UCAR > Communications > Staff Notes > May 1997 Search

May 1997

The U.S. Weather Research Program is off and running

What makes a good weather forecast? According to Rit Carbone, it has to be more than accurate and reliable; it must be specific. "We're forecasting isobars on the weather map pretty well," says Rit, "but they don't describe a line of thunderstorms or a severe frontal rainband. These are the things that affect people's lives, not the presence of low pressure. We have to predict the right phenomenon at the right place. When Philly gets the weather forecast for Baltimore, it can be worse than having no forecast at all."

A better forecast for the general public--as well as for weather-sensitive economic sectors--is the ultimate goal of a very complex multiagency program finally gaining steam after years of tepid financial support. Rit is lead scientist for the U.S. Weather Research Program (USWRP), which held several key organizational meetings at NCAR last month.

High on ambition but low on direct funding, the USWRP is attempting to leverage and coordinate current research at its participating laboratories. The directors of 23 labs or divisions within NOAA, NASA, the Naval Research Laboratory, and NCAR (including ATD, CGD, ESIG, MMM, and RAP) assembled at the Foothills Lab on 15-16 April to discuss research now under way that could be coordinated to help meet USWRP goals. Two weeks earlier, a group composed mainly of social scientists and end users met at the Mesa Lab to discuss how weather information is used and where USWRP energies might be focused for the best societal benefit (see sidebar).

Meanwhile, NSF's Stephan Nelson is helping to administer new and reprogrammed funds contributed by all participating agencies for fresh research tied to USWRP goals. NCAR's portion of those funds was divvied up through an internal proposal process and the awarding of matching in-house grants.

Why now?

The hidden costs of winter storms may be greater than their immediate damage. For instance, payroll surveys--a part of the U.S. labor statistics that reflect the condition of the national economy--can be influenced significantly by snowstorms that throw people out of work for days. The storms themselves remain hard to pin down as well. While the development of major winter storms such as the March 1993 East Coast blizzard can often be seen days in advance, their exact locations are harder to predict--sometimes leading to large snowfalls in places where rain had been forecast, or vice versa. Research on winter weather tied to USWRP goals includes radar-based algorithms developed at NCAR's Research Applications Program for improving short-term snowfall predictions. (Photo courtesy Carlye Calvin, taken at her home in Nederland on 24 April 1997.)

According to Rit, "a conjunction of meteorological stars" led to the current blossoming of the USWRP. "Some of what the STORM program sought to accomplish 15 or 20 years ago now can be done with modest incremental effort."

STORM was the Stormscale Operational and Research Meteorology program, whose project office was based at NCAR through much of the 1980s. "The USWRP is basically a renaming and reformulation of that program," says Rit. A large group of atmospheric scientists, including former UCAR president Bob White, conceived of STORM and developed research plans in the late 1970s and early 1980s as a way to step up the exchange between research and operations in the soon-to-be-modernized National Weather Service (NWS).

Hurricanes and their landfall are a prime focus of the USWRP. NOAA's Hurricane Research Division would like to bring two-day track forecasts to the present accuracy of one-day outlooks. "If you have two daylight periods . . . there are a lot more things you can do to minimize damage," says HRD director Hugh Willoughby. However, changes in hurricane intensity still cannot be reliably forecast by present-day computer models. Above: an infrared satellite view of Hurricane Andrew as it churns ashore south of Miami on 24 August 1992. (Photo courtesy NOAA.)

"It was a good idea, but the system wasn't ready for it," Rit says. "The science was immature, some of the tools weren't there, and I'd say the dispositions of both the research and operations communities weren't completely amenable. Some researchers didn't envy the prospect of getting their hands dirty [with operations-oriented work] 15 or 20 years ago." By the same token, he adds, "some operations people didn't think the pointy-headed researchers knew much about day-to-day weather and forecasting."

Neither STORM nor the early USWRP gained significant material support from Washington, says Rit. "There was token recognition late in the Bush administration through endorsement of an unfunded strategic plan and a so-called implementation plan. That was its status when I came along."

Rit Carbone
In 1993, Rit, a senior scientist since 1987, left his post as ATD director and joined MMM to restart his personal program of research. By arrangement with NOAA and NSF, he also agreed to spend half of his time as lead scientist in developing what he calls "a fresh definition of issues and opportunities concerning research related to weather prediction." (Former NOAA scientist Bill Hooke leads the main USWRP office in Washington.) When he came on board, Rit felt the USWRP program at long last had a chance of gaining momentum. "The opportunity looked substantial enough to try to launch this program, because the ingredients were there."

One ingredient is the recent deployment of over 100 science and operations officers (SOOs), one for each local NWS office. Each SOO has a background in both research and operational meteorology and receives some additional training at UCAR through the COMET program. "The SOOs strengthen our ability to do research that involves both the academic community and [NWS] field offices," explains Rit.

At the same time, the massive, decade-long overhaul of the NWS observational networks has been nearing maturity. Scientists now have access to the data they've needed in order to look at weather on the scale it occurs--the mesoscale--in addition to the larger synoptic scale mapped by traditional surface and upper-air observing systems. Computer models also are becoming better equipped to look at the atmosphere more closely. Operational models of the 1980s traced conditions at points typically separated by 300 kilometers or more. The newest NWS operational model has a resolution of 28 km, with a sharpening to 8 km on the national scale possible by the end of the decade.

Now the challenge is to figure out how best to use the new data sources and computer models, along with newly applied mathematical techniques to help the models assimilate (or incorporate) the data. This is where the USWRP comes in. The program's stated focus is to "determine the best practicable mix of observations, data assimilation schemes, and forecast models for operations beyond the year 2000."

While synoptic-scale features can be described fairly well through basic atmospheric dynamics, the smaller scales are more directly tied to earth/atmosphere exchanges of heat and moisture and to cloud and precipitation physics. "The balance of forces changes," says Rit, "so that if you want to forecast how much it's going to rain, it requires far better representation of these processes in limited-area forecast models."

Where the attention is going

The USWRP's current areas of emphasis are

Studies related to the importance and mix of specific observations as they affect 0- to 4-day forecast accuracy over North America. This includes the assessment of forecast sensitivity or uncertainty and the use of "adaptive" observing and modeling grids to focus attention on areas Rit refers to as belly buttons. "These are thought to have greatly elevated sensitivity to the future state of the atmosphere, so that if the analysis of initial conditions is slightly off in these spots, it can lead to large forecast errors."

In theory, once a model is run, scientists could use adjoint models and other specialized techniques to work backwards in time and find the atmospheric belly buttons that are influencing the model's forecast. Then special observations would promptly be taken at those points and the new information added to the model to produce an improved multiday forecast. The model itself could "adapt" by putting a grid of increased resolution atop the belly-button points.

"NCAR researchers have been on the cutting edge of adaptive techniques," says Rit. Chris Snyder (MMM) is working on fundamental issues that underlie the targeting of special observations. Ron Errico (CGD), Xiaolei Zou (MMM), Andrew Crook and Jenny Sun (MMM/RAP), and others are exploring advanced data-assimilation issues related to adaptive grids. Joe Klemp and Bill Skamarock (MMM) are working on dynamically nested models that automatically increase their resolution at locations and times where experience shows the added information is most likely to reap a benefit. For instance, a model's grid might tighten where a low-pressure center is rapidly deepening and rain and snow are starting to develop.

Studies related to quantitative precipitation forecasts (QPFs). Present-day computer models provide site-specific estimates of rainfall and snowfall for 12-hour periods extending out to several days. However, because actual rain and snow patterns are so closely tied to mesoscale conditions, these QPFs are often in error on both amounts and locations. The USWRP will study how precipitation is formed, how it is depicted in models, and how new data can be used to improve model QPFs.

Extreme heat kills an average of nearly 400 people each year in the United States. Aside from their direct impact on people, heat waves produce tremendous stress on utilities and agriculture. Advance warnings of extreme heat are currently limited in scope. Participants in the USWRP societal-impacts meeting at NCAR recommended applying new scientific understanding from a variety of disciplines to add outlooks, advisories, watches, and warnings to the current heat-wave forecast system. They also called for enhanced education and prevention efforts. "Here is an area of quick payoff," says Chris Adams (Cooperative Institute for Research in the Atmosphere).

Such forecasting will remain a challenge even with the 8-km forecast model to come, says Rit, because "even in a mesoscale model, convection [showers and thunderstorms] is parameterized, not explicitly represented." The resulting output is in "a kind of no-man's land" between the broad-brush rainfall estimates now available and the prediction of individual storm cells, which may require even finer-mesh models, the use of statistical techniques, or both.

Studies related to hurricane forecasts near landfall. "This is the one area where the [societal] impacts are reasonably well documented," says Rit. As such, it's where many USWRP scientists feel they can produce the most immediate, demonstrable benefit through improved forecasts. The research topics will include adaptive observing techniques, the mechanisms underlying such landfall hazards as the storm surge, and the initialization of numerical hurricane models with radar, satellite, and targeted in-place data, such as reports from dropwindsondes. ATD's recently developed sondes based on Global Positioning System (GPS) technology are now being deployed from hurricane-hunter planes; they may eventually be dropped from remotely piloted aircraft.

The progress of hurricane forecasting in recent years has been checkered. Time and location estimates for hurricane landfall continue to improve, yet the growth and decay patterns of tropical cyclones over the open ocean continue to defy forecasters' best efforts. Though the public may not realize it, says Rit, "right now we have essentially no skill in forecasting hurricane intensity changes."

Next steps

The building blocks for last month's USWRP meetings at NCAR were a series of prospectus reports issued by large working groups over the past two years and published in the Bulletin of the American Meteorological Society beginning in July 1995. The prospectus teams were Rit's idea, a way to "bring information from some of the best researchers to the table." Each report is an extended essay spotlighting the areas of progress and opportunity in a given niche of atmospheric science--"not a plan," says Rit, "but a statement of physical science issues and opportunities: 'From our perspective, here's what we think is important.' "

In the past six months, flooding has ravaged the United States to the tune of several billion dollars and dozens of deaths. The USWRP is paying particular attention to the societal aspects of flood prediction and warning. Current data on flood-related damages and injuries are woefully insufficient. According to Mary Fran Myers, codirector of the University of Colorado's Natural Hazards Center, "We may not know what the real damages are." Also, she adds, "The aging infrastructure in this country is going to be more and more vulnerable to [flood] disasters." Above: a scene from the record Mississippi River flooding of 1993. (Photo by Curt Zukosky.)

Guided by these reports, the lab directors and the USWRP science advisory committee used their NCAR meetings to figure out how ongoing research fits into the larger scheme of the USWRP. "About 99 percent of the research we discussed is not directly funded by USWRP, and yet it's on the order of $100 million in weather-prediction-related research," says Rit. Among the decisions made at the meeting was to proceed with development of a five-year plan for research on prediction of hurricanes and specific hazards at landfall.

The USWRP is being mirrored on a global scale by the World Weather Research Program. This multinational effort, conceived in 1995, is similar in its mission and approach to the U.S. program. Scientists representing the interests of nations on all continents met in Toulouse, France, last fall to map out scientific objectives. In June the program plan goes to the World Meteorological Organization's executive council (representing 32 nations) and next January it is up for approval by the WMO Commission on Atmospheric Sciences and all WMO member states. The two topics now being emphasized are landfalling tropical cyclones and warm-season convection. Another possible initiative is high-impact urban weather. Efforts are also being coordinated or supported on mountain-induced heavy precipitation and in-flight icing.

Rit hopes that the USWRP and its global analogue will help steer atmospheric research toward end points that make a real difference to society. "We think we can take things to a new level in the next decade. However, recognizing we can't do everything, we're getting the maximum guidance possible from the socioeconomic research and user communities. We're trying to find problems that are both tractable and beneficial. It's that intersection that will make the USWRP worth doing." •BH

NCAR's piece of the pie

Six proposals submitted by NCAR scientists were approved in January for two years of funding from the NSF-managed USWRP grants program. The grants are being matched by NCAR, resulting in total support ranging from $250,000 to $1.1 million for each proposal.

Title Divisions Principal investigator
Radar/satellite precipitation estimation RAP, ATD Ed Brandes
Improved characterization of water vapor profiles in the atmosphere and an investigation into their impact on quantitative precipitation forecasts ATD, MMM Mike Hardesty
Numerical precipitation forecasting program MMM Joe Klemp
Development of a community mesoscale data assimilation system and its research applications in support of USWRP MMM, CGD, ATD Bill Kuo
Research in adaptive observations MMM Chris Snyder
An evaluation of model and radar-based convective precipitation estimates for use in runoff-model predictions of flash floods in complex terrain RAP, MMM Tom Warner
Thunderstorm and quantitative precipitation nowcasting ATD, RAP, MMM Jim Wilson

Roger Pielke

The human side of the equation

When was the last time you saw an insurance executive, a policy manager, a psychologist, a petrophysicist, and a pathologist assembled at NCAR? Precisely the point, says Roger Pielke, Jr. The ESIG political scientist and colleagues brought these diverse people and 20 others to the Mesa Lab on 2-4 April for a USWRP-sponsored workshop on the social and economic impacts of weather. It was three days of unusually frank and open-minded debate on exactly what it is that society needs from atmospheric scientists.

"The physical science, social science, and user groups are working together to define and implement the research. ESIG has done this sort of thing in the past in the area of climate, but not weather," says Roger. Not that social scientists haven't looked at meteorological issues. However, Roger says, "most natural hazards research has focused more on earthquakes and floods than other hazards." The Federal Emergency Management Agency, he adds, has "traditionally focused on what happens after an event, which doesn't leave a lot of room for seeing how users actually use weather information."

Usefulness was a paramount concern raised in the meeting. After all, notes Roger, "If you have the best information in the world and nobody uses it, it has no value." But the notion of usefulness also pertained to the USWRP itself. Jeff Kimpel, the new director of the National Severe Storms Laboratory and a cochair of the NCAR meeting, kicked things off by stating, "We've had some difficulty [explaining] what the benefits of the USWRP would be to the people of this nation."

There are, in fact, potentially giant payoffs to improved weather forecasts. Storms are a natural focus. David Epps (BP Exploration, Inc.) noted that in extreme years, storms in the Gulf of Mexico can produce up to 60 days of delay a year in offshore deep-water drilling for oil and gas, increasing costs at the rate of $250,000 per day. Tropical cyclones defer production (and reduce federal revenue), with shutdowns typically lasting five to seven days. Lopping three days from that window, says Epps, could save up to $75 million per year.

Day in and day out, it's heat and cold that take the biggest toll on people and industry. Improved temperature prediction, especially in the longer time ranges, could save untold millions of dollars and many lives. Mike Smith, whose Wichita-based firm, WeatherData, Inc., issues forecasts for freight trains, noted that extreme heat and cold can produce track damage and cause derailments as easily as a severe storm can. On the workshop's second day, when a forecast high of around 5 degrees C (41 degrees F) fell far short of the day's actual warm-up to near 20 degrees C (68 degrees F), Nick Keener (Duke Power Co.) asserted that the cost of such an error to a utility company making production decisions could be as much as several hundred thousand dollars.

Current data indicate that hot and cold weather combined kill an average of more than 1,000 U.S. residents each year, far more than tornadoes or hurricanes. However, the death tolls from all of these events are difficult to assess. Gib Parrish (Centers for Disease Control), who assembles official statistics on weather-related mortality, lamented the fact that "there is no generally accepted definition of what constitutes a weather-related death, injury, or illness." For instance, his office has recorded deaths ranging from a workman asphyxiated by a storm-weakened tree trunk that toppled onto him 15 days after a tornado occurred, to a murder-suicide in temporary housing six months after a hurricane, to a would-be looter shot and killed by a homeowner.

Three working groups closed the workshop with a range of recommendations, including:

The good news, says Roger, is that social science is well suited to provide the USWRP with the guidance it needs, as long as physical and social scientists work together with the ultimate users of weather products. "There's an army of natural hazards researchers, thousands of them, out there. It's not just cost/benefit work--the social sciences have many more tools available to determine the value of information. There's plenty of work to be done." •BH

In this issue...
Other issues of Staff Notes Monthly


Edited by Bob Henson, bhenson@ucar.edu