unmanned aerial system (uas) platforms for cotton breeding...
TRANSCRIPT
Unmanned Aerial System (UAS) Platforms for
Cotton Breeding: Findings and Challenges
Beltwide Cotton ConferencesCotton Agronomy and PhysiologyDallas, TXJanuary 5th, 2017
Murilo M. Maeda, Juan Landivar, Jinha Jung, AnjinChang, Junho Yeom, Andrea Maeda, Josh McGinty,Juan Enciso, Wayne Smith, David Stelly, Steve Hague,and Jane Dever
EnvironmentGenotype
Phenotype
• Cultivar Selection• Growth Habit• Stress Tolerance• Insect Tolerance• Disease Resistance• Earliness• Health Status
Sensors & Platforms DJI Phantom 4
◉ RGB sensor◉ 12 Mega Pixel
3DR X8+◉ Tetracam ADC Snap◉ FLIR Vue Pro R (data
not shown)
Data Collected &2016 Study Temporal plant growth patterns
◉ Plant height◉ Canopy cover◉ Biomass
Plant health (temporal)◉ Reflectance, NDVI◉ Canopy temperature, tIR
Maturity and yield parameters◉ Bloom count, maturity◉ Open boll count, yield
Growth Analysis◉ Sigmoidal growth models, growth rates
Simulation◉ GOSSYM, a process level, physiological model
2016 Study
• 31 Entries (Dever, Hague, Smith & Stelly)• 2 Water regimes• 4 Replications• 2 Row plots• 10 Grids per row
9,800 Measurements flight ‐1 parameter ‐1
Methodologies
<DSM> <DEM> <CHM>
Canopy Height Model
Canopy Cover
Plot Boundaryand
Grid Structure
CottonBoll Count Analysis
Plant Height: Observed vs. Estimated
y = 0.9278x + 0.111R² = 0.9078
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60
UA
S (m
)
Ground (m)
Plant Height 1 to 1
y = 0.8224x + 0.2538R² = 0.9217
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60
UA
S (m
)
Ground (m)
y = 0.847x + 0.2082R² = 0.9109
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60
UA
S (m
)
Ground (m)
Data Processing:Feature Extraction (Ph, CC, CV)
Max
Half Max
Max
Day
Duration
Growth Curve (Sigmoidal) Growth Rate Curve (1st Derivative)
Data Processing:Feature Extraction (NDVI and ExGI)
Max
Day
Slope 1 Slope2
Duration 1 Duration 2
Intersection
Area 2Area 1
UAS-derivedGrowth, Efficiency, and Yield ParametersRaw Data Parameter Description
1 Early Growth Duration ‐ Days2 Early Growth ‐ Rate3 Relative Growth Rate ‐ Early4 Mid‐growth ‐ Linear ‐ Day
Plant Height 5 Mid‐growth ‐ Linear ‐ Rate Max6 Relative Growth Rate ‐ Late7 Late Growth Duration ‐ Days8 Late Growth ‐ Rate9 Duration Max. Growth‐ No of Days10 Early Growth Duration ‐ Days11 Early Growth ‐ Rate12 Relative Growth Rate ‐ Early13 Mid‐growth ‐ Linear ‐ Day
Canopy Cover 14 Mid‐growth ‐ Linear ‐ Rate Max15 Relative Growth Rate ‐ Late16 Late Growth Duration ‐ Days17 Late Growth ‐ Rate18 Duration Max. Growth‐ No of Days19 Early Growth Duration ‐ Days20 Early Growth ‐ Rate21 Relative Growth Rate ‐ Early22 Mid‐growth ‐ Linear ‐ Day
Canopy Volume 23 Mid‐growth ‐ Linear ‐ Rate Max24 Relative Growth Rate ‐ Late25 Late Growth Duration ‐ Days26 Late Growth ‐ Rate27 Duration Max. Growth‐ No of Days
28 No. Third week of Bloom29 No. Fourth week of Bloom
Bloom count 30 No. Fifth week of Bloom31 Rate of Bloom 1st to 2nd Week32 Rate of Bloom 2st to 3nd Week33 No. of Bolls34 Average Bolls area ‐ Size
Boll Count 35 Total Boll Area36 Average Boll Volume37 Total boll volume38 Maximum Value39 Time of Max, days40 Early Slope ‐ Square to Bloom
NDVI 41 Duration Max.‐ No of Days 42 Late Slope ‐ Bloom to Cutout 43 Area of Early Season
44 Area of Late Season45 Duration of Early, days46 Duration of Late, days47 Maximum Value48 Time of Max, days49 Early Slope ‐ Square to Bloom50 Duration Max.‐ No of Days
Greeness Index 51 Late Slope ‐ Bloom to Cutout 52 Area of Early Season
53 Area of Late Season54 Duration of Early, days55 Duration of Late, days56 Early ‐ Squaring57 Mid ‐ Bloom
NIR 58 Late ‐ Cutout59 Maximum Value60 Water Stress Index
Yield PredictionSeedcotton per row (dryland)
6 parameters, including:‐ Growth‐ Canopy Efficiency‐ Bolls
y = 1.0337x ‐ 0.0651R² = 0.7397
0
2
4
6
8
10
12
0 1 2 3 4 5 6 7 8 9 10
Mea
sured (lb
s/ro
w)
Predicted (lbs/row)
Trait X Genotype Data Matrix
Data VisualizationMaps
258 838(count)
11.9 24.3(cm2)
Number of Bolls Average Boll Area
(lbs/ac)low high
Lint Yield
Team Presetations Juan Landivar, Jinha Jung, Andrea Maeda, Long Huynh, and Murilo Maeda. Integration of
Unmanned Aerial System (UAS) Data and Process Based Simulation Models to Forecast CropGrowth and Yield. Cotton Physiology Conference.
Murilo M. Maeda, Juan Landivar, Jinha Jung, Anjin Chang, Junho Yeom, Andrea Maeda, JoshMcGinty, Juan Enciso, Wayne Smith, David Stelly, Steve Hague, and Jane Dever. UnmannedAerial System (UAS) Platforms for Cotton Breeding: Findings and Challenges. CottonPhysiology Conference.
Junho Yeom, Jinha Jung, Anjin Chang, Juan Landivar, and Murilo Maeda. Open Cotton BollDetection Methodology Using Unmanned Aerial System (UAS). Cotton EngineeringConference.
Jinha Jung, Juan Landivar, Anjin Chang, Junho Yeom, and Murilo Maeda. Unmanned AerialSystem (UAS)‐Based Asymmetric Cotton Growth Model for High Throughput Phenotyping.Cotton Engineering Conference .
Anjin Chang, Jinha Jung, Murilo Maeda, Juan Landivar, Henrique Carvalho, and Junho Yeom.Unmanned Aerial System (UAS)‐Based Cotton Canopy Temperature Measurement System.Cotton Engineering Conference.
Thank You!
Dr. Juan LandivarTexas A&M AgriLife
Corpus Christi
Dr. Jinha JungTexas A&M-CCCorpus Christi
Andrea Maeda, M.Sc.Texas A&M AgriLife
Corpus Christi
Dr. Josh McGintyTexas A&M AgriLife
Corpus Christi
Dr. Anjin ChangTexas A&M-CCCorpus Christi
Dr. Juan EncisoTexas A&M AgriLife
Weslaco
Dr. Wayne SmithTexas A&M
College Station
Dr. Junho YeomTexas A&M-CCCorpus Christi
Dr. David StellyTexas A&M
College Station
Dr. Steve HagueTexas A&M
College Station
Dr. Jane DeverTexas A&M AgriLife
Lubbock