Perspectivas y avances del uso de UAV en AP. Bruno Basso

Transcription

1 Perspectivas y avances del uso de UAV en AP Bruno Basso

2 Outline of the presentation Remote and proximal sensing Pro and Cons of UAV Available UAVs Technologies Applications of UAVs in Precision Agriculture How to link UAV with crop modeling

3 Electromagnetic spectrum and spectral signatures of plants

4 Healthy vegetation Absorbs Red light (RED= m) Reflects Near Infrared light (NIR= m). Vegetation in stress Absorbs Near Infrared light (NIR= m). NDVI Reflects Red light (RED= m)

5 NDVI Normalized Difference Vegetation Index NDVI = (NIR-RED) / (NIR+RED) NDVI increases linearly with LAI (for LAI < 3) LAI

6 Unmanned Aerial Vehicles According to AUVSI, following the FAA integration of UAVs into national airspace by Sept. 2015, UAVs will have $13.6 Billion impact on US economy (growth in precision agriculture) UAVs have many advantages but there is still work to do on them before they can be used directly by farmers

7 Pro and Cons of UAV UAV Pros: - Response time - Precision - Resolution - User Control Cons: - Stability ( if windy) - Costs (hardware) - Regulations (COA) - Training Satellite Satellite Pros: - Processing time - Established - Availability - Coverage Cons: - Cloud cover - Lacks user control - Return time - Costs (images)

8 Published Applications of UAVs in Agriculture Detection of crop/tree chlorosis (Zarco-Tejada & Berni, 2008) Detection of water spills in irrigated orchards (Zarco-Tejada & Berni, 2008) Estimating leaf carotenoid content in vineyards (Zarco-Tejada et al., 2013) Leaf Area Index (Berni & Zarco-Tejada, 2009; Zarco-Tejada & Berni, 2008) Water stress (Berni & Zarco-Tejada, 2009; Suárez, 2008; Suárez, 2009; Zarco-Tejada & Berni, 2008; Zarco-Tejada et al., 2012) Chlorophyll content (Berni & Zarco, 2009; Zarco-Tejada & Berni, 2008; Uto et al., 2013) Canopy temperature (Zarco-Tejada & Berni, 2008) Remote sensing of N stress levels (Teoh & Hassan, 2012; Tremblay et al., 2011) Weed Management (Torres-Sánchez et al. 2013)

9 Two Two Common UAV Airframes Fixed-wing Rotorcraft Vertical Takeoff and Landing (VTOL) Grenzdörffer et al., 2008 Nebiker et al., 2008 Berni & Zarco-Tejada, 2009 Nebiker et al., 2008

10 UAV model: mdr microdrones.com md is equipped with RGB digital video-camera, thermal camera, multispectral, laser scanner. Spatial resolution < 1 in (1-7cm) Payload 3 lbs, Flying time minute with 1 battery

11

12 Sensors Visible Multispectral Tetracam Sony, microdrones microdrones Headwall Photonics Thermal (IR) Hyperspectral

13 Laser scanner imagery from UAV (microdrone md-1000)

14 Thermal Imagery

15 Phantom 2+ = Cost ~ $1200

16

17

18

19

20 Stitching

21

22 Multispectral broad-band vegetation indices Mulla 2013

23 Hyperspectral narrow-band vegetation indices Mulla 2013

24 RED-Edge If chlorophyll decreases, the red-edge moves towards shorter wavelength and reflectance increases at 720 nm Barnes, et al., 2000; Fritzgerald et al., 2008; Cammarano, Basso et al., 2011, 2014

25 The chlorophyll vegetation indices Chlorophyll Indices Formula NDRE1 (NIR Combined - RE) / (NIR Chlorophyll + RE) Indices NDRE2 (RE - Red) / (RE NDRE1 + Red) / NDVI CRM NIR / RE - 1 NDRE2 / NDVI CGM NIR / Green - 1 NDRE1 / GC CVI (NIR / Green) *(Red / Green) MTCI NDRE2 / GC (NIR - RE) / (RE - Red) CARI (RE - Red) TCARI * (RE / Green) OSAVI TCARI MCARI MCARI2 3 * ((RE - Red) MCARI * (RE / OSAVI - Green) * (RE / Red)) ((RE - Red) MCARI * (RE -/ Green)) MTVI * (RE / Red) 1.5*(2.5*(NIR-Red) *(NIR-Green)) / ((2 * NIR +1)² -(6*NIR - 5*Red^0.5 - MCARI / MTVI2 0.5))^0.5 Structural Indices NDVI OSAVI MTVI MTVI2 GC Formula (NIR - Red) / (NIR + Red) (1+0,16)*(NIR-Red)/(NIR+RED+0,16) 1,5*(1,2*(NIR-Green)-2,5*(Red-Green))/((2*(2*NIR+1)^0.5-(6*NIR-5*(Red)^0.5)-0,5)) (1.5 * (1.2 * (NIR - Green) * (Red - Green))) / ( (2 * NIR + 1)^2 - (6 * NIR - 5 * (Red^0.5)) - 0.5)^0.5 Ground cover calculated with RapidEye method

26 Vegetation Indices

27 Assessment of crop N status at the time of second N application Canopy Chlorophyll Content Index (CCCI) (Barnes et al., 2000): CCCI = (NDRE NDRE min )/(NDRE max - NDRE min ) Canopy Nitrogen Index (CNI) (Rodrigez et al., 2006): %N min = e^[1.55+(-7.38*10-3 )*biomass] %N max = e^[1.68+(-1.52*10-3 )*biomass] Equation [1]: N [g N m -2 ]= [(%N max - %N min )*(1.86*CCCI 0.346) + % N min ]*(dry biomass/100) Cammarano et al AJ

28

29 Case study: VRN N rates T1 90 Kg ha -1 T2 60 Kg ha -1 T3 30 Kg ha -1 Basso et al., 2012, 2013

30 Testing SALUS Years 0 N Measured 0 N Simulated 90 N Measured 90 N Simulated (kg/ha) (kg/ha) 1991/ / / / / / / / / / RMSE* 308 kg/ha 407 kg/ha R.E.** 14 % 12 % Basso et al., Eur. J. Agr. 2010

31 Testing SALUS 0 N 90 N Observed Mean Simulated Mean RMSE R.E. Observed Simulated RM R.E. Mean Mean SE (%) (%) N uptake Soil N Growth stages Kg N ha Kg N ha End of juvenile stage(zado ck 10) Anthesis (zadock 60) Maturity(z adock 87) 15-Mar 13-Mar 15-Mar 13-Mar 21-Apr 20-Apr 21-Apr 20-Apr 01-Jun 30-May 01-Jun 30-May Basso et al., Eur. J. Agr. 2010

32 NDVI 14 March 2009 Electrical Resistivity Tomography

33 14 March April 18 May Satellite

34 Yield Maps Media dei 5 anni

35

36 Yield Maps N rates T1 90 Kg ha -1 T2 60 Kg ha -1 T3 30 Kg ha -1

37 MTCI Class CCCI NDVI Class Confusion Matrix Nitrogen Rates Classified CCCI NDVI Index Ground Truth (Nitrogen Rates) Plots T3 (30Kg ha -1 ) T2 (60Kg ha -1 ) T1 (90Kg ha -1 ) Class1 (Low VI values) Class2 (Medium VI values) Class3 (High VI values) Difference CCCI NDVI Overall Accuracy= 71% Tot Confusion Matrix analysis Confusion Matrix Nitrogen Rates Classified MTCI Index Ground Truth (Nitrogen Rates) Plots T3 (30Kg ha -1 ) T2 (60Kg ha -1 ) T1 (90Kg ha -1 ) Class1 (Low VI values) Class2 (Medium VI values) Class3 (High VI values) Difference MTCI Overall Accuracy =50% Tot T1 90 Kg ha -1 T2 60 Kg ha -1 T3 30 Kg ha -1

38 Net Revenue ($ ha -1) Strategic and tactical N management using spatially explicit crop modeling kg N ha High Yield Zone Medium Yield Zone Low Yield Zone kg N ha kg N ha Nitrate Leaching (kg N ha -1 ) Dual criteria optimization through tested model determines the N rate that minimizes nitrate leaching and increases net revenues for farmers (Basso et al., 2011; Eur J. Agron 35: )

39 Conclusions UAVs will revolutionize data collection in agriculture and significantly improve the efficiency of input applications at the field/farm scale UAV need to be able to deliver maps that can be used by farmers to implement changes in their managament practices over space and time The integration of UAV with crop modeling is the key to understand a complex systems like crop production in space and time.