Tip:
Highlight text to annotate it
X
>> Who's going to give it to explain with?
>> Excuse me, this is the operator.
I would like to inform all participants, the conference is being recorded.
Given the objections, please disconnect at this time.
Thank you.
>> Hello everybody.
I want to thank you all for taking time out of your schedules
to listen and I suppose watching on this.
This is Bob Dobos speaking here at the National Soil Survey Center.
I guess up here, we have Mike Robotham and a few other folks in the crowd here, so,
you know, don't say anything nasty.
And I want to say if you have questions or comments
as I'm talking along, please just let me know.
You know, I don't get to-- I've been out of shape about that.
What I want to talk to you about today is update
of the National Commodity Crop Productivity Index which is something I've been working
with for a while and-- oops, and just trying to get a way of doing, you know, productivity index
and we'll talk about the whys and everything here.
Here's the outline of what I want to talk about today.
Talk a little bit about on the background, you know, why?
I always like to ask, why you want to do this, what is it, how does it work.
I mean, I don't know if some of you have probably seen it before sort
of some of the data.
So, just kind of let you know what's different.
And I'm going to go out on a limb a little bit and talk about how good is it
and talk just very briefly about what we want to do.
Man, I'll clue you in now.
We'd like to get this on to the Datamart and there's a variety
of reasons why it hasn't been done so yet, but we got further talk, we're going to get
in on to the background of NCCPI.
The reason why we wanted to do this is because we need to be able to array the soils nationwide
on the basis of their inherent productivity.
A lot of states like Iowa, few others have their own state crop productivity indices
and in those states they work very well and we're not trying to replace those.
But most of them, I'm not sure all, but I'm pretty sure probably all,
once you get them outside of their area, they kind of fall apart,
and NCCPI is designed, hopefully, to not do that.
And currently, this current edition is just for dry land agriculture.
I've been working on an irrigated version and it's coming along.
We'll be hearing more about it probably pretty soon.
And so, so in a theory of this thing is we want to use the use-invariant soil properties
because it's basically what the NASIS database is made up of is use-invariant soil properties.
Those are hoped, at least, or thought to be a major factor in production of a crop.
Of course, you have fertility and other management concerns.
But if you hold those constant, then perhaps the use-invariant properties
of a soil have an impact on the productivity of that soil, the ease of management
of the soil, and just the bottom line.
You know, you might be able to get as much yield out of a-- you know, one soil and other.
But if you're using twice as much fertilizer or something like that, then, you know,
the bottom line is different and this is kind of aimed at the bottom line as well.
Another operating theory is a crop is grown on a soil,
on a landscape, and it's subject to a climate.
So you can have a beautiful deep loess soil, but if it never gets any rain then you don't get a--
you don't have any dry land production off of it.
And the same thing goes.
You can have a beautiful deep loess soil, fine silt,
but if the mean annual soil temperature is zero Celsius, it's going to be difficult
to get any productivity off of it.
So the climate, although, you know, wasn't really something they wanted to put into,
the model becomes very important because you just can't grow a crop without a climate.
And then, also, kind of going down the climate road, NCCPI is a three-part model
because we want to deal with frigid soils or the crops that are grown
on frigid soils, winter wheat, some corn.
Crops that are grown on mesic soils, winter wheat, corn, soybeans.
And crops that grow on thermic, on thermic area, so that's cotton, really.
Although, wheat and corn will also grow there but they're out of their--
they're out of their comfort zone, really, in a thermic environment.
So thermic crop that were after is cotton which introduces a whole bunch of stuff
but that's something we can't talk about here.
Projected users, you know, you don't want to do an interpretation if there's nobody going
to use it and it turns out that there's actually quite a few users and more on the way.
Farm service agency likes it as part of rental rate calculation.
I've dealt with the risk management agency because they can kind of get an idea of,
you know, how insurance claims are going if they're being subjected to fraud, really.
The ERS folks have used it to help in projections and, you know?
Part of real estate people like it because you can get an idea of quantitatively, you know,
getting a value on farm A versus farm B. Other people that I've been with asking me
on this are land assessors where they have that real estate assessors,
but also seed companies have been-- like NCCPI data and also quite a few university people,
Cornell and Penn State and actually University of Nebraska also.
So it's actually got pretty-- a broad user group.
So what is this thing?
Well, NCCPI is, you know, it's a fuzzy system model that uses the data
and relationships that are in the NASIS database.
All of the data comes out of NASIS, whether that's good, bad or indifferent,
it doesn't allow us, meaning the soil survey program, to have control over the data.
Everything that is in there, if there's something I see is wrong or peculiar,
nothing is every wrong, some things are peculiar, in theory, I can call somebody
on the phone and get it changed to be less peculiar, at least in theory.
Also, if we have all the data in one place, then if somebody in California runs the model,
they get the same thing at the same time with somebody in Florida.
And so, it has a good consistency.
It's been vetted by a lot of people across the country and.
So really, one of the things I like about it is because we actually control all the data.
We don't have to worry about going out and fetching data from outside sources.
It's-- for what it is, you know, you have to realize it's an index that is to array soils
on the basis over inherent productivity.
We're not really trying to predict the yield because in order to really predict the yield,
at least for a particular year, you need a whole lot more data
than what's in our soil survey database.
But, we have-- we can look at soil properties versus, you know, a membership function.
Actually, we're asking the question-- or answering the question,
soil is a member of this set of highly productive soils if it has, you know,
an AWC of 30 centimeters per centimeter and so on.
And so we have membership functions.
Well in this place, the condition right here, this situation that I'm showing,
corn yield is the membership function, you know, and that we're just looking
at available water holding capacity, nothing-- nothing major thing there.
Actually, it's major component, but we use the relationships in the database.
Some soil, landscape, climate parameters have a bigger impact than others,
but some are only really, you know, really seem to kick in the extreme
like bulk density doesn't seem to do much until it's just too dense.
For cotton, the pH is almost is, meaningless [inaudible],
plus it's about over 8-1/2 or below 3-1/2.
So, you know, some things are really only important in the extreme.
But when they're important, they're really important.
They limit the growth of the crop almost entirely.
And so, in order to construct the model in NASIS,
we kind of look at the shape of the curves.
You can see this here.
You have a definite minimum-- maximum here, well around,
just a little under one perhaps or about one.
And this is kind of an interesting variable.
You see it's the log of the product
of the saturated hydraulic conductivity in the linear extensibility.
What's that all about?
Well, in order to make something that's going to work around the country, you know,
vertisols are important part of the farming community in Texas.
And since we're dealing with saturated hydraulic conductivity, well,
a vertisol saturated hydraulic conductivity is very low but it has a lot
of cracks because the LEP is pretty high.
And so what we're really after with this is just the gas transmission.
So that's why we have the KSAT times a linear extensibility and you can see.
And then in the logarithm of it, that just presses it down into a usable framework
or usable piece of real estate on a graph.
So you see a lot of interesting relationships on just this one graph.
You know, you have a definite maximum here of rising limb and a pretty drop rough falling limb
as soil becomes faster and faster in their permeability.
And so, just relationships that are in the database.
Some things aren't independent of one another which always makes it interesting
because rainfall and for mean annual air temperature and mean annual precipitation
and the domain of corn walk hand in hand.
And so-- but what is really limiting to the corn growth is probably not the rainfall
as much as the temperature.
And so it's just, you know, lots of relationships, lots of interdependencies.
And so, one of the things I want to look at is just the shape of the curve and that's part
of the problem with this is also is one of the opportunities is using the data,
the yield data that's in the database.
The whole reason for having the index is because the yield data is flaky.
So here we are with the yield data coming up in the development of the model, but really,
just looking at trends and where things are peaked out or where they are at minimum.
And so, we're looking at a rain or comparing soil properties to productivity.
And we've looked at that enough then we can finally decide to put that into as a fuzzy set
or an evaluation in the NASIS database.
So you see, there is our nice curve here, peak around one.
So we translate that into an evaluation, and peak around one gently rising
and steeply falling after the maximum.
So we capture those relationships and those trends in the data
and trends in the relationship really.
I don't have any R-square stuff on here but just this little thing here, this Spline function
as an R-square of like 10 percent.
So from our per yield production, 10 percent of the variation is just based
on KSAT times the LEP from 50 to 100 centimeters so that's kind of interesting.
What sort of data does it use?
The NCCPI uses quite of a bit of data in the database.
One of the biggest things is the Root Zone Available Water Holding Capacity.
We also look at the Bulk Density because that's one of those things that most
of the time isn't a big player but in the extreme, it's important.
Saturated Hydraulic Conductivity is like that also to an extent.
Linear extensibility is kind of interesting because that's a two-tailed curve.
If you look, LEP can be too low, for high productivity and it can be high.
So it's, you know, kind of like not necessarily a bell-shaped but it's got two tails.
Rock Fragment Content, Rooting Depth, Rooting Depth has been pretty disappointing.
The data there is pretty flaky and I don't know quite why.
We look at the sand, silt, and clay percentages for physical data.
Chemical data, we look at the CEC, pH, Organic Matter Content,
SAR, Gypsum, Electrical Conductivity.
Climate data or landscape data, I'm sorry.
Slope Gradient and Shape are important.
Ponding, Flooding, Water Table, Erosion, Surface Stones, Rock Outcrop and other phase features
like channeled, whatever that means, you know, it's used in some places around the country.
They're not everywhere but it seems in some places, it's correlation thing
and in some places, it's very important.
In other places, it's not recognized the same way so that's been kind of the challenge.
Climate data and so the model uses, Mean Annual Precip, Mean Annual Air, Frost Free Days,
MLRA and the Soil Temperature Regime.
The MLRA is kind of a peculiar thing because for winter week,
the best area for winter week is in a xeric climate.
And so, I needed to be able to capture the areas where it's a zero climate.
Well, sometimes that shows up in the Soil Taxonomy name.
In other places, a lot of times it doesn't but it's usually more
or less constantly within an MLRA.
So I picked up the xeric nature using just a look up table of Major Land Resource Areas.
It's kind of clunky but it seems to work.
So how does it work?
Well, NCCPI looks an awful lot like the story index, the fuzzy numbers or the scores
for the property are-- they're multiplied together.
One good thing I think is one low property score can drag down the whole score.
So if you have a really high or really low pH where everything else being just fine,
the pH is what's going to drive the bus or drive the score for the overall index.
It uses a lot of hedges in the fuzzy system to modify the fuzzy numbers and its three factors,
corn and soybeans, winter wheat or small grains and cotton.
So, the highest of those three as taken is the score for a component.
But most of the time for scientific analysis, really,
one score or the other is more informative than the overall score.
Although for programs, then I think the overall score is probably more useful.
Oh, that's a pretty busy looking graph there, graphic but that's the corn and soybeans model.
And see there is a lot of different things going on, the chemical properties.
The water holding capacity is broken out.
The water supplying subrule as its own subrule is a major part of productivity.
Climate, landscape, interpretable component just is a switch to be able
to not let certain components repeat and produce an index.
I don't know.
Sometimes, you know, tax on above family are-- seemed to be far and the other place is not
so where the miscellaneous area is usually not some tax on both family are too.
Physical properties, one of the new things here I did for version 2 is I used to just add
or subtract a twist plow a positives and the twist negatives.
And the reason why these are twists is because there's not a really good way
of making a Spline fine ocean or some of these other things.
So they are more of a Boolean, you know, it's either channeled or it's not channeled.
And so, using the not hedge is where it's one minus the fuzzy number.
So if there's zero crisp negative attributes and one minus zero, well,
it's just one so the product of one times any other number is just the number you started
out with.
So that's what-- that's how that actually made quite a difference rather than just adding
or subtracting it from the score.
It's multiplied to the inverse of it-- not the inverse but, well,
it's kind of the inverse but not in this way of thinking.
It is just multiplied by the overall score.
And this is just a graphic that shows the overall model.
It's just the three pieces, corn, soybeans, small grains, cotton or operator.
It means that we take the highest of the three.
And if you can come up with a better way of doing that, let me know.
You can't really, you know, you don't want to use the lowest because over most of the country,
cotton comes out as a big fat zero because cotton has got to be thermic or it doesn't grow.
The waited average-- maybe a weighted average of the two, there are three that are relevant.
I don't know but right now, I just use the highest one.
So what's different between version 1 and version 2?
I borrowed this thing called sufficiency from Missouri and it's kind of interesting concept,
and it's basically what sufficiency does is prorate a data element,
see available water holding capacity with depth under the assumption that water close
to the surface in a soil is more valuable to a crop than water at some depth
like a 150 or 100-- 200 centimeters.
And so it's kind of prorates the value of that property with depth.
But available water holding capacity is one that really came out.
It was like, you know, 5 percent difference in the R-square
between just a flat available water hold--
recent available water holding capacity like I've always use it
and using the root zone sufficiency.
Particularly for corn, less for smaller grains
and a little bit less for cotton because, you know?
They're simply-- some of these crops don't require the water that corn does.
And I've mentioned that the score from negative soil attribute is handled differently now.
I've done some different things with the way seasonal soil wetness,
particularly in cotton [background noise] 'cause--
[ Noise & Inaudible Remark ]
-- It's kind of interesting.
This mostly happens in cotton.
Cotton is really kind of whole-- is a very adaptive, an adaptable crop so,
ranging from 2000 millimeters of rain down to about 500 millimeters of rain.
So, it covers so much territory that the genetics of it has got to be different
from place to place and its response to pH is different.
Although, we just call it cotton so we don't have this pronunciation here, cotton.
Some cotton apparently can handle a low pH
and other cotton can handle a high pH. It's not the same cotton plant though.
But there's a few other different ones.
I think that's different, I think there's probably a few others but those are the big--
>> Bob, you're breaking up.
>> Well, it's not me son.
I don't know what's going on here.
Somebody has got-- and I don't-- maybe--
[ Noise & Inaudible Remark ]
So how good is it?
Well, that's a big question.
That's the right answer.
We have three different lines on here.
The orange line is just a linear fit.
This is my idealized thing because I wanted some of, that's a linear relationship
between the cotton index and the yield or the corn or the--
whichever index to the yield, excuse me.
And so, that's the orange line.
I guess that's orange.
The red line is smoothing spline that's--
[noise] and so, it's basically a linear regression using these squares.
The green line or the teal line is an orthogonal fit.
Okay, that's all pretty dysenteric.
There might be somebody in the crowd who can understand this but the reason
for that difference is that linear regression assumes that there's no variation in X
which means you've measured X without any variance.
If you say it's one, it's one.
If you say it's two, it's two.
But we know it's a soil data, we're looking at estimates of estimates,
so we got a lot of variance in there.
And so, a linear regression, if you're using soil properties
out of NASIS's linear regressions, you're violating the rules there.
And that's why you see the red line doesn't seem to follow the axis of the point,
it's because it's-- there's too much--
the variance isn't accounted for with the linear regression.
When you're fitting X, you're fitting Y to X. So, the green line,
does this sound about right Clayton [phonetic]?
The green line here, the teal line we're fitting X to Y. So we're looking at more,
the index being the dependent variable and the yield being the independent variable.
And so, it has a slightly different look.
So I figure if I look at the volume of points within this, then that's, you know,
which is showing, you know, that we're doing a pretty good job of holding down the variance.
But you see, there's a lot of points all over the place.
And sometimes when I get really bored, I start packing out, what's going on?
Why is that this point out here in left field?
What's going on there?
Is that, you know, a breakdown in a model?
Am I missing something important?
Or is it fat-finger error, or just-- what's going on there?
And so, we come up with the "Poster Child" for "data harmonization."
So, when we look all this one point, it's way out in the left field here,
because it's actually frequently flooded.
And we don't usually have, you know, better productivity
on something that's floods every month of the year.
And so, these 900 bushels per acre or pounds per acre is probably a little optimistic
for that particular soil.
So we go to the next one that's over here and we see that they had a 165 frost-free days.
I can't remember which component it is.
Well probably, it will be good to see any ways but-- by cotton's entry level,
you can even be able to get a crop, there's around 180 frost-free days.
It likes to have at least 200.
So, we probably had a map unit that was dragged, drag, hold,
correlated from the northerner's place down to a southerner place.
The yield was touched but the frost-free days weren't.
You weren't-- with the frost-free day value wasn't change.
And so, there's a kind of a harmonization or data checking issue.
So, we'll look at a couple more of these things.
'Cause really, we really got going on this 'cause I just wondering why is the R-square,
when everything looks so good in the model,
why is the R-square less 0.5 with this-- with the yield data?
And so, I think well, sometimes the yield data needs to be updated.
Here's this soil [inaudible], some of the older surveys, 300 pounds per acre,
this is cotton again, some of the newer surveys, well 900.
So that's the up and down variation.
You see up and down the yield scale variation.
From left to right variation, it's a little more-- can be a little more problematic.
So what's the left and right variation?
Well, sometimes, you know, if you had slope,
soil slope is increasing then you expect the index to go
down the slope and that's got to be okay.
Sometimes, soils are eroded and we'll see one of those.
Sometimes, it's, you know, the yield data is also not just vantages but also,
states have different conventions of how they want to have yield populated,
or how they even think of the yield.
For instance, if you have the famous soil in Iowa, then they're looking
like 220 bushes per acre is the populated yield for Tama.
So you step across that line into Illinois and you drop down to about 180 bushes per acre,
same soil, same everything really, just the difference in the conventions of the state.
I will not be surprised if the land grant, you know,
the extension people will have some influence to every kind.
Some places, they might advocate a little more fertilizer use and others are yeah, whatever.
And so, you can have yield differences there, up and down.
The vertical difference, this one was pretty amazing.
And so, I thought I would track it down.
And here is what we found out with this one.
But the only difference in these components, are the frost-free days.
When you look at the distribution of the points, you know, you have one out here by itself.
It's actually several points.
There's about 30 or 40 squares lighted up.
But the distribution, the same kind of distribution,
that's the only thing that's different between those soils is the frost-free days and whether
that should be or not, I don't know.
But certainly, the ecologic variations aren't being accounted for.
That, you know, the frost-free days score of 0.26 as opposed to 0.90 something.
So that's another data checking issue.
And here's a couple more of these, I won't do this too much more.
Cecil, some of you have met cecil so you know what cecil is,
just a nice big blast right in the middle.
So you know, there's some variation left to right, it can be a cecil or side erosion
and sloped issues, and stuff like that so you expect some left to right variation,
if you have different vantages, different philosophies of date of population
or year of population to give you some of the vertical variation.
Same thing over here on Amarillo, nice, big [inaudible].
Pretend someone is trailing up here, I don't know quite exactly what's going on with those.
I don't know if those were actually a yield that somebody reported somewhere
or if it's just something peculiar going on.
But all these kind of things helped to decrease the R-square of the model that actually,
I think, captures the salient points, salient features, salient relationships
of the soil properties to crop productivity pretty well.
And there's Uchee, as data is harmonized, the shapes, maxima, minima,
and various curves probably have to be re-evaluated or that, either that
or we get a lot more of data from independent sources
but then just getting a hold of that data has been difficult.
It's out there but quite a task to get it together.
All right, and what's the future of this thing?
Well, as I mentioned before, the next step is to get it under Datamart.
And that will give a real chance for some serious bedding.
And also, this is kind of a heads up to look at the output of the model, you know,
how it's arraying the soils in your areas.
So as-- which you're familiar and just say, "Well, you know, this doesn't look right here,"
or, "It looks pretty good", you know, if it doesn't look right, let me know.
And, you know, we can see what-- see if there's something missing or what's going
on to be able to adjust for that.
To learn more about the NCCPI, if we go to the technical references page on soils.usda.gov
down towards the bottom of that page, there's the NCCPI user's guide that you can look at.
That's a couple of three years old now but really, some things have changed
but the philosophy is still pretty much the same.
Okay, then I just want to have-- that that does it for me there.
I just want to say thanks for listening.
And if you have any questions in the caption and you can't see these pictures, use no till
or the ag police are going to bust you.
This is a police tractor.
I've never seen one of those before, police tractor.
So anybody have any questions, or comments, or anything for me?
>> Bob, quick question, this is Lesley Glover Adam [phonetic], Arizona.
>> Okay, yes?
>> I noticed with the-- with your climate model
that you were using thermic for the time and production?
Did you also include hyperthermic--
>> Sir, yes sir.
>> Okay, yeah.
>> Thermic and hotter.
They also take in isothermic and hyperthermic but I don't think it gets isomesic.
I think mostly because the isomesic in this country is probably in Hawaii
and then I don't think they just-- they don't grow cotton in Hawaii.
>> Not that high-- its high elevation, yeah.
>> And so even though it could possibly grow there, it might not get hot enough
but we're saved by the geographics of--
it just doesn't grow and where that temperature even exists for isometric.
>> Hey Mark, this is Paul Benedict.
Also, some of your desert areas are going to have--
they may be thermic but their nights will be too cold for the cotton to grow well.
>> I think most of those places are going to be a little bit on the extra dry side too.
So I can't remember what the cut off is.
It's like if I want to say 4 or 500 millimeters of rainfall--
>> Right, which brings out the issue of irrigated, you know,
I know you've been working on an irrigated--
>> Yeah.
>>-- index too.
What's the status on that one?
>> Well, I've got that as good as I can get it by my own.
I've got my straw men set up again, and it's ready for some pot shots.
I suppose we'll get more of that out-- more out on that later.
But if you-- anybody with some NASIS access can look at it.
I have reports written so you can easily--
and for dry land too, there's a lot of report is written on Pangea if you have NASIS.
You don't even need skills for these.
All you need is access because they have it down to the point where you--
all you got to do is, you know, hit the button that says, run against--
run offline against the National Database.
And if you can just remember the first two letters that your state FIPS,
you can put that in with an asterisk and run, you know, a variety of reports
with just those two pieces of data.
>> I thought like somebody at the headquarters could run that one?
>> I don't know, I think so, yeah.
No, actually yeah.
But the other thing I wanted to mention on this is to really get some analysis
like you saw on my graphs they have in this.
I used JMP Statistical Analysis Software, I mean, and that's something we've got to get
into pretty much everybody that wants its hands.
I wouldn't say everybody needs to have it but if you have a, you know, want for it, you know,
we need to get that to people because it's really neat to be able to see relationships
of soil properties against other soil properties because when things are, you know,
somebody is a little bit off the reservation, boy, it really shows up in a hurry
because you have a big cluster points with a couple of outliers
and you think, "Well, what's going on there?
Is that a fat finger thing or is somebody, you know,
freelancing a little too hard, or what's going on there?"
And you can, you know, you can-- being able to see your relationships
between properties is pretty good.
They also have, you know, scripts to do that, to pull properties, you know, in groups.
And then-- and also to map properties, just to soil property output to--
with a mapping the key and just put it in a like a GIS product with it.
All kinds of things, so you can-- going to get the idea
of how soil properties are affecting land use or in this particular case,
soil productivity to GIS products and as well as the statistical analysis software
because Excel is, I don't know, then the later versions of Excel get
over 300 data points on a graph, I don't think so.
With JMP, you can put in.
I have as like 2 or 300,000 data points for some of these graphs that I'll show you.
>> Bob, this is Karyo Lewis [phonetic] out in San Joaquin Valley up in California.
>> Yes.
>> And-- yeah.
Hi, I had a question, recently a few of the--
from risk management agency folks have been talking about dry land wheat?
>> Uh-hmm.
>> And especially in the thermic condition because that's where a lot
of ours is growing out here in the xeric area.
>> Yeah.
>> I was just wondering if you would address that a little bit further because you mentioned
that without of the comfort zone and you're probably right.
>> Well, I think the xeric is probably going to override the thermic
because just having wheat is, you know, winter wheat it's got the, the it's--
it's in the ground when the ground is moist.
And about the time the water turns off, it's, you know, ready to be harvested or close to it,
so it's kind of a really nice fit for a xeric climate.
I'm not pretty sure how the thermic works on that although we can look at the wheat index
and see how that kind of goes, you know, because-- yeah.
>> Yeah, I think in our case, a lot of them were pushing it down on the lower end maybe,
you know, below 10 inches of rain so those areas are getting pretty marginal and, you know--
>> Yeah that's still--
>> You know, the vast majority of things out here, of course, are irrigated, you know,
one county would, you know, close to 7 billion dollars in ag production, so that kind of thing.
But there are those edges where the dry land has grown.
So-- thanks.
>> Yes sir.
[ Pause ]
Any other questions out there?
>> Is your PowerPoint available anywhere?
It can be.
Do I make that generally available, Mark?
How do you handle that?
>> We'll be posting this by tomorrow afternoon.
>> All righty.
>> PowerPoint and the recording.
>> Okay there's your answer there.
I was afraid that you're going to videotape this and then at which case, I was going to say,
"No thanks, Mark, where are they posted, Mark?"
>> You'll get that information in the email.
>> RSVP.
>> Hey Mark, I got one more question.
You know, with the drought being a common thing in the Southern Texas this year, you know,
if you looked at a-- given any kind of a confidence, what's the right term, you know,
it's accurate, 80 percent of the time or--
>> Yeah. You know, I've dealt with at some-- thought about at some, as you can tell,
you know, we're looking at static time of data just to get the index.
You know, we're not really trying to predict a yield.
>> Right.
>> So we get those off of our back.
So I think in most cases, you're looking--
it's probably pretty good, plus or minus, or half a unit.
So if you had a .6, yeah that's a .55 to a .647 in that ball park.
I have done some experiments where I've changed the rainfall but, you know,
like I say that nothing doesn't co-- everything co-varies.
So, well, you know, you don't just decrease the rainfall or increase the rainfall
because when you do those, you're probably also having some mean annual air temperature
or frost-free day effects as well.
And at least in corn, you know, in the domain of corn,
as you increase the temperature, the rainfall also goes up.
I don't know if that's always going to be true but that seems to be kind of hold true
with what the climate change people say.
You know, they might get more severe storms but in general,
the rainfall increases with the temperature.
And so, it gets to be a little bit-- more of a mental gymnastics exercise to try to think,
"Well, you know, if I'm going to increase my rainfall, then as the temperature going to go up
or down, stay the same or, you know, so I guess you'd have
to put together several scenarios for that."
[ Pause ]
>> Anything else?
[ Pause ]
Any questions from the Peanut Gallery [phonetic] here?
No? Well, thanks again for listening.
And if you have questions on this, please give me a call, you know.
If you see something that looks flaky or something you just plainly don't understand,
and there's a lot of refined structure we didn't talk about today.
So if you see something you-- what the heck is he doing here?
If you, you know, just give me a holler and we'll talk about it.
>> Bob, this is Clive Owen [phonetic]
>> Yes Clive Owen.
>> The irrigated productivity index, are you going to put a bullet
from that and request a review by state?
>> I think so.
Is that-- that's about right, Mike.
We'll probably have a boat--
>> Yeah, it should be out soon.
I think we're-- Bob, this is Mike [inaudible].
Bob says, we're pretty-- we're darn close to good to go.
So we'll get that out the door here pretty quick.
>> Okay.
>> Any feedback is much, much appreciated.
So--
>> Yes. But the irrigated thing and, you know, once you start filling water on, you know,
a lot soil properties, you know, unless they're in the extreme,
they just don't make much difference anymore.
You know, it would-- but still, I think though, that the management has to be more difficult
or more informed as the soil properties become less and less-- [Silence]