In the past few years, you may have noticed a new term sneaking into television and National Weather Service forecasts. Somewhere in the barrage of jargon, you might have heard or read something like, "The model says that the front should intensify...." As pilots, we're most likely to see references to "models" in the prognostic discussions that are appended to the 36- and 48-hour low-level prognostic charts.
The term "model" refers to computer-generated, artificial recreations of atmospheric conditions. Analysis and forecasting by computer involves dumping thousands of meteorological observations into computers, then waiting while various algorithms do their work. The end result is a computerized best guess of future weather. Now, it's high- powered supercomputers that give meteorological guidance to the National Weather Service's forecasters. To be sure, human meteorologists make the final calls on official forecasts, but that's only after they've consulted one or more of the latest models. The days of weathermen's relying on gut intuition are definitely over — and have been since the early 1980s.
This is all for the good. With the rapid increase in meteorological observations, there's no way any forecaster could monitor all of them, perform all the necessary arcane calculations, and come up with a sturdy forecast — all within a matter of a few hours. Super computers, like the Cray C90s at the National Meteorological Center (NMC) in Suitland, Maryland, can put out several models per day. Those C90s don't fool around, either. They run at up to three gigaflops per second. In plain English, this equates to three billion calculations per second.
Each model is programmed differently so that a variety of objectives and contingencies can be analyzed as thoroughly as possible. Meteorology is acronym-happy, so each model also has a distinctive moniker. There's the LFM (Limited Fine Mesh model), the NGM (Nested Grid Model), the MRF (Medium Range Forecast model), and the hurricane model, to name a few. There's also an aviation (AVN) model, tailored to analyze and forecast adverse flying weather.
To find out more about aviation models, I visited the National Oceanic and Atmospheric Administration's Forecast Systems Laboratory (FSL) in Boulder, Colorado. There, I was given a glimpse into the future of computerized forecasting (or numerical prediction, as it's called in the trade).
The FSL is the nation's meteorological brain trust and think tank. Its director, A. E. (Sandy) MacDonald, opened his talk with an explanation of why the FSL's research products are so valuable. "Though they're much better than they used to be," he remarked, "forecasts are still too general. We need to develop more effective ways of handling smaller-scale phenomena. That's why we're emphasizing the analysis and forecasting of mesoscale events such as major storms, icing, turbulence, and convection."
MacDonald's emphasis on mesoscale phenomena (generally construed as weather affecting areas running from 100 to 1,000 nautical miles in area) is made manifest by work done in the FSL's Forecast Research Division. There, Research Chief Thomas W. Schlatter and Chief of Regional Analysis and Prediction Stanley G. Benjamin have been perfecting the input for what is called the Aviation Gridded Forecast System (AGFS). This project, done in conjunction with the FAA, is designed to run on the Weather Service Forecast Office (WSFO) workstations of the future — as part of an initiative better known to some as AWIPS (Advanced Weather Interactive Processing System). From there, AGFS-derived forecasts will be passed along to flight service stations and the DUAT data stream, where pilots will be treated to more accurate and timely forecasts.
The AGFS is also scheduled to be installed at the National Aviation Weather Advisory Unit (NAWAU) in Kansas City. There, the AGFS will help NAWAU generate more accurate area forecasts, icing sigmets, and other severe weather advisories.
What data is fed into the AGFS, and what models will run it? Once again, the FSL has been on the job, trying to include more data, increase the number of forecasts per day, and produce more accurate short- term forecasts of smaller-scale aviation weather hazards.
Enter the RUC (Rapid Update Cycle) model, another FSL-invented product. Benjamin noted that the RUC differs from the other models in that it spits out short-range forecasts (out to 12 hours) every three hours. Compared to the current AVN model, which forecasts out to 72 hours once a day, or the MRF, which goes out to 10 days on a daily basis, the RUC is a speed demon.
RUC takes advantage of today's increasing number of available observations. RUC input includes rawinsonde balloon measurements, late- breaking ASOS observations, upper-level wind flows from the FSL-initiated wind profiler demonstration network (see " Wx Watch: Bye-Bye Balloons," May 1994 Pilot), plus recent in-flight observations from ACARS (Aeronautical Radio, Inc.'s [ARINC's] Communications, Addressing, and Reporting System)- equipped airliners. ACARS reports include wind speed and direction, outside air temperature, and pressure altitude, and provide some 13,000 reports per day automatically from airliners in flight. How can general aviation pilots help data volume and accuracy? Submit as many detailed voice pireps as you can.
Imagine a grid network spread over the entire contiguous United States. Each grid is 60 kilometers, or about 32 nm, square. Now imagine a vertical stack of 25 of these grid networks, starting at the surface and extending up to 53,000 feet or more. Next, think of 85,000 bits of weather data sprinkled over each of these layers. Feed this mountain of information into a computer loaded with atmospheric equations, then stand back for forecast outputs. That's how the RUC — and most other models — do their work.
By the way, RUC has evolved into more than just a research project. It now runs operationally at the NMC, where the Crays perform RUC forecasts in under 10 minutes.
The FSL has other aviation irons in the fire. One, a workstation called the Aviation Impact Variables (AIV) Editor, is now being evaluated at the NAWAU. The AIV editor lets meteorologists observe RUC output in visual form on a monitor. This includes such elements as cloud coverage, cloud tops, and areas likely to produce icing. At this time, thunderstorm prediction isn't RUC's strong point; a 60-km grid isn't small enough to detect the kind of small convective currents that can explode into storm cells.
The AIV editor is a unique system because it lets meteorologists change atmospheric variables with slider bars on the computer screen. If a meteorologist wants to change one or more atmospheric variables, he or she simply points, moves, and clicks with a mouse.
Using this method, forecasters can interact with, say, an icing algorithm "on the fly." "Let's say the forecaster wants to identify areas with the potential for clear ice," explained meteorologist Dennis M. Rodgers. "He can search for areas with the highest relative humidities and define the areas having temperatures in the 0 to -10 degrees Celsius range." A click of the mouse, and voila, the suspect area is highlighted in red.
Rodgers demonstrated that meteorologists can also use the AIV editor to define a route of flight and cruising altitude, then have a vertical profile of the flight path plotted on screen. This feature has immense promise as a high-quality briefing tool, since areas of hazardous forecast weather can be pinpointed along the flight path. To top it all off, one of the AIV editor's visual fields lets the operator tilt the weather imagery so that a number of visual angles can be presented for analysis. It's this kind of technology that FSL hopes will make its way into AWIPS workstations by the end of the century.
Knowing about mesoscale weather is great, but in aviation it's the small-scale hazards — things like microbursts, localized heavy icing, wind shear, and storm cells — that are a pilot's worst enemy. And the large- scale models run at the NMC often aren't accurate enough to really nail down the more localized weather extremes.
The FSL has another program — the Local Analysis and Prediction System (LAPS) — that focuses on the 330-square-nm area surrounding Denver and Boulder. LAPS uses a 5.4-square-nm grid, with 17 vertically stacked grid layers. LAPS meteorologist Daniel L. Birkenheuer gave a brief rundown on the LAPS world.
Data for LAPS comes from a vast network of portable automated mesonets, ASOS observations, wind profilers, satellite information, a nearby Doppler radar, and ACARS reports. Portable mesonets (PAMs) are rather like mini-ASOS stations on wheels. But unlike ASOS, the PAMs give a stream of information at five-minute intervals. The ASOS observations are updated hourly.
LAPS provides real-time analysis, as well as a daily 12-hour forecast. For its model, LAPS uses the Regional Atmospheric Modeling Systems (RAMS), a product of the Colorado State University. This model has a cloud physics package that does a neat job of showing predicted moisture transport and cloud formation. In real time, it's fascinating to watch the local surface winds shift direction slightly, right on the computer screen.
A data-rich system like LAPS would be fantastic if implemented on a national scale. So far, however, it's been adaptable only to those regions with enough raw data to adequately feed the model. The Denver WSFO now runs a LAPS display, and the University of Oklahoma and the Massachusetts Institute of Technology's Lincoln Laboratories have also experimented with LAPS. The FSL is currently talking to FAA officials in Seattle and Salt Lake City about installing a LAPS in those locations. Word has it that a LAPS will be installed in Atlanta in October 1995, where it will serve as the official weather-watcher of the 1996 Olympic Games.
I watched a predictive model run on Birkenheuer's computer, and the level of detail was amazing. As a day progressed, clouds rose from the Rockies, were blown east, curled in the currents of the midday heat, then began to dissipate as afternoon passed into evening.
I asked Birkenheuer where LAPS got its start. "Oh, that was back in '79 under PROFS. But it really got going during STORM-FEST in '92," he said. By this time I'd overdosed on acronyms. When I asked what the acronyms meant, Birkenheuer also seemed puzzled. Swivelling in his chair, he went to his mousepad and called up — I kid you not — an acronym glossary. It took up the entire screen, and it was a big, big screen. PROFS stands for Program for Regional Observing and Forecasting Services, and STORM- FEST means Stormscale Operational and Research Meteorology-Fronts Experimental Systems Test. Whew!
The FSL has a casual aura, and informality rules. But it's a productive environment, and the intensity of the work is also evident. MacDonald was awarded the Department of Commerce's gold medal for leadership in developing programs to improve the nation's weather services through technology transfer. The fellow I just passed in the hall is Dr. Russell B. Chadwick who, along with two other co-workers, won another gold medal for creating the world's first major wind profiler network. Five other FSL leaders (Joan Brundage, Carl Bullock, Thomas LeFebvre, Joseph Wakefield, and Dennis Walts) were given the bronze medal for developing a new weather forecasting workstation. When the National Weather Service's modernization program is finished, we can thank experts like those mentioned in this article for the look and feel of the end-user products.
Related Articles
