-N, Adapt-N from (mumbles)
[00:05:28.110]Granular, FarmersEdge, Flurosat.
[00:05:30.510]And I'm sure that more and more of these softwares
[00:05:32.940]and platforms are gonna come up popular,
[00:05:35.570]but all in the background,
[00:05:37.030]they're starting to run a crop model.
[00:05:39.130]So you will have a simulation of your cropping system
[00:05:43.170]in the background.
[00:05:44.510]So what is the best model?
[00:05:46.520]I always make this joke in terms of like,
[00:05:49.000]well, we kind of pick the bagel
[00:05:51.420]that best suit our purpose, right?
[00:05:53.640]So this is the same.
[00:05:54.930]You might drive the truck for going to the farm, right?
[00:05:58.050]And then your car for going to the city.
[00:06:00.490]So this is kind of what we're trying to learn is like,
[00:06:03.210]what model is better for,
[00:06:06.090]depending on the logistics of your farm
[00:06:08.520]and depending what environment are we at.
[00:06:11.720]So that's kind of what we're testing.
[00:06:13.400]So a great opportunity to test these models
[00:06:16.440]via On-Farm Research Network.
[00:06:18.710]So today I'm gonna share two of the pilot studies
[00:06:22.430]that we did last year,
[00:06:23.960]and that we are gonna plan to conduct
[00:06:25.990]for the next three years.
[00:06:28.710]So in this study,
[00:06:30.880]we have two pilot sites in Southern Nebraska,
[00:06:35.220]and we have one dryland popcorn,
[00:06:37.940]and one irrigated corn,
[00:06:40.130]so very contrasting situations.
[00:06:43.610]And this is sort of
[00:06:44.443]how the layout of the experiment looks like.
[00:06:48.150]You might see a lot of colors here.
[00:06:49.620]I will walk you through.
[00:06:51.740]So what you see the strip that is pink,
[00:06:55.190]that is the model Adapt-N from (mumbles)
[00:06:58.290]that the grower wanted to test against Granular software,
[00:07:03.100]that they have their own proprietary crop model.
[00:07:06.920]And then the blocks that you see in different colors,
[00:07:09.890]those are different nitrogen rates.
[00:07:12.810]And we call it nitrogen ramps or blocks.
[00:07:16.080]Basically, increments of nitrogen
[00:07:18.470]until you saturate the response of yield to nitrogen.
[00:07:22.610]That is key here
[00:07:24.280]because that will allow us to calculate
[00:07:26.440]what was your optimal N rate.
[00:07:28.360]So the N rate that gives you the maximum profit
[00:07:31.610]at each part of the field.
[00:07:33.610]And then at the end of the system,
[00:07:35.160]we can compare what Granular or Adapt-N give us
[00:07:40.110]as optimal N rate against our kind of true
[00:07:43.770]for that particular year.
[00:07:46.280]So this is kind of for giving you an overview
[00:07:49.240]of what the kind of data
[00:07:50.760]that we are getting for those blocks.
[00:07:52.910]So what I'm showing in this picture is yield on the Y-axis
[00:07:57.700]and in the X-axis, nitrogen.
[00:08:00.130]So this is the curve that we are expecting to have
[00:08:03.100]in different parts of the field.
[00:08:05.260]And in this example,
[00:08:06.520]that is coming from another field
[00:08:08.500]that we also run some pilot studies,
[00:08:12.170]we have a different EONR maximizing your profit
[00:08:16.610]in each of the soils that we have in that field.
[00:08:19.090]So this is the kind of information that we are gathering
[00:08:21.920]with these studies.
[00:08:23.490]So now let's jump into the prescriptions.
[00:08:26.180]So for this field,
[00:08:27.510]we gather all the managements, soil bubbles, yield history,
[00:08:31.960]and we input everything into the two software platforms.
[00:08:36.590]We work with Granular,
[00:08:37.860]and they show us this prescription on the left,
[00:08:41.770]and then in Adapt-N,
[00:08:43.060]we got the prescription on the right.
[00:08:45.650]Two things to notice.
[00:08:47.350]Total amount of nitrogen spent in the field
[00:08:49.640]was pretty similar.
[00:08:51.190]Although the distribution of fertilizer
[00:08:53.690]across the two fields was different.
[00:08:55.890]So what we are starting to learn here
[00:08:58.120]is that the way that they configure soils
[00:09:01.320]and the way that they attribute different N rates
[00:09:03.730]across the field are different.
[00:09:05.710]But now we are diving into the reasons of why
[00:09:08.560]and what are the impacts of those different recommendations.
[00:09:13.200]So when we look at all the data aggregated,
[00:09:15.740]so this is after we harvest the field,
[00:09:18.380]we view and monitor.
[00:09:21.430]We analyze the total nitrogen rate used,
[00:09:24.420]the corn yield, nitrogen use efficiency
[00:09:26.900]and partial profit.
[00:09:27.990]So imagine that you summarize all those strips
[00:09:30.670]that I show you at the beginning,
[00:09:32.710]and now we calculate all the different inputs and outputs
[00:09:36.620]for each of the treatments.
[00:09:38.530]So what we found in this case
[00:09:40.450]is that there were no significant difference
[00:09:42.420]between the two models.
[00:09:44.170]So they've overalled the field,
[00:09:46.170]they yield the same,
[00:09:47.540]they spent pretty much the same amount of nitrogen,
[00:09:50.600]and we had a really high nitrogen use efficiency
[00:09:53.730]that was around .82, .86.
[00:09:57.300]So that was one example.
[00:09:59.610]Now, jumping into the second pilot study that we had.
[00:10:04.270]This was in dry land,
[00:10:05.760]and keep in mind that this was popcorn,
[00:10:08.860]one of the crops that is not quite calibrated
[00:10:12.930]within these platforms,
[00:10:14.620]but what we can see is that Adapt-N recommended
[00:10:17.880]a little bit lower N rate across the field.
[00:10:21.130]And again, the distribution of fertilizer
[00:10:23.960]is contrasting as well.
[00:10:26.430]So when we look at the results of this trial,
[00:10:29.700]we found that the total amount of nitrogen
[00:10:32.320]that you can see it in the top panel on the left,
[00:10:35.820]the total amount recommended by Granular in this case
[00:10:38.520]was about 50 pounds (mumbles)
[00:10:41.650]And we think that this was related with the fact that
[00:10:43.870]we targeted 100 bushel yield for this popcorn,
[00:10:48.070]and it ended up being a super windy season, super dry,
[00:10:52.350]so really our yield goal was pretty lower.
[00:10:56.650]And then for the rest of the thing,
[00:10:58.160]we did not find either for the yield or the partial profit,
[00:11:02.780]we did not find any significant difference.
[00:11:05.760]So a lot to dive into this data.
[00:11:08.130]I just wanna like share with you
[00:11:09.680]kind of the first opportunities here that we're finding.
[00:11:14.830]And if you're interested on comparing
[00:11:17.190]some of these softwares,
[00:11:18.720]we will have more opportunities for you in a little bit,
[00:11:21.810]but please don't hesitate to ask and contact me.
[00:11:25.023](bright upbeat outro music)
Log in to post comments