Grapevine Remote sensing Analysis of Phylloxera Early Stress (GRAPES)


California winegrape cultivation is vital to the State's economy and provides the main support for the nation's $12 billion retail wine industry. In 1994 the wine industry provided 194,000 direct jobs nationally, with total wages of $3.5 billion. Tax assessments on wine remitted $1.6 billion to federal, state and local governments (source: Barsby & Associates).

Since the late 1980's, California wine growers have been faced with destruction of their vines by infestation of a root louse named phylloxera (biotype B). The louse kills vines by feeding on their roots. There is no way to eradicate the pest, and infested areas must eventually be replanted on a phylloxera-resistant or tolerant rootstock. The infestation is present in eight California counties, and is particularly severe in Napa and Sonoma Counties where thousands of acres of premiere vineyards have already been destroyed or are scheduled for future replacement.


During the 1993-1995 time period, NASA Ames Research Center ( Ecosystem Science and Technology Branch) collaborated with industry and university partners to develop and transfer the use of remote sensing and associated computerized technologies as a tool for vineyard managers to use in addressing the phylloxera problem. NASA's partners on this project included the University of California Cooperative Extension (Napa County), University of California Davis (Entomology Dept.), California State University Chico (School of Agriculture) and the Robert Mondavi Winery. Staff from each organization brought unique expertise to the project, working together in the field, laboratory and computer room. The work was co-funded by NASA's Office of Advanced Concepts and Technology and the Robert Mondavi Winery. Project results are being made available to the wine industry, commercial remote sensing product vendors, agricultural community and general public through invited oral presentations and written reports.

The typically rapid progression of phylloxera infestation was characterized for the 1989-1993 time period across a ~1000 acre Mondavi property using retrospective color infrared aerial photography.

During the 1993 growing season, field and aircraft data were collected from Napa Valley test sites with special sensors designed to study earth resources, including plant stress manifested as reductions in vegetation canopy density. Infestations are detectable in this remotely sensed imagery, even in the early stage when phylloxera are underground eating vine roots but the above ground plant still appears healthy.

Airborne image data were acquired in 1993, 1994, and 1995 over the study ranch. These images (310K) are only part of the ranch. Each image was acquired with a different sensor with different spatial (pixel) resolutions. The 1993 image has a spatial resolution of about 2m, the 1994 image about 5m, and the 1995 image about 3m. The images from 1993 and 1994 were normalized to each other and used to compute a vegetation index. This index image was then divided into twelve ranges of values, each representing a different amount of vegetation biomass or canopy density. The images were then colored so the areas with the little vegetation or bare soil are dark red, then through tan, yellow, light green, and deep green, corresponding to increasing canopy cover and plant health. Black rectangles in these images denote the field work (85K) study plots, where leaves and root diggings were collected. The 1993 binned image (8K) shows some damaged areas, but is otherwise pretty uniform. The 1994 binned image (10K) shows that only one year later the damage has spread and now affects much of the block. This decrease in vegetation canopy was counterbalanced by an increase in density in the uninfested parts of the block, resulting in an overall increase in canopy for the block as a whole. The difference image (12K) highlights the areas with a decrease in plant canopy in red and orange, and yellow, while areas with little change are pale green and those with an increased canopy in are shown in blues. The images can also be viewed as a group.

Some work has also been done to predict the extent of the damage within the vineyard blocks. This analysis was done by looking at the pixels in proximity to the 1993 stressed areas (3K). If the predicted distance was too small, then there was an error due to underestimating the spread. If the predicted distance was too great, then there was an error due to overestimating. For the study block the minimum error of about 26% occurred at about 8m. The 1994 prediction (3K) compares reasonably well with the image of the 1994 stressed areas (3K). The error may be reduced if other variables are taken into account, such as soil type or elevation, and etc. The images can also be viewed as a group. Because the spread of phylloxera is exponential, the predicted distance varies from block to block depending on the initial level of investation.

Vineyard managers work on the scale of the block, so the difference image was summarized by using the mean difference in each block. In this image (3K) the color scheme is the same as in the per pixel difference image above, but here the red areas are blocks that been pulled. The orange blocks should be pulled next and the yellow blocks are questionable. The pale green block is the study area and had a small increase, while the cyan block had a large increase.

Airborne imagery was collected over several vineyard regions in Napa and Sonoma counties 1995. An example of these 3m spatial resolution data from Carneros Region (74K) shows mostly variation in soil conditions. Except for the triangle in the upper right, there is not much vegetation canopy in this area. This image has also been processed into a binned vegetation index image (31K). Some 0.5m data were also acquired over the study area (82K). This image is close to the spatial resolution of color infrared photography. At this resolution the individual vines can be easily recognized. An interesting feature in this image is the violet rectangle in the bottom right. Note that lines on the tennis court can be distinguished.


By using remote sensing and associated analysis techniques, growers can attain earlier knowledge on the rate of spread of the infestation, and the rate of decline for affected vines. It is anticipated that this source of information will allow for more informed replanting decisions, helping California wineries retain market share.

Further Reading

Several publications related to this research have also been written.

Related Project

The 1997 Canopy Remote-sensing for Uniformly Segmented Harvest (CRUSH) project demonstrated a potential contribution of airborne imagery to harvest planning.

For more information about the GRAPES project please contact:

Lee F. Johnson
Senior Research Scientist
CSU Monterey Bay, Earth Systems Science & Policy
NASA Ames Research Center
Moffett Field CA 94035-1000

Back to the Ecosystem Science and Technology Branch page

This page has had accesses since 25 March 1996.

Last updated: September 1998

Please send any comments or suggestions concerning this page to
Brad Lobitz (blobitz(at)