Embed Size (px)
Citation preview
YIELD EDITORBETTER DATA, FASTER
Scott Drummond and Ken SudduthUSDA-ARS Cropping Systems and Water Quality Research UnitColumbia, MO
From Sensors to Yield Maps
Why clean yield data?
Yield data ALWAYS contain problems Operational problems Data collection errors Calibration issues
Sensor system related problems Sensor failure, noise, accuracy issues Timing/filtering issues for individual sensorsMultiple sensors can magnify problems
Rapid change in velocity
Corn Yield(bu/ac)
174 to 2963 165 to 174 153 to 165 145 to 153 138 to 145 132 to 138 126 to 132 120 to 126 113 to 120 6 to 113
Incorrect delay time(s)
Corn Yield(bu/ac)
124 to 180 110 to 124 101 to 110 95 to 101 89 to 95 83 to 89 76 to 83 68 to 76 52 to 68 30 to 52
Delay Time = 9 s Delay Time = 14 s
Ramping at crop edge
-50 0 50 100 150 200Time from Crop Edge (s)
0
4
8
12
16Gr
ain
Flow
Rat
e (lb
/s)
Entering CropExiting CropHeader Up/Down
11 Second Shift Applied
Ramping at crop edge
Corn Yield(bu/ac)
124 to 207 110 to 124 102 to 110 95 to 102 90 to 95 84 to 90 77 to 84 69 to 77 55 to 69 1 to 55
No edge treatment Edge cleaned
Ramping at crop edgeKriged
No edge treatmentKriged
Edge cleanedDifference
(bu/ac)
Mean = 106.4STD = 32.3
Mean = 108.1STD = 31.7
>10% affected bymore than 5 bu/ac
-5
0
5
10
15
20
25
30
35
40
45
50
55
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
Unknown swath width
Soybean Yield(bu/ac) 29 to 54 28 to 29 26 to 28 25 to 26 25 to 25 24 to 25 23 to 24 22 to 23 18 to 22 0 to 18
Dealing with yield map errors
All yield maps contain some errors that should be removed before analysis
Manufacturers’ software does a good job of dealing with many simpler problems
They have a tougher time with “trial-and-error” type settings (i.e. delay time – set, view, adjust)
Manual editing (point, transect, area, etc.) must be done elsewhere (generally in a GIS)
A software package was needed that provided all of these tools in one place
Yield Editorversion 1.0, developed 2003
Yield Editor has been widely used by farmers, students, researchers, consultants, and others
Total downloads of the software are now well over 5000
Yield Editorversion 2.0
In 2010 we began developing a new version of Yield Editor that would automate more of the yield data cleaning process Automated delay time determination Overlapped travel and partial swaths removed Unrealistic localized yield values removed “Smart” selection of existing filter parameters Batch mode operation Currently over 11,000 total downloads
Yield Editorversion 2.0
Automated yield cleaning expert (AYCE)
Delay time determination Using relative spatial consistency
Lee, D. H., K. A. Sudduth, S. T. Drummond, S. O. Chung, and D. B. Myers. 2012. Automated yield map delay identification using phase correlation methodology. Trans. ASABE 55(3): 743-752.
Han’s bitmap overlap filter
Han, S., S.M. Schneider, S.L. Rawlins, and R.G. Evans. 1997. A bitmap method for determining effective combine cut width in yield mapping. Trans. ASAE 40(2): 485-490.
Localized standard deviation filter
Localized STD Filter Settings:Grid Cell = 5 Header WidthsLimit = 3 Standard Deviations
Cell Results:Mean = 165 STD = 15Retain 120-210 in this cellCyan points removed
Crop Yield 31 58 86 113 141 168 196 223 Cell Results:
Mean = 172 STD = 9Retain 145-199 in this cellNo points removed
Summary
Yield data contains large numbers of errors. These errors will affect your precision management. Yield Editor 2.0 can automate many common yield
cleaning tasks Having a way to document changes and retain the
original data is important.
