Application of Remote Sensing to Plant Breeding · “Application of Remote Sensing to ... Remote sensing Using instruments mounted on satellites, in planes, etc. to produce images

“Application of Remote Sensing to Plant Breeding”

William Schapaugh, Brent Christenson, Hatice Aslan, Randi Clark, Kevin Price, Nan An, Vara Prasad and John Boyer

Remote sensing Using instruments mounted on satellites, in planes, etc. to produce images

Introduction

Cellular leaf structure and its interaction with visible and infrared radiation

Temperature Reflectance

Color spectrum seen by passing white light through prism

Electromagnetic

Spectrum

http://www.trimble.com/Agriculture/greenseeker.aspx

Ritewing Zephyr II

DJI S800 Spreading Wing

hexacopter JDrone hexacopter

Workstation

Small Unmanned Aircraft Systems (sUAS)

10

Green

Blue UV Red

Red Edge

Red Shoulder

Water Absorption Near-

infrared (NIR)

Plateau

Typical spectral reflectance from a living leaf

1 2 3 4 5 6 7 8

Red Wavelength

Creating a Normalized Difference Vegetation Index (NDVI)

NIR

=

+

NDVI

Agronomy Journal, 2008

Where, NIR is near infrared light at 770±12nm generated by sensor and VIS is visible red waveband at 660±12nm.

Introduction

Genetic Gain - MG III Cultivars

* Spectral reflectance data collected

Introduction

Genetic Gain - MG IV Cultivars

* Spectral reflectance data collected

20

25

30

35

40

45

50

55

60

65

70

1920 1930 1940 1950 1960 1970 1980 1990 2000 2010

Yie

ld (

bu

ac-1

)

Year of Release

MGIII MGIV

Seed Yield and Year of Release

Mean spectral response curves of MGIV genotypes. Wavebands are 10nm intervals and reflectance is the percentage of a white reference panel.

Private 4-23

Private 4-22

70 50 30 10

MGIII

MGIII

-1.00

-0.80

-0.60

-0.40

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

40

5

43

5

46

5

49

5

52

5

55

5

58

5

61

5

64

5

67

5

70

5

73

5

76

5

79

5

82

5

85

5

88

5

91

5

94

5

97

5

10

05

10

35

10

65

10

95

11

25

11

55

11

85

12

15

12

45

12

75

13

05

Co

rre

lati

on

Co

effi

cie

nt

(r)

Waveband (nm)

Correlation of Waveband and Yield

MG3

MG4

BLUE GREEN RED Near-Infrared

Model Variable (nm)

MGIII 715

1100

MGIV 715

915

R² = 0.83

R² = 0.86

10

15

20

25

30

35

40

45

50

55

60

25 30 35 40 45 50 55 60 65 70

Esti

mat

ed

Yie

ld (

bu

ac-1

)

Observed Yield (bu ac-1)

MGIII and MGIV Yield Estimation Models

MGIII

MGIV

MGIII

0.75

0.80

0.85

0.90

0.95

25 30 35 40 45 50 55 60 65

Cal

cula

ted

Ind

ice

Val

ue

Observed Yield (bu ac-1)

MGIV

NDVI 900,680

NDVI 780,670

NDVI 815,615

NDVI 760,615

GNDVI 780,550

WI 970/900

r = 0.79

r = 0.90

r = 0.87

r = 0.82

r = 0.83

r = 0.78

29.5

30.0

30.5

31.0

31.5

32.0

32.5

500 1000 1500 2000 2500 3000 3500 4000

Can

op

y t

emp

eratu

re (

oC

)

Seed yield (bu ac-1)

Canopy Temperature vs. Seed Yield

MG III

r = -0.87**

Tukey’s HSD = 0.3

MG IV

r = -0.79**

Tukey’s HSD = 1.1

Results

16 32 48 64

Select these varieties

Eliminate lower yielding varieties

• Higher yielding varieties have lower reflectance values in the VIS and red-edge spectra portions and higher values in the NIR portion of the spectra and lower canopy temperatures than lower yielding varieties.

• Canopy temperature and canopy reflectance measurements can account for a large portion of variability in seed yield among varieties.

• Attempting to build upon these results to develop applications in high throughput phenotyping.

Conclusions

Acknowledgements Agronomy • William Schapaugh • Brent Christensen • Hatice Aslan • Randi Clark • Kevin Price • Nan An • Vara Prasad • Russell Dille • M. Djanaguiraman • Nathan Keep Statistics • John Boyer • Leigh Murray • Nick Bleodow Plant Pathology • Tim Todd • Tom Oakley Vet. Medicine • Deon van der Merwe

K-State Research and Extension Soybean Breeding and Genetics

Thanks for your support!

Questions?