Spatiotemporal Prediction of Soil Properties and States in Variably Saturated Landscapes

Trenton Franz Author

02/15/2018 Added

53 Plays

Description

2018 Water Seminar

Searchable Transcript

Search:

[00:00:00.152]Hello, can everybody hear me?
[00:00:01.640]All right, excellent.
[00:00:02.951]So, I'm Trenton Franz, I recognize a lot of you
[00:00:06.520]from the department or from my various classes.
[00:00:10.569]Today I'll be speaking about some
[00:00:14.039]of the hydrogeophysical work that I do.
[00:00:17.189]Given the theme of the seminar series,
[00:00:20.880]I'm gonna try it in, tie it in to, excuse me,
[00:00:23.800]irrigation technology and some of the other seminars
[00:00:27.739]that you guys have already watched.
[00:00:29.920]A little bit of background about myself
[00:00:32.080]is I joined the department about five years ago.
[00:00:34.989]I did my undergraduate at the University of Wyoming,
[00:00:38.339]then went on to do a Ph.D at Princeton University,
[00:00:41.971]and then followed that up with a post-doc
[00:00:44.499]at the University of Arizona and then joined SNR
[00:00:48.499]about five years ago.
[00:00:51.400]I'm a hydrogeophysicist, I'll get into exactly
[00:00:54.578]what that means.
[00:00:55.896]I'm also a faculty fellow at the Daugherty Water
[00:00:59.093]for Food program.
[00:01:02.707]So how does hydrogeophysics really tie
[00:01:06.117]in to the theme of irrigation?
[00:01:08.477]And how can we use that to help make a decisions?
[00:01:11.427]This is a, when I'm thinking about research,
[00:01:14.517]I'm really trying to put myself in the shoes
[00:01:16.656]of your how can we use this kind of technology
[00:01:19.375]to help schedule irrigation.
[00:01:22.747]Thinking about things like how much we should water
[00:01:25.053]and where and how that changes in time and space,
[00:01:28.867]and sort of where some of the opportunities
[00:01:31.067]and what are some of the challenges of using some
[00:01:33.368]of these newer techniques around hydrogeophysics
[00:01:37.048]to give us some new information, particularly
[00:01:39.968]about soil properties and soil states
[00:01:43.747]that we actually need to make an optimal
[00:01:46.877]irrigation decisions.
[00:01:49.437]So this talk will get into some of that
[00:01:51.237]as well as more of a sort of a general background
[00:01:54.216]on hydrogeophysics and soil moisture in general
[00:01:58.636]and how that sort of impacts how we think about things
[00:02:03.557]like weather and just general ecosystems
[00:02:07.048]and how they function.
[00:02:09.717]So as you've seen a lot in the previous seminars
[00:02:12.735]as we're talking a lot about center pivot irrigation.
[00:02:16.077]So this is just a map showing various center pivots.
[00:02:20.237]Those of you who don't know what that looks like by now,
[00:02:22.768]this is what a sprinkler system looks like
[00:02:25.976]and so the first thing that I sort of notice in this
[00:02:27.967]is that there's large spatial areas that these systems
[00:02:32.037]cover, but often as we're limited by point scale
[00:02:36.376]observations and so we have this disconnect
[00:02:40.076]both in spatial scales as well as temporal scales
[00:02:43.716]to make some of these decisions.
[00:02:47.288]So irrigation scheduling 101, so I think about this
[00:02:50.075]from a soil moisture perspective and so when I think
[00:02:54.408]about irrigation as sort of the simplest way to do that
[00:02:58.008]if I can get this mouse working, is on this graph here
[00:03:02.917]or these two graphs, I have a center pivot time series
[00:03:07.128]of soil moisture so that's daily values.
[00:03:11.237]And on the y-axis we have volume metric water contents.
[00:03:15.048]So this is how much water is in the soil.
[00:03:17.687]In addition on this graph, in green I have something
[00:03:19.968]called field capacity which is the point
[00:03:23.256]at which gravity is really have after a few days
[00:03:26.814]of a big rain event, is that's the sort of the storage value
[00:03:31.256]where gravity is no longer acting significantly
[00:03:33.792]on that bucket of soil.
[00:03:37.005]The red line is the wilting point so this is the point
[00:03:39.005]at which plants will have permanent effects in terms
[00:03:43.565]of their water stress and so a nice watering strategy
[00:03:47.845]for most crops is when you get to about halfway
[00:03:50.605]between the wilting point and field capacity is you should
[00:03:53.565]trigger an irrigation event.
[00:03:55.474]In addition, the size of that irrigation event
[00:03:58.334]should fill up that entire bucket.
[00:04:00.423]So we look at these two time series as we see
[00:04:02.633]that this particular irrigator did do a very good job
[00:04:06.673]of keeping the soil moisture status at about 50%
[00:04:10.772]during the July and August period.
[00:04:13.540]This event could be a rain event, we went oh, sort of
[00:04:16.953]over topped field capacity so there mighta been
[00:04:18.892]some runoff or potentially leakage during that event.
[00:04:23.532]And when we compare that to the dry land time series
[00:04:27.063]is we see there are significant stress periods
[00:04:29.812]that we made up with irrigation.
[00:04:31.953]And so in a perfect system thinking about irrigation
[00:04:34.952]as a bucket and how much we, how large that bucket
[00:04:37.553]and how much water we have in that bucket is we can
[00:04:40.183]make optimal irrigation decisions.
[00:04:43.261]Also assuming we know something about the future
[00:04:46.161]of rainfall and potentially evapo-transpiration.
[00:04:49.760]We'll get back to this here in a few slides.
[00:04:53.441]And so the hypothesis that I'm really working with is
[00:04:56.943]that it's actually very challenging to actually measure
[00:04:59.712]soil moisture at the same scale at which we make
[00:05:03.273]decisions particularly when we use low cost sensors
[00:05:07.033]that really give us point scale information.
[00:05:10.743]And so a fundamental problem is how do we take some
[00:05:13.532]of these low cost sensors that act at a point
[00:05:17.401]and scale them up to areas where we actually make
[00:05:20.932]these decisions about irrigation.
[00:05:23.192]And so geophysics is one tool that allows us to derive
[00:05:27.332]some of these scaling functions as well as give us
[00:05:30.281]information through a different lens
[00:05:33.103]at a different spatial scale.
[00:05:36.572]Alright, so why is irrigation so important when we look
[00:05:39.239]at a map of irrigated acres across the U.S.,
[00:05:44.321]isn't Nebraska's we're number one in irrigation
[00:05:47.743]with about 9 million irrigated acres as of 2007.
[00:05:53.593]And in terms of water use for the state is about 90%
[00:05:57.091]goes for agricultural water use in the state.
[00:06:01.233]So it's a huge volume of the water that we use
[00:06:03.361]as humans goes to irrigation.
[00:06:05.633]So there's potentially a lot of water savings
[00:06:09.273]that we can attack through this efficiency of irrigation.
[00:06:15.892]So where is irrigation happening is we can look
[00:06:18.501]at a map of the density of irrigation wells.
[00:06:22.713]So there's about 93,000 in this state, a huge economic
[00:06:26.353]investment and most of that, a lot of those wells went
[00:06:29.921]in the 1970's with the widespread use of center pivot
[00:06:34.361]irrigation.
[00:06:35.361]And so there's a huge investment in center pivot
[00:06:37.941]irrigation.
[00:06:39.233]In total I think there's about 80,000 center pivots
[00:06:42.313]in the state.
[00:06:45.290]So where are we drawing all this water from?
[00:06:47.050]So the high plains aquifer, hopefully some of you have seen
[00:06:49.292]some slides on this.
[00:06:51.222]So it's a fantastic resource for this state and others.
[00:06:56.072]So it extends through eight states.
[00:06:59.490]On this map we show the spatial extent of the aquifer
[00:07:03.153]as well as the saturated thickness.
[00:07:05.543]That's very important is what is the size
[00:07:07.953]of that reservoir or that aquifer.
[00:07:10.292]So we're very blessed in the sand hills to have a very deep
[00:07:13.681]aquifer at over 1,000 feet and when you get into some
[00:07:16.862]of the middle and southern high plains part of the aquifer
[00:07:19.480]is that shrinks down to a couple of hundred feet.
[00:07:26.513]Uh-oh, broke it, there we go.
[00:07:29.033]So this aquifer system has about 3 billion acre feet
[00:07:34.412]of water and so it provides water to about two
[00:07:37.540]and a half million people and we think about what is
[00:07:40.903]feeding this aquifer?
[00:07:42.273]On average we get about one inch of recharge a year
[00:07:46.812]distributed over the entire aquifer.
[00:07:49.092]And so that means if we're pulling out 10 or 15 inches
[00:07:51.961]of water, is we have to be judicious in our use
[00:07:55.161]of that aquifer and we can't be doing that everywhere
[00:07:57.623]or otherwise we'd have no aquifer system left.
[00:08:02.281]So there's been quite a bit of work on looking
[00:08:05.240]at the draw down in the aquifer.
[00:08:07.861]If I can get to my next slide, maybe.
[00:08:14.292]I'll try the mouse, there we go.
[00:08:17.081]So we can look at water level changes.
[00:08:18.983]So this is a map of before industrial development
[00:08:23.553]and widespread use of irrigation and then sort
[00:08:27.353]of the yearly update of water level changes.
[00:08:31.161]So this is just showing water level change.
[00:08:34.353]So this red area is in feet here, so about 50 foot
[00:08:38.251]of draw down over in the upper Republican reaches.
[00:08:44.841]And we can look at this across the entire high plains system
[00:08:47.552]and we see parts of Kansas and the panhandle of Texas
[00:08:51.232]over 150 feet of draw down over time.
[00:08:56.241]And so really, the metric that you should be looking
[00:08:58.423]at is the ratio of draw down versus saturated thickness
[00:09:03.012]to get at things like percent depleted.
[00:09:06.993]So in the north, southern high plains, you're looking
[00:09:10.092]at well over 50% depletion of that available aquifer
[00:09:14.692]whereas in Nebraska and other parts of Northern Kansas
[00:09:19.062]is you have a lot less percent use of that actual
[00:09:22.401]aquifer system.
[00:09:25.731]And so, a little bit more, recent paper is we can look
[00:09:28.159]at the state of aquifers across the entire U.S.
[00:09:31.143]So this shows five major aquifer systems as well as the
[00:09:36.273]average depth to ground water.
[00:09:38.092]So in addition to the high plains being a huge volume
[00:09:40.541]of water it's also very shallow in terms of where the
[00:09:44.841]actual water table is. And that's very important
[00:09:47.241]for energy costs and how much energy and money
[00:09:50.532]it takes to pump that water.
[00:09:53.103]Also in this slide, we also have another graph
[00:09:57.732]showing what is our rate of water decline
[00:10:01.440]across the U.S.?
[00:10:02.952]So the top graph shows shallow wells whereas the
[00:10:06.572]lower graph shows deep wells.
[00:10:08.741]And so this is really, sort of the, you know, flashing light
[00:10:12.531]is when we have significant draw down over time
[00:10:16.359]in these deep well systems.
[00:10:18.041]So we see this red blob really popping up
[00:10:22.041]in the southern high plains aquiver at about a meter
[00:10:24.991]per year of decline between 1940 and 2015.
[00:10:29.340]We also see other aquifers around the U.S. having similar
[00:10:32.711]rates of decline.
[00:10:34.260]So this is certainly a worrisome trend on this rate
[00:10:38.180]of decline as some of these key aquifer systems
[00:10:42.703]So just summarizing, the motivation, a lot of you probably
[00:10:45.561]heard about growing up population
[00:10:47.778]so we expect about a, you know, a doubling
[00:10:50.217]of serial production by 2050 and so I'm really curious
[00:10:54.409]about how are we gonna do that in terms of the availability
[00:10:57.881]of water as well as protecting some of the integrity
[00:11:01.074]of our ecosystems as we use up more and more
[00:11:04.665]water for human use.
[00:11:07.774]So about on average globally about 70%
[00:11:10.051]of water consumption is used for agriculture.
[00:11:15.044]In terms of food production, about 40% comes from
[00:11:19.073]irrigation agriculture where aggregated or, excuse me,
[00:11:22.356]irrigated land only covers about 20% of the land surface.
[00:11:25.878]So we get about two times as much food production
[00:11:28.526]for an irrigated patch as we do a rain fed area.
[00:11:31.197]So certainly that's, you know, very very nice that we can
[00:11:35.193]use less land to get more food production
[00:11:38.235]if we include irrigation in that strategy.
[00:11:43.050]In terms of efficiency, again, if we're thinking about water
[00:11:47.339]that is used for evapotranspiration and yield gain
[00:11:50.969]is when you look at some of these large numbers
[00:11:53.811]only about 60 percent, excuse me, about 60% goes to
[00:11:57.496]non-productive evo transpiration so that waters
[00:12:00.757]not technically wasted.
[00:12:02.565]Some of it is returned back to the system so it's
[00:12:05.226]still beneficial water.
[00:12:06.616]But certainly there's some gains to be made
[00:12:09.267]in terms of the efficiency of this, of this process.
[00:12:14.387]Also sort of a last point that I think a lot about is
[00:12:17.106]these land surface models.
[00:12:19.507]I'll get into that, what I mean by that here in a second.
[00:12:21.445]But until very recently, so there's a nice 2016 paper
[00:12:25.917]that basically has argued that up to now we really
[00:12:29.637]haven't included irrigation in any of these models
[00:12:33.256]and a lot of these get used for weather forecasting.
[00:12:36.135]And so this feedback of what's happening at the land surface
[00:12:38.877]and the atmosphere is we're not taking that into account
[00:12:41.837]in a lot of these land surface models.
[00:12:44.885]So that's being resolved, it's actually quite challenging
[00:12:48.347]to put some of these schemes into land surface model
[00:12:52.475]especially operationally but that certainly has a
[00:12:55.113]feedback on the atmospheric prophecies.
[00:12:59.195]Alright, so gettin' back to my hypothesis again, is we have
[00:13:03.016]this disconnect between these low cost centers
[00:13:05.776]that we often use to schedule irrigation and the areas
[00:13:09.885]that we actually wanna make these decisions on.
[00:13:13.422]So the solution is actually quite easy.
[00:13:15.485]You know, apply water only when and where we need to.
[00:13:19.016]If it's raining outside, you should turn off your sprinkler.
[00:13:22.405]If it's raining in the next county
[00:13:24.727]and you know it's raining, maybe you should manually
[00:13:26.569]shut off your pivot.
[00:13:28.047]But it turns out that that's actually quite challenging
[00:13:31.420]in terms of being able to get these observations,
[00:13:33.939]both rainfall, evapotranspiration, and also having some
[00:13:38.081]of the technology to have, to make that remote decision
[00:13:42.655]in operation as well as be confident
[00:13:44.563]in making that decision.
[00:13:48.020]And so, one limitation is that as we talked about
[00:13:51.523]in my hydrology class last couple of, the last week.
[00:13:54.932]Is that evapotranspiration measurements are really
[00:13:57.871]representative of an area or an entire field.
[00:14:01.462]So we have a slew of different methods
[00:14:03.374]that we can use that.
[00:14:04.964]But when we match that up with what we can do
[00:14:06.774]in terms of low cost soil moisture sensors
[00:14:09.284]is those are really limited to a point a estimate
[00:14:12.421]so that we can do.
[00:14:13.273]And so this is a pretty widely used sensor in Nebraska.
[00:14:16.900]This is a rometer watermark sensor and so it sees the volume
[00:14:20.538]of about a basketball in terms of its measurement scale.
[00:14:24.917]And so the key question is how representative are these
[00:14:27.756]point observations and how can we potentially scale
[00:14:30.858]those up into areas that we actually make decisions on.
[00:14:36.528]And so that's really sort of the crux of what I study.
[00:14:40.566]It's sort of the hydrology, it's the, you know,
[00:14:42.777]the basic conservation equations, mass, energy,
[00:14:46.117]momentum.
[00:14:46.957]Thinking about some of the ecology agronomy
[00:14:49.576]of this system or the mechanisms and processes
[00:14:51.669]driving that and then using geophysics to actually
[00:14:54.837]get us some better observations and working between
[00:14:58.146]these three different areas to come up with some
[00:15:00.446]nice applied solutions to some of these problems.
[00:15:06.248]And so a couple main points from this lecture
[00:15:09.957]is that really, when we think about the next generation
[00:15:14.514]of land surface modeling, whether that be
[00:15:16.914]from hydrologic as well as biogeochemical cycles,
[00:15:20.383]I won't talk about that too much today, is that our
[00:15:23.234]poor parametrization of what's happening with the soil
[00:15:26.064]and flow of water through that is really a huge
[00:15:29.714]limitation that we have that really needs
[00:15:31.816]to be addressed in the next set of land surface models.
[00:15:37.416]Second is that hydrogeophysics does offer us a suite
[00:15:40.794]of tools that allow us to get some of these spatial
[00:15:43.943]data sets that we really need to make these optimal
[00:15:46.415]irrigation decisions.
[00:15:48.394]However, this, you know, comes at a cost both
[00:15:50.685]in terms of time, effort as well as money and so we have
[00:15:53.903]to find certain situations that it makes sense
[00:15:56.925]to apply these geophysical techniques.
[00:16:00.536]Third is that we really need to consider both space
[00:16:02.824]and time and so there's some statistical tools
[00:16:05.742]that we've developed that we can use to really merge
[00:16:09.856]some low cost point sensor information with some
[00:16:12.975]of these spatial data sets to come up with a better
[00:16:15.704]spatial and temporal model of what's happening
[00:16:19.416]with the system.
[00:16:21.456]Alright, so how do we think about putting together a
[00:16:26.005]simple model for what's happening in the subsurface?
[00:16:29.193]So I'll start with the bucket model so the diagram
[00:16:32.043]on the left shows sort of the cartoon version
[00:16:35.434]of what's happening.
[00:16:37.103]So we have our intermittent rainfall, a stochastic
[00:16:39.704]process so we have, we know it's gonna rain, you know
[00:16:42.944]probably about, you know, four or five inches
[00:16:44.816]in June but we don't know when that rainfall
[00:16:46.734]is gonna arrive and what amount that rainfall
[00:16:49.314]is gonna be.
[00:16:50.147]Is it gonna be a few large storms or a bunch
[00:16:52.154]of little storms?
[00:16:53.656]And so we need to be thinking about this rainfall
[00:16:57.776]coming in as a stochastic process through probability
[00:17:01.483]of these kind of events.
[00:17:03.765]And so when it rains that water gets partitioned
[00:17:06.114]into a few different places, you know, it hits the canopy
[00:17:08.884]and may evaporate.
[00:17:10.056]Some of it may run off the surface.
[00:17:12.274]Some of it may end up as recharge or leakage
[00:17:15.624]through the system and some of it may sit
[00:17:18.184]in this bucket that we have for water use later on.
[00:17:21.976]And so we can turn that into a very simple
[00:17:23.554]mathematical model.
[00:17:24.994]So this is just the model that I'm working with.
[00:17:26.776]So this is the Z is the size of our bucket, theta
[00:17:30.525]is the volumetric water content, T is time,
[00:17:33.583]and then we have this loss function of this system
[00:17:36.646]that's shown by the graph.
[00:17:38.278]And so this is the sum of both evapotranspiration
[00:17:41.529]and leakage.
[00:17:42.729]So we'll start at the sort of, very wet end of the graph.
[00:17:45.948]So if we have very wet conditions, is we're gonna have
[00:17:49.068]both evapotranspiration as well as water loss
[00:17:53.209]that's draining through that bucket.
[00:17:54.849]That's the L term.
[00:17:56.558]And as we start to get under drier conditions, we're
[00:17:59.108]sort of in this key zone for irrigation scheduling
[00:18:02.946]between field capacity and about half way to wilting point
[00:18:06.649]is this is where we're at maximum evapotranspiration.
[00:18:09.628]So this is really where we wanna be.
[00:18:11.620]We're not having too much water draining through the soil
[00:18:14.208]and we're not also effecting the vegetation
[00:18:17.248]in terms of the stress on the crop.
[00:18:20.357]And as we get to lower water contents we start to see
[00:18:23.238]reduced evapotranspiration, so we're getting into stressed
[00:18:25.985]conditions and eventually it gets so dry
[00:18:28.638]that you actually can't pull that water out of the system
[00:18:32.100]through the pressure forces that are acting on it.
[00:18:35.398]So that's a very simple model and so you can integrate
[00:18:39.788]this curve and you can come up with a nice,
[00:18:42.049]closed form solution of what's happening in terms
[00:18:44.457]of your model.
[00:18:48.777]In reality, when we think about a more complex system
[00:18:52.438]as we often use the Richards Equation.
[00:18:54.617]So, I'm not gonna go into too many details, but this is
[00:18:56.177]what that equation looks like.
[00:18:58.828]But the main point of this slide is that we now have
[00:19:01.737]two functions that we need to consider when we're thinking
[00:19:04.954]about soil processes.
[00:19:06.046]We have a relationship between pressure or tension
[00:19:08.827]in the soil versus water content and then the flux
[00:19:12.238]of water moving out of that system as a function
[00:19:14.637]of water content.
[00:19:16.337]And so, a key thing is that this is on a log scale
[00:19:19.577]and I usually get this nice, non-linear function.
[00:19:22.278]And so we really need about four or five parameters
[00:19:25.318]to describe the shapes of these curves when we
[00:19:28.438]start to think about a physically based model
[00:19:31.118]of this system.
[00:19:33.177]So, where did we get this kind of information from?
[00:19:35.337]And so one source is the U.S.D.A. soil texture diagrams.
[00:19:40.665]For those of you not familiar with this, is if we go out
[00:19:43.217]to a site, take soil samples, we can bring them back
[00:19:45.926]to the lab, sieve those samples and we can get
[00:19:48.297]a distribution of weight percent of sand, silt, clay,
[00:19:52.117]bulk density, and certain properties that we can
[00:19:55.526]easily measure.
[00:19:57.518]And the U.S.D.A has put together this triangle
[00:20:01.086]where you have your three textural properties
[00:20:04.135]as you can plot those on this triangle graph
[00:20:07.285]and you can come up with a general description of what
[00:20:09.806]that soil is.
[00:20:10.748]So there's about 12 key textural classes that people
[00:20:14.276]will describe for the type of soil.
[00:20:18.388]So that's what shown on this slide diagram, but when
[00:20:21.164]we think about turning this into actual properties
[00:20:24.897]or processes like infiltration and drainage,
[00:20:28.438]is it turns out there's a lot more gradation
[00:20:32.187]as well as some of these boundaries don't necessarily
[00:20:34.278]line up with the physical processes.
[00:20:37.468]And so the graph C here shows sort of the infiltration
[00:20:40.358]curve over laid with the soil texture triangle, drainage
[00:20:45.188]and then we do things like infiltration and drainage
[00:20:47.566]or infiltration plus drainage as you have different soils
[00:20:52.286]acting differently depending on what they're exposed
[00:20:55.676]to in terms of it's precipitation and drainage time.
[00:21:00.006]And so it's a, sort of, good first sort of step
[00:21:02.838]as these textural triangles but when we really
[00:21:05.216]get into some of the finer scale information,
[00:21:08.718]is we need more information.
[00:21:12.068]So this is an interesting study that Nearing led,
[00:21:16.318]sort of making this point where they were looking
[00:21:19.005]at weather forecasting skill of some of these
[00:21:22.165]continental scale models and so they ran some of these
[00:21:25.577]type of models and then they compared the results
[00:21:29.606]with a continental network of scanned sites.
[00:21:32.708]So the red dots show soil moisture at different depths.
[00:21:36.566]And Ameriflux sites are where we have these Edicovariance
[00:21:38.878]towers where you get evapotranspiration estimates
[00:21:42.758]of the land surface.
[00:21:45.396]And so, using some information theory is what Nearing
[00:21:49.166]showed is that basically is where is our loss
[00:21:52.016]of information in the system.
[00:21:53.457]Is it the model, is it the parameters, is it
[00:21:55.588]boundary conditions?
[00:21:57.308]As it turns out for soil moisture, both at the surface
[00:22:01.217]and at depth is that a lot of our information loss
[00:22:04.297]about 60% happens with the parameters
[00:22:07.887]that we choose to put into this model.
[00:22:10.687]Whereas evapotranspiration is really, it's our
[00:22:12.767]uncertainty in what's happening in terms of our
[00:22:15.737]atmospheric conditions that pressure temperature,
[00:22:19.186]relative humidity and some of the other factors
[00:22:22.137]going on.
[00:22:23.187]And so this was a really nice way to show that basically,
[00:22:26.097]hey, we need better soil parametrization in these models
[00:22:30.199]if we're gonna want to improve some of our forecast skill
[00:22:33.176]and get better information.
[00:22:35.075]And so that's really sort of the cutting edge,
[00:22:37.155]is how do we link up some of these databases
[00:22:40.105]and information and put this into actual parameters
[00:22:43.736]in some of these forecast models.
[00:22:46.187]Alright, so besides sort of the available U.S.D.A.
[00:22:49.955]type of information, there's other techniques
[00:22:52.707]that we can use to get at spatial information.
[00:22:55.966]And so we can use light or specifically we can use
[00:22:58.755]radiation at different wave lengths to get at some
[00:23:02.507]of these, pieces of information.
[00:23:05.595]And so this graph shows the electro-magnetic spectrum
[00:23:10.486]and so we should all be familiar with visible light.
[00:23:14.466]So that happens between about 400 and 700 nanometers.
[00:23:18.611]But we also have a whole range of other wavelengths
[00:23:21.435]that we can view what's happening in our world
[00:23:26.626]through different wave lengths .
[00:23:29.065]And so my researcher is really using different
[00:23:31.955]parts of this spectrum and viewing the land service
[00:23:35.805]at different wave lengths.
[00:23:39.296]Alright, so physicists have figured this out actually
[00:23:42.288]and gotten lots and lots of funding for that.
[00:23:45.168]And so they have a constellation of these different
[00:23:48.478]satellites that are orbiting the Earth or on the ground.
[00:23:52.447]And so they have a series of telescopes that they can
[00:23:55.317]look at the same features.
[00:23:57.608]So this is just a series of maps of the crab nebula
[00:24:01.327]at different wave lengths.
[00:24:02.968]And from these series of images they can figure out,
[00:24:05.288]okay what is that actual constellation made of
[00:24:08.516]and what are some of the flows of these elements
[00:24:12.827]inside of the system moving back and forth.
[00:24:15.206]And so the point is that we can do the same thing
[00:24:17.526]by looking basically at telescopes
[00:24:19.987]from space or on the ground back at the Earth's surface
[00:24:23.877]and use those different wavelengths to come up
[00:24:26.166]with a better information about what's happening.
[00:24:29.477]And so we have a couple of different sensors
[00:24:31.427]that can do that.
[00:24:32.260]We have our TDR sensors that sort of happen
[00:24:34.587]at the point scale.
[00:24:35.878]We have our satellite sensors which use radar
[00:24:39.336]that we can orbit around and look
[00:24:41.227]at the Earth's Surface.
[00:24:42.307]It is the SMAP satellite for those interested.
[00:24:46.616]We also have some on the ground sensors
[00:24:48.606]that we can fill in some of these gaps.
[00:24:50.536]So one is electromagnetics and the other one
[00:24:54.638]is this cosmic-ray neutron probe that I work with
[00:24:57.725]and so we're using a different parts of the electromagnetic
[00:25:01.317]spectrum to get the same images of the same place
[00:25:04.867]on Earth and then turn that into useful information.
[00:25:10.166]So this is sort of the, one of my main research thrusts
[00:25:14.198]is using and developing this particular sensor.
[00:25:17.518]So this is the cosmic-ray neutron probe, so this is what
[00:25:21.246]a sensor looks like on the ground so it's a gas-filled
[00:25:25.096]detector.
[00:25:25.947]So this one is filled with boron tri-fluoride
[00:25:28.965]and what happens is neutrons are coming in from outer space
[00:25:32.147]interacting with water and hydrogen in the soil
[00:25:35.746]and some of those get spit back out into the air
[00:25:39.307]and then eventually detected by our device.
[00:25:42.885]And so we're just simply counting neutrons at certain
[00:25:46.278]energy bands and then converting that into soil moisture
[00:25:50.295]information.
[00:25:51.678]And so the cool thing about this detector is
[00:25:53.867]that the neutrons are mixing over about a 200 to 250 meter
[00:25:58.015]radius.
[00:25:59.176]And so neutrons are doing all of this averaging
[00:26:01.998]of local conditions for us.
[00:26:04.736]And so this is about the same footprint as a center
[00:26:07.718]pivot irrigation system.
[00:26:10.558]So the stationary probe, it takes us about an hour
[00:26:14.096]to get an observations.
[00:26:15.718]We have to count a certain number of neutrons,
[00:26:17.486]however, we can put a bunch of these detectors together,
[00:26:21.125]put them on the back of a ATV or in the back of a vehicle
[00:26:24.254]and we can get down to about a 30 second or one minute
[00:26:27.805]observations.
[00:26:29.046]And so I hire students over the summer to drive these things
[00:26:32.595]back around like a lawn mower across the actual land
[00:26:36.086]surface.
[00:26:38.665]So where did this, sort of information come from?
[00:26:41.825]So this was part of my post-doc was helping build up
[00:26:44.702]this global network of these detectors.
[00:26:47.503]So right now there's about 200 of them operating
[00:26:49.862]on all seven continents.
[00:26:52.314]And so they count neutrons over an hour then they report
[00:26:55.195]those neutron counts and then we turn that into a soil
[00:26:58.016]moisture product.
[00:27:00.075]And so there's this, the idea was to eventually use some
[00:27:03.337]of this information in real-time forecasting.
[00:27:08.686]So that's where we have some of these detectors.
[00:27:11.995]So some of my summer is spent around, you know,
[00:27:15.697]traveling to these various networks and working
[00:27:18.314]with their scientists and helping turn this
[00:27:20.944]into useful information.
[00:27:23.465]So a little bit more about the physics real quickly is
[00:27:26.265]that how this works is, again, we count neutrons
[00:27:29.446]over a certain energy band so this is the epithermal
[00:27:33.766]band.
[00:27:35.137]So if we have very wet soils, as we have suppressed
[00:27:36.955]neutrons counts.
[00:27:38.337]And if we have drier soils we get more neutron counts
[00:27:41.085]and so there is a direct relationship between neutron
[00:27:44.274]counts and what the water content is in the actual soil.
[00:27:51.355]Alright, so that's essentially what this information is,
[00:27:54.566]is we have basically two neutron detectors.
[00:27:57.795]One underneath a center pivot, shown by the top graph.
[00:28:01.143]Turn that information into water content.
[00:28:04.945]And we also have one in this dry land corner
[00:28:07.937]about 10 kilometers away showing, you know,
[00:28:10.977]obviously much drier conditions
[00:28:12.465]where you don't have irrigation.
[00:28:15.285]So, that's great, unfortunately the technology is
[00:28:18.909]about 20 times more expensive than what most
[00:28:22.366]producers can afford.
[00:28:23.614]So I have convinced one farmer in the state
[00:28:25.915]to purchase a sensor so that was great.
[00:28:28.795]But other than our one farmer, as we have to think
[00:28:33.126]a little bit more about how we were actually gonna
[00:28:34.755]make this information available and useful.
[00:28:38.725]And so I just wanted to show a study by my former
[00:28:42.507]master's student, Katie Thinkenbiner, finished up last year.
[00:28:46.045]As she was working at Palman Farms in Central Nebraska,
[00:28:50.234]and so this diagram shows the elevation of the site.
[00:28:54.245]And so it's a very undulating site.
[00:28:56.406]It has a full variable rate irrigation system.
[00:28:59.606]There's just a picture for what the site looks like.
[00:29:03.425]And so what Katie did was did a series
[00:29:06.185]of geophysical maps as well as soil core property
[00:29:09.145]estimation to get some spatial information.
[00:29:12.795]And so we can look at what the USDA tell us
[00:29:15.245]about and there are seven different soil types
[00:29:17.934]in this particular field.
[00:29:20.795]So again is a 160 acre field, you have seven different
[00:29:24.257]soil types ranging from basically beach sand at the top
[00:29:28.326]of the hill, down to a silty clay.
[00:29:30.545]So a huge amount of spatial variability in terms of the
[00:29:33.697]soil properties.
[00:29:38.545]So these are just an example of some of these maps.
[00:29:41.006]So this is our electrical conductivity map, so that's that
[00:29:43.895]yellow tube that we drag around behind.
[00:29:46.846]And so she just hooked up to the back of the ATV
[00:29:49.105]and drove around.
[00:29:50.474]Takes about an hour and 20 minutes to cover
[00:29:53.337]the entire field.
[00:29:54.777]And so we have our electrical conductivity map
[00:29:57.326]and then in addition we also did a series of moisture
[00:30:00.561]maps with the cosmic-ray probe.
[00:30:02.023]So here's ten of those images from 2015 to 2016.
[00:30:08.292]In addition, we also had a stationary probe
[00:30:11.452]that was recording every hour, so that's shown by the black
[00:30:14.732]line here going up and down.
[00:30:16.292]And the red stars are where we did these spatial
[00:30:19.532]surveys and I just plotted the spatial mean of these
[00:30:23.772]particular images.
[00:30:25.323]So we have continuous information, essentially at a point
[00:30:29.092]or integrated over an area and we have a series
[00:30:31.783]of spatial maps not at the times we actually wanna make
[00:30:35.792]an irrigation decision.
[00:30:37.183]So how do we combine this continuous information in time
[00:30:41.103]and some of these spatial maps that we have
[00:30:43.343]to something that's actually useful?
[00:30:47.193]Alright, so to do that is I will reference this sort
[00:30:51.284]of random sort of analogy is when I was working
[00:30:57.473]on my PhD I would often go to Kenya every couple
[00:31:01.833]of months for six weeks on end.
[00:31:04.172]And so I would load up on DVD's so I could watch
[00:31:07.343]that at night while I was working on my PhD.
[00:31:11.604]And so I'd often visit Chinatown in New York
[00:31:14.261]and I was always sort of surprised that they can get,
[00:31:18.752]you know, instead of one movie on a single DVD is they could
[00:31:22.244]smash five or six full-length movies on a single DVD.
[00:31:27.412]It'd have about the same amount of quality in terms of
[00:31:30.692]what you were seeing.
[00:31:31.583]So, you know, we're looking at Harry Potter, you know,
[00:31:34.061]maybe Ron and Hermoine don't exactly look like themselves
[00:31:37.092]but you get most of the picture of what's happening.
[00:31:40.492]And so it turns out, that there's some very useful
[00:31:42.862]techniques for compression that we can use both for an image
[00:31:46.890]or a series of images if we're thinking about a movie.
[00:31:50.031]And to extract that key information from this particular
[00:31:54.921]information.
[00:31:56.780]And so this is essentially the problem that we have
[00:31:58.600]in remote sensing or ground based sensing is we have
[00:32:01.412]this spirit, this spirit snapshots of space
[00:32:05.540]and what's happening but really we wanna pull out
[00:32:07.612]is the underlying spatial pattern of what's happening.
[00:32:11.281]And so this is the formal technique that we can use
[00:32:13.481]that where we can take basically, a time series
[00:32:16.070]of spatial maps and remove the key components of what's
[00:32:19.791]happening in that to really isolate the underlying
[00:32:23.361]spatial patterns.
[00:32:24.921]So this is empirical or orthogonal functions.
[00:32:27.820]I'm not gonna go into the details but it's a great
[00:32:30.652]technique to do deconstruction of spatial,
[00:32:34.179]temporal information.
[00:32:36.749]And so what we do is we run these series of maps,
[00:32:40.252]So we did ten maps, I just showing 7 here
[00:32:42.890]and it turns out that instead of having these 8 maps
[00:32:47.339]of information, is we can actually compress that
[00:32:50.479]into a set of one coefficients that explains about 90%
[00:32:54.957]of the pattern.
[00:32:56.428]So instead of having, you know, 10 Harry Potter movies,
[00:32:59.719]is we can compress that down into, instead of having ten
[00:33:03.077]different DVD's we can press all ten movies onto a single
[00:33:05.708]disk by using this set of loadings.
[00:33:09.658]And so what Katie looked at in her master's work is,
[00:33:13.138]alright, how good are these set of spatial patterns
[00:33:16.999]that we've identified at predicting spatial properties
[00:33:19.607]of things that we care about in terms of water flow
[00:33:23.487]through soils.
[00:33:25.788]And so Katie went out and she collected a series
[00:33:27.639]of these sampling rings.
[00:33:29.195]So these are basically 4-inch rings so we just distributed,
[00:33:34.324]basically uniformly across these fields.
[00:33:36.703]We did 31 as shown by the black dots down there.
[00:33:41.506]She brought these cores back to our laboratory.
[00:33:44.356]So what she does is you wet them up from below
[00:33:48.786]you drill a couple holes and you put
[00:33:50.606]two tensiometers inside of this soil core
[00:33:54.123]I then take that base and throw it on a scale
[00:33:58.605]and you let the basically the core dry out through
[00:34:02.425]evaporation over three or four day period.
[00:34:04.686]And you record the tension at two different heights
[00:34:07.606]and you can turn that into this information that you want
[00:34:10.726]about water flow through soil.
[00:34:12.841]Since this device is a decagon high prop and so
[00:34:15.932]our four devices are upstairs if you're interested
[00:34:18.633]in looking more at 'em.
[00:34:21.141]And so this is the kind of information
[00:34:22.948]that you can generate as you have a curve between
[00:34:26.127]water content and what the actual tension
[00:34:28.847]or pressure inside of your soil is.
[00:34:32.199]And so this unit is in Log 10, cause again we have
[00:34:34.428]to think about a log scale, so many orders
[00:34:36.529]of magnitude.
[00:34:38.278]And the three curves that we have is we have a sand,
[00:34:41.987]a loam and a silty loam.
[00:34:43.447]Again, all from the same field and so you can pick out
[00:34:46.398]the pressure values that are associated with field
[00:34:50.059]capacity and wilting point.
[00:34:51.758]And the difference between those gives you
[00:34:53.539]your available water.
[00:34:54.891]So this is really the critical piece of information
[00:34:57.691]that we're after in terms of soil properties
[00:35:00.319]or what are these two points and how far apart are these
[00:35:03.388]such that we can make optimal irrigation recommendation
[00:35:08.070]off of that.
[00:35:10.131]Alright, so again we can do this
[00:35:12.091]for 31 points, take the core
[00:35:14.910]back to our lab, you know, it takes about a day
[00:35:17.219]of prep, four of five days for a student
[00:35:20.070]to run the analysis, you know.
[00:35:22.150]Another day to actually process the data and so it takes
[00:35:26.440]a long time to actually do these observations.
[00:35:29.829]Kind of a pain.
[00:35:31.361]And so instead of doing a whole bunch of these in space
[00:35:34.119]is we wanna optimize this process.
[00:35:37.201]And where do we take samples and how can we minimize
[00:35:39.370]the number that we need to describe what's happening
[00:35:42.280]in this field.
[00:35:43.766]So this is just showing a scatter gram basically
[00:35:47.011]of these key properties so this is the top graph
[00:35:50.519]shows field capacity versus different environmental
[00:35:54.108]co-variants.
[00:35:55.281]So maybe elevation is a good predictor of field capacity.
[00:35:58.390]Maybe wetness index, so that's where you are sort
[00:36:02.219]of are you in a locally low spot of your field?
[00:36:04.322]A locally high spot?
[00:36:06.242]We can use the electrical conductivity map
[00:36:08.522]so this is widely used commercially
[00:36:10.970]in the state as doing EC maps.
[00:36:13.402]Then they could also use our Soil Moisture maps to make
[00:36:17.178]these decisions.
[00:36:18.261]So it turns out that EOF is a pretty good predictor.
[00:36:22.181]I'm showing on the lower diagram here is we can make
[00:36:25.642]some pretty low polynomial relationships between
[00:36:29.322]these geophysical information and what the lab cores
[00:36:32.741]are telling us.
[00:36:35.271]And so that's what we've done here is basically
[00:36:36.729]taken a series of these cores, made a relationship
[00:36:39.402]between the geophysics and turned that into spatial maps
[00:36:44.471]which now a producer could actually use to make
[00:36:47.621]irrigation decisions.
[00:36:49.202]So again, field capacity, wilting point,
[00:36:51.711]and available water are really the key metrics
[00:36:54.905]that any person will be after in terms
[00:36:58.436]of making a prescription maps.
[00:37:01.375]Alright so, is this a feasible strategy?
[00:37:04.855]And so we've really had to think about, sort of this,
[00:37:07.943]sort of effort versus what we're getting out
[00:37:10.695]of in terms of water, excuse me, reduction in pumping,
[00:37:14.655]or irrigation savings.
[00:37:17.525]And so it's sort of the business as usual
[00:37:20.036]is let's take some USDA information, maybe we'll invest
[00:37:23.756]a little bit into soil sampling, some people
[00:37:27.436]and some consultants actually will go out
[00:37:29.546]and do an EC map and I guess what we're saying is
[00:37:32.167]that basically as we add more and more of these layers,
[00:37:35.487]as you're getting some return on the amount of
[00:37:39.236]irrigation that you might save by adding more effort
[00:37:43.047]to that system.
[00:37:45.114]And so it's really what are the key systems
[00:37:47.927]that we can identify where, you know, it makes sense
[00:37:50.927]to invest $15 an acre into doing geophysical maps
[00:37:55.237]and lab corps versus the price that you're actually
[00:37:57.957]getting out of that kind of information.
[00:38:00.705]So certainly for, you know, high value crops
[00:38:03.236]potentially, you know vegetables or other things
[00:38:05.146]in California, vineyards are sort of a land
[00:38:08.836]of geophysicists.
[00:38:10.298]They often end up working in vineyards just cause they
[00:38:12.418]can afford some of the techniques.
[00:38:16.058]And, but I guess the argument is that as you know,
[00:38:17.847]water becomes more, and more limiting,
[00:38:20.096]is I think there's gonna be a need for more
[00:38:21.745]and more types of these data sets in the future.
[00:38:26.385]So, sort of wrapping up here, and one last quick study
[00:38:31.197]is Justin Gibson a current PhD student, he's
[00:38:35.205]continuing this strategy and building on it.
[00:38:38.927]We're actually now looking at several more fields
[00:38:42.247]as well as trying to come up with these four key
[00:38:46.626]soil parameters that are really put into these
[00:38:49.203]land surface models.
[00:38:51.284]This is out near Ogallala and so we did a series
[00:38:54.686]of geophysical maps at three different fields.
[00:38:57.553]And again, we find that, you know, out four or five
[00:39:00.624]geophysical maps run 'em through the statistical
[00:39:03.294]analysis is we can explain a lot of the underlying
[00:39:06.675]spatial patterns with that technique,
[00:39:09.766]So this is just what some of the raw data looks like.
[00:39:12.195]Again, we have our curve of water content
[00:39:16.416]versus log of pressure and then we can also relate
[00:39:19.286]that to these environmental co-variants.
[00:39:21.716]Again, elevation, bulk conductivity, and then potentially
[00:39:25.945]soil moisture.
[00:39:29.026]Alright, so this is a busy table so just sort
[00:39:31.204]of the key points is that, basically what we're
[00:39:34.655]finding is that you don't need a huge number of samples
[00:39:39.354]to describe the spatial variability in a field.
[00:39:42.596]And what we found is that basically in these 168 acre fields
[00:39:46.266]is about five to seven samples was adequate to describe
[00:39:50.775]that spatial variability.
[00:39:52.594]If you selected those locations based off
[00:39:55.026]of geophysical maps, and we did this using a cross
[00:39:59.604]validation analysis.
[00:40:01.592]And so typically right now if you're taking soil cores,
[00:40:05.145]you talk to a crop consultant looking at soil fertility,
[00:40:08.124]as they'll probably, as a rule of thumb, go out
[00:40:10.772]and do samples every two and a half acres.
[00:40:13.732]And so you're looking at about 60 samples
[00:40:16.593]for 160 acre field.
[00:40:19.793]Cause were basically were showin' for these fields,
[00:40:21.483]you could reduce that by 90% if you do a few
[00:40:24.612]geophysical maps and then select better sampling
[00:40:27.903]locations.
[00:40:28.754]So that's a huge amount of labor savings by just having
[00:40:32.403]some opreory information with the geophysics.
[00:40:36.583]Second it turns out that we can get about half
[00:40:38.962]of the information that we're after with the geophysics.
[00:40:41.604]So we can do a really good job at explaining
[00:40:44.092]bulk density and a shape factor but some parts
[00:40:48.223]of the curve, we don't do quite as well at.
[00:40:51.034]And that's just the nature that the geophysics is
[00:40:53.684]we can only get in the field and sort of, you know,
[00:40:55.782]moderately wet conditions to very dry conditions, not a lot
[00:40:59.284]of producers are happy with you driving an ATV
[00:41:01.943]across their very wet saturated field, makes sense.
[00:41:05.823]And so we're thinking about other sort of, techniques
[00:41:09.004]that we can use to get at this information.
[00:41:12.533]Alright, so summarizing is basically we found
[00:41:15.442]that four geophysical maps allows us to describe
[00:41:18.935]the spatial variability within a field.
[00:41:21.334]And so this, we prefer to do sort of over a range
[00:41:24.524]of water contents.
[00:41:25.564]So, you know, maybe medium conditions to very dry
[00:41:27.914]conditions and so how are we gonna do this moving forward?
[00:41:31.354]It's, you know, very expensive to have, you know,
[00:41:33.422]your own ATV out there constantly mapping,
[00:41:36.463]and so we're thinking about ways
[00:41:38.172]that we can maybe combine this
[00:41:39.382]with, you know, self-driving vehicles.
[00:41:41.881]There's a lot of interest in that right now
[00:41:43.991]in agriculture for picking particular crops.
[00:41:49.023]Also attaching it to existing farm equipment.
[00:41:51.772]And so maybe put one on a community sprayer or
[00:41:54.002]detasseler.
[00:41:55.255]So there's some opportunities for this opportunistic
[00:41:59.092]sensing is what we call this.
[00:42:02.691]A second in terms of sampling location is having these
[00:42:05.143]geophysical maps as a guide to where to pick locations
[00:42:08.589]and how to scale that as we found for these three fields,
[00:42:11.602]is very effective, reducing the effort by
[00:42:14.663]about 90% and so that's something that we're gonna be
[00:42:18.232]looking more at is maybe we can move some of these costs
[00:42:22.162]and some soil sampling and laboratory analysis
[00:42:24.892]into these geophysical maps to help, to help
[00:42:28.460]guide that location.
[00:42:30.858]In terms of what's happening
[00:42:32.709]in terms of modeling approaches, is that I would just
[00:42:36.045]argue that these geophysical layers are different
[00:42:38.890]wave lengths so there's some very useful
[00:42:41.338]information in them when we think about applying
[00:42:44.029]some of these advanced machine learning, deep learning
[00:42:47.909]type of algorithms is these might be very good layers
[00:42:50.890]to discriminate different process ese in terms
[00:42:53.370]of the computational science aspect.
[00:42:57.978]Alright, so again as we can take, do a series
[00:43:00.400]of geophysical maps, pull out the underlying patterns,
[00:43:03.896]and then use that to select lab corps locations,
[00:43:07.200]do a few lab corps and potentially generate
[00:43:09.578]these very useful layers.
[00:43:11.720]Or really getting at these producer management zones.
[00:43:15.589]Which you'll hear a lot about in, sort of the
[00:43:17.829]crop consulting world.
[00:43:20.570]And finally, is that as we think that there's some
[00:43:23.418]also some techniques that we can combine
[00:43:25.229]with unmanned aircraft particularly using, you know,
[00:43:29.490]visible light or near infrared or some other wave
[00:43:32.149]lengths to help get us at this wet end of the curve
[00:43:35.309]to really try and complete sort of our total understanding
[00:43:38.538]of water flow through soil.
[00:43:41.610]So with that, I will finish early.
[00:43:44.098]No one will probably object to that so thank you
[00:43:47.239]for your attention.
[00:43:49.237]Hi Dr. Franz
[00:43:50.280]Hey Penny.
[00:43:51.120]It's your old friend Penny.
[00:43:53.469]One of the assumptions that you made that I've
[00:43:55.340]been thinking about without being able to go too far
[00:43:57.461]with it, is that precipitation by and large is
[00:43:59.984]stochastic.
[00:44:01.730]As you know, there are many who would disagree with that
[00:44:03.949]at this point and would say that it is basically
[00:44:06.280]not stochastic.
[00:44:07.930]If it not stochastic, is that gonna effect any
[00:44:10.741]of the process ese that you've been able to derive,
[00:44:13.549]to deal with this basic question that you've posed?
[00:44:16.960]Sure, I guess we could have a longer argument
[00:44:20.547]about stochastic versus following a known probability
[00:44:25.629]distribution.
[00:44:26.709]So I guess how I use that is that we have a pretty good
[00:44:30.609]estimate of what that distribution of rainfall is
[00:44:33.200]so when we wanna model that system is we wanna pick
[00:44:36.680]an underlying distribution in the sample
[00:44:39.019]from that distribution and run that through our models
[00:44:41.469]so we can do a range of scenarios.
[00:44:43.800]So, I don't wanna get too much into the semantics
[00:44:47.141]of, you know, having known probabilities and sampling
[00:44:50.450]from that versus what a stochastic process is.
[00:44:53.480]So, happy to chat more in class tomorrow or another
[00:44:57.239]time on that one.
[00:44:58.378]So, thanks.
[00:45:08.187]Hi Trenton, as far as the mapping that you've
[00:45:11.410]discussed here.
[00:45:13.341]I'm not gonna pretend that I understand it completely
[00:45:15.960]but once you generate a map of say a given center pivot
[00:45:19.909]that your happy with and it seems to represent
[00:45:21.760]conditions correctly, how long is that map good for?
[00:45:25.650]How often do you have to do this?
[00:45:27.499]Taking into account different farming practices.
[00:45:30.101]Everything, all those other factors.
[00:45:32.821]Great, excellent question.
[00:45:34.449]So, we don't fully know to be honest.
[00:45:37.501]So we've done a series of maps over a couple of years.
[00:45:42.730]Generated these property distribution then compared
[00:45:46.090]that to maps later on so there is some stability
[00:45:50.040]in the type of information.
[00:45:54.344]This really has to do with are you sampling the full
[00:45:56.848]range of expected patterns but if you get into a drought
[00:46:00.717]year, three years down the road, how does that sort of
[00:46:03.717]map hold for those conditions?
[00:46:06.395]We don't totally know that.
[00:46:09.627]However, you know, it has to do with sort of the time
[00:46:10.996]scale that soils form so we think, you know,
[00:46:16.098]just based on sort of my knowledge of soil process ese,
[00:46:18.008]you know, these maps are gonna last, you know, a fairly
[00:46:19.978]long time, you know.
[00:46:21.077]Maybe bulk density and the near surface is gonna be
[00:46:23.389]affected by wheel traffic and the type of crop rotation.
[00:46:27.508]But, you know, soils take a long time to actually form
[00:46:32.356]so we're fairly confident that a series of maps
[00:46:36.698]over a course of a year from dry to sort of intermediate
[00:46:40.828]conditions should be a pretty good road map
[00:46:43.127]of what's happening and that should last, you know,
[00:46:45.756]most, you know, decades I would say.
[00:46:49.268]But hard to say depending on the crop rotation
[00:46:52.596]and how intense that is.
[00:46:54.595]You may have some effects on bulk density that could
[00:46:56.847]then propagate through he rest of the soil process ese, so.
[00:47:00.538]That sort of answer your question or, follow up?
[00:47:05.847]It just all gets back to, you know, you talk
[00:47:07.836]about the expense of it, that's probably gonna be
[00:47:09.636]one of the first things a farmer's gonna ask is, okay
[00:47:11.846]how often do I have to do this to get the maximum
[00:47:14.148]benefit outta the technique.
[00:47:15.887]Yep, yeah I would say, you know, as a rule of thumb,
[00:47:18.756]four maps, couple dry, couple wet or intermediate
[00:47:23.818]it's gonna give you most of that information.
[00:47:25.378]Your not gonna have to do that for probably several
[00:47:28.120]decades but that's just a shot in the dark there so.
[00:47:34.386]Other questions?
[00:47:37.207]Okay, thank you.
[00:47:38.704]Alright, thanks
[00:47:40.066](audience applause)

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Spatiotemporal Prediction of Soil Properties and States in Variably Saturated Landscapes

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments