Transcript Yield Monitors

Yield Monitors and Maps
BAE 4213
April 12, 2007
Randy Taylor
Biosystems and Ag Engineering
What Are the Tasks?
 Measure grain flow

Mass or Volume Flow
Sensor
 Measure ground
speed

Existing ground speed
sensor or position
sensor signal
 Program harvest
width

Programmed as a
constant value or
changed on-the-go
 Combine position

GPS Position Sensor
Flow Sensors
Yield Monitor Errors
 How do we calculate yield
mass
mass
Yield 

area length  width
 Yield errors must be related to one of these
3 measurements: mass, length, width
 For a yield monitor
 Mass is determined from the flow sensor
 Width is a programmed constant
 Length is determined from speed
Width
 When do errors occur?
 header not full (i.e. harvest width does not
match header width)
 How do we fix it?
 Adjust on the go => bad idea
 How much error are we really talking
about?
 U of Missouri research found it was 8-12% in
drilled beans if they assumed constant full
header
 How much do you have to reduce harvest
width to get area (field) to be accurate?
Distance Errors
 UNL Research harvesting up & down slope
found no significant difference in mass
accumulation.
 However they found a 42’ difference going
uphill verses down on a 6% slope
 Though GPS was the intended speed signal,
differences in end points was not observed
in a GIS
 The greater distance measurements going
uphill cause a reduction in calculated yield
Mass Flow Measurement Errors
 Combine Dynamics
 Calibration
Combine Dynamics
Crop is cut or removed from plant
Conveyed to feeder house in the header
Conveyed to threshing unit (cylinder or rotor)
~80% of separation should occur during threshing
~20% of grain goes on to separation (rotor or straw
walkers)
 Grain that falls on the cleaning shoe should pass through
near the front of the shoe
 Grain that goes through the returns





All of these affect the grain
flow relative to its former
location in the field
Mass Flow Sensors
Lag/Resonance Time
Sensor Calibration
Response to mass flow is non linear
Diaphragm vs Triangular
Can get a very good fit with linear
Operating at points away from one
calibration can cause errors
 Where do we see these?




 Start and stop grain flow
Transitional Mass Flow
What Causes Error?
20
R2 = 0.78
10
R2 = 0.61
R2 = 0.53
Error, %
0
0
-10
10 2
20
R = 0.86
30
-20
-30
-40
Average Mass Flow, lbs/s
40
50
Ranking Plots
40
Yield Monitor Rank
35
30
Project 2
Project 3
Project 4
Project 5
Project 6
Ideal
25
20
15
10
5
0
0
10
20
Actual Rank
30
40
Using YM for OFR
 50% of the error between weigh wagon and
yield monitor weights was due to mass flow
 Correlation between yield monitor and
weigh wagon weights was 0.97
 Regression results lead to the same
conclusions regarding the treatments
 Challenging to rank treatments with YM
data
What Can a Yield Map Tell Us?






Soil fertility, type, etc.
Disease or insect pressure
Variety differences
Poorly drained areas
Compacted areas
Does not point out the yield limiting
variable, it only indicates the
response to it
Using YM Data
1.
2.
3.
4.
Diagnosing Crop Production
Estimating Nutrient Removal
On-Farm Research
Establishing Yield Potential (Goals)
1. Diagnosing Crop Production
 Probably the most widespread use for
yield maps today
 Print maps to keep records on
 Select appropriate ranges
 Number of ranges
 Spread (don’t create or exaggerate
variability)
 Color scheme
Problem Diagnoses
Wire worm infestation
Crop drowned
Presenting Yield Maps
 5 – 6 ranges or groups maximum
 Based on





Natural Break
Even Intervals
Predefined Crop
Standard Deviation
Percent of Average
 Color Scheme
Dryland Wheat
Even Intervals
1996
1997
Dryland Wheat
Predefined Crop
1996
1997
Dryland Wheat
Percent of Average
1996
1997
Normalized Yield (96-97)
Data Aggregation
 Point data
 Contouring
 Some type of interpolation
 Likely have minimal or confusing choices
 Grid
 Interpolated
 Averaged
 Summed
Points versus Interpolation
How many
of the dark
blue points
are zero
yield?
Header Status
Raised the
mean yield
about 5
bu/ac, but
did it really
make a
difference?
Irrigated Corn/Beans
Normalized Yield
1996 Beans/Corn
Beans
Corn
1997 Corn
Average of Two Years
Interpreting Patterns
 Straight lines are manmade
 Parallel with travel
 At an angle with travel patterns
 Irregular patterns are generally
naturally occurring
 Lines
 Areas/patches
Sand Pivot (1996-97 Crops)
Yield Variability
 Many causes of yield variability
 Yield monitors and maps don’t
determine the cause
 Yield maps display the location and
magnitude (area and degree)
 This information should lead to better
decisions
Yield Variability
 That which can be changed
 Fertility
 That which must be managed
 Soil physical properties
3. On-Farm Research
 Has the potential to expand
knowledge about individual farms
 Comparison of varieties, tillage
practices, fertility rates, etc.
 Not as easy as it may seem
 What do you want to know?
 Why do you want to know it?
Layering Maps
Yield
Topsoil
Population
1998 Corn - Osage County
165
Yield, bpa
160
155
150
145
22500
25500
28500
140
135
0
2
4
6
Topsoil, inches
8
10
4. Prescribing Spatial Inputs
 Some input recommendation models
require the use of a crop yield goal
 Development of a nutrient
recommendation map may require
the use of a yield goal map
 How can you generate variable yield
goals?
Yield Stability Analysis
 Data were obtained
with various yield
monitors
 Converted to point
yield and
unrealistic values
were removed
 Data were block
averaged to 180
foot cells
‘Whisker Plots’ of YM Data
0.4
0.2
0.0
-0.2
0
20
40
60
80
100
120
140
160
180
-0.4
-0.6
-0.8
-1.0
0.0
-0.2
0
20
40
60
80
100
120
140
160
-0.4
-0.6
-0.8
Rank
0.4
Mean Relative Difference
0.2
-1.0
Rank
0.2
0.0
-0.2
Mean Relative Difference
Mean Relative Difference
0.4
0
20
40
60
80
100
-0.4
-0.6
-0.8
-1.0
Rank
120
140
160
180
 Points are the mean
relative difference for
each cell
 Bars are the standard
deviation of yield
through time.
180
Classification Maps
Mean Relative Difference
 Standard statistical analysis offers minimal
insight into spatial data
 Low yielding cells tend to be more variable
 There is a better opportunity to classify
consistently low yielding areas
 Because like classed cells were spatially
contiguous, this method showed more
promise than typical methods
Conclusions
 Yield monitor data can be used for
anything that yield data are used for
1.
2.
3.
4.
Diagnosing Crop Production
Estimating Nutrient Removal
On-Farm Research
Establishing Yield Potential (Goals)