Remote Scouting of Insect Damage in Potatoes

Transcription

1 Remote Scouting of Insect Damage in Potatoes Ian MacRae, Timothy Baker Dept. of: Entomology, Univ. of Minnesota Potato Remote Sensing Conference Madison, WI. Nov14, Use hyperspectral sensors to identify wavelengths of interest, this can determine what wavelengths to look at with airborne sensors. 1

2 Select wavelengths Alves, T.M., Macrae, I.V. and Koch, R.L., Soybean aphid (Hemiptera: Aphididae) affects soybean spectral reflectance. Journal of economic entomology, 108(6), pp Wide variability of UAS available, we use multirotor systems for loiter time over plots Fixed wing systems more applicable in commercial operations 3-D Robotics Solo DJI S

3 Imaging sensors Multi-sensor array (Sentera Quad) RGB + 3 specific bandwidth sensors Provide ~cm resolution at 40m altitude. FLIR Vue Pro R cameras good resolution, small, lightweight so fits small UAS Thermal IR cameras small enough to be mounted on small UAS Chaerle, Laury, and Dominique Van Der Straeten Imaging Techniques And The Early Detection Of Plant Stress. Trends In Plant Science 5 (11): doi: /s (00)

4 Software Tools Stitching with Agisoft PhotoScan (Agrisoft LLC, St Petersburg, RU) Image / Data Analysis Image analysis ENVI, ArcGIS Python scripting for data manipulation, preparation Depends on what analysis is necessary Colorado Potato Beetle Poster Child for defoliating insects (and resistance!) Current control based largely on use of at-plant neonicotinoid insecticides development of resistance and potential regulatory restrictions leading to re-adoption of threshold based foliar treatments 4

5 Defoliation estimates Percent defoliation estimated weekly Plots flown simultaneously Analysis compared visual defoliation estimates to those derived from aerial imagery Canopy segmentation Process - Identify Imagery from TetraCam ADC used analyses spectral reflections of from PixelWrench to do supervised classification representative nonbased on min & max red & NIR values (e.g. plant areas to create red<6, NIR <20), non-vegetative pixels rendered selection sieve values to a user-specified solid color while leaving for selection decisions pixels representing vegetative material Delineate the area of unchanged. interest and perform analysis. % canopy coverage then calculated, because images are geocoded, actual area can also be determined 5

6 Results % calculated canopy coverage regressed against % observed defoliation to assess fit of remote sensing methodology to observational estimates P<0.001, R2=0.58 6

7 Image analysis Decided to use Visible Atmospheric Resistant Index (VARI*) to utilize visible data Stitched image uploaded into ArcGIS 10.3 and plots bounded by polygons and center 2 rows highlighted. The stitched image was clipped to produce a raster of only the plots to be analyzed. Supervised classification was applied on center two rows *Gitelson, A., et al. "Vegetation and Soil Lines in Visible Spectral Space: A Concept and Technique for Remote Estimation of Vegetation Fraction." International Journal of Remote Sensing 23 (2002):

8 Analysis Maximum likelihood classification was conducted. All pixels were included in the classification, i.e., no values remained unclassified due to low probability. Resulting raster image displaying calculated areas of vegetation and soil were converted to a polygon shapefile The raster image was overlaid with the original layer with the plot center rows to retain plot numbers. Model fit Correlated to ground observed estimates r 2 = 0.797, p <

9 Regression Statistics Multiple R R Square Adjusted R Square Standard Error Observations

10 Biological issues Varietal impact Success of RS of Sugarbeet Root Maggot variable depending on variety Certain diseases as well Confounding Stressors Disease & insect damage affecting same wavelengths Management tactics can influence reflectance Insecticide (Alves, T.M., Marston, Z.P., MacRae, I.V. and Koch, R.L., Effects of Foliar Insecticides on Leaf-Level Spectral Reflectance of Soybean. Journal of Economic Entomology.) Preliminary data indicates Aphoil as well (applied to prevent aphid virus vectors from probing in seed potatoes) Acknowledgements - Beetle Counters & Defoliation Estimators (AKA The MN SpudBug Crew) Nate Russart, Josie Dillon, Christopher Follette, Christian Halos, Guthrie Dingman, Abbie Anderson, Alex McGregor, Nicolai Broekmeier, Joey Koening, Lynn Haake, & Allison Larson Remote Sensing Timothy Baker & Nicole Dudycha The admin & staff at the UMN Sand Plains Research Farm (Becker), & the northwest Research & Outreach Ccenter (Crookston) This research was supported by funding from the MN Dept of Agriculture Crop Research Program, MN Dept. of Ag / USDA Specialty Crop Block Grant, the Northern Plains Potato Growers Assoc. and MN Area II Potato Growers Assoc. 10