Precision Farming Final

PLSC 244 Cal Poly Winter 2025 Dr. Wong

What is the primary function of Yield Monitoring?

The first step in the development of a field map

• a measure for how well your plants are growing (same with remote sensing and soil sampling)

• your report card = tell you how well your management decisions are impacting your production

• tells you if you’re doing a good job in a ways that tangible for people to understand

• to provide immediate feedback on how your plants are growing (nutrients, moisture, pest pressures)

• provide groundwork for site specific management

What is yield monitoring?

• it was originally designed for grain production: sensors mounted on a combine measuring yield as the crop is harvested (grain harvester combines)

• is coupled with a GPS logging location where data can be mapped to create an image of changes in overall yield step by step and to build an idea of foot by foot productivity

what does yield represent?

volume (amount of material) per unit area

why do yield maps have different colors?

• it is a way to distinguish the differing levels of productivity

• each color represents either above or below average levels

what does yield mapping not tell you?

what the problem is, it can show you that there is a problem but not why there’s a problem

what is whole field yield?

‎‎farmers have been doing this for a long time so there was a sense on how well the field did but you didn’t know where in the field your crop was doing well or poorly because you used to just total up the averages from the truck weight tickets

What are plot yield monitors?

• small plot harvest with very precise information bc actually weighing material but in order to do that it would need to stop

• would stop frequently to weigh the yield and would take too long so growers didn’t want it until it could go faster

what are the components of a yield monitor required for getting an instantaneous yield?

• flow sensor

• moisture sensor

• ground speed sensor

• header position switch

• processor/display console

what function does the flow sensor play in yield monitors?

calculates the material flowing through the machine

• but how do you do this while you’re moving?

What is the function of a moisture sensor in yield monitors?

yield is based on a moisture set on a certain level, for examples corns is 15.5%

• need to be able to adjust to that moisture level in order to estimate your marketable yield

what is the function of a ground speed sensor in a yield monitor?

tells you how much area you’re covering so with the data on how much area your covering with the amount of material flowing through the machine you can calculate yield.

what is the function of the processor/display console in yield monitors?

had to create a computer to collect all this information and to determine yield per unit area. able to determine on the fly its traveling speed, width, and mass and can tell you in an instant how much you’re averaging

• constantly collecting information

What is the function of the header position switch in yield monitors?

It knows that if your implement is down to collect data but knows that if its lifted it can signal to stop data collection so it doesn’t mess with your average calculations

• basically an on and off system for yield monitoring

What are the four ways we are able to measure flow?

1. Mass flow

2. Radiometric

3. Load Cell

4. Volume Measurement

How do mass flow sensors work?

• most complicated but all about deflection

• transports the grain to a collection point for just clean grain

• when the augur fils up it gets transported to the top of the machine

• once at the top its thrown into a collection bucket but you can’t just drop it, you need to distribute it evenly

• at the top of the clean grain elevator it flings the grain then hits a deflecting plate with a spring behind it

• the more mass, the more the plate is deflected so if its calibrated you can know much much yield it is based on how much its pushed because grain is being flung at the same velocity but the amount of grain being flung at a time is changing

• most commonly used

how do radiometric sensors work?

• uses a radioactive isotope (americum 241) and has a radioactive detector

• as grain goes across it deflects the amount radioactive material that can be read by the detector so if you have more grain, less radioactive material is being read

• use how much of the radioactive isotope is being detected to predict your yield

How does a load cell work?

• like a scale

• combine moves grain into a trough at the bottom of a machine and moves it across to the clean grain elevator

• they put load cells on the augur that goes across the trough to weigh it

how do volumetric measurements work in the context of yield monitoring?

• how much light is captured on the other side

• issue is that if your not on flat ground and theres a tilt, grain can cover the detector without filling the whole thing which will skew your yield estimations

what is the most common method for moisture sensing for yield monitoring?

A capacitance system which is a type of moisture sensor that creates a dielectric field with positives and negatives on either side. you’re going to run something through the field. water carries charge well so if you have high amount of wet grain your capacitance level goes up because you have great flow in the dielectric, if you have low water the capacitance will be less because the total amount of energy flowing between the two side is less. can use the different levels of capacitance to calculate moisture

• physical and mathematical determination

• issue: they’re not constantly reading, they have to cycle because the shoot that creates the dielectric has to be full. if theres empty spaces it cant get a good measurement because theres nothing to carry electricity.

• it has to fill, measure, and then cleans

• not instantaneous but for the resolution we need its enough

what else do we need for successful yield monitoring?

• harvest rate/speed

• swath

Why is swath an important feature to account for in yield monitoring?

theoretically doesn’t change because the width of your combine doesn’t change but our fields aren’t perfectly square so the swath will shrink as you approach a row that dwindles when you get close to the edge

what do you need to calculate an instantaneous yield?

flow rate, moisture, harvest rate and swath

• relatively simple

What is Auto Swath?

one of the first harvesters to account for swath in the computer

How it Works:

• when you start planting, the planters on the john deeres figures out how far you’ve planted the seeds and maps/calculates every place where seed has been planted and creates a map

• when you go to harvest you take the planting map and load it into the combine so it can monitor its width and know where you did and didn’t plant which allows it to ignore the information that didn’t have anything planted

• calculates a yield based off of expected plantings

how is accuracy and precision in yield monitoring accounted for?

1. header position switch and auto swath ensure you aren’t collecting data when nothing is being harvested

2. Operation Delay

   • takes 5-7 seconds between harvest and weight/yield

   • have to wait for up to 7 seconds to get your first reading

   • implemented this operational delay to account for that so when you get your first reading it knows that its the yield for the 10 ft behind you

   • issue with the differences in how people drive which creates differences in operational delay that can shift your information. not devastating but is accurate enough

what kind of crops can you find commercial yield monitoring systems for?

• cotton

• silage

• peanut*

• potato*

• carrot*

• tomato*

• sugar cane*

• grapes*

  • = builds on a simple design using load cells

Yield monitoring for cotton

• been around a while, shortly after grain

• volumetric flow system

  • sensor mounted on front and rear chute where the wet cotton balls are being thrown

    • the sensors count the # of cotton balls coming in

    • how much light is being blocked by the cotton balls

  • monitor mounts in cab

  • calculates volumetric flow

  • flown in an airstream, not on an elevator so doesnt have the same pitfalls that come with using an elevator as opposed to being shot up by air

Silage yield monitoring

monitor’s total amount of material passing through the cutter

• sucked into a bunch of blades

• depending on the volume, you have to open or close more

• they stuck a sensor onto the drive spring to measure deflection

total moisture by measuring capacitance can also help determine yield

Grape yield monitoring

advanced technology viticulture

• load cell system

• place a load cell under the conveyer belt after grapes are dumped from bucket

• system also works for processing tomatoes, potatoes, carrots, almonds

good system if you can calibrate them

Citrus yield monitoring

different systems for use with mechanical harvesters

• Mass system: as oranges are picked they’re moved onto a conveyer belt and uses a load cell to calculate the mass for juicing oranges

• Counting: counted the fruit harvested

• Oxbo harvesters: deflection of weight off dampener to calculate yield but you also need to calculate your area

Almonds yield monitoring

crop dried on ground

• pick up machine has a series of load cells that as it went along, it got heavier

  • getting half the tree at a time

• can’t monitor on a per tree basis but we can manage for a resolution of tree blocks in order to implement variable rate technology

possible errors

• doesnt always know where it is because of canopy but was solved by using RFID tags which the tractor was able to count to keep track of where it was

Broccoli yield monitoring

• harvested by hand as opposed to machine

• comparison between weighed and machine weighed

• load cell monitoring installed on “harvest aid” - long pieces of equipment with conveyer belts on them, with workers putting broccoli on it

• only took a load cell reading x number of feet, then takes the weight, zeros itself and goes again so harvest area would be 4 rows 10ft, giving you zones of yield you can map

• yield would be captured multiple times

• one of the simplest and easiest ways to tell you how you’re farming

how are fruits (like apples and avocados) harvested in the context of yield monitoring?

• use RFID tags on the bins and trees

• map trees using GPS

• harvester will dump harvest in nearest bin so you can use the yield of that bin to track the production of those surrounding trees

• every bin gets weighed

• not very complicated

what are the main takeaways of yield monitoring sytems?

as long as you know where you are harvesting from, so it is not impossible to figure out, so as long as you monitor that path you can figure out yield

why do you want to have georeferenced yield data?

• represents a relatively cheap method of adding “a layer” of information- has to follow a specific method for it to be useful

  • but data itself has a certain amount of error and has to be displayed in a proper method in order for you to understand it

• can show you things that are so obvious but also show not so obvious things which is where the value lies

• important to know how to present and interpret your data

what are some of the methods you can use to effectively present your field data?

• data aggregation

• number of zones

• color schemes

Why isn’t yield monitoring not a perfect science and how can we account for that?

• 10%-15% of raw data contains errors

• these errors can be built into the system but its important to be consistent with your collection so the errors you know you need to account for are the same

• data scrubbing is based on “set” fudge factors

• tend to over or under estimate the actual error

• most data scrubbing is done post-harvest which can increase possibility for errors

Where are the areas of error in yield monitoring?

• time delays

• position offsets

• wobbly GPS

• over/under laps

• double harvest

• field mix up

• missing data

• data spikes

• data overload

• multi harvester

what is the yield scrubbing program?

program created to account for the numerous ways error can occur and filter through that data using scrubbers (fudge factors) but those fudge factors can change a little so this program allowed you to change these fudge factors a little bit to try to account for these kinds of errors

allows you to use filtering algorithms to “clean” field data

• multiple filters can be used

• can adjust amount of filtering used

allowed you to increase or decrease fudge factors

• change the delay time of how long does it takes for the grain to enter and the grain is processed - play around with it until the edges are smooth/dark - if you don’t do that then it looks like every other row there is nothing then something

• gives you a more accurate representation of what was harvested where

what are the ways we aggregate data and why do we do that?

1. equal count

2. equal interval count

allows u to change how our data is read because we have so much of it it needs to be boiled down into the most important pieces

what are the parameters we use to assess the success of our yield data?

if its above, at, or below average bc looking at total yield makes it too busy

• LOOK FOR THE DATA SET THAT SHOWS YOU DIFFERENCES

  • can calculate this based off of equal count = taking in percentile sections

  • equal count interval = yield greater than 50 is high and less than 25 is low

• yield data there to tell you if you have a problem but not necessarily what the problem is

• if trends are very linear its probably something you did

• does your yield data align with your soil and remote sensing data

what are the things you should take into account when presenting you data?

• number of zones

  • how many ways do you want to break up your yield?

  • 5 classes: 2 lows, one medium, two high

• color scheme

  • always look at the legend first

what should you be taking into account when interpreting your data?

• remember that a yield map shows variation but it doesn’t show what caused the variation

• value of map is in how it validates management decisions i.e. do yields increase/decrease, stay the same

• yield mapping is a marathon not a sprint because on year is only one snapshot of one environmental condition

should everybody use yield mapping?

• maybe if you field doesn’t show any spatial variability but that highly unlikely because even if there aren’t drastic ones, there small ones and the game is knowing how much swing is too much swing to the point of economically hindering you

• what causes the variability?

  • if it’s uncontrollable then it doesn’t really matter

• not all spatial variability is created equally

what are the possible causes of variation?

• can help you describe causes or explain causes

• look at whether or not the change is straight - if it was then caused by you biological systems don’t work in straight lines

yield frequency maps

• every crop has a different way of calculating yield

• deals well with rotations

  • simple mathematical process - come up with a yield distribution +1, 0, -1 for both one - make two different maps overlap them

• can create relative yield maps over time and can consistently see the areas of high and lows

what is remote sensing?

the power of observation

what is the simpliest form of remote sensing?

• drive by scouting

• been used since WW1 - guy in a balloon taking a photo

  • 1930s aerial photos

• past farming practices can predict what is happening today

• Missouri study: 1939-1999

  • a photo can predict what will happen

what are the basics of remote sensing?

• measurements of energy that is reflected or emitted from objects

• energy - electromagnetic

  • wide array of wavelengths

  • not all this energy is useful to us

• radio wavelengths

• as wavelengths shorten, the amount of energy it produced increases

• ultraviolet to inferred is where most of the energy comes from, especially around the visible spectrum

  • white light: can be broken into smaller areas which can provide different types of light

what is the active sense system?

• no external energy is needed, everything comes internally

• radar is a form of active sensing

• cons: trying to active sense from a satellite is difficult - especially to send a wavelength down to earth and back to the satellite

what is the passive sense system?

• rely mostly on passive sensing

• utilize the suns energy and collecting the amount of energy that is lost or reflected we can start to predict growth

what are the available wavelengths of remote sensing?

• not all wavelengths are available for remote sensing

• the natural world absorbs a lot of energy

  • water

  • particles (dust)

  • soil

  • atmosphere

  • the only energy that provides with info is a narrow band

  • sun does not produce enough microwaves strong enough to utilize effectively (need an active sensor for this)

what are the types of scattering?

• Rayleigh scattering

• Mie scattering

• non selective scattering

what are the primary characteristics of Rayleigh scattering?

• dominant scattering mechanism

• very small particles scatter short wavelengths

  • likes to scatter short wavelengths the most, like blue

  • describes why on a clear day the sky looks blue

what are the primary characteristics of Mie scattering?

• when larger particles are present

• larger particles interact with light

• affecting longer wavelengths

• like a beautiful sunset - with reds and oranges

  • the bigger the sunset - bigger the particles - reds and oranges are scattered more

• why a cloud is white: has a large enough particles (water), that all of the wavelengths violet to red scattered = white (kind of the opposite of the prism)

what is passive sensing?

• ag applications utilize wavelengths between UV and IR (including area with visible light)

• other sensors are used but require active sensors

• anything besides these will need active sensors

what is being sensed in passive sensing?

• solar radiation will be:

  • absorbed: energy lost within the system, a fair amount of light is absorbed by soils - so there are certain remote sensing that you can do

  • transmitted: passed through, can eventually be absorbed or reflected = a lot of noise

  • reflected: comes back to you

How does EM light interact with a leaf?

• comes into contact with the plant

• most green light is reflected - more chlorophyll is present within the plant, the more green light present '

• basically predicting how much chlorophyll is present or active and how the plant is doing

• most of blue and red is absorbed

• how much reflectance is occurring at each wavelength

• depending on the total amount of chlorophyll, there are shifts of energy being reflected or absorbed

what is the value of analyzing spectral signature for remote sensing?

• important to build relationships/correlation on wavelengths and types of plant stress

  • how each wavelengths interacts with tissue

• the spectral analysis for each crop is different

• spectral analysis also changes with age and bearing fruit (sometimes)

• species, age and stress are the three things we need to compare

how does remote sensing of plant stress generally work?

• every time the plant is stressed there must be a response

• many stresses can lead to the exact response

  • its your job to figure out why the plant is struggling i.e. need to ground truth your data, remote sensing points you to the right place

What are common features of modern aerial photos?

• they’re true color photos

  • crop canopy

  • signs of soil variability

what are some common features of using google earth pro?

• can see layers of photos

• compaction from laser leveling crew

• predicting a growth problem based on a true color image

• if it was natural = across the field in irregular patterns

how is thermal imaging utilized in remote sensing?

• areas of the earth or plants can accumulate heat

• works at night

• used to predict drought stress

• darker colors = lower temps

• provides information that the naked eye can not see

• *effect of increased temp due to the closure of stomata (from the increase rate of CO2 - plant has enough so won’t open stomata = decreased transpiration = increased heat), appears to be lighter

what are the ways data can be collected for remote sensing?

• ground based remote sensors

• aerial based remote sensors

• satellite based remote sensors

how do ground based remote sensors work?

• hanging a crane over the top of the field, collecting microwave data (kinda stupid)

• other ones that can be attached to a sprayer, sensing data - light from weeds vs crops that you want

how do aerial based remote sensors work?

• drones - dr. wong no likey - very finite value

• planes

what is IFOV and why is it valuable?

instantaneous field of view

• how much of the land can we see?

• allows us to determine max resolution that our camera can get - how big of an area can i zoom into

• area of the earth seen at time x and altitude y

• has NADA

What is NADA?

the thing that is directly below the camera that allows for the instantaneous field view in remote sensing because it can swing, increasing the area that we can view. However, the resolution is the best directly below it

• biggest problem with droned: drone flying at 400-500 ft can only take a portion of the field, will have to move because you can’t swing the camera

what are the ways performance is measured in remote sensing?

• spatial resolution = how fine of a piece can you see

• spectral resolution = what radiations can you view

• radiometric resolution = how finely can you see the differences

• temporal resolution = how many times can you take a pic i.e. how long does it take between pictures/how long it takes the camera to come back to the same spot

what is one of the most significant differences between drones, planes, and satellites when it comes to spatial resolution?

they will have a different resolution/pixel size because a drone can stay low

what changes between aerial and satellite spatial resolution?

pixel size which tell us how much we can focus on

what are the differences of satellite capabilities regarding spatial resolution?

• Landsat 7+ETM = 30 meters

  • 40 meters = half a football field

  • old landsat used to observe as low as one football field/acre

• SPOT-5 = 20 meters

• IKONOS = 4 to 1 meters

• Quickbird = 0.61 meters

in what context can a low resolution be helpful ?

generally just useful for topography which is why we prefer medium resolution which is pretty darn good

why is the Quickbird satellite imagery the most effective for spatial resolution?

it can zoom in the most and maintain its crispyness

What is the GEO-eye 1?

a satellite that has a 41 cm spatial resolution

15.2 km swath width (how much it moves) = 350,000 sq km/day

What is the World View 4?

another satellite that can be useful for spatial resolution

• 30 cm

• 14.5 km

• 680,000 sq km/day (way bigger than CA)

what are the main takeaways to consider with spatial resolution?

• its how small of a point you can see

• the important question is how small of a point do you want because does it really matter if you can see a single leaf?

• as a farmer is it worth managing based on every plant? generally not because that’s also a lot of information

• this determines the size at which you need to manage differences for

how does Black and White spectral resolution work?

when you take a B and W photo, it represents every single wavelengths in the visible spectrum

not segregating out the different electromagnetic energies

spectral resolution thats really large

separates every single visible wavelength into either black or white

what do the black and white sections in a spectral image represent?

black = absence of color

white = presence of visible light

how does color work in spectral resolution?

the visible spectrum gets broken down into three set wavelengths (red, green, blue)

what kinds of applications will remote sensing most often offer?

• true color

• multi spec

• hyper spec

what is the true color application?

looks at the three specific spectrums RGB

What does multi spec and hyper spec do that true color applications dont?

breaks the electromagnetic spectra into even smaller pieces

how does multispec break down the em spectra?

instead of RGB, you’re looking at 4-16 different wavelengths from infered to ultraviolot

• youre going to pick out specific spectra that you think are of value in the plant

• would include all ROYGBIV

• can capture more bands per image which allows you to tell the difference between different landscapes like a grassland and a forest

how does hyper spec break down the em spectra?

• break it down into 100s of wavelengths

• what if you measured from 400-799 nm? every single frequency on the visible spectrum

• can capture the most bands per image which allows you to tell the differences between species, colors, and the changes in the shaded of those colors

what is the issue with hyperspec?

it collects so much data that you can’t use unless you’ve studied it for years which is why peoples are hired to interpret their data

• most farmers don’t have the time or expertise to make use of that much information on their own

what is radiometric resolution?

how many words/adjectives you need yo describe a color which gives you more specific information

• increasing the specificity of how you describe something

ability to distinguish between energy levels

what are the differences betwen a 2 bit and 4 bit radiometric resolution?

• how many bits do you use to describe something

• 2 bits = 2 color descriptors

• 4 bits = 16 color descriptors

how many radiometric resolution bits do satellites tend to work in?

8 and 14 bits

• 8 bits = 256 bits/different shades

• 14 bits = 16,386 different shades/intensities

whats the point of having more bits ?

the more you have the finer the image comes out

• increasing color in different shades increases clarity

• more bits = more data = more required storage

  • gete get really big fast

what are the resolution tradeoffs?

• as you increase spatial resolution you immediately lower radio metric because of the amount of light available for you to reflect off of goes down

• high spatial = low radiometric spectra

• increase in radiometric without affecting spatial reduces spectral resolution

• this makes it important to ask yourself what kind of outcome you want/prefer?

what is temporal resolution?

calculates by how often does the remote sensing device revisit any given location i.e. how often does that satellite, drone, plane take that picture

• mostly designed for satellites which travel north-south

• this frequency all depends on elevation and orbital speed!

• satellites we use for studying plants are in sun synchronous orbit

what does it mean to be in sun synchronous orbit?

satellites come over the areas we’re interested in during the day time because most of the data collection we do is passive i.e. we rely on the sun

when considering drones, planes, etc what does your revisit time/temporal resolution depend on?

how much you want to pay bc eahc one has a different temporal resolution

what are the different unmanned aerial vehicles as remote sensing platforms?

• DJI phantoms are the more popular

• Quest UAV

• Alava Ingenieros

• CropCam

• DragonFly

• Yamaha RMAX

what are the two kinds of platforms UAVs can have?

1. Fixed wing

2. Rotor

what are the advantages of fixed wing platforms?

• larger payload = carrying camera which can be heavy and the fixed wing can support it better

• longer flight times because it can glide

• comes together to form two elements

what are the disadvantages of fixed wing platforms?

• less maneuverable

• hight space requirement

• harder to learn how to fly

what are the advantages of a rotor platform?

• highly maneuverable

  • easier to operate

what are the disadvantages of a rotor platform?

shorter flight times = need to use more energy to keep in the air and move through the air

what is the FAA Modernization and Reform Act change in regards to remote sensing?

• you need to have a pilots license to sell the data you collect with a drone of a field

• covered the operation of recreational UAV operations

• did not allow operations for business ventures

• all drowns but be flown under 400 ft, anything above that would be entering into airspace where you’re not allowed

• can’t be near people or public parks where people are present

• can only carry up to 55 lbs.

• cannot fly for payment