A Novel Methodology for Improving Plant Pest Surveillance in Vineyards and Crops Using UAV-Based Hyperspectral and Spatial Data
<p>Predictive detection model workflow.</p> "> Figure 2
<p>Ground control point (GCP) target and white reference board (<b>left</b>), MicaSense reflectance reference board (<b>top right</b>), and Spectralon white reference (<b>bottom right</b>).</p> "> Figure 3
<p>Hyperspectral white reference processing. (<b>a</b>) Saturated values around the highlighted area; (<b>b</b>) Maximum values are not saturated and this white target portion can be used for obtaining reflectance; (<b>c</b>) Cropped portion of the white reference board image with higher radiance values.</p> "> Figure 4
<p>Selection of individual plants or grapevines for spectral signature extraction based on vigour assessment into four classes. (<b>a</b>) Selected grapevines from an attribute table where vigour is 2/5 (<b>b</b>) Four different polygon areas for each class based on grapevine location and vigour assessment.</p> "> Figure 5
<p>(<b>a</b>) Headwall Nano hyperspectral sensor on-board a S800 unmanned aerial vehicle (UAV); (<b>b</b>) Custom-made two-axis gimbal hosting the hyperspectral camera (SolidWorks 3D model).</p> "> Figure 6
<p>Generated orthomosaic of studied sites; two vineyards in the Yarra valley, Victoria, Australia. Bottom images show enlarged regions of the vineyard with clearly visible plants.</p> "> Figure 7
<p>Result of the digital surface model (DSM) and digital terrain model (DTM) of Site 1. Markers are the ground control points (GCP).</p> "> Figure 8
<p>Results of traditional expert-based assessment of grapevines vigour per panel (groups of four to six grapevines, fragment transposed to cardinal directions).</p> "> Figure 9
<p>Result of the unbiased DVM for Site 1.</p> "> Figure 10
<p>Expert assigned classes of grapevines vigour per panel (<b>a</b>) and the results of a remotely-sensed vigour assessment of individual grapevines for December 2016 (<b>b</b>) and February 2017 (<b>c</b>) for a block with the Chardonnay variety.</p> "> Figure 11
<p>Expert-assigned classes of grapevines vigour per panel (<b>a</b>) and the results of a remotely-sensed vigour assessment of individual grapevines for February 2017 (<b>b</b>), site 2 with the <span class="html-italic">V. vinifera</span> Roussanne variety.</p> "> Figure 12
<p>Vegetation indices for block 3 with the highest correlation to vigour assessment. (<b>a</b>) PI2; (<b>b</b>) PI5; (<b>c</b>) NDVI; and (<b>d</b>) OSAVI<sub>H</sub>. All of the indices are based on hyperspectral imagery collected in February 2017.</p> "> Figure 13
<p>Mean spectral signature for different levels of vigour of the grapevine for the <span class="html-italic">V. vinifera</span> Chardonnay variety measured in (<b>a</b>) December 2016; and (<b>b</b>) February 2017 for wavelengths from 400 to 1000 nm.</p> "> Figure 14
<p>Mean spectral signature for different levels of vigour of the grapevine for the <span class="html-italic">V. vinifera</span> Chardonnay variety measured in (<b>a</b>) December 2016 and (<b>b</b>) February 2017 for wavelengths from 400 nm to 700 nm.</p> "> Figure 15
<p>Pearson’s correlation coefficient matrix showing the strength of the relationship between the February 2017 multispectral vegetation indices and expert and digital vigour assessment. Orange colours indicate negative correlation, blue colours indicate positive correlation, and intensity of colour indicates relative strength.</p> "> Figure 16
<p>Scatter plot of correlation for data presented in <a href="#sensors-18-00260-f015" class="html-fig">Figure 15</a>, generated from multispectral imagery in February 2017.</p> "> Figure 17
<p>Pearson’s correlation coefficient matrix showing correlation between the hyperspectral vegetation indices and expert vigour assessment for data collected in December 2016. Orange colour indicates negative correlation; blue colour indicate positive correlation with the intensity of colour indicating relative strength and highlighted red font are the indices that correlate positively with the digital vigour model (DVM) and expert vigour assessment.</p> "> Figure 18
<p>Scatter plot of correlation for data presented in <a href="#sensors-18-00260-f017" class="html-fig">Figure 17</a>, generated from hyperspectral imagery in December 2016.</p> "> Figure A1
<p>Mean spectral signature for different levels of vigour of the grapevine for the Chardonnay variety measured in December 2016 for wavelengths from 400 nm to 700 nm.</p> ">
Abstract
:1. Introduction
2. Methods
2.1. Predictive Detection Model Workflow
- Tree number
- Latitude
- Longitude
- Block
- Row
- Panel
- Tree variety
- Expert visual vigour assessment
- Digital vigour model
- Multispectral derived indices and bands
- Hyperspectral derives indices and bands
- EM38 data
2.2. Orthorectification, Hyperspectral Imagery Processing, and GIS Tools
2.3. Georeferencing
2.4. Ground Control Points (GCP) and White Reference Boards
2.5. Vigour Assessment
2.6. Hyperspectral Processing
2.7. Mean Spectral Signatures for Different Grapevine Types
2.8. Vegetation Indices
3. UAV and Sensors
3.1. UAV
3.2. High Resolution RGB Camera
3.3. Multispectral Camera
3.4. Hyperspectral Sensor
3.5. Expert Visual Vigour Assessment and EM-38 Data
4. Field Experiments
5. Results and Discussion
5.1. Visual Vigour Assessment Results
5.2. Digital Vigour Model (DVM)
5.3. Hyperspectral Analysis
5.4. Correlation Analysis of Different Variables Using Attribute Tables
- Vigour in both December 2016 and February 2017 has the highest positive correlation with DVM, and to NDREM, but only to NDVIM, NDVIGreenM, and OSAVIM in December. However, these are relatively minor relationships, with the linear relationship between 0.23 and 0.3.
- EM38 has no relationship to any of the vegetation indices, Vigour, or DVM.
- Certain indices are extremely positively correlated for December and February (indicated by dark blue boxes in Figure 15). These are: BLUE, GREEN, RED, and Red Edge (RE) multispectral bands; NIR with RE bands, OSAVIM, MCARIM, TCARIM, MCARI1M, and MCARI2M; NDVIM with GREEN, NDRE, OSAVIM and MCARIM1; NDVIGreenM with NDRE and OSAVIM; and TCARIM with MCARIM, MCARI1M, and MCARI2M.
- Vigour showed the highest positive relationship (r > 0.4) with the vegetation indices PI1, PI3, PI4, PI5, NDVI, NDVIGreenH, and OSAVI, as well as with DVM (r = 0.4).
- Similar to the multispectral data, certain vegetation indices were positively correlated for both December and February; PI1, PI3, PI4, PI5, NDVI, NDVIGreenH, MCARIH, TCARIH, and OSAVIH with each other; MCARI/OSAVI with BAND800; and BAND670 with BAND504.
- The EM38 data showed no relationship with any of the vegetation indices, vigour, or DVM.
6. Conclusions and Further Research
Acknowledgments
Author Contributions
Conflicts of Interest
Appendix A
References
- Mulla, D.J. Twenty five years of remote sensing in precision agriculture: Key advances and remaining knowledge gaps. Biosyst. Eng. 2013, 114, 358–371. [Google Scholar] [CrossRef]
- Marciniak, M. Use of remote sensing to understand the terroir of the Niagara Peninsula. Applications in a Riesling vineyard. J. Int. Sci. Vigne Vin 2015, 49, 1–26. [Google Scholar] [CrossRef]
- Mathews, A.J. A Practical UAV Remote Sensing Methodology to Generate Multispectral Orthophotos for Vineyards: Estimation of Spectral Reflectance. Int. J. Appl. Geospat. Res. 2015, 6, 65–87. [Google Scholar] [CrossRef]
- Zhang, C.; Kovacs, J.M. The application of small unmanned aerial systems for precision agriculture: A review. Precis. Agric. 2012, 13, 693–712. [Google Scholar] [CrossRef]
- Blanchfield, A.L.; Robinson, S.A.; Renzullo, L.J.; Powell, K.S. Phylloxera-infested grapevines have reduced chlorophyll and increased photoprotective pigment content—Can leaf pigment composition aid pest detection? Funct. Plant Biol. 2006, 33, 507–514. [Google Scholar] [CrossRef]
- Skinkis, P.; Walton, V.; Kaiser, C. Grape Phylloxera, Biology and Management in the Pacific Northwest; Oregon State University, Extension Service: Corvallis, OR, USA, 2009; pp. 7–9. [Google Scholar]
- Benheim, D.; Rochfort, S.; Robertson, E.; Potter, I.D.; Powell, K.S. Grape phylloxera (Daktulosphaira vitifoliae)—A review of potential detection and alternative management options. Ann. Appl. Biol. 2012, 161, 91–115. [Google Scholar] [CrossRef]
- Powell, K. A Holistic Approach to Future Management of Grapevine Phylloxera; Springer: Dordrecht, The Netherlands, 2012; ISBN 9789400740327. [Google Scholar]
- Powell, K.S.; Korosi, G.A.; Mackie, A.M. Monitoring grape phylloxera populations using simple non- destructive trapping systems. Acta Hortic. 2009, 816, 29–34. [Google Scholar] [CrossRef]
- Giblot-Ducray, D.; Correll, R.; Collins, C.; Nankivell, A.; Downs, A.; Pearce, I.; Mckay, A.C.; Ophel-Keller, K.M. Detection of grape phylloxera (Daktulosphaira vitifoliae Fitch) by real-time quantitative PCR: Development of a soil sampling protocol. Aust. J. Grape Wine Res. 2016, 22, 469–477. [Google Scholar] [CrossRef]
- Renzullo, L.J.; Blanchfield, A.L.; Powell, K.S. Insights into the early detection of grapevine phylloxera from in situ hyperspectral data. Acta Hortic. 2007, 733, 59–74. [Google Scholar] [CrossRef]
- Wildman, W.E.; Nagaoka, R.T.; Lider, L.A. Monitoring spread of grape phylloxera by color infrared aerial photography and ground investigation. Am. J. Enol. Vitic. 1983, 34, 83–94. [Google Scholar]
- Lobitz, B.; Johnson, L.; Hlavka, C.; Armstrong, R.; Bell, C. Grapevine Remote Sensing Analysis of Phylloxera Early Stress (GRAPES): Remote Sensing Analysis Summary; NASA Ames Research Center: Mountain View, CA, USA, 1997; pp. 1–36.
- Powell, K.S.; Hackworth, P.; Lewis, M.; Lamb, D.W.; Edwards, J. Identification of phylloxera from high resolution infrared aerial imagery: A comparative study between airborne imagery types. Aust. N. Z. Grapegrow. Winemak. 2004, 488, 51–54. [Google Scholar]
- Renzullo, L.J. Characterizing grapevine spectral response to phylloxera infestation. Boll. Comunità Sci. Australas. 2004, 1, 28–32. [Google Scholar]
- Frazier, P.; Whiting, J.; Powell, K.; Lamb, D. Characterising the development of grape phylloxera infestation with multi-temporal near-infrared aerial photography. Aust. N. Z. Grapegrow. Winemak. 2004, 485, 133–136. [Google Scholar]
- Bellvert, J.; Zarco-Tejada, P.J.; Girona, J.; Fereres, E. Mapping crop water stress index in a Pinot-noir vineyard: Comparing ground measurements with thermal remote sensing imagery from an unmanned aerial vehicle. Precis. Agric. 2014, 15, 361–376. [Google Scholar] [CrossRef]
- Mathews, A.J. Object-based spatiotemporal analysis of vine canopy vigor using an inexpensive unmanned aerial vehicle remote sensing system. J. Appl. Remote Sens. 2014, 8, 085199. [Google Scholar] [CrossRef]
- Bachmann, F.; Herbst, R.; Gebbers, R.; Hafner, V.V. Micro Uav Based Georeferenced Orthophoto Generation in Vis + Nir for Precision Agriculture. ISPRS Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2013, XL-1/W2, 11–16. [Google Scholar] [CrossRef]
- Honkavaara, E.; Kaivosoja, J.; Pellikka, I.; Pesonen, L.; Saari, H.; Salo, H.; Hakala, T.; Marklelin, L.; Solutions, M. Hyperspectral Reflectance Signatures and Point Clouds for Precision Agriculture by Light Weight Uav Imaging System. ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci. 2012, I, 353–358. [Google Scholar] [CrossRef]
- Honkavaara, E.; Saari, H.; Kaivosoja, J.; Pölönen, I.; Hakala, T.; Litkey, P.; Mäkynen, J.; Pesonen, L. Processing and assessment of spectrometric, stereoscopic imagery collected using a lightweight UAV spectral camera for precision agriculture. Remote Sens. 2013, 5, 5006–5039. [Google Scholar] [CrossRef] [Green Version]
- De Castro, A.I.; Ehsani, R.; Ploetz, R.C.; Crane, J.H.; Buchanon, S. Detection of laurel wilt disease in avocado using low altitude aerial imaging. PLoS ONE 2015, 10, 1–13. [Google Scholar] [CrossRef] [PubMed]
- Psirofonia, P.; Samaritakis, V.; Eliopoulos, P.; Potamitis, I. Use of Unmanned Aerial Vehicles for Agricultural Applications with Emphasis on Crop Protection: Three Novel Case—Studies. Int. J. Agric. Sci. Technol. 2017, 5, 30–39. [Google Scholar] [CrossRef]
- Habili, N.; Oorloff, J. Scyllarus: From Research to Commercial Software. In Proceedings of the ASWEC 2015 24th Australasian Software Engineering Conference, Adelaide, Australia, 28 September–1 October 2015; pp. 119–122. [Google Scholar]
- Rouse, J.W.; Haas, R.H.; Schell, J.A. Monitoring the Vernal Advancement and Retrogradation (Greenwave Effect) of Natural Vegetation; NASA’s Goddard Space Flight Center: Greenbelt, MD, USA, 1974; pp. 1–8.
- Rodriguez, D.; Fitzgerald, G.J.; Belford, R.; Christensen, L.K. Detection of nitrogen deficiency in wheat from spectral reflectance indices and basic crop eco-physiological concepts. Aust. J. Agric. Res. 2006, 57, 781–789. [Google Scholar] [CrossRef]
- Daughtry, C. Estimating Corn Leaf Chlorophyll Concentration from Leaf and Canopy Reflectance. Remote Sens. Environ. 2000, 74, 229–239. [Google Scholar] [CrossRef]
- Haboudane, D.; Miller, J.R.; Pattey, E.; Zarco-Tejada, P.J.; Strachan, I.B. Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop canopies: Modeling and validation in the context of precision agriculture. Remote Sens. Environ. 2004, 90, 337–352. [Google Scholar] [CrossRef]
- Haboudane, D.; Miller, J.R.; Tremblay, N.; Zarco-Tejada, P.J.; Dextraze, L. Integrated narrow-band vegetation indices for prediction of crop chlorophyll content for application to precision agriculture. Remote Sens. Environ. 2002, 81, 416–426. [Google Scholar] [CrossRef]
- Rondeaux, G.; Steven, M.; Baret, F. Optimization of soil-adjusted vegetation indices. Remote Sens. Environ. 1996, 55, 95–107. [Google Scholar] [CrossRef]
- Zarco-Tejada, P.J.; Berjón, A.; López-Lozano, R.; Miller, J.R.; Martín, P.; Cachorro, V.; González, M.R.; De Frutos, A. Assessing vineyard condition with hyperspectral indices: Leaf and canopy reflectance simulation in a row-structured discontinuous canopy. Remote Sens. Environ. 2005, 99, 271–287. [Google Scholar] [CrossRef]
Vegetation Index | Equation | Reference |
---|---|---|
Normalised Difference Vegetation Index (NDVI) | NDVIH = (R800 − R670/(R800 + R670) NDVIM = (R840 − R668/(R840 + R668) | [25] |
Normalised Difference Vegetation Index (NDVIGreen) (Green band) | NDVIGreen H = (R800 − R551/(R800 + R551) NDVIGreen M = (R840 − R560/(R840 + R560) | |
Normalised Difference Red Edge (NDRE) | NRDEM = (R840 − R717)/(R840 + R717) | [26] |
Modified Cab Absorption in Reflectance Index (MCARI) | MCARIH = [(R700 − R670) − 0.2(R700 − R551)] (R700/R670) MCARIM = [(R717 − R668) − 0.2(R717 − R560)] (R717/R668) | [27] |
Modified Chlorophyll Absorption in Reflectance Index (MCARI1) | MCARI1 M =1.2 [2.5 (R840 − R668) − 1.3(R840 − R560)] | [28] |
Modified Chlorophyll Absorption in Reflectance Index (MCARI2) | [28] | |
Transformed CARI (TACRI) | TCARIH = 3[(R700 − R670) − 0.2(R700 − R551)(R700/R670)] TCARIM = 3[(R717 − R668) − 0.2(R717 − R560)(R717/R668)] | [29] |
Optimised Soil-Adjusted Vegetation Index (OSAVI) | OSAVIH = (1 + 0.16)(R800 − R670)/(R800 + R670 + 0.16) OSAVIM = (1 + 0.16)(R840 − R668)/(R840 + R668 + 0.16) | [30] |
Blue/Green and Blue/Red Pigment indices | BGI2 M = R475/R560 BRI2 M = R475/R668 | [31] |
Phylloxera index 1 (PI1) | PI1 = (R522 − R504)/(R522 + R504) | (This study) |
Phylloxera index 2 (PI2) | PI2 = (R551 − R562)/(R551 + R562) | |
Phylloxera index 3 (PI3) | PI3 = (R700 − R680)/(R700 + R680) | |
Phylloxera index 4 (PI4) | PI4 = (R782 − R700)/(R782 + R700) | |
Phylloxera index 5 (PI5) | PI5 = (R782 − R671)/(R782 + R671) | |
Phylloxera index 6 (PI6) | PI4 = (R680 − R563)/(R680 + R563) |
Class | Vigour | Criteria | Phylloxera Presence Conjecture |
---|---|---|---|
5 | High | Plant or grapevines to or above a given height e.g., top supportive wire | Healthy (probably no infestation e.g., phylloxera) |
4 | Medium-high | Plant just below a given height e.g., top supportive wire | Mild symptoms (probably no infestation or early stages of impact e.g., phylloxera) |
3 | Medium | Plant height below middle wire and above bottom wire | Intermediate impact (probably low levels of infestation e.g., phylloxera) |
2 | Low | Short plants e.g., grapevines. Plants below bottom wire | Severe symptoms of infestation (e.g., phylloxera, surrounding (3–4) plants also likely to be infested) |
1 | No vigour | Dead plant | Extreme symptoms of infestation (e.g., phylloxera has been affecting the plant for years) |
© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Share and Cite
Vanegas, F.; Bratanov, D.; Powell, K.; Weiss, J.; Gonzalez, F. A Novel Methodology for Improving Plant Pest Surveillance in Vineyards and Crops Using UAV-Based Hyperspectral and Spatial Data. Sensors 2018, 18, 260. https://doi.org/10.3390/s18010260
Vanegas F, Bratanov D, Powell K, Weiss J, Gonzalez F. A Novel Methodology for Improving Plant Pest Surveillance in Vineyards and Crops Using UAV-Based Hyperspectral and Spatial Data. Sensors. 2018; 18(1):260. https://doi.org/10.3390/s18010260
Chicago/Turabian StyleVanegas, Fernando, Dmitry Bratanov, Kevin Powell, John Weiss, and Felipe Gonzalez. 2018. "A Novel Methodology for Improving Plant Pest Surveillance in Vineyards and Crops Using UAV-Based Hyperspectral and Spatial Data" Sensors 18, no. 1: 260. https://doi.org/10.3390/s18010260