Back to Science for Everyone
Equipped with more powerful computers, more precise weather data, and more accurate long-range weather predictions, hydrologists at the United States' National Weather Service (NWS) believe they are prepared to extend river stage and discharge forecasts further into the future. Automated weather and river measurements, when coupled with new computer models, could allow hydrologists to predict the ebb and flow of rivers as much as several months in advance, instead of the current one, two, and three days.
Long-range prediction is the goal of the Advanced Hydrologic Prediction System (AHPS), an NWS program that will be tested in March 1997 during a demonstration on the Des Moines river basin from southern Minnesota through Iowa. AHPS is being designed to capitalize on the modernization of NWS equipment and computer models—principally, the installation of 140 NEXRAD radars, the Automated Surface Observing System (ASOS), and the Advanced Weather Interactive Processing System (AWIPS)—with the expectation that meteorologists will be able to give hydrologists more reliable predictions of short-term weather patterns and long-term seasonal trends.
"To the people of the world, it sounds like witchcraft," says John Ingram, program manager for AHPS at the NWS Office of Hydrology. "But we are going to be providing forecasts days, weeks, months, even a season into the future."
Today, the River Forecast Centers of NWS provide mostly short-range forecasts, such as daily stages on large rivers, flood warnings, and deterministic predictions of river stage at specific points along a river once flood stage is reached. But the future promises probabilistic forecasts, according to Ingram, whereby hydrologists will predict in percentages the likelihood that water will flow by a point along a river at a certain stage or volume during various intervals of a day, week, or month.
"This is something that has been raised by emergency managers," Ingram notes. "They have asked us: 'when you put out your forecast, can you tell us a relative level of uncertainty?'" By providing water resource managers with information on the chance of a river being above or below its usual flow, AHPS should allow them to better mitigate the effects of changes in stream flow—particularly floods and droughts—on irrigation, drinking water, fisheries, shipping, and hydropower, and to allow for the sustainable use of water. The developers of AHPS expect that other federal agencies—such as the U.S. Geological Survey, the Army Corps of Engineers, the Federal Emergency Management Agency, the Environmental Protection Agency, the U.S. Department of Agriculture, and other sections of the National Oceanic and Atmospheric Administration—will contribute data to and rely on predictions from the AHPS. In addition to these federal participants, NWS is working with state and regional departments of natural resources and of emergency management, with city water managers and public works officials, and with constituents in industry, agriculture, manufacturing, boating, and recreation.
In a survey conducted last spring and summer by hydrologists Mike Longnecker and John Feldt of NWS's Des Moines office, 60% of potential users of the AHPS asked for extended forecasts out to 5 days, and 40% sought forecasts out to 10 days. Having been hit hard by the Great Flood of 1993, Longnecker notes, local water resource managers are eager to have the advance notice that AHPS might provide in order to better prepare for water disasters. According to the National Research Council, floods and flash floods in the United States cause about $1 billion in property damage and claim more than 100 lives each year. About 75% of all disaster areas declared by the U.S. federal government are due to floods.
"These probabilistic forecasts are really going to lend themselves to predicting potential floods because we'll be able to provide a flow or a stage at a given point," says Dean Braatz, the Hydrologist in Charge at NWS's North Central River Forecast Center in Chanhassen, Minnesota, which will run the streamflow models for the Des Moines River demonstration. "We can give water resource managers a 25% or 50% or 75% exceedence probability, so then they can make intelligent decisions based on more complete hydrologic information." For instance, Braatz suggests that if such forecasts had been in effect in 1993, hydrologists might have been able to pinpoint earlier the signature of a major flood. With an indication of magnitude of the flood, city planners might have taken steps to avoid disasters such as the overflow of the sewage treatment plant in Des Moines.
Rather than constructing an entirely new hydrologic system, the developers of AHPS are relying on the existing infrastructure of NWS and new approaches to using climatic data. The centerpiece of the system is a new computer model being developed by the NWS Office of Hydrology and Riverside Technologies, Inc. The system is designed to incorporate historical data sets on stream flow, temperatures, and precipitation; current snow water contents; quantitative precipitation forecasts; knowledge of soil moisture conditions; and seasonal forecasts from the models of NWS's Climate Prediction Centers.
When all of those factors are combined with up-to-the-minute weather data, the River Forecast Centers will be able to generate time-series forecasts of stage and discharge at selected points along a river. Whereas today the RFCs deliver stage and crest forecasts to local NWS forecast offices, and they in turn issue flood watches and warnings, in the near future, they will predict the likelihood that any given reach along a river will be flooded and to what depth. By combining geographic and topographic information with modeled projections of river flow, AHPS will provide "inundation maps" for entire swatches of a river's banks, instead of stage values for individual gauged points on a rivers.
AHPS will get its trial run during the 1997 snow melt and runoff season (March to May) in the Des Moines River basin. According to Longnecker, information such as charts, maps, hydrographs, and other visual displays produced during the demonstration will be provided to public and private users via the World Wide Web. NWS hydrologists will continually update information on stage hydrograph displays, flash flood guidance, quantitative precipitation forecasts, radar-detected rainfall, soil conditions, and the probability that streams will exceed their banks in any given area.
Testing AHPS in Iowa will be the first step toward predicting hydrologic response on all the streams and rivers that feed the Mississippi River, and eventually on the mighty river itself. The Minnesota River basin is likely to have similar, enhanced forecasts by the spring of 1998 if the 1997 demonstration is successful, Braatz notes.—Michael Carlowicz