An App for Predicting Crop Disease Spread

by Joy Drohan

A simple smartphone app could halt the spread of one of the most devastating viral diseases of cereal crops worldwide. At least that’s the hope of Penn State researchers who are creating an app that can demonstrate the value of predicting where and how quickly the barley yellow dwarf virus (BYDV) will spread within individual fields of wheat, barley, rice, and oats. Just as important, grain growers’ reaction to such an information system will be evaluated.

Satellite image of land surface temperatures of the United States. PHOTO: NASA

This satellite image is a composite of land surface temperatures and an example of data scientists might use in development of predictive apps. The image was produced by INTREPID from data taken by the National Oceanic and Atmospheric Administration's (NOAA) Advanced Very High Resolution Radiometer (AVHRR) instrument. INTREPID is NASA's International Research Partnership for Infectious Diseases program. PHOTO: NASA

Ed Rajotte, professor of entomology, and Cristina Rosa, a research associate specializing in insect-transmitted plant viruses, are leading the development of this prototype smartphone app through a $420,000 grant from the U.S. Department of Agriculture’s National Institute of Food and Agriculture. 

They envision growers, crop consultants, and other agricultural professionals reporting crop and pest data for particular fields via smartphone to be combined with weather-based models. The reporters will then be able to view online maps showing the locations of insects that spread the disease and forecasts of their population movements and changes.

BYDV is spread by more than 20 species of aphids that transmit the virus to plants when they insert their mouthparts into the sticky phloem. The infection zaps the plants’ resources, leaving them less able to photosynthesize. The result is a yellowing of the leaves and reduced growth. 

This damage can cause yield losses of up to 70 percent, and the problem is expected to grow with climate change. The disease is not well reported in Pennsylvania, so its symptoms are often mistaken for poor nutrition. 

According to Rosa, there are currently no control options once BYDV infects plants, so growers target the insects that transmit it to prevent its establishment. They scout for aphids and apply pesticides when it seems to make economic sense. However, aphid scouting is time consuming and costly, economic returns are uncertain, and the overuse of pesticides can have detrimental environmental effects.

“If BYDV was devastating every year, we wouldn’t need a system like this,” Rajotte says. Because the disease is variable and patchy, a warning system makes sense.


Joseph Walls, Cristina Rosa, and Mitzy Porras work to strengthen disease-forecasting model. PHOTO: STEVE WILLIAMS

LR: Joseph Walls, Cristina Rosa, and Mitzy Porras. Walls and Porras, a Fulbright scholar from Columbia, work with Rosa, a research associate specializing in insect-transmitted plant viruses, to help adapt and strengthen the disease-forecasting model. They will also help collect and integrate supplementary data from Pennsylvania fields. PHOTO: STEVE WILLIAMS


Constructing a Disease Model for Pennsylvania

Rajotte’s team will work with ZedX, a private information technology company in Bellefonte, Pennsylvania, to design and build the app. ZedX is the developer and provider of several web-based commercial and government-backed information technology platforms that integrate weather, crops, and pest information. They also offer precision agriculture services to assist producers in field-level crop management.


Joe Russo, founder of ZedX. PHOTO: PAT LITTLE

Joe Russo, founder of ZedX. PHOTO: PAT LITTLE

To get started, Penn State and ZedX researchers will construct a bydv disease-forecasting model using 20 years of data on weather, crops, aphid growth, and virus spread in Italy. An Italian colleague, Piero Caciagli, of the Institute of Plant Virology in Turin, Italy, provided the data set.

“We don’t yet know the accuracy of the model or its usefulness in the United States,” says Rosa. “But it should give us enough information to build the smartphone delivery system and gauge the reaction of a group of growers in south-central Pennsylvania who will test a pilot app.”

Two entomology graduate students, Mitzy Porras, a Fulbright scholar from Colombia, and Joseph Walls, will help adapt and strengthen the disease-forecasting model. They will also help collect and integrate supplementary data from Pennsylvania fields.

The disease-forecasting model will serve as the basis for the app, which will include capacity for data entry in the field, geospatial mapping, and crop treatment recommendations at a scale of one square kilometer in Pennsylvania. Observers around the state will enter data into the app via smartphone. Data will include the species of pest, crop cultivar and stage, and the location of the field. ZedX will then generate high-resolution weather data for the field. By configuring the bydv model with the app-collected observations and using the site-specific weather data as input, aphid movement and population changes can be simulated across the Commonwealth.

“Every person with a smartphone can become an observer, and we’ve never had that before,” says Joe Russo, founder of ZedX. “It’s a different way of looking at your farm when you can get information field by field at a very rapid pace. You have to adjust your decision making for the tool.”

The app’s output should be useful for short- to long-range crop management planning. In the short term, it could help a grower decide whether to spray pesticide or delay planting to avoid a coming wave of aphids. In the medium term, it could help growers decide which variety of a crop to plant. If populations of BYDV-carrying aphids are expected to be high based on meteorological trends, growers can sometimes choose a resistant plant variety. In the long term, policy makers could use forecasts from tools such as this to decide, for example, how much wheat to import. 


Overturning Barriers to Adoption

From prior research, Rajotte, Rosa, and other project collaborators know that various barriers stand in the way of growers’ adoption of new technologies. 

First, the scientific barrier. The technology must be relevant and scientifically based. The Penn State team brings credibility and experience from many years of modeling other crop pests and diseases. That experience, plus 20 years of bydv and associated crop and meteorological data, ensures a successful app and forecasting system.

Second, the sociological barriers. Growers and other potential users have a tremendous variety of experience with smartphones, apps, and geospatial mapping. Two rural sociology graduate students will assess how growers want to get information, what they like and dislike about the app, how it could be tweaked to be more intuitive and useful, and whether growers change their crop management practices based on the app’s output. The students will help evaluate the app’s success by comparing the knowledge, goals, and pest management decisions of growers with and without access to the app. 

“Farmers are running complicated businesses,” says Rajotte. “They’ve been doing things a certain way for years. We need to show them that local, on-the-spot information makes for better farming. The emphasis of this project is to determine how growers want to receive information like this.”

Bridging the gap between traditional book-based learning and cutting-edge, real-time technology is a major goal of this project. That’s why the new tool will become part of the college’s integrated pest management course. Introducing tools like this to the world’s future farmers and farm decision makers when they’re in college makes their use on the farm later more likely. It also builds the future agricultural technology workforce. The project team also will test the app with extension educators and crop consultants. 

Third, the technological barriers, which include making the app work across various phone models and operating systems, overcoming the limitations on cellular service in rural areas, and dealing with the need to update the BYDV model at least daily. Russo says that through experience gained in past projects, the team has pretty well addressed these hurdles. 

Finally, the longevity barrier. Growers have to see the technology as having a long, useful life and reliable customer service. That’s where ZedX comes in as a private company. “We need the private sector to deliver the tool to farmers after the grant period,” says Rajotte. 

Russo says that ZedX hopes eventually to offer a commercialized application that will track bydv and other diseases as part of a web-based subscription service. 

Rajotte and Rosa emphasize that this two-year project will not produce a finalized perfectly predictive app. ZedX will still need to refine the disease model and the app. But some best practices for communicating with the grower audience via smartphone app should be clearer. 

“It’s easy for us to make phone apps,” Russo says. “The tough part is getting the right design for the right audience for the practices and crops and pests involved.”


Building on Experience 

Penn State has worked before with ZedX to develop other pest prediction tools. One of these is pa pipe (, a web-based state-level pest prediction system for insects, diseases, and weeds done in collaboration with the Pennsylvania Department of Agriculture. Until ZedX’s multidisease scouting application becomes available, bydv tracking information could be displayed on pa pipe.

Rajotte and other researchers at Penn State also collaborated with ZedX on the highly successful and cost-effective web-based international platform for Asian soybean rust prediction ( Soybean rust entered the United States in 2004. Through the platform, soybean extension specialists, agricultural service companies, independent consultants, producers, and other stakeholders could track the disease from its first day of entry into North America. The BYDV project team is modifying the soybean rust web platform for application to BYDV. 

It turns out that soybean rust has not been a devastating pathogen so far in the United States, but when it showed up, says Rajotte, “the fear factor was high.” So for peace of mind, without the web-based tool telling growers in which counties the fungus had actually been found, they would have sprayed a lot of fungicide to protect their crops. 

A study by usda’s Economic Research Service estimated that in 2005 alone, a year with low rust infection, use of the web-based tool saved U.S. soybean growers at least two times and up to 60 times more than it cost to develop the tool. Savings came mainly in the form of fungicides not sprayed. Similar savings may be possible with the bydv app. And those figures do not account for unquantifiable benefits such as protecting human health, beneficial insects, and water quality by reducing pesticide use. 

By taking advantage of experience and knowledge gained through the soybean rust platform and by tweaking the elements of this existing system, Rajotte and Rosa hope to make rapid progress with BYDV and, in the end, help growers to adopt web-based pest prediction systems.


Bird Cherry-oat Aphids on leaf. PHOTO: CRISTINA ROSA



Other personnel include John Tooker, assistant professor of entomology; Beth Gugino, assistant professor of plant pathology; and Carolyn Sachs, professor of rural sociology.