|
 |
|
Re-Thinking Drought Prediction
But what seems like an amusing notion grew out of a serious investigation of a grim issue for the country's agricultural regions: trying to forecast droughts. Nationwide drought losses can total up to 8 billion dollars a year. Great Basin farming states like Nebraska have suffered such severe droughts that old-timers remember when the very crops they should have been growing had to be shipped in to feed starving residents and their livestock. Even now, several farming states are suffering through a drought that started in 2000. Cyclical weather patterns guarantee there will be other droughts in the decades to come. Where, when and how severe are critical questions for farmers and citizens who depend on the farm economy. Waltman and his fellow Digital Government researchers Nebraska computer scientists Stephen Goddard and Stephen E. Reichenbach received a DG grant to create a National Agricultural Decision Support System, which would allow better drought forecasting so that states might be able to avoid devastating losses. They are doing their work on behalf of the U.S. Department of Agriculture (USDA) Risk Management Agency (RMA), which underwrites farm insurance policies. The purpose of the partnership is the development of risk management tools for use directly by agricultural producers to minimize their production risks, according to Ron Lundine, a Risk Management Specialist at the RMA. The memories of droughts past are not just anecdotal. "Everything is already there in a government database," says Waltman of quantified information about precipitation and temperature from agencies such as the USGS, NOAA and the High Plains Regional Climate Center. In some cases, there is more than a century's worth of records, with statistics going back to the 1880s. But the information is uncoordinated and covers a vast spatio-temporal range, over thousands of miles and many decades. The researchers realized they were like farm families thinking of appearing on "Antiques Roadshow." If there is generations worth of stuff stored in dozens of barns, what's the most efficient way to go through it? And what if you should discover that what one relative categorized as a "farm implement," another called a "hoe"? The answer was a complicated combination of database integration and data mining, taking data from all the different monitoring systems to be able to portray droughts at different time scales, including severity, intensity, and geospatial occurrence of drought. "We even created two new data-mining algorithms to identify relationships between different types of episodes that may not overlap in time," says Goddard. Waltman gives an example of the power of the new system: "We did data mining to look at El Niño/La Niña processes relative to droughts. If you have a La Niña setting up April to December, the following spring growing season you can count on a drought in Nebraska. If you know that, you can make some mitigation changes in farming, like planting early to get root development or switching to a shorter season corn hybrid, or switching to another, more drought resistant crop." This, explains Waltman, is why the wine grape project was actually as serious as the rest. Wine grapes turn out to be both drought resistant and lucrative: "You need a little bit of trickle irrigation in the beginning, but the root systems on grapes go down twenty feet, so you can deep-mine the water. And you can get $1,000 an acre for grapes, where you're only getting $300 an acre for corn." Palmer Drought Severity Index, which was created in 1965 by the late W. C. Palmer, a government climatologist. Palmer designed a uniform scale to compare the actual effects of droughts, so that "extreme" drought meant just that, whether in Santa Fe or Seattle - whatever the residents' subjective experience. The PDSI is far from an academic exercise, it is often used to calculate whether to initiate government drought relief. Unfortunately for those affected, the limited computing resources of the time restricted both the amount of data Palmer could include and the complexity of his mathematical calculations. There have been many criticisms of the Index over time (For a discussion see: http://nadss.unl.edu/PDSIReport/pdsi/problems.html). Goddard and Nathan Wells, an outstanding UNL undergraduate who was selected as a member of USA Today's 2002 Academic Second Team, created the first update to the index in the four decades since its creation. "We went back to the original literature and looked at his constraints," says Goddard. "We looked at what he would have done if he'd had today's computers. With our way of calibrating, you get a much more comparable spatial index. " The Web back-end for the project is innovative as well, for it Goddard designed a four-layer distributed architecture. It is an open standards restructuring of proprietary geographic information systems that should ease interoperability. Unfortunately, the front-end may seem overly complex to the average farmer, since the current interface is designed to be understood by professional climate/soil researchers and officials at concerned government agencies. Planned is an additional interface that will be targeted towards growers, more directly concerned with whether to plant corn or soy than with interpreting the Palmer Drought Severity Index. "All of our partnerships are designed for the benefit of individual farmers," says Lundine. "This project will incorporate a lot more usable drought information than farmers can normally get on Web sitesÑso it should be quite useful." | ||||||
|
This site is maintained by the Digital Government Research Center at the University of Southern California's Information Sciences Institute. |
|
• CONTACT • POLICIES | ||
| | |||||
