1. Sensor Readings in Corn as Affected by Time of Day
Peter Scharf
University of Missouri
With thanks to Larry Mueller,
Joao Medeiros, Bettina Coggeshall,
and Amanda Coggeshall

2. Why consider time of day?
Most of our readings were collected within one hour in our calibration research
Our method requires comparison with a reference area
If spectral properties change during the day, the time between measuring the reference area and fertilizing may introduce error
There is some evidence in the literature that spectral properties change during the day

3. Literature: color vs. time
Hoel & Solhaug, 1998:
Wheat SPAD highest at dusk & dawn
SPAD dropped when plants were transferred from low to high intensity light in growth chambers
5 units in 30 minutes in shade plant
6% in one hour in wheat
SPAD goes back up when light intensity is dropped
Brugnoli & Bjorkman, 1992:
Sunflower transmittance higher at higher light intensity
Due to chloroplast movement (arrangement & orientation)

4. Methods Experiments conducted in 2004 near Columbia
Sensors were mounted in place over individual plants
Sensor measurements were taken hourly from 6am to 9pm (approx. weekly)
Experiment 1: June 16 (V8) to June 29 (V12)
Experiment 2: July 7 (V6) to July 28 (V14)

5. NDVI vs. time: green is most sensitive

6. Green NDVI consistently drops a lot in the middle of the day

7. Green NDVI consistently drops a lot in the middle of the day

8. Regular NDVI also consistently drops in the middle of the day

9. Yellow NDVI also drops in mid-day, but perhaps the least of the three

10. What’s going on?--NIR

11. What’s going on?--VIS

12. What’s going on?--SPAD

13. Why were NDVI values lower in the middle of the day?
SPAD suggests that these may change late in day
Ideas:
Change in canopy architecture
Change in internal leaf structure
Change in leaf pigment concentration or arrangement
Influence of shadows (readings were in north-south rows: shadows in the background change dramatically during the day)

14. Why were NDVI values lower in the middle of the day?
Ideas:
Change in canopy architecture
Change in internal leaf structure
Change in leaf pigment concentration or arrangement
Influence of shadows (readings were in north-south rows: shadows in the background change dramatically during the day)
We were curious about
this and decided to test it

15. Influence of shadows
Greenseeker green & red, Crop Circle, stationary over soybean canopy
Take reading in sun, then move shade over sensor & canopy, take reading, repeat
Ten pairs afternoon 8/9/2005
Ten pairs morning 8/10/2005
Photo & measurements: Larry Mueller

16. Influence of shadows
Greenseeker green NDVI was consistently much higher when measured in shadow
Mean in sun, in shade
Different with p =
Greenseeker red NDVI was consistently higher when measured in shadow
Mean in sun, in shade
Different with p = 0.04

17. Influence of shadows
Crop Circle yellow NDVI was not different measured in sun or shade
Mean in sun, in shade
p = 0.27

18. What about visible/NIR ratio?
Visible/NIR ratio can be calculated exactly from NDVI—the exact same information is contained in both parameters
So I won’t show vis/NIR vs. time of day
But proportionally, differences are larger in vis/NIR than in NDVI analogs

19. Summary
We thought that with active light sensors, we would not have problems with reflectance measurements changing as light conditions changed (time of day, clouds)
With all three sensors, NDVI values were lower in mid-day and higher near dawn and dusk
This pattern was much more pronounced with the Greenseeker green sensor than the other two

20. Summary
Some of the diurnal change in NDVI may be due to actual changes in the crop
SPAD is higher in evening
Canopy architecture may also change
Change in value for NDVI (or vis/NIR) during the course of the day is an obstacle to making good N rate decisions
Need to solve this to be a useful on-farm tool
Frequent reference reading is one possibility (N applied across rows with airplane?)
Another possibility is to use a correction function

21. Summary
We also observed differences in sensor readings in sun vs. shade
Biggest in green Greenseeker
Small in red Greenseeker
No problem in Crop Circle
Change in sensor readings when a cloud comes over is also an obstacle to making good N rate decisions
We also have data from passive sensors in these experiments—not yet analyzed, but we will eventually compare them with active sensors