Current research has identified spectral differences between virus-infected and uninfected vines. These specific variations occur in the visible (550 nm), near infrared (900 nm), and mid infrared (1600 nm and 2200 nm) regions of the electromagnetic spectrum. Leaf Roll is currently associated with 8 different viruses and is vectorized by small insects known as mealybugs. This specific research was conducted by Washington State University's Department of Plant Pathology. These spectral signatures provide a potential tool to efficiently identify and assess the spread Leaf Roll Virus within specific vineyards.
Through the the scope of free satellite imagery (specifically ASTER), I plan on exploring the spread rates of Leaf Roll Virus between the years 2010 and 2011. I am interested in vineyards located on the Central Coast of California, specifically vineyards with ground-identified, viral infection. By examining a time series of two images, I am looking to identify spreading patterns or noticeable trends. This research can not only aid vineyard management, but also protect something as large as the California Wine Industry.
Study Area
- Southern San Luis Obispo County
- Rincon Vineyard, Talley Estate
- Arroyo Grande, California
 |
| Fig. 1, Southern San Luis Obispo County coastal view |
 |
| Fig. 2, Talley Vineyards from above |
 |
| Fig. 3, Vectorized map of the East Rincon Vineyard |
Data Analysis
Because my studying will involve only passive remote sensing techniques, all imagery analyzed will be taken on cloudless days ranging from the months of May to October. This annual period is associated with the most foliage and therefore the strongest reflectance. Viewing this specific season as a time series from September 2010 to present will ideally yield change in the Normalized Difference Vegetation Index that is potentially associated with Leaf Roll Virus infection. The images were taken by NASA's Advanced Spaceborne Thermal Emissions Reflection Radiometer (ASTER). My specific ASTER imagery was ordered and provided free of charge through NASA's Reverb Echo Data Portal. These images were captured September 9th, 2010 and September 12, 2011. Each image was imported into Envi and viewed within the red and infrared spectral bands. Data analysis for each ASTER image was completed through this specific algorithm.
Algorithm
1) Upload Aster Images in Envi and display via bands 2 and 3
2) Save image as an Envi file
3) Upload new Envi file and transform into a Normalized Difference Vegetation Index (NDVI)
4) Overlay a density slice on the newly created NDVI
5) Through tools, utilize the spatial pixel editor to scan ground tested infected vines to determine common NDVI values
6) Assign new ranges within the Density Slice to appropriately break up the range of NDVI pixel values.
 |
| Fig. 4, Density Slice Ranges |
After these steps have been processed, I will run a change detection command from between the years of 2010 and 2011 to visually quantify the percent change potentially associated with Leaf Roll Activity. Ideally I am looking for uneven NDVI values and noticeable spreading within the most recent images.
Results
Viewing the NDVI density slices for 2010 and 2011, it is evident that change has occurred. Increases and decreases in overall greenness exist throughout the entire East Rincon Vineyard. Noticeable trends are a homogenous increase in NDVI value (on average .15 to .20) within the central, most southern, and most eastern blocks of the vineyard. This indicates primarily healthy vines within this portion. Interestingly, the only decrease within these specific blocks occurred on the western boundaries. The western and southwestern blocks of the vineyard have experienced a noticeable decrease in NDVI ranges (on average .18 to .13). It is important to note that the ground-tested, infected region (see fig. 3) experienced clusters of this specific decrease in NDVI value. The change detection map summarizes these trends into greenness increases or decreases between the years 2010 and 2011. The two vineyard adjacent to the infected block experienced similar patches of decrease, potentially indicating the spread of infected vines.
 |
| Fig. 5, ASTER Image of San Luis Obispo County, September 9th 2010 |
 |
| Fig. 6, ASTER Image of San Luis Obispo County, September 12th 2011 |
 |
| Fig. 7, Talley Estate Vineyards in ASTER bands 2&3 (9-9-2010) |
 |
| Fig. 8, Talley Estate Vineyards in ASTER bands 2&3 (9-12-2011) |
 |
| Fig. 9, NDVI Talley Estate Vineyards (9-9-2010) |
 |
| Fig. 10, NDVI Talley Estate Vineyards (9-12-2011) |
 |
| Fig. 11, East Rincon Vineyard Density Slice of NDVI (9-9-2010) |
 |
| Fig. 12, East Rincon Vineyard Density Slice of NDVI (9-12-2011) |
 |
| Fig. 13, East Rincon Vineyard Change Detection Map from 2010-2011 |
Discussion
Up until now, vineyard management has relied on time and labor intensive ground testing to determine viral infection within individual blocks of vines (Blanchfield et al. 3). It is important to note that remote sensing is still a very new approach for the detection of the Leaf Roll Virus, indicating that the procedure of detection is still much in development. My study utilized only free, online imagery, and no on site survey data. Although this method is time and money efficient, my results are restricted to a resolution of 15m x 15m (ASTER bands 2&3) and potentially missed smaller deviations or stress indicators. Due to advances in remote sensing technology, high resolution imagery such as IKONOS or QuickBird have become ideal for agriculture or viticulture surveillance (Blanchfield et al. 4). This equates to a major increase in spatial resolution, and with that comes the ability to detect phenomena that vary on the scale of centimeters to meters (Franklin and Wulder 179). This advanced imagery belongs to corporations and is priced at values that exceeded my current budget.
Experts agree that the red and near infrared bands are the best spectral regions for detecting 'weak spots' or vegetative stress. These were the specific bands that I chose to activate in my remote sensing image analysis. More importantly, these bands allow for the generation and application of vegetation indices capable of detecting differences within the "chlorophyll red-edge" of leaves (Naidu et al. 39). My primary tool for gauging these chlorophyll fluctuations was the Normalized Vegetation Difference Index (NDVI). With ENVI software, creating an NDVI is a relatively simple procedure because the algorithm is already built into the program. In an experiment by Naidu et al. (2009), the NDVI, Visibile Atmospherically Resistant Index, Modified Absorption in Reflectance Index, Red-edge Vegetative Stress Index, Photochemical Reflectance Index and Water Index were all taken and tested for their detection accuracy of both infected and noninflected vines. Overall the Photochemical Reflectance index was individually the most accurate achieving 72%. A combination of Red-edge Vegetative Stress index with Photochemical Reflectance Index increased this accuracy to 78%. Although no findings were published on the accuracy of NDVI alone, a combination of NDVI with Photochemical Reflectance Index yielded a 73% detection accuracy (Naidu et al. 44).While conducting my research, I was unaware of the accuracy associated with these more specific indices.
My final change detection map yielded patches of the East Rincon Vineyard that have experienced decrease in chlorophyll activity or overall greenness between September 2010 and 2011. Many abiotic and biotic factors can influence this reduction such as "water stress, nutrient deficiencies, or drainage issues" (Blanchfield et al. 4). With this being said, I can't say that all of the blue regions on my change detection map equate specifically to Leaf Roll Disease. Ground surveying is required to definitively confirm infection, but this map does provide focus areas or patches within the vineyard that should be tested. This saves the vineyard management time and money by immediately directing their attention. Overall, my remote sensing approach could be drastically improved by investing in high resolution imagery, employing the Photochemical Reflectance index alongside an NDVI, and utilizing on site survey measurements. In the future I expect to see many more applications of remote sensing within viticulture.
References
Franklin, S.E., Wulder, M.A. 2002. Remote sensing methods in medium spatial resolution satellite data land cover of large areas. Progress in Physical Geography 26 (2), 173-205.
Blanchfield, A.L., Robinson, S.A., Renzullo, L.J., Powell, K.S. 2006. Phylloxera infested grapevines have reduced chlorophyll and increased photoprotective pigment content- Can leaf composition aid pest detection? Functional Plant Biology 33, 507-517.
Rumpf, T., Mahlein, A. K., Steiner, U., Oerke, E.C., Dehne, H.W., Plumer, L. 2010. Early detection and classification of plant diseases with Support Vector Machines based on hyperspectral reflectance. Computers and Electronics in Agriculture 74, 91-99.
Naidu, R.A., Perry, E.M., Pierce, F.J., Mekuria, T. 2009. The potential of spectral reflectance techniques for the detection of Grapevine leafroll-associated virus-3 in two red-berried wine grape cultivars. Computers and Electronics in Agriculture 66, 38-45.
Jenson, J.R. 2007. Remote Sensing of the Environment: An Earth Resource Perspective. second ed. Pearson Education, Inc., New Jersey.
Reverb Echo: The Next Generation Earth Science Discovery Tool.
http://reverb.echo.nasa.gov/reverb/#utf8=%E2%9C%93&spatial_map=satellite&spatial_type=rectangle
Envi Tutorials. http://www.ittvis.com
